diff --git a/expat/.gitignore b/expat/.gitignore
index e6794547..b7071987 100755
--- a/expat/.gitignore
+++ b/expat/.gitignore
@@ -28,3 +28,6 @@ expat.pc
/callgraph.svg
/libexpat.so.*
/run.sh
+build__R*
+coverage__R*
+source__R*
diff --git a/expat/clean_coverage.sh b/expat/clean_coverage.sh
new file mode 100755
index 00000000..bf02fbff
--- /dev/null
+++ b/expat/clean_coverage.sh
@@ -0,0 +1,3 @@
+rm -r build__*
+rm -r coverage__*
+rm -r source__*
diff --git a/expat/lib/xmlparse.c b/expat/lib/xmlparse.c
index d26ebd3c..7dc36eb1 100644
--- a/expat/lib/xmlparse.c
+++ b/expat/lib/xmlparse.c
@@ -1870,9 +1870,22 @@ XML_Parse(XML_Parser parser, const char *s, int len, int isFinal)
if (errorCode == XML_ERROR_NONE) {
switch (ps_parsing) {
case XML_SUSPENDED:
+ /* It is hard to be certain, but it seems that this case
+ * cannot occur. This code is cleaning up a previous parse
+ * with no new data (since len == 0). Changing the parsing
+ * state requires getting to execute a handler function, and
+ * there doesn't seem to be an opportunity for that while in
+ * this circumstance.
+ *
+ * Given the uncertainty, we retain the code but exclude it
+ * from coverage tests.
+ *
+ * LCOV_EXCL_START
+ */
XmlUpdatePosition(encoding, positionPtr, bufferPtr, &position);
positionPtr = bufferPtr;
return XML_STATUS_SUSPENDED;
+ /* LCOV_EXCL_STOP */
case XML_INITIALIZED:
case XML_PARSING:
ps_parsing = XML_FINISHED;
@@ -3061,9 +3074,17 @@ doContent(XML_Parser parser,
return XML_ERROR_NO_MEMORY;
break;
default:
+ /* All of the tokens produced by XmlContentTok() have their own
+ * explicit cases, so this default is not strictly necessary.
+ * However it is a useful safety net, so we retain the code and
+ * simply exclude it from the coverage tests.
+ *
+ * LCOV_EXCL_START
+ */
if (defaultHandler)
reportDefault(parser, enc, s, next);
break;
+ /* LCOV_EXCL_STOP */
}
*eventPP = s = next;
switch (ps_parsing) {
@@ -3342,8 +3363,23 @@ storeAtts(XML_Parser parser, const ENCODING *enc,
((XML_Char *)s)[-1] = 0; /* clear flag */
id = (ATTRIBUTE_ID *)lookup(parser, &dtd->attributeIds, s, 0);
- if (!id || !id->prefix)
- return XML_ERROR_NO_MEMORY;
+ if (!id || !id->prefix) {
+ /* This code is walking through the appAtts array, dealing
+ * with (in this case) a prefixed attribute name. To be in
+ * the array, the attribute must have already been bound, so
+ * has to have passed through the hash table lookup once
+ * already. That implies that an entry for it already
+ * exists, so the lookup above will return a pointer to
+ * already allocated memory. There is no opportunaity for
+ * the allocator to fail, so the condition above cannot be
+ * fulfilled.
+ *
+ * Since it is difficult to be certain that the above
+ * analysis is complete, we retain the test and merely
+ * remove the code from coverage tests.
+ */
+ return XML_ERROR_NO_MEMORY; /* LCOV_EXCL_LINE */
+ }
b = id->prefix->binding;
if (!b)
return XML_ERROR_UNBOUND_PREFIX;
@@ -3720,8 +3756,16 @@ doCdataSection(XML_Parser parser,
}
return XML_ERROR_UNCLOSED_CDATA_SECTION;
default:
+ /* Every token returned by XmlCdataSectionTok() has its own
+ * explicit case, so this default case will never be executed.
+ * We retain it as a safety net and exclude it from the coverage
+ * statistics.
+ *
+ * LCOV_EXCL_START
+ */
*eventPP = next;
return XML_ERROR_UNEXPECTED_STATE;
+ /* LCOV_EXCL_STOP */
}
*eventPP = s = next;
@@ -3781,8 +3825,20 @@ doIgnoreSection(XML_Parser parser,
eventEndPP = &eventEndPtr;
}
else {
+ /* It's not entirely clear, but it seems the following two lines
+ * of code cannot be executed. The only occasions on which 'enc'
+ * is not 'parser->m_encoding' are when this function is called
+ * from the internal entity processing, and IGNORE sections are an
+ * error in internal entities.
+ *
+ * Since it really isn't clear that this is true, we keep the code
+ * and just remove it from our coverage tests.
+ *
+ * LCOV_EXCL_START
+ */
eventPP = &(openInternalEntities->internalEventPtr);
eventEndPP = &(openInternalEntities->internalEventEndPtr);
+ /* LCOV_EXCL_STOP */
}
*eventPP = s;
*startPtr = NULL;
@@ -3815,8 +3871,16 @@ doIgnoreSection(XML_Parser parser,
}
return XML_ERROR_SYNTAX; /* XML_ERROR_UNCLOSED_IGNORE_SECTION */
default:
+ /* All of the tokens that XmlIgnoreSectionTok() returns have
+ * explicit cases to handle them, so this default case is never
+ * executed. We keep it as a safety net anyway, and remove it
+ * from our test coverage statistics.
+ *
+ * LCOV_EXCL_START
+ */
*eventPP = next;
return XML_ERROR_UNEXPECTED_STATE;
+ /* LCOV_EXCL_STOP */
}
/* not reached */
}
@@ -4058,15 +4122,14 @@ entityValueInitProcessor(XML_Parser parser,
result = processXmlDecl(parser, 0, start, next);
if (result != XML_ERROR_NONE)
return result;
- switch (ps_parsing) {
- case XML_SUSPENDED:
- *nextPtr = next;
- return XML_ERROR_NONE;
- case XML_FINISHED:
+ /* At this point, ps_parsing cannot be XML_SUSPENDED. For that
+ * to happen, a parameter entity parsing handler must have
+ * attempted to suspend the parser, which fails and raises an
+ * error. The parser can be aborted, but can't be suspended.
+ */
+ if (ps_parsing == XML_FINISHED)
return XML_ERROR_ABORTED;
- default:
- *nextPtr = next;
- }
+ *nextPtr = next;
/* stop scanning for text declaration - we found one */
processor = entityValueProcessor;
return entityValueProcessor(parser, next, end, nextPtr);
@@ -4389,8 +4452,14 @@ doProlog(XML_Parser parser,
&dtd->paramEntities,
externalSubsetName,
sizeof(ENTITY));
- if (!entity)
- return XML_ERROR_NO_MEMORY;
+ if (!entity) {
+ /* The external subset name "#" will have already been
+ * inserted into the hash table at the start of the
+ * external entity parsing, so no allocation will happen
+ * and lookup() cannot fail.
+ */
+ return XML_ERROR_NO_MEMORY; /* LCOV_EXCL_LINE */
+ }
if (useForeignDTD)
entity->base = curBase;
dtd->paramEntityRead = XML_FALSE;
@@ -4950,8 +5019,29 @@ doProlog(XML_Parser parser,
: !dtd->hasParamEntityRefs)) {
if (!entity)
return XML_ERROR_UNDEFINED_ENTITY;
- else if (!entity->is_internal)
- return XML_ERROR_ENTITY_DECLARED_IN_PE;
+ else if (!entity->is_internal) {
+ /* It's hard to exhaustively search the code to be sure,
+ * but there doesn't seem to be a way of executing the
+ * following line. There are two cases:
+ *
+ * If 'standalone' is false, the DTD must have no
+ * parameter entities or we wouldn't have passed the outer
+ * 'if' statement. That measn the only entity in the hash
+ * table is the external subset name "#" which cannot be
+ * given as a parameter entity name in XML syntax, so the
+ * lookup must have returned NULL and we don't even reach
+ * the test for an internal entity.
+ *
+ * If 'standalone' is true, it does not seem to be
+ * possible to create entities taking this code path that
+ * are not internal entities, so fail the test above.
+ *
+ * Because this analysis is very uncertain, the code is
+ * being left in place and merely removed from the
+ * coverage test statistics.
+ */
+ return XML_ERROR_ENTITY_DECLARED_IN_PE; /* LCOV_EXCL_LINE */
+ }
}
else if (!entity) {
dtd->keepProcessing = dtd->standalone;
@@ -5423,11 +5513,15 @@ appendAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata,
&& (poolLength(pool) == 0 || poolLastChar(pool) == 0x20))
break;
n = XmlEncode(n, (ICHAR *)buf);
- if (!n) {
- if (enc == encoding)
- eventPtr = ptr;
- return XML_ERROR_BAD_CHAR_REF;
- }
+ /* The XmlEncode() functions can never return 0 here. That
+ * error return happens if the code point passed in is either
+ * negative or greater than or equal to 0x110000. The
+ * XmlCharRefNumber() functions will all return a number
+ * strictly less than 0x110000 or a negative value if an error
+ * occurred. The negative value is intercepted above, so
+ * XmlEncode() is never passed a value it might return an
+ * error for.
+ */
for (i = 0; i < n; i++) {
if (!poolAppendChar(pool, buf[i]))
return XML_ERROR_NO_MEMORY;
@@ -5501,8 +5595,26 @@ appendAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata,
break;
}
if (entity->open) {
- if (enc == encoding)
- eventPtr = ptr;
+ if (enc == encoding) {
+ /* It does not appear that this line can be executed.
+ *
+ * The "if (entity->open)" check catches recursive entity
+ * definitions. In order to be called with an open
+ * entity, it must have gone through this code before and
+ * been through the recursive call to
+ * appendAttributeValue() some lines below. That call
+ * sets the local encoding ("enc") to the parser's
+ * internal encoding (internal_utf8 or internal_utf16),
+ * which can never be the same as the principle encoding.
+ * It doesn't appear there is another code path that gets
+ * here with entity->open being TRUE.
+ *
+ * Since it is not certain that this logic is watertight,
+ * we keep the line and merely exclude it from coverage
+ * tests.
+ */
+ eventPtr = ptr; /* LCOV_EXCL_LINE */
+ }
return XML_ERROR_RECURSIVE_ENTITY_REF;
}
if (entity->notation) {
@@ -5529,9 +5641,21 @@ appendAttributeValue(XML_Parser parser, const ENCODING *enc, XML_Bool isCdata,
}
break;
default:
+ /* The only token returned by XmlAttributeValueTok() that does
+ * not have an explicit case here is XML_TOK_PARTIAL_CHAR.
+ * Getting that would require an entity name to contain an
+ * incomplete XML character (e.g. \xE2\x82); however previous
+ * tokenisers will have already recognised and rejected such
+ * names before XmlAttributeValueTok() gets a look-in. This
+ * default case should be retained as a safety net, but the code
+ * excluded from coverage tests.
+ *
+ * LCOV_EXCL_START
+ */
if (enc == encoding)
eventPtr = ptr;
return XML_ERROR_UNEXPECTED_STATE;
+ /* LCOV_EXCL_STOP */
}
ptr = next;
}
@@ -5664,12 +5788,15 @@ storeEntityValue(XML_Parser parser,
goto endEntityValue;
}
n = XmlEncode(n, (ICHAR *)buf);
- if (!n) {
- if (enc == encoding)
- eventPtr = entityTextPtr;
- result = XML_ERROR_BAD_CHAR_REF;
- goto endEntityValue;
- }
+ /* The XmlEncode() functions can never return 0 here. That
+ * error return happens if the code point passed in is either
+ * negative or greater than or equal to 0x110000. The
+ * XmlCharRefNumber() functions will all return a number
+ * strictly less than 0x110000 or a negative value if an error
+ * occurred. The negative value is intercepted above, so
+ * XmlEncode() is never passed a value it might return an
+ * error for.
+ */
for (i = 0; i < n; i++) {
if (pool->end == pool->ptr && !poolGrow(pool)) {
result = XML_ERROR_NO_MEMORY;
@@ -5690,10 +5817,18 @@ storeEntityValue(XML_Parser parser,
result = XML_ERROR_INVALID_TOKEN;
goto endEntityValue;
default:
+ /* This default case should be unnecessary -- all the tokens
+ * that XmlEntityValueTok() can return have their own explicit
+ * cases -- but should be retained for safety. We do however
+ * exclude it from the coverage statistics.
+ *
+ * LCOV_EXCL_START
+ */
if (enc == encoding)
eventPtr = entityTextPtr;
result = XML_ERROR_UNEXPECTED_STATE;
goto endEntityValue;
+ /* LCOV_EXCL_STOP */
}
entityTextPtr = next;
}
@@ -5791,8 +5926,25 @@ reportDefault(XML_Parser parser, const ENCODING *enc,
eventEndPP = &eventEndPtr;
}
else {
+ /* To get here, two things must be true; the parser must be
+ * using a character encoding that is not the same as the
+ * encoding passed in, and the encoding passed in must need
+ * conversion to the internal format (UTF-8 unless XML_UNICODE
+ * is defined). The only occasions on which the encoding passed
+ * in is not the same as the parser's encoding are when it is
+ * the internal encoding (e.g. a previously defined parameter
+ * entity, already converted to internal format). This by
+ * definition doesn't need conversion, so the whole branch never
+ * gets executed.
+ *
+ * For safety's sake we don't delete these lines and merely
+ * exclude them from coverage statistics.
+ *
+ * LCOV_EXCL_START
+ */
eventPP = &(openInternalEntities->internalEventPtr);
eventEndPP = &(openInternalEntities->internalEventEndPtr);
+ /* LCOV_EXCL_STOP */
}
do {
ICHAR *dataPtr = (ICHAR *)dataBuf;
@@ -5961,9 +6113,30 @@ getContext(XML_Parser parser)
len = dtd->defaultPrefix.binding->uriLen;
if (namespaceSeparator)
len--;
- for (i = 0; i < len; i++)
- if (!poolAppendChar(&tempPool, dtd->defaultPrefix.binding->uri[i]))
- return NULL;
+ for (i = 0; i < len; i++) {
+ if (!poolAppendChar(&tempPool, dtd->defaultPrefix.binding->uri[i])) {
+ /* Because of memory caching, I don't believe this line can be
+ * executed.
+ *
+ * This is part of a loop copying the default prefix binding
+ * URI into the parser's temporary string pool. Previously,
+ * that URI was copied into the same string pool, with a
+ * terminating NUL character, as part of setContext(). When
+ * the pool was cleared, that leaves a block definitely big
+ * enough to hold the URI on the free block list of the pool.
+ * The URI copy in getContext() therefore cannot run out of
+ * memory.
+ *
+ * If the pool is used between the setContext() and
+ * getContext() calls, the worst it can do is leave a bigger
+ * block on the front of the free list. Given that this is
+ * all somewhat inobvious and program logic can be changed, we
+ * don't delete the line but we do exclude it from the test
+ * coverage statistics.
+ */
+ return NULL; /* LCOV_EXCL_LINE */
+ }
+ }
needSep = XML_TRUE;
}
@@ -5975,8 +6148,15 @@ getContext(XML_Parser parser)
PREFIX *prefix = (PREFIX *)hashTableIterNext(&iter);
if (!prefix)
break;
- if (!prefix->binding)
- continue;
+ if (!prefix->binding) {
+ /* This test appears to be (justifiable) paranoia. There does
+ * not seem to be a way of injecting a prefix without a binding
+ * that doesn't get errored long before this function is called.
+ * The test should remain for safety's sake, so we instead
+ * exclude the following line from the coverage statistics.
+ */
+ continue; /* LCOV_EXCL_LINE */
+ }
if (needSep && !poolAppendChar(&tempPool, CONTEXT_SEP))
return NULL;
for (s = prefix->name; *s; s++)
@@ -6647,8 +6827,20 @@ poolCopyString(STRING_POOL *pool, const XML_Char *s)
static const XML_Char *
poolCopyStringN(STRING_POOL *pool, const XML_Char *s, int n)
{
- if (!pool->ptr && !poolGrow(pool))
- return NULL;
+ if (!pool->ptr && !poolGrow(pool)) {
+ /* The following line is unreachable given the current usage of
+ * poolCopyStringN(). Currently it is called from exactly one
+ * place to copy the text of a simple general entity. By that
+ * point, the name of the entity is already stored in the pool, so
+ * pool->ptr cannot be NULL.
+ *
+ * If poolCopyStringN() is used elsewhere as it well might be,
+ * this line may well become executable again. Regardless, this
+ * sort of check shouldn't be removed lightly, so we just exclude
+ * it from the coverage statistics.
+ */
+ return NULL; /* LCOV_EXCL_LINE */
+ }
for (; n > 0; --n, s++) {
if (!poolAppendChar(pool, *s))
return NULL;
@@ -6745,8 +6937,15 @@ poolGrow(STRING_POOL *pool)
// to avoid dangling pointers:
const ptrdiff_t offsetInsideBlock = pool->ptr - pool->start;
- if (blockSize < 0)
- return XML_FALSE;
+ if (blockSize < 0) {
+ /* This condition traps a situation where either more than
+ * INT_MAX/2 bytes have already been allocated. This isn't
+ * readily testable, since it is unlikely that an average
+ * machine will have that much memory, so we exclude it from the
+ * coverage statistics.
+ */
+ return XML_FALSE; /* LCOV_EXCL_LINE */
+ }
bytesToAllocate = poolBytesToAllocateFor(blockSize);
if (bytesToAllocate == 0)
@@ -6767,8 +6966,18 @@ poolGrow(STRING_POOL *pool)
int blockSize = (int)(pool->end - pool->start);
size_t bytesToAllocate;
- if (blockSize < 0)
- return XML_FALSE;
+ if (blockSize < 0) {
+ /* This condition traps a situation where either more than
+ * INT_MAX bytes have already been allocated (which is prevented
+ * by various pieces of program logic, not least this one, never
+ * mind the unlikelihood of actually having that much memory) or
+ * the pool control fields have been corrupted (which could
+ * conceivably happen in an extremely buggy user handler
+ * function). Either way it isn't readily testable, so we
+ * exclude it from the coverage statistics.
+ */
+ return XML_FALSE; /* LCOV_EXCL_LINE */
+ }
if (blockSize < INIT_BLOCK_SIZE)
blockSize = INIT_BLOCK_SIZE;
diff --git a/expat/lib/xmlrole.c b/expat/lib/xmlrole.c
index a7c56302..c809ee51 100644
--- a/expat/lib/xmlrole.c
+++ b/expat/lib/xmlrole.c
@@ -170,7 +170,14 @@ prolog1(PROLOG_STATE *state,
case XML_TOK_COMMENT:
return XML_ROLE_COMMENT;
case XML_TOK_BOM:
- return XML_ROLE_NONE;
+ /* This case can never arise. To reach this role function, the
+ * parse must have passed through prolog0 and therefore have had
+ * some form of input, even if only a space. At that point, a
+ * byte order mark is no longer a valid character (though
+ * technically it should be interpreted as a non-breaking space),
+ * so will be rejected by the tokenizing stages.
+ */
+ return XML_ROLE_NONE; /* LCOV_EXCL_LINE */
case XML_TOK_DECL_OPEN:
if (!XmlNameMatchesAscii(enc,
ptr + 2 * MIN_BYTES_PER_CHAR(enc),
@@ -1285,6 +1292,26 @@ declClose(PROLOG_STATE *state,
return common(state, tok);
}
+/* This function will only be invoked if the internal logic of the
+ * parser has broken down. It is used in two cases:
+ *
+ * 1: When the XML prolog has been finished. At this point the
+ * processor (the parser level above these role handlers) should
+ * switch from prologProcessor to contentProcessor and reinitialise
+ * the handler function.
+ *
+ * 2: When an error has been detected (via common() below). At this
+ * point again the processor should be switched to errorProcessor,
+ * which will never call a handler.
+ *
+ * The result of this is that error() can only be called if the
+ * processor switch failed to happen, which is an internal error and
+ * therefore we shouldn't be able to provoke it simply by using the
+ * library. It is a necessary backstop, however, so we merely exclude
+ * it from the coverage statistics.
+ *
+ * LCOV_EXCL_START
+ */
static int PTRCALL
error(PROLOG_STATE *UNUSED_P(state),
int UNUSED_P(tok),
@@ -1294,6 +1321,7 @@ error(PROLOG_STATE *UNUSED_P(state),
{
return XML_ROLE_NONE;
}
+/* LCOV_EXCL_STOP */
static int FASTCALL
common(PROLOG_STATE *state, int tok)
diff --git a/expat/lib/xmltok.c b/expat/lib/xmltok.c
index cdf0720d..db4a5c8c 100644
--- a/expat/lib/xmltok.c
+++ b/expat/lib/xmltok.c
@@ -1019,7 +1019,11 @@ streqci(const char *s1, const char *s2)
if (ASCII_a <= c1 && c1 <= ASCII_z)
c1 += ASCII_A - ASCII_a;
if (ASCII_a <= c2 && c2 <= ASCII_z)
- c2 += ASCII_A - ASCII_a;
+ /* The following line will never get executed. streqci() is
+ * only called from two places, both of which guarantee to put
+ * upper-case strings into s2.
+ */
+ c2 += ASCII_A - ASCII_a; /* LCOV_EXCL_LINE */
if (c1 != c2)
return 0;
if (!c1)
@@ -1291,7 +1295,7 @@ XmlUtf8Encode(int c, char *buf)
};
if (c < 0)
- return 0;
+ return 0; /* LCOV_EXCL_LINE: this case is always eliminated beforehand */
if (c < min2) {
buf[0] = (char)(c | UTF8_cval1);
return 1;
@@ -1314,7 +1318,7 @@ XmlUtf8Encode(int c, char *buf)
buf[3] = (char)((c & 0x3f) | 0x80);
return 4;
}
- return 0;
+ return 0; /* LCOV_EXCL_LINE: this case too is eliminated before calling */
}
int FASTCALL
@@ -1465,6 +1469,9 @@ XmlInitUnknownEncoding(void *mem,
else if (c < 0) {
if (c < -4)
return 0;
+ /* Multi-byte sequences need a converter function */
+ if (!convert)
+ return 0;
e->normal.type[i] = (unsigned char)(BT_LEAD2 - (c + 2));
e->utf8[i][0] = 0;
e->utf16[i] = 0;
diff --git a/expat/lib/xmltok_impl.c b/expat/lib/xmltok_impl.c
index 5f779c05..4fa1ff67 100644
--- a/expat/lib/xmltok_impl.c
+++ b/expat/lib/xmltok_impl.c
@@ -1198,8 +1198,14 @@ PREFIX(attributeValueTok)(const ENCODING *enc, const char *ptr,
const char *start;
if (ptr >= end)
return XML_TOK_NONE;
- else if (! HAS_CHAR(enc, ptr, end))
- return XML_TOK_PARTIAL;
+ else if (! HAS_CHAR(enc, ptr, end)) {
+ /* This line cannot be executed. The incoming data has already
+ * been tokenized once, so incomplete characters like this have
+ * already been eliminated from the input. Retaining the paranoia
+ * check is still valuable, however.
+ */
+ return XML_TOK_PARTIAL; /* LCOV_EXCL_LINE */
+ }
start = ptr;
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
@@ -1258,8 +1264,14 @@ PREFIX(entityValueTok)(const ENCODING *enc, const char *ptr,
const char *start;
if (ptr >= end)
return XML_TOK_NONE;
- else if (! HAS_CHAR(enc, ptr, end))
- return XML_TOK_PARTIAL;
+ else if (! HAS_CHAR(enc, ptr, end)) {
+ /* This line cannot be executed. The incoming data has already
+ * been tokenized once, so incomplete characters like this have
+ * already been eliminated from the input. Retaining the paranoia
+ * check is still valuable, however.
+ */
+ return XML_TOK_PARTIAL; /* LCOV_EXCL_LINE */
+ }
start = ptr;
while (HAS_CHAR(enc, ptr, end)) {
switch (BYTE_TYPE(enc, ptr)) {
@@ -1614,6 +1626,14 @@ PREFIX(predefinedEntityName)(const ENCODING *UNUSED_P(enc), const char *ptr,
return 0;
}
+/* This function does not appear to be called from anywhere within the
+ * library code. It is used via the macro XmlSameName(), which is
+ * defined but never used. Since it appears in the encoding function
+ * table, removing it is not a thing to be undertaken lightly. For
+ * the moment, we simply exclude it from coverage tests.
+ *
+ * LCOV_EXCL_START
+ */
static int PTRCALL
PREFIX(sameName)(const ENCODING *enc, const char *ptr1, const char *ptr2)
{
@@ -1677,14 +1697,21 @@ PREFIX(sameName)(const ENCODING *enc, const char *ptr1, const char *ptr2)
}
/* not reached */
}
+/* LCOV_EXCL_STOP */
static int PTRCALL
PREFIX(nameMatchesAscii)(const ENCODING *UNUSED_P(enc), const char *ptr1,
const char *end1, const char *ptr2)
{
for (; *ptr2; ptr1 += MINBPC(enc), ptr2++) {
- if (end1 - ptr1 < MINBPC(enc))
- return 0;
+ if (end1 - ptr1 < MINBPC(enc)) {
+ /* This line cannot be executed. THe incoming data has already
+ * been tokenized once, so imcomplete characters like this have
+ * already been eliminated from the input. Retaining the
+ * paranoia check is still valuable, however.
+ */
+ return 0; /* LCOV_EXCL_LINE */
+ }
if (!CHAR_MATCHES(enc, ptr1, *ptr2))
return 0;
}
diff --git a/expat/tests/runtests.c b/expat/tests/runtests.c
index 480167a6..0e496a70 100644
--- a/expat/tests/runtests.c
+++ b/expat/tests/runtests.c
@@ -26,6 +26,7 @@
#include "minicheck.h"
#include "memcheck.h"
#include "siphash.h"
+#include "ascii.h" /* for ASCII_xxx */
#ifdef XML_LARGE_SIZE
#define XML_FMT_INT_MOD "ll"
@@ -143,8 +144,19 @@ static unsigned long dummy_handler_flags = 0;
#define DUMMY_UNPARSED_ENTITY_DECL_HANDLER_FLAG (1UL << 11)
#define DUMMY_START_NS_DECL_HANDLER_FLAG (1UL << 12)
#define DUMMY_END_NS_DECL_HANDLER_FLAG (1UL << 13)
+#define DUMMY_START_DOCTYPE_DECL_HANDLER_FLAG (1UL << 14)
+#define DUMMY_END_DOCTYPE_DECL_HANDLER_FLAG (1UL << 15)
+#define DUMMY_SKIP_HANDLER_FLAG (1UL << 16)
+#define DUMMY_DEFAULT_HANDLER_FLAG (1UL << 17)
+static void XMLCALL
+dummy_xdecl_handler(void *UNUSED_P(userData),
+ const XML_Char *UNUSED_P(version),
+ const XML_Char *UNUSED_P(encoding),
+ int UNUSED_P(standalone))
+{}
+
static void XMLCALL
dummy_start_doctype_handler(void *UNUSED_P(userData),
const XML_Char *UNUSED_P(doctypeName),
@@ -285,6 +297,106 @@ dummy_default_handler(void *UNUSED_P(userData),
int UNUSED_P(len))
{}
+static void XMLCALL
+dummy_start_doctype_decl_handler(void *UNUSED_P(userData),
+ const XML_Char *UNUSED_P(doctypeName),
+ const XML_Char *UNUSED_P(sysid),
+ const XML_Char *UNUSED_P(pubid),
+ int UNUSED_P(has_internal_subset))
+{
+ dummy_handler_flags |= DUMMY_START_DOCTYPE_DECL_HANDLER_FLAG;
+}
+
+static void XMLCALL
+dummy_end_doctype_decl_handler(void *UNUSED_P(userData))
+{
+ dummy_handler_flags |= DUMMY_END_DOCTYPE_DECL_HANDLER_FLAG;
+}
+
+static void XMLCALL
+dummy_skip_handler(void *UNUSED_P(userData),
+ const XML_Char *UNUSED_P(entityName),
+ int UNUSED_P(is_parameter_entity))
+{
+ dummy_handler_flags |= DUMMY_SKIP_HANDLER_FLAG;
+}
+
+/* Useful external entity handler */
+typedef struct ExtOption {
+ const char *system_id;
+ const char *parse_text;
+} ExtOption;
+
+static int XMLCALL
+external_entity_optioner(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *systemId,
+ const XML_Char *UNUSED_P(publicId))
+{
+ ExtOption *options = (ExtOption *)XML_GetUserData(parser);
+ XML_Parser ext_parser;
+
+ while (options->parse_text != NULL) {
+ if (!strcmp(systemId, options->system_id)) {
+ ext_parser =
+ XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ return XML_STATUS_ERROR;
+ if (_XML_Parse_SINGLE_BYTES(ext_parser, options->parse_text,
+ strlen(options->parse_text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ return XML_STATUS_ERROR;
+ XML_ParserFree(ext_parser);
+ return XML_STATUS_OK;
+ }
+ options++;
+ }
+ fail("No suitable option found");
+ return XML_STATUS_ERROR;
+}
+
+/*
+ * Parameter entity evaluation support.
+ */
+#define ENTITY_MATCH_FAIL (-1)
+#define ENTITY_MATCH_NOT_FOUND (0)
+#define ENTITY_MATCH_SUCCESS (1)
+static const XML_Char *entity_name_to_match = NULL;
+static const XML_Char *entity_value_to_match = NULL;
+static int entity_match_flag = ENTITY_MATCH_NOT_FOUND;
+
+static void XMLCALL
+param_entity_match_handler(void *UNUSED_P(userData),
+ const XML_Char *entityName,
+ int is_parameter_entity,
+ const XML_Char *value,
+ int value_length,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId),
+ const XML_Char *UNUSED_P(notationName))
+{
+ if (!is_parameter_entity ||
+ entity_name_to_match == NULL ||
+ entity_value_to_match == NULL) {
+ return;
+ }
+ if (!strcmp(entityName, entity_name_to_match)) {
+ /* The cast here is safe because we control the horizontal and
+ * the vertical, and we therefore know our strings are never
+ * going to overflow an int.
+ */
+ if (value_length != (int)strlen(entity_value_to_match) ||
+ strncmp(value, entity_value_to_match, value_length)) {
+ entity_match_flag = ENTITY_MATCH_FAIL;
+ } else {
+ entity_match_flag = ENTITY_MATCH_SUCCESS;
+ }
+ }
+ /* Else leave the match flag alone */
+}
+
/*
* Character & encoding tests.
*/
@@ -378,6 +490,16 @@ START_TEST(test_bom_utf16_le)
}
END_TEST
+/* Parse whole buffer at once to exercise a different code path */
+START_TEST(test_nobom_utf16_le)
+{
+ char text[] = " \0<\0e\0/\0>\0";
+
+ if (XML_Parse(parser, text, sizeof(text)-1, XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
static void XMLCALL
accumulate_characters(void *userData, const XML_Char *s, int len)
{
@@ -389,9 +511,15 @@ accumulate_attribute(void *userData, const XML_Char *UNUSED_P(name),
const XML_Char **atts)
{
CharData *storage = (CharData *)userData;
- if (storage->count < 0 && atts != NULL && atts[0] != NULL) {
+
+ /* Check there are attributes to deal with */
+ if (atts == NULL)
+ return;
+
+ while (storage->count < 0 && atts[0] != NULL) {
/* "accumulate" the value of the first attribute we see */
CharData_AppendXMLChars(storage, atts[1], -1);
+ atts += 2;
}
}
@@ -430,6 +558,40 @@ _run_attribute_check(const XML_Char *text, const XML_Char *expected,
#define run_attribute_check(text, expected) \
_run_attribute_check(text, expected, __FILE__, __LINE__)
+typedef struct ExtTest {
+ const char *parse_text;
+ const char *encoding;
+ CharData *storage;
+} ExtTest;
+
+static void XMLCALL
+ext_accumulate_characters(void *userData, const XML_Char *s, int len)
+{
+ ExtTest *test_data = (ExtTest *)userData;
+ accumulate_characters(test_data->storage, s, len);
+}
+
+static void
+_run_ext_character_check(const XML_Char *text,
+ ExtTest *test_data,
+ const XML_Char *expected,
+ const char *file, int line)
+{
+ CharData storage;
+
+ CharData_Init(&storage);
+ test_data->storage = &storage;
+ XML_SetUserData(parser, test_data);
+ XML_SetCharacterDataHandler(parser, ext_accumulate_characters);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ _xml_failure(parser, file, line);
+ CharData_CheckXMLChars(&storage, expected);
+}
+
+#define run_ext_character_check(text, test_data, expected) \
+ _run_ext_character_check(text, test_data, expected, __FILE__, __LINE__)
+
/* Regression test for SF bug #491986. */
START_TEST(test_danish_latin1)
{
@@ -587,18 +749,27 @@ END_TEST
START_TEST(test_utf16)
{
/*
- some text
- */
+ * some {A} text
+ *
+ * where {A} is U+FF21, FULLWIDTH LATIN CAPITAL LETTER A
+ */
char text[] =
"\000<\000?\000x\000m\000\154\000 \000v\000e\000r\000s\000i\000o"
"\000n\000=\000'\0001\000.\000\060\000'\000 \000e\000n\000c\000o"
"\000d\000i\000n\000g\000=\000'\000U\000T\000F\000-\0001\000\066"
"\000'\000?\000>\000\n"
- "\000<\000d\000o\000c\000 \000a\000=\000'\0001\0002\0003\000'"
- "\000>\000s\000o\000m\000e\000 \000t\000e\000x\000t\000<\000/"
- "\000d\000o\000c\000>";
+ "\000<\000d\000o\000c\000 \000a\000=\000'\0001\0002\0003\000'\000>"
+ "\000s\000o\000m\000e\000 \xff\x21\000 \000t\000e\000x\000t\000"
+ "<\000/\000d\000o\000c\000>";
+ char expected[] = "some \357\274\241 text";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetUserData(parser, &storage);
+ XML_SetCharacterDataHandler(parser, accumulate_characters);
if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1, XML_TRUE) == XML_STATUS_ERROR)
xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
}
END_TEST
@@ -625,6 +796,34 @@ START_TEST(test_utf16_le_epilog_newline)
}
END_TEST
+/* Test that an outright lie in the encoding is faulted */
+START_TEST(test_not_utf16)
+{
+ const char *text =
+ ""
+ "Hi";
+
+ /* Use a handler to provoke the appropriate code paths */
+ XML_SetXmlDeclHandler(parser, dummy_xdecl_handler);
+ expect_failure(text,
+ XML_ERROR_INCORRECT_ENCODING,
+ "UTF-16 declared in UTF-8 not faulted");
+}
+END_TEST
+
+/* Test that an unknown encoding is rejected */
+START_TEST(test_bad_encoding)
+{
+ const char *text = "Hi";
+
+ if (!XML_SetEncoding(parser, "unknown-encoding"))
+ fail("XML_SetEncoding failed");
+ expect_failure(text,
+ XML_ERROR_UNKNOWN_ENCODING,
+ "Unknown encoding not faulted");
+}
+END_TEST
+
/* Regression test for SF bug #481609, #774028. */
START_TEST(test_latin1_umlauts)
{
@@ -649,6 +848,121 @@ START_TEST(test_latin1_umlauts)
}
END_TEST
+/* Test that an element name with a 4-byte UTF-8 character is rejected */
+START_TEST(test_long_utf8_character)
+{
+ const char *text =
+ "\n"
+ /* 0xf0 0x90 0x80 0x80 = U+10000, the first Linear B character */
+ "";
+ expect_failure(text,
+ XML_ERROR_INVALID_TOKEN,
+ "4-byte UTF-8 character in element name not faulted");
+}
+END_TEST
+
+/* Test that a long latin-1 attribute (too long to convert in one go)
+ * is correctly converted
+ */
+START_TEST(test_long_latin1_attribute)
+{
+ const char *text =
+ "\n"
+ "\n";
+ const char *expected =
+ /* 64 characters per line */
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNO"
+ "\xc3\xa4";
+
+ run_attribute_check(text, expected);
+}
+END_TEST
+
+
+/* Test that a long ASCII attribute (too long to convert in one go)
+ * is correctly converted
+ */
+START_TEST(test_long_ascii_attribute)
+{
+ const char *text =
+ "\n"
+ "\n";
+ const char *expected =
+ /* 64 characters per line */
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "01234";
+
+ run_attribute_check(text, expected);
+}
+END_TEST
+
/* Regression test #1 for SF bug #653180. */
START_TEST(test_line_number_after_parse)
{
@@ -878,6 +1192,14 @@ END_TEST
* Element event tests.
*/
+static void XMLCALL
+start_element_event_handler(void *userData,
+ const XML_Char *name,
+ const XML_Char **UNUSED_P(atts))
+{
+ CharData_AppendXMLChars((CharData *)userData, name, -1);
+}
+
static void XMLCALL
end_element_event_handler(void *userData, const XML_Char *name)
{
@@ -1028,6 +1350,31 @@ START_TEST(test_xmldecl_misplaced)
}
END_TEST
+START_TEST(test_xmldecl_invalid)
+{
+ expect_failure("\n",
+ XML_ERROR_XML_DECL,
+ "Failed to report invalid XML declaration");
+}
+END_TEST
+
+START_TEST(test_xmldecl_missing_attr)
+{
+ expect_failure("\n\n",
+ XML_ERROR_XML_DECL,
+ "Failed to report missing XML declaration attribute");
+}
+END_TEST
+
+START_TEST(test_xmldecl_missing_value)
+{
+ expect_failure("\n"
+ "",
+ XML_ERROR_XML_DECL,
+ "Failed to report missing attribute value");
+}
+END_TEST
+
/* Regression test for SF bug #584832. */
static int XMLCALL
UnknownEncodingHandler(void *UNUSED_P(data),const XML_Char *encoding,XML_Encoding *info)
@@ -1090,28 +1437,28 @@ END_TEST
/* Regression test for SF bug #620106. */
static int XMLCALL
-external_entity_loader_set_encoding(XML_Parser parser,
- const XML_Char *context,
- const XML_Char *UNUSED_P(base),
- const XML_Char *UNUSED_P(systemId),
- const XML_Char *UNUSED_P(publicId))
+external_entity_loader(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId))
{
- /* This text says it's an unsupported encoding, but it's really
- UTF-8, which we tell Expat using XML_SetEncoding().
- */
- const char *text =
- ""
- "\xC3\xA9";
+ ExtTest *test_data = (ExtTest *)XML_GetUserData(parser);
XML_Parser extparser;
extparser = XML_ExternalEntityParserCreate(parser, context, NULL);
if (extparser == NULL)
fail("Could not create external entity parser.");
- if (!XML_SetEncoding(extparser, "utf-8"))
- fail("XML_SetEncoding() ignored for external entity");
- if ( _XML_Parse_SINGLE_BYTES(extparser, text, strlen(text), XML_TRUE)
+ if (test_data->encoding != NULL) {
+ if (!XML_SetEncoding(extparser, test_data->encoding))
+ fail("XML_SetEncoding() ignored for external entity");
+ }
+ if ( _XML_Parse_SINGLE_BYTES(extparser,
+ test_data->parse_text,
+ strlen(test_data->parse_text),
+ XML_TRUE)
== XML_STATUS_ERROR) {
- xml_failure(parser);
+ xml_failure(extparser);
return XML_STATUS_ERROR;
}
XML_ParserFree(extparser);
@@ -1125,10 +1472,17 @@ START_TEST(test_ext_entity_set_encoding)
" \n"
"]>\n"
"&en;";
+ ExtTest test_data = {
+ /* This text says it's an unsupported encoding, but it's really
+ UTF-8, which we tell Expat using XML_SetEncoding().
+ */
+ "\xC3\xA9",
+ "utf-8",
+ NULL
+ };
- XML_SetExternalEntityRefHandler(parser,
- external_entity_loader_set_encoding);
- run_character_check(text, "\xC3\xA9");
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader);
+ run_ext_character_check(text, &test_data, "\xC3\xA9");
}
END_TEST
@@ -1147,36 +1501,6 @@ START_TEST(test_ext_entity_no_handler)
END_TEST
/* Test UTF-8 BOM is accepted */
-static int XMLCALL
-external_entity_loader_set_bom(XML_Parser parser,
- const XML_Char *context,
- const XML_Char *UNUSED_P(base),
- const XML_Char *UNUSED_P(systemId),
- const XML_Char *UNUSED_P(publicId))
-{ /* This text says it's an unsupported encoding, but it's really
- UTF-8, which we tell Expat using XML_SetEncoding().
- */
- const char *text =
- "\xEF\xBB\xBF" /* BOM */
- ""
- "\xC3\xA9";
- XML_Parser extparser;
-
- extparser = XML_ExternalEntityParserCreate(parser, context, NULL);
- if (extparser == NULL)
- fail("Could not create external entity parser.");
- if (!XML_SetEncoding(extparser, "utf-8"))
- fail("XML_SetEncoding() ignored for external entity");
- if ( _XML_Parse_SINGLE_BYTES(extparser, text, strlen(text), XML_TRUE)
- == XML_STATUS_ERROR) {
- xml_failure(extparser);
- return XML_STATUS_ERROR;
- }
-
- XML_ParserFree(extparser);
- return XML_STATUS_OK;
-}
-
START_TEST(test_ext_entity_set_bom)
{
const char *text =
@@ -1184,40 +1508,55 @@ START_TEST(test_ext_entity_set_bom)
" \n"
"]>\n"
"&en;";
+ ExtTest test_data = {
+ "\xEF\xBB\xBF" /* BOM */
+ ""
+ "\xC3\xA9",
+ "utf-8",
+ NULL
+ };
- XML_SetExternalEntityRefHandler(parser,
- external_entity_loader_set_bom);
- run_character_check(text, "\xC3\xA9");
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader);
+ run_ext_character_check(text, &test_data, "\xC3\xA9");
}
END_TEST
/* Test that bad encodings are faulted */
-static int XMLCALL
-external_entity_loader_bad_encoding(XML_Parser parser,
- const XML_Char *context,
- const XML_Char *UNUSED_P(base),
- const XML_Char *UNUSED_P(systemId),
- const XML_Char *UNUSED_P(publicId))
+typedef struct ext_faults
{
- /* Claim this is an unsupported encoding */
- const char *text =
- ""
- "u";
- XML_Parser extparser;
+ const char *parse_text;
+ const char *fail_text;
+ const char *encoding;
+ enum XML_Error error;
+} ExtFaults;
- extparser = XML_ExternalEntityParserCreate(parser, context, NULL);
- if (extparser == NULL)
- fail("Could not create external entity parser.");
- if (!XML_SetEncoding(extparser, "unknown"))
- fail("XML_SetEncoding unknown encoding failed");
- if (_XML_Parse_SINGLE_BYTES(extparser, text, strlen(text),
+static int XMLCALL
+external_entity_faulter(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId))
+{
+ XML_Parser ext_parser;
+ ExtFaults *fault = (ExtFaults *)XML_GetUserData(parser);
+
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ fail("Could not create external entity parser");
+ if (fault->encoding != NULL) {
+ if (!XML_SetEncoding(ext_parser, fault->encoding))
+ fail("XML_SetEncoding failed");
+ }
+ if (_XML_Parse_SINGLE_BYTES(ext_parser,
+ fault->parse_text,
+ strlen(fault->parse_text),
XML_TRUE) != XML_STATUS_ERROR)
- fail("Unsupported encoding not faulted");
- if (XML_GetErrorCode(extparser) != XML_ERROR_UNKNOWN_ENCODING)
- xml_failure(extparser);
+ fail(fault->fail_text);
+ if (XML_GetErrorCode(ext_parser) != fault->error)
+ xml_failure(ext_parser);
- XML_ParserFree(extparser);
+ XML_ParserFree(ext_parser);
return XML_STATUS_ERROR;
}
@@ -1228,12 +1567,40 @@ START_TEST(test_ext_entity_bad_encoding)
" \n"
"]>\n"
"&en;";
+ ExtFaults fault = {
+ "u",
+ "Unsupported encoding not faulted",
+ "unknown",
+ XML_ERROR_UNKNOWN_ENCODING
+ };
- XML_SetExternalEntityRefHandler(parser,
- external_entity_loader_bad_encoding);
- if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
- XML_TRUE) != XML_STATUS_ERROR)
- fail("Bad encoding should not have been accepted");
+ XML_SetExternalEntityRefHandler(parser, external_entity_faulter);
+ XML_SetUserData(parser, &fault);
+ expect_failure(text,
+ XML_ERROR_EXTERNAL_ENTITY_HANDLING,
+ "Bad encoding should not have been accepted");
+}
+END_TEST
+
+/* Try handing an invalid encoding to an external entity parser */
+START_TEST(test_ext_entity_bad_encoding_2)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "&entity;";
+ ExtFaults fault = {
+ "",
+ "Unknown encoding not faulted",
+ "unknown-encoding",
+ XML_ERROR_UNKNOWN_ENCODING
+ };
+
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_faulter);
+ XML_SetUserData(parser, &fault);
+ expect_failure(text, XML_ERROR_EXTERNAL_ENTITY_HANDLING,
+ "Bad encoding not faulted in external entity handler");
}
END_TEST
@@ -1275,28 +1642,6 @@ START_TEST(test_wfc_undeclared_entity_standalone) {
}
END_TEST
-static int XMLCALL
-external_entity_loader(XML_Parser parser,
- const XML_Char *context,
- const XML_Char *UNUSED_P(base),
- const XML_Char *UNUSED_P(systemId),
- const XML_Char *UNUSED_P(publicId))
-{
- char *text = (char *)XML_GetUserData(parser);
- XML_Parser extparser;
-
- extparser = XML_ExternalEntityParserCreate(parser, context, NULL);
- if (extparser == NULL)
- fail("Could not create external entity parser.");
- if ( _XML_Parse_SINGLE_BYTES(extparser, text, strlen(text), XML_TRUE)
- == XML_STATUS_ERROR) {
- xml_failure(extparser);
- return XML_STATUS_ERROR;
- }
- XML_ParserFree(extparser);
- return XML_STATUS_OK;
-}
-
/* Test that an error is reported for unknown entities if we have read
an external subset, and standalone is true.
*/
@@ -1305,11 +1650,14 @@ START_TEST(test_wfc_undeclared_entity_with_external_subset_standalone) {
"\n"
"\n"
"&entity;";
- char foo_text[] =
- "";
+ ExtTest test_data = {
+ "",
+ NULL,
+ NULL
+ };
XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
- XML_SetUserData(parser, foo_text);
+ XML_SetUserData(parser, &test_data);
XML_SetExternalEntityRefHandler(parser, external_entity_loader);
expect_failure(text,
XML_ERROR_UNDEFINED_ENTITY,
@@ -1317,6 +1665,26 @@ START_TEST(test_wfc_undeclared_entity_with_external_subset_standalone) {
}
END_TEST
+/* Test that external entity handling is not done if the parsing flag
+ * is set to UNLESS_STANDALONE
+ */
+START_TEST(test_entity_with_external_subset_unless_standalone) {
+ const char *text =
+ "\n"
+ "\n"
+ "&entity;";
+ ExtTest test_data = { "", NULL, NULL };
+
+ XML_SetParamEntityParsing(parser,
+ XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE);
+ XML_SetUserData(parser, &test_data);
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader);
+ expect_failure(text,
+ XML_ERROR_UNDEFINED_ENTITY,
+ "Parser did not report undefined entity");
+}
+END_TEST
+
/* Test that no error is reported for unknown entities if we have read
an external subset, and standalone is false.
*/
@@ -1325,14 +1693,15 @@ START_TEST(test_wfc_undeclared_entity_with_external_subset) {
"\n"
"\n"
"&entity;";
- char foo_text[] =
- "";
+ ExtTest test_data = {
+ "",
+ NULL,
+ NULL
+ };
XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
- XML_SetUserData(parser, foo_text);
XML_SetExternalEntityRefHandler(parser, external_entity_loader);
- if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text), XML_TRUE) == XML_STATUS_ERROR)
- xml_failure(parser);
+ run_ext_character_check(text, &test_data, "");
}
END_TEST
@@ -1349,15 +1718,24 @@ START_TEST(test_not_standalone_handler_reject)
"\n"
"\n"
"&entity;";
- char foo_text[] =
- "";
+ ExtTest test_data = {
+ "",
+ NULL,
+ NULL
+ };
XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
- XML_SetUserData(parser, foo_text);
+ XML_SetUserData(parser, &test_data);
XML_SetExternalEntityRefHandler(parser, external_entity_loader);
XML_SetNotStandaloneHandler(parser, reject_not_standalone_handler);
- if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text), XML_TRUE) != XML_STATUS_ERROR)
- fail("NotStandalone handler failed to reject");
+ expect_failure(text, XML_ERROR_NOT_STANDALONE,
+ "NotStandalone handler failed to reject");
+
+ /* Try again but without external entity handling */
+ XML_ParserReset(parser, NULL);
+ XML_SetNotStandaloneHandler(parser, reject_not_standalone_handler);
+ expect_failure(text, XML_ERROR_NOT_STANDALONE,
+ "NotStandalone handler failed to reject");
}
END_TEST
@@ -1374,15 +1752,21 @@ START_TEST(test_not_standalone_handler_accept)
"\n"
"\n"
"&entity;";
- char foo_text[] =
- "";
+ ExtTest test_data = {
+ "",
+ NULL,
+ NULL
+ };
XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
- XML_SetUserData(parser, foo_text);
XML_SetExternalEntityRefHandler(parser, external_entity_loader);
XML_SetNotStandaloneHandler(parser, accept_not_standalone_handler);
- if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text), XML_TRUE) == XML_STATUS_ERROR)
- xml_failure(parser);
+ run_ext_character_check(text, &test_data, "");
+
+ /* Repeat wtihout the external entity handler */
+ XML_ParserReset(parser, NULL);
+ XML_SetNotStandaloneHandler(parser, accept_not_standalone_handler);
+ run_character_check(text, "");
}
END_TEST
@@ -1401,38 +1785,6 @@ START_TEST(test_wfc_no_recursive_entity_refs)
END_TEST
/* Test incomplete external entities are faulted */
-typedef struct ext_faults
-{
- const char *parse_text;
- const char *fail_text;
- enum XML_Error error;
-} ExtFaults;
-
-static int XMLCALL
-external_entity_faulter(XML_Parser parser,
- const XML_Char *context,
- const XML_Char *UNUSED_P(base),
- const XML_Char *UNUSED_P(systemId),
- const XML_Char *UNUSED_P(publicId))
-{
- XML_Parser ext_parser;
- ExtFaults *fault = (ExtFaults *)XML_GetUserData(parser);
-
- ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
- if (ext_parser == NULL)
- fail("Could not create external entity parser");
- if (_XML_Parse_SINGLE_BYTES(ext_parser,
- fault->parse_text,
- strlen(fault->parse_text),
- XML_TRUE) != XML_STATUS_ERROR)
- fail(fault->fail_text);
- if (XML_GetErrorCode(ext_parser) != fault->error)
- xml_failure(ext_parser);
-
- XML_ParserFree(ext_parser);
- return XML_STATUS_ERROR;
-}
-
START_TEST(test_ext_entity_invalid_parse)
{
const char *text =
@@ -1444,19 +1796,22 @@ START_TEST(test_ext_entity_invalid_parse)
{
"<",
"Incomplete element declaration not faulted",
+ NULL,
XML_ERROR_UNCLOSED_TOKEN
},
{
"<\xe2\x82", /* First two bytes of a three-byte char */
"Incomplete character not faulted",
+ NULL,
XML_ERROR_PARTIAL_CHAR
},
{
"\xe2\x82",
"Incomplete character in CDATA not faulted",
+ NULL,
XML_ERROR_PARTIAL_CHAR
},
- { NULL, NULL, XML_ERROR_NONE }
+ { NULL, NULL, NULL, XML_ERROR_NONE }
};
const ExtFaults *fault = faults;
@@ -1501,6 +1856,107 @@ START_TEST(test_dtd_default_handling)
}
END_TEST
+/* Test handling of attribute declarations */
+typedef struct AttTest {
+ const XML_Char *definition;
+ const XML_Char *element_name;
+ const XML_Char *attr_name;
+ const XML_Char *attr_type;
+ const XML_Char *default_value;
+ int is_required;
+} AttTest;
+
+static void XMLCALL
+verify_attlist_decl_handler(void *userData,
+ const XML_Char *element_name,
+ const XML_Char *attr_name,
+ const XML_Char *attr_type,
+ const XML_Char *default_value,
+ int is_required)
+{
+ AttTest *at = (AttTest *)userData;
+
+ if (strcmp(element_name, at->element_name))
+ fail("Unexpected element name in attribute declaration");
+ if (strcmp(attr_name, at->attr_name))
+ fail("Unexpected attribute name in attribute declaration");
+ if (strcmp(attr_type, at->attr_type))
+ fail("Unexpected attribute type in attribute declaration");
+ if ((default_value == NULL && at->default_value != NULL) ||
+ (default_value != NULL && at->default_value == NULL) ||
+ (default_value != NULL && strcmp(default_value, at->default_value)))
+ fail("Unexpected default value in attribute declaration");
+ if (is_required != at->is_required)
+ fail("Requirement mismatch in attribute declaration");
+}
+
+START_TEST(test_dtd_attr_handling)
+{
+ const char *prolog =
+ "\n";
+ AttTest attr_data[] = {
+ {
+ "\n"
+ "]>"
+ "",
+ "doc",
+ "a",
+ "(one|two|three)", /* Extraneous spaces will be removed */
+ NULL,
+ XML_TRUE
+ },
+ {
+ "\n"
+ "\n"
+ "]>"
+ "",
+ "doc",
+ "a",
+ "NOTATION(foo)",
+ NULL,
+ XML_FALSE
+ },
+ {
+ "\n"
+ "]>"
+ "",
+ "doc",
+ "a",
+ "NOTATION(foo)",
+ "bar",
+ XML_FALSE
+ },
+ {
+ "\n"
+ "]>"
+ "",
+ "doc",
+ "a",
+ "CDATA",
+ "\xdb\xb2",
+ XML_FALSE
+ },
+ { NULL, NULL, NULL, NULL, NULL, XML_FALSE }
+ };
+ AttTest *test;
+
+ for (test = attr_data; test->definition != NULL; test++) {
+ XML_SetAttlistDeclHandler(parser, verify_attlist_decl_handler);
+ XML_SetUserData(parser, test);
+ if (_XML_Parse_SINGLE_BYTES(parser, prolog, strlen(prolog),
+ XML_FALSE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ if (_XML_Parse_SINGLE_BYTES(parser,
+ test->definition,
+ strlen(test->definition),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ XML_ParserReset(parser, NULL);
+ }
+}
+END_TEST
+
/* See related SF bug #673791.
When namespace processing is enabled, setting the namespace URI for
a prefix is not allowed; this test ensures that it *is* allowed
@@ -1733,6 +2189,31 @@ START_TEST(test_good_cdata_utf16)
}
END_TEST
+START_TEST(test_good_cdata_utf16_le)
+{
+ /* Test data is:
+ *
+ *
+ */
+ const char text[] =
+ "<\0?\0x\0m\0l\0"
+ " \0v\0e\0r\0s\0i\0o\0n\0=\0'\0\x31\0.\0\x30\0'\0"
+ " \0e\0n\0c\0o\0d\0i\0n\0g\0=\0'\0u\0t\0f\0-\0""1\0""6\0'"
+ "\0?\0>\0\n"
+ "\0<\0a\0>\0<\0!\0[\0C\0D\0A\0T\0A\0[\0h\0e\0l\0l\0o\0]\0]\0>\0<\0/\0a\0>\0";
+ const char *expected = "hello";
+
+ CharData storage;
+ CharData_Init(&storage);
+ XML_SetUserData(parser, &storage);
+ XML_SetCharacterDataHandler(parser, accumulate_characters);
+
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text) - 1, XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
/* Test UTF16 conversion of a long cdata string */
/* 16 characters: handy macro to reduce visual clutter */
@@ -1841,6 +2322,34 @@ START_TEST(test_multichar_cdata_utf16)
}
END_TEST
+/* Test that an element name with a UTF-16 surrogate pair is rejected */
+START_TEST(test_utf16_bad_surrogate_pair)
+{
+ /* Test data is:
+ *
+ *
+ *
+ * where {BADLINB} is U+10000 (the first Linear B character)
+ * with the UTF-16 surrogate pair in the wrong order, i.e.
+ * 0xdc00 0xd800
+ */
+ const char text[] =
+ "\0<\0?\0x\0m\0l\0"
+ " \0v\0e\0r\0s\0i\0o\0n\0=\0'\0\x31\0.\0\x30\0'\0"
+ " \0e\0n\0c\0o\0d\0i\0n\0g\0=\0'\0u\0t\0f\0-\0""1\0""6\0'"
+ "\0?\0>\0\n"
+ "\0<\0a\0>\0<\0!\0[\0C\0D\0A\0T\0A\0["
+ "\xdc\x00\xd8\x00"
+ "\0]\0]\0>\0<\0/\0a\0>";
+
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text) - 1,
+ XML_TRUE) != XML_STATUS_ERROR)
+ fail("Reversed UTF-16 surrogate pair not faulted");
+ if (XML_GetErrorCode(parser) != XML_ERROR_INVALID_TOKEN)
+ xml_failure(parser);
+}
+END_TEST
+
START_TEST(test_bad_cdata)
{
@@ -2207,12 +2716,16 @@ START_TEST(test_set_foreign_dtd)
"\n";
const char *text2 =
"&entity;";
- char dtd_text[] = "";
+ ExtTest test_data = {
+ "",
+ NULL,
+ NULL
+ };
/* Check hash salt is passed through too */
XML_SetHashSalt(parser, 0x12345678);
XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
- XML_SetUserData(parser, dtd_text);
+ XML_SetUserData(parser, &test_data);
XML_SetExternalEntityRefHandler(parser, external_entity_loader);
/* Add a default handler to exercise more code paths */
XML_SetDefaultHandler(parser, dummy_default_handler);
@@ -2239,6 +2752,137 @@ START_TEST(test_set_foreign_dtd)
}
END_TEST
+/* Test foreign DTD handling with a failing NotStandalone handler */
+START_TEST(test_foreign_dtd_not_standalone)
+{
+ const char *text =
+ "\n"
+ "&entity;";
+ ExtTest test_data = {
+ "",
+ NULL,
+ NULL
+ };
+
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetUserData(parser, &test_data);
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader);
+ XML_SetNotStandaloneHandler(parser, reject_not_standalone_handler);
+ if (XML_UseForeignDTD(parser, XML_TRUE) != XML_ERROR_NONE)
+ fail("Could not set foreign DTD");
+ expect_failure(text, XML_ERROR_NOT_STANDALONE,
+ "NotStandalonehandler failed to reject");
+}
+END_TEST
+
+/* Test invalid character in a foreign DTD is faulted */
+START_TEST(test_invalid_foreign_dtd)
+{
+ const char *text =
+ "\n"
+ "&entity;";
+ ExtFaults test_data = {
+ "$",
+ "Dollar not faulted",
+ NULL,
+ XML_ERROR_INVALID_TOKEN
+ };
+
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetUserData(parser, &test_data);
+ XML_SetExternalEntityRefHandler(parser, external_entity_faulter);
+ XML_UseForeignDTD(parser, XML_TRUE);
+ expect_failure(text,
+ XML_ERROR_EXTERNAL_ENTITY_HANDLING,
+ "Bad DTD should not have been accepted");
+}
+END_TEST
+
+/* Test foreign DTD use with a doctype */
+START_TEST(test_foreign_dtd_with_doctype)
+{
+ const char *text1 =
+ "\n"
+ "]>\n";
+ const char *text2 =
+ "&entity;";
+ ExtTest test_data = {
+ "",
+ NULL,
+ NULL
+ };
+
+ /* Check hash salt is passed through too */
+ XML_SetHashSalt(parser, 0x12345678);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetUserData(parser, &test_data);
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader);
+ /* Add a default handler to exercise more code paths */
+ XML_SetDefaultHandler(parser, dummy_default_handler);
+ if (XML_UseForeignDTD(parser, XML_TRUE) != XML_ERROR_NONE)
+ fail("Could not set foreign DTD");
+ if (_XML_Parse_SINGLE_BYTES(parser, text1, strlen(text1),
+ XML_FALSE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+
+ /* Ensure that trying to set the DTD after parsing has started
+ * is faulted, even if it's the same setting.
+ */
+ if (XML_UseForeignDTD(parser, XML_TRUE) !=
+ XML_ERROR_CANT_CHANGE_FEATURE_ONCE_PARSING)
+ fail("Failed to reject late foreign DTD setting");
+ /* Ditto for the hash salt */
+ if (XML_SetHashSalt(parser, 0x23456789))
+ fail("Failed to reject late hash salt change");
+
+ /* Now finish the parse */
+ if (_XML_Parse_SINGLE_BYTES(parser, text2, strlen(text2),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test XML_UseForeignDTD with no external subset present */
+static int XMLCALL
+external_entity_null_loader(XML_Parser UNUSED_P(parser),
+ const XML_Char *UNUSED_P(context),
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId))
+{
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_foreign_dtd_without_external_subset)
+{
+ const char *text =
+ "]>\n"
+ "&foo;";
+
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetUserData(parser, NULL);
+ XML_SetExternalEntityRefHandler(parser, external_entity_null_loader);
+ XML_UseForeignDTD(parser, XML_TRUE);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+START_TEST(test_empty_foreign_dtd)
+{
+ const char *text =
+ "\n"
+ "&entity;";
+
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_null_loader);
+ XML_UseForeignDTD(parser, XML_TRUE);
+ expect_failure(text, XML_ERROR_UNDEFINED_ENTITY,
+ "Undefined entity not faulted");
+}
+END_TEST
+
/* Test XML Base is set and unset appropriately */
START_TEST(test_set_base)
{
@@ -2695,15 +3339,17 @@ START_TEST(test_ext_entity_invalid_suspended_parse)
{
"<",
"Incomplete element declaration not faulted",
+ NULL,
XML_ERROR_UNCLOSED_TOKEN
},
{
/* First two bytes of a three-byte char */
"\xe2\x82",
"Incomplete character not faulted",
+ NULL,
XML_ERROR_PARTIAL_CHAR
},
- { NULL, NULL, XML_ERROR_NONE }
+ { NULL, NULL, NULL, XML_ERROR_NONE }
};
ExtFaults *fault;
@@ -2749,6 +3395,7 @@ START_TEST(test_explicit_encoding)
}
END_TEST
+
/* Test handling of trailing CR (rather than newline) */
static void XMLCALL
cr_cdata_handler(void *userData, const XML_Char *s, int len)
@@ -3349,21 +3996,31 @@ START_TEST(test_byte_info_at_end)
END_TEST
/* Test position information from errors */
+#define PRE_ERROR_STR ""
START_TEST(test_byte_info_at_error)
{
- const char *text = "";
+ const char *text = PRE_ERROR_STR POST_ERROR_STR;
if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
XML_TRUE) == XML_STATUS_OK)
fail("Syntax error not faulted");
if (XML_GetCurrentByteCount(parser) != 0)
fail("Error byte count incorrect");
- if (XML_GetCurrentByteIndex(parser) != 7)
+ if (XML_GetCurrentByteIndex(parser) != strlen(PRE_ERROR_STR))
fail("Error byte index incorrect");
}
END_TEST
+#undef PRE_ERROR_STR
+#undef POST_ERROR_STR
/* Test position information in handler */
+typedef struct ByteTestData {
+ int start_element_len;
+ int cdata_len;
+ int total_string_len;
+} ByteTestData;
+
static void
byte_character_handler(void *userData,
const XML_Char *s,
@@ -3372,17 +4029,21 @@ byte_character_handler(void *userData,
#ifdef XML_CONTEXT_BYTES
int offset, size;
const char *buffer;
- intptr_t buflen = (intptr_t)userData;
+ ByteTestData *data = (ByteTestData *)userData;
buffer = XML_GetInputContext(parser, &offset, &size);
if (buffer == NULL)
fail("Failed to get context buffer");
+ if (offset != data->start_element_len)
+ fail("Context offset in unexpected position");
+ if (len != data->cdata_len)
+ fail("CDATA length reported incorrectly");
+ if (size != data->total_string_len)
+ fail("Context size is not full buffer");
if (XML_GetCurrentByteIndex(parser) != offset)
fail("Character byte index incorrect");
if (XML_GetCurrentByteCount(parser) != len)
fail("Character byte count incorrect");
- if (buflen != size)
- fail("Buffer length incorrect");
if (s != buffer + offset)
fail("Buffer position incorrect");
#else
@@ -3392,21 +4053,31 @@ byte_character_handler(void *userData,
#endif
}
+#define START_ELEMENT ""
+#define CDATA_TEXT "Hello"
+#define END_ELEMENT ""
START_TEST(test_byte_info_at_cdata)
{
- const char *text = "Hello";
+ const char *text = START_ELEMENT CDATA_TEXT END_ELEMENT;
int offset, size;
+ ByteTestData data;
/* Check initial context is empty */
if (XML_GetInputContext(parser, &offset, &size) != NULL)
fail("Unexpected context at start of parse");
+ data.start_element_len = strlen(START_ELEMENT);
+ data.cdata_len = strlen(CDATA_TEXT);
+ data.total_string_len = strlen(text);
XML_SetCharacterDataHandler(parser, byte_character_handler);
- XML_SetUserData(parser, (void *)strlen(text));
+ XML_SetUserData(parser, &data);
if (XML_Parse(parser, text, strlen(text), XML_TRUE) != XML_STATUS_OK)
xml_failure(parser);
}
END_TEST
+#undef START_ELEMENT
+#undef CDATA_TEXT
+#undef END_ELEMENT
/* Test predefined entities are correctly recognised */
START_TEST(test_predefined_entities)
@@ -3499,6 +4170,27 @@ START_TEST(test_invalid_tag_in_dtd)
}
END_TEST
+/* Test entities not quite the predefined ones are not mis-recognised */
+START_TEST(test_not_predefined_entities)
+{
+ const char *text[] = {
+ "&pt;",
+ "&amo;",
+ "&quid;",
+ "&apod;",
+ NULL
+ };
+ int i = 0;
+
+ while (text[i] != NULL) {
+ expect_failure(text[i], XML_ERROR_UNDEFINED_ENTITY,
+ "Undefined entity not rejected");
+ XML_ParserReset(parser, NULL);
+ i++;
+ }
+}
+END_TEST
+
/* Test conditional inclusion (IGNORE) */
static int XMLCALL
external_entity_load_ignore(XML_Parser parser,
@@ -3547,6 +4239,116 @@ START_TEST(test_ignore_section)
}
END_TEST
+static int XMLCALL
+external_entity_load_ignore_utf16(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId))
+{
+ const char text[] =
+ /* ]]> */
+ "<\0!\0[\0I\0G\0N\0O\0R\0E\0[\0"
+ "<\0!\0E\0L\0E\0M\0E\0N\0T\0 \0e\0 \0"
+ "(\0#\0P\0C\0D\0A\0T\0A\0)\0*\0>\0]\0]\0>\0";
+ XML_Parser ext_parser;
+
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ fail("Could not create external entity parser");
+ if (_XML_Parse_SINGLE_BYTES(ext_parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+
+ XML_ParserFree(ext_parser);
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_ignore_section_utf16)
+{
+ const char text[] =
+ /* */
+ "<\0!\0D\0O\0C\0T\0Y\0P\0E\0 \0d\0 "
+ "\0S\0Y\0S\0T\0E\0M\0 \0'\0s\0'\0>\0\n\0"
+ /* &en; */
+ "<\0d\0>\0<\0e\0>\0&\0e\0n\0;\0<\0/\0e\0>\0<\0/\0d\0>\0";
+ const XML_Char *expected =
+ "]]>\n&en;";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetUserData(parser, &storage);
+ XML_SetExternalEntityRefHandler(parser,
+ external_entity_load_ignore_utf16);
+ XML_SetDefaultHandler(parser, accumulate_characters);
+ XML_SetStartDoctypeDeclHandler(parser, dummy_start_doctype_handler);
+ XML_SetEndDoctypeDeclHandler(parser, dummy_end_doctype_handler);
+ XML_SetElementDeclHandler(parser, dummy_element_decl_handler);
+ XML_SetStartElementHandler(parser, dummy_start_element);
+ XML_SetEndElementHandler(parser, dummy_end_element);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+static int XMLCALL
+external_entity_load_ignore_utf16_be(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId))
+{
+ const char text[] =
+ /* ]]> */
+ "\0<\0!\0[\0I\0G\0N\0O\0R\0E\0["
+ "\0<\0!\0E\0L\0E\0M\0E\0N\0T\0 \0e\0 "
+ "\0(\0#\0P\0C\0D\0A\0T\0A\0)\0*\0>\0]\0]\0>";
+ XML_Parser ext_parser;
+
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ fail("Could not create external entity parser");
+ if (_XML_Parse_SINGLE_BYTES(ext_parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+
+ XML_ParserFree(ext_parser);
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_ignore_section_utf16_be)
+{
+ const char text[] =
+ /* */
+ "\0<\0!\0D\0O\0C\0T\0Y\0P\0E\0 \0d\0 "
+ "\0S\0Y\0S\0T\0E\0M\0 \0'\0s\0'\0>\0\n"
+ /* &en; */
+ "\0<\0d\0>\0<\0e\0>\0&\0e\0n\0;\0<\0/\0e\0>\0<\0/\0d\0>";
+ const XML_Char *expected =
+ "]]>\n&en;";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetUserData(parser, &storage);
+ XML_SetExternalEntityRefHandler(parser,
+ external_entity_load_ignore_utf16_be);
+ XML_SetDefaultHandler(parser, accumulate_characters);
+ XML_SetStartDoctypeDeclHandler(parser, dummy_start_doctype_handler);
+ XML_SetEndDoctypeDeclHandler(parser, dummy_end_doctype_handler);
+ XML_SetElementDeclHandler(parser, dummy_element_decl_handler);
+ XML_SetStartElementHandler(parser, dummy_start_element);
+ XML_SetEndElementHandler(parser, dummy_end_element);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
/* Test mis-formatted conditional exclusion */
START_TEST(test_bad_ignore_section)
{
@@ -3557,14 +4359,23 @@ START_TEST(test_bad_ignore_section)
{
"",
"Invalid XML character not faulted",
+ NULL,
XML_ERROR_INVALID_TOKEN
},
- { NULL, NULL, XML_ERROR_NONE }
+ {
+ /* FIrst two bytes of a three-byte char */
+ "\n"
+ "\n"
+ "\n"
+ "%e1;\n";
+ XML_Parser ext_parser;
+
+ if (systemId == NULL)
+ return XML_STATUS_OK;
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ fail("Could not create external entity parser");
+ if (!strcmp(systemId, "004-1.ent")) {
+ if (_XML_Parse_SINGLE_BYTES(ext_parser, text1, strlen(text1),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(ext_parser);
+ }
+ else if (!strcmp(systemId, "004-2.ent")) {
+ ExtFaults *fault = (ExtFaults *)XML_GetUserData(parser);
+ enum XML_Status status;
+ enum XML_Error error;
+
+ status = _XML_Parse_SINGLE_BYTES(ext_parser,
+ fault->parse_text,
+ strlen(fault->parse_text),
+ XML_TRUE);
+ if (fault->error == XML_ERROR_NONE) {
+ if (status == XML_STATUS_ERROR)
+ xml_failure(ext_parser);
+ } else {
+ if (status != XML_STATUS_ERROR)
+ fail(fault->fail_text);
+ error = XML_GetErrorCode(ext_parser);
+ if (error != fault->error &&
+ (fault->error != XML_ERROR_XML_DECL ||
+ error != XML_ERROR_TEXT_DECL))
+ xml_failure(ext_parser);
+ }
+ }
+
+ XML_ParserFree(ext_parser);
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_external_entity_values)
+{
+ const char *text =
+ "\n"
+ "\n";
+ ExtFaults data_004_2[] = {
+ {
+ "",
+ NULL,
+ NULL,
+ XML_ERROR_NONE
+ },
+ {
+ "",
+ "Invalid token not faulted",
+ NULL,
+ XML_ERROR_INVALID_TOKEN
+ },
+ {
+ "'wombat",
+ "Unterminated string not faulted",
+ NULL,
+ XML_ERROR_UNCLOSED_TOKEN
+ },
+ {
+ "\xe2\x82",
+ "Partial UTF-8 character not faulted",
+ NULL,
+ XML_ERROR_PARTIAL_CHAR
+ },
+ {
+ "\n",
+ NULL,
+ NULL,
+ XML_ERROR_NONE
+ },
+ {
+ "",
+ "Malformed XML declaration not faulted",
+ NULL,
+ XML_ERROR_XML_DECL
+ },
+ {
+ /* UTF-8 BOM */
+ "\xEF\xBB\xBF",
+ NULL,
+ NULL,
+ XML_ERROR_NONE
+ },
+ {
+ "\n$",
+ "Invalid token after text declaration not faulted",
+ NULL,
+ XML_ERROR_INVALID_TOKEN
+ },
+ {
+ "\n'wombat",
+ "Unterminated string after text decl not faulted",
+ NULL,
+ XML_ERROR_UNCLOSED_TOKEN
+ },
+ {
+ "\n\xe2\x82",
+ "Partial UTF-8 character after text decl not faulted",
+ NULL,
+ XML_ERROR_PARTIAL_CHAR
+ },
+ {
+ "%e1;",
+ "Recursive parameter entity not faulted",
+ NULL,
+ XML_ERROR_RECURSIVE_ENTITY_REF
+ },
+ { NULL, NULL, NULL, XML_ERROR_NONE }
+ };
+ int i;
+
+ for (i = 0; data_004_2[i].parse_text != NULL; i++) {
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_valuer);
+ XML_SetUserData(parser, &data_004_2[i]);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ XML_ParserReset(parser, NULL);
+ }
+}
+END_TEST
+
+/* Test the recursive parse interacts with a not standalone handler */
+static int XMLCALL
+external_entity_not_standalone(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *systemId,
+ const XML_Char *UNUSED_P(publicId))
+{
+ const char *text1 =
+ "\n"
+ "\n"
+ "%e1;\n";
+ const char *text2 = "";
+ XML_Parser ext_parser;
+
+ if (systemId == NULL)
+ return XML_STATUS_OK;
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ fail("Could not create external entity parser");
+ if (!strcmp(systemId, "foo")) {
+ XML_SetNotStandaloneHandler(ext_parser,
+ reject_not_standalone_handler);
+ if (_XML_Parse_SINGLE_BYTES(ext_parser, text1, strlen(text1),
+ XML_TRUE) != XML_STATUS_ERROR)
+ fail("Expected not standalone rejection");
+ if (XML_GetErrorCode(ext_parser) != XML_ERROR_NOT_STANDALONE)
+ xml_failure(ext_parser);
+ XML_SetNotStandaloneHandler(ext_parser, NULL);
+ XML_ParserFree(ext_parser);
+ return XML_STATUS_ERROR;
+ }
+ else if (!strcmp(systemId, "bar")) {
+ if (_XML_Parse_SINGLE_BYTES(ext_parser, text2, strlen(text2),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(ext_parser);
+ }
+
+ XML_ParserFree(ext_parser);
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_ext_entity_not_standalone)
+{
+ const char *text =
+ "\n"
+ "";
+
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_not_standalone);
+ expect_failure(text, XML_ERROR_EXTERNAL_ENTITY_HANDLING,
+ "Standalone rejection not caught");
+}
+END_TEST
+
+static int XMLCALL
+external_entity_value_aborter(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *systemId,
+ const XML_Char *UNUSED_P(publicId))
+{
+ const char *text1 =
+ "\n"
+ "\n"
+ "\n"
+ "%e1;\n";
+ const char *text2 =
+ "";
+ XML_Parser ext_parser;
+
+ if (systemId == NULL)
+ return XML_STATUS_OK;
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ fail("Could not create external entity parser");
+ if (!strcmp(systemId, "004-1.ent")) {
+ if (_XML_Parse_SINGLE_BYTES(ext_parser, text1, strlen(text1),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(ext_parser);
+ }
+ if (!strcmp(systemId, "004-2.ent")) {
+ XML_SetXmlDeclHandler(ext_parser, entity_suspending_xdecl_handler);
+ XML_SetUserData(ext_parser, ext_parser);
+ if (_XML_Parse_SINGLE_BYTES(ext_parser, text2, strlen(text2),
+ XML_TRUE) != XML_STATUS_ERROR)
+ fail("Aborted parse not faulted");
+ if (XML_GetErrorCode(ext_parser) != XML_ERROR_ABORTED)
+ xml_failure(ext_parser);
+ }
+
+ XML_ParserFree(ext_parser);
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_ext_entity_value_abort)
+{
+ const char *text =
+ "\n"
+ "\n";
+
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser,
+ external_entity_value_aborter);
+ resumable = XML_FALSE;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+START_TEST(test_bad_public_doctype)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "";
+
+ /* Setting a handler provokes a particular code path */
+ XML_SetDoctypeDeclHandler(parser,
+ dummy_start_doctype_handler,
+ dummy_end_doctype_handler);
+ expect_failure(text, XML_ERROR_PUBLICID, "Bad Public ID not failed");
+}
+END_TEST
+
+/* Test based on ibm/valid/P32/ibm32v04.xml */
+START_TEST(test_attribute_enum_value)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "This is a \n \n\nyellow tiger";
+ ExtTest dtd_data = {
+ "\n"
+ "\n"
+ "",
+ NULL,
+ NULL
+ };
+
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader);
+ XML_SetUserData(parser, &dtd_data);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ /* An attribute list handler provokes a different code path */
+ XML_SetAttlistDeclHandler(parser, dummy_attlist_decl_handler);
+ run_ext_character_check(text, &dtd_data,
+ "This is a \n \n\nyellow tiger");
+}
+END_TEST
+
+/* Slightly bizarrely, the library seems to silently ignore entity
+ * definitions for predefined entities, even when they are wrong. The
+ * language of the XML 1.0 spec is somewhat unhelpful as to what ought
+ * to happen, so this is currently treated as acceptable.
+ */
+START_TEST(test_predefined_entity_redefinition)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "'";
+ run_character_check(text, "'");
+}
+END_TEST
+
+/* Test that the parser stops processing the DTD after an unresolved
+ * parameter entity is encountered.
+ */
+START_TEST(test_dtd_stop_processing)
+{
+ const char *text =
+ "\n"
+ "]>";
+
+ XML_SetEntityDeclHandler(parser, dummy_entity_decl_handler);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ if (dummy_handler_flags != 0)
+ fail("DTD processing still going after undefined PE");
+}
+END_TEST
+
+/* Test public notations with no system ID */
+START_TEST(test_public_notation_no_sysid)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "]>\n";
+
+ dummy_handler_flags = 0;
+ XML_SetNotationDeclHandler(parser, dummy_notation_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ if (dummy_handler_flags != DUMMY_NOTATION_DECL_HANDLER_FLAG)
+ fail("Notation declaration handler not called");
+}
+END_TEST
+
+static void XMLCALL
+record_element_start_handler(void *userData,
+ const XML_Char *name,
+ const XML_Char **UNUSED_P(atts))
+{
+ CharData_AppendString((CharData *)userData, name);
+}
+
+START_TEST(test_nested_groups)
+{
+ const char *text =
+ "\n"
+ ""
+ "]>\n"
+ "";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetElementDeclHandler(parser, dummy_element_decl_handler);
+ XML_SetStartElementHandler(parser, record_element_start_handler);
+ XML_SetUserData(parser, &storage);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckString(&storage, "doce");
+ if (dummy_handler_flags != DUMMY_ELEMENT_DECL_HANDLER_FLAG)
+ fail("Element handler not fired");
+}
+END_TEST
+
+START_TEST(test_group_choice)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "\n"
+ "\n"
+ "]>\n"
+ "\n"
+ "\n"
+ "This is a foo\n"
+ "\n"
+ "\n";
+
+ XML_SetElementDeclHandler(parser, dummy_element_decl_handler);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ if (dummy_handler_flags != DUMMY_ELEMENT_DECL_HANDLER_FLAG)
+ fail("Element handler flag not raised");
+}
+END_TEST
+
+static int XMLCALL
+external_entity_public(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *systemId,
+ const XML_Char *publicId)
+{
+ const char *text1 = (const char *)XML_GetUserData(parser);
+ const char *text2 = "";
+ const char *text = NULL;
+ XML_Parser ext_parser;
+ int parse_res;
+
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ return XML_STATUS_ERROR;
+ if (systemId != NULL && !strcmp(systemId, "http://example.org/")) {
+ text = text1;
+ }
+ else if (publicId != NULL && !strcmp(publicId, "foo")) {
+ text = text2;
+ }
+ else
+ fail("Unexpected parameters to external entity parser");
+ parse_res = _XML_Parse_SINGLE_BYTES(ext_parser, text, strlen(text),
+ XML_TRUE);
+ XML_ParserFree(ext_parser);
+ return parse_res;
+}
+
+START_TEST(test_standalone_parameter_entity)
+{
+ const char *text =
+ "\n"
+ "'>\n"
+ "%entity;\n"
+ "]>\n"
+ "";
+ char dtd_data[] =
+ "\n";
+
+ XML_SetUserData(parser, dtd_data);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_public);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test skipping of parameter entity in an external DTD */
+/* Derived from ibm/invalid/P69/ibm69i01.xml */
+START_TEST(test_skipped_parameter_entity)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "]>\n"
+ "";
+ ExtTest dtd_data = { "%pe2;", NULL, NULL };
+
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader);
+ XML_SetUserData(parser, &dtd_data);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetSkippedEntityHandler(parser, dummy_skip_handler);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ if (dummy_handler_flags != DUMMY_SKIP_HANDLER_FLAG)
+ fail("Skip handler not executed");
+}
+END_TEST
+
+/* Test recursive parameter entity definition rejected in external DTD */
+START_TEST(test_recursive_external_parameter_entity)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "]>\n"
+ "";
+ ExtFaults dtd_data = {
+ "\n%pe2;",
+ "Recursive external parameter entity not faulted",
+ NULL,
+ XML_ERROR_RECURSIVE_ENTITY_REF
+ };
+
+ XML_SetExternalEntityRefHandler(parser, external_entity_faulter);
+ XML_SetUserData(parser, &dtd_data);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ expect_failure(text,
+ XML_ERROR_EXTERNAL_ENTITY_HANDLING,
+ "Recursive external parameter not spotted");
+}
+END_TEST
+
+/* Test undefined parameter entity in external entity handler */
+static int XMLCALL
+external_entity_devaluer(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *systemId,
+ const XML_Char *UNUSED_P(publicId))
+{
+ const char *text =
+ "\n"
+ "\n"
+ "%e1;\n";
+ XML_Parser ext_parser;
+ int clear_handler = (intptr_t)XML_GetUserData(parser);
+
+ if (systemId == NULL || !strcmp(systemId, "bar"))
+ return XML_STATUS_OK;
+ if (strcmp(systemId, "foo"))
+ fail("Unexpected system ID");
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ fail("Could note create external entity parser");
+ if (clear_handler)
+ XML_SetExternalEntityRefHandler(ext_parser, NULL);
+ if (_XML_Parse_SINGLE_BYTES(ext_parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(ext_parser);
+
+ XML_ParserFree(ext_parser);
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_undefined_ext_entity_in_external_dtd)
+{
+ const char *text =
+ "\n"
+ "\n";
+
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_devaluer);
+ XML_SetUserData(parser, (void *)(intptr_t)XML_FALSE);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+
+ /* Now repeat without the external entity ref handler invoking
+ * another copy of itself.
+ */
+ XML_ParserReset(parser, NULL);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_devaluer);
+ XML_SetUserData(parser, (void *)(intptr_t)XML_TRUE);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+
+static void XMLCALL
+aborting_xdecl_handler(void *UNUSED_P(userData),
+ const XML_Char *UNUSED_P(version),
+ const XML_Char *UNUSED_P(encoding),
+ int UNUSED_P(standalone))
+{
+ XML_StopParser(parser, resumable);
+ XML_SetXmlDeclHandler(parser, NULL);
+}
+
+/* Test suspending the parse on receiving an XML declaration works */
+START_TEST(test_suspend_xdecl)
+{
+ const char *text = long_character_data_text;
+
+ XML_SetXmlDeclHandler(parser, aborting_xdecl_handler);
+ resumable = XML_TRUE;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_SUSPENDED)
+ xml_failure(parser);
+ if (XML_GetErrorCode(parser) != XML_ERROR_NONE)
+ xml_failure(parser);
+ /* Attempt to start a new parse while suspended */
+ if (XML_Parse(parser, text, strlen(text), XML_TRUE) != XML_STATUS_ERROR)
+ fail("Attempt to parse while suspended not faulted");
+ if (XML_GetErrorCode(parser) != XML_ERROR_SUSPENDED)
+ fail("Suspended parse not faulted with correct error");
+}
+END_TEST
+
+/* Test aborting the parse in an epilog works */
+static void XMLCALL
+selective_aborting_default_handler(void *userData,
+ const XML_Char *s,
+ int len)
+{
+ const char *match = (const char *)userData;
+
+ if (match == NULL ||
+ (strlen(match) == (unsigned)len &&
+ !strncmp(match, s, len))) {
+ XML_StopParser(parser, resumable);
+ XML_SetDefaultHandler(parser, NULL);
+ }
+}
+
+START_TEST(test_abort_epilog)
+{
+ const char *text = "\n\r\n";
+ char match[] = "\r";
+
+ XML_SetDefaultHandler(parser, selective_aborting_default_handler);
+ XML_SetUserData(parser, match);
+ resumable = XML_FALSE;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ fail("Abort not triggered");
+ if (XML_GetErrorCode(parser) != XML_ERROR_ABORTED)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test a different code path for abort in the epilog */
+START_TEST(test_abort_epilog_2)
+{
+ const char *text = "\n";
+ char match[] = "\n";
+
+ XML_SetDefaultHandler(parser, selective_aborting_default_handler);
+ XML_SetUserData(parser, match);
+ resumable = XML_FALSE;
+ expect_failure(text, XML_ERROR_ABORTED, "Abort not triggered");
+}
+END_TEST
+
+/* Test suspension from the epilog */
+START_TEST(test_suspend_epilog)
+{
+ const char *text = "\n";
+ char match[] = "\n";
+
+ XML_SetDefaultHandler(parser, selective_aborting_default_handler);
+ XML_SetUserData(parser, match);
+ resumable = XML_TRUE;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_SUSPENDED)
+ xml_failure(parser);
+}
+END_TEST
+
+START_TEST(test_unfinished_epilog)
+{
+ const char *text = "<";
+
+ expect_failure(text, XML_ERROR_UNCLOSED_TOKEN,
+ "Incomplete epilog entry not faulted");
+}
+END_TEST
+
+START_TEST(test_partial_char_in_epilog)
+{
+ const char *text = "\xe2\x82";
+
+ /* First check that no fault is raised if the parse is not finished */
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_FALSE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ /* Now check that it is faulted once we finish */
+ if (XML_ParseBuffer(parser, 0, XML_TRUE) != XML_STATUS_ERROR)
+ fail("Partial character in epilog not faulted");
+ if (XML_GetErrorCode(parser) != XML_ERROR_PARTIAL_CHAR)
+ xml_failure(parser);
+}
+END_TEST
+
+START_TEST(test_hash_collision)
+{
+ /* For full coverage of the lookup routine, we need to ensure a
+ * hash collision even though we can only tell that we have one
+ * through breakpoint debugging or coverage statistics. The
+ * following will cause a hash collision on machines with a 64-bit
+ * long type; others will have to experiment. The full coverage
+ * tests invoked from qa.sh usually provide a hash collision, but
+ * not always. This is an attempt to provide insurance.
+ */
+#define COLLIDING_HASH_SALT 0xffffffffff99fc90
+ const char * text =
+ "\n"
+ "\n"
+ "This is a foo\n"
+ "\n"
+ "\n"
+ "\n"
+ "This triggers the table growth and collides with b2\n"
+ "\n";
+
+ XML_SetHashSalt(parser, COLLIDING_HASH_SALT);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+#undef COLLIDING_HASH_SALT
+
+/* Test resuming a parse suspended in entity substitution */
+static void XMLCALL
+start_element_suspender(void *UNUSED_P(userData),
+ const XML_Char *name,
+ const XML_Char **UNUSED_P(atts))
+{
+ if (!strcmp(name, "suspend"))
+ XML_StopParser(parser, XML_TRUE);
+}
+
+START_TEST(test_suspend_resume_internal_entity)
+{
+ const char *text =
+ "HiHo'>\n"
+ "]>\n"
+ "&foo;\n";
+ const char *expected1 = "Hi";
+ const char *expected2 = "HiHo";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetStartElementHandler(parser, start_element_suspender);
+ XML_SetCharacterDataHandler(parser, accumulate_characters);
+ XML_SetUserData(parser, &storage);
+ if (XML_Parse(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_SUSPENDED)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, "");
+ if (XML_ResumeParser(parser) != XML_STATUS_SUSPENDED)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected1);
+ if (XML_ResumeParser(parser) != XML_STATUS_OK)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected2);
+}
+END_TEST
+
+/* Test syntax error is caught at parse resumption */
+START_TEST(test_resume_entity_with_syntax_error)
+{
+ const char *text =
+ "Hi'>\n"
+ "]>\n"
+ "&foo;\n";
+
+ XML_SetStartElementHandler(parser, start_element_suspender);
+ if (XML_Parse(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_SUSPENDED)
+ xml_failure(parser);
+ if (XML_ResumeParser(parser) != XML_STATUS_ERROR)
+ fail("Syntax error in entity not faulted");
+ if (XML_GetErrorCode(parser) != XML_ERROR_TAG_MISMATCH)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test suspending and resuming in a parameter entity substitution */
+static void XMLCALL
+element_decl_suspender(void *UNUSED_P(userData),
+ const XML_Char *UNUSED_P(name),
+ XML_Content *model)
+{
+ XML_StopParser(parser, XML_TRUE);
+ XML_FreeContentModel(parser, model);
+}
+
+START_TEST(test_suspend_resume_parameter_entity)
+{
+ const char *text =
+ "'>\n"
+ "%foo;\n"
+ "]>\n"
+ "Hello, world";
+ const char *expected = "Hello, world";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetElementDeclHandler(parser, element_decl_suspender);
+ XML_SetCharacterDataHandler(parser, accumulate_characters);
+ XML_SetUserData(parser, &storage);
+ if (XML_Parse(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_SUSPENDED)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, "");
+ if (XML_ResumeParser(parser) != XML_STATUS_OK)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test attempting to use parser after an error is faulted */
+START_TEST(test_restart_on_error)
+{
+ const char *text = "<$doc>";
+
+ if (XML_Parse(parser, text, strlen(text), XML_TRUE) != XML_STATUS_ERROR)
+ fail("Invalid tag name not faulted");
+ if (XML_GetErrorCode(parser) != XML_ERROR_INVALID_TOKEN)
+ xml_failure(parser);
+ if (XML_Parse(parser, NULL, 0, XML_TRUE) != XML_STATUS_ERROR)
+ fail("Restarting invalid parse not faulted");
+ if (XML_GetErrorCode(parser) != XML_ERROR_INVALID_TOKEN)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test that angle brackets in an attribute default value are faulted */
+START_TEST(test_reject_lt_in_attribute_value)
+{
+ const char *text =
+ "'>]>\n"
+ "";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Bad attribute default not faulted");
+}
+END_TEST
+
+START_TEST(test_reject_unfinished_param_in_att_value)
+{
+ const char *text =
+ "]>\n"
+ "";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Bad attribute default not faulted");
+}
+END_TEST
+
+START_TEST(test_trailing_cr_in_att_value)
+{
+ const char *text = "";
+
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Try parsing a general entity within a parameter entity in a
+ * standalone internal DTD. Covers a corner case in the parser.
+ */
+START_TEST(test_standalone_internal_entity)
+{
+ const char *text =
+ "\n"
+ "\n"
+ " '>\n"
+ " \n"
+ " %pe;\n"
+ "]>\n"
+ "";
+
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test that a reference to an unknown external entity is skipped */
+START_TEST(test_skipped_external_entity)
+{
+ const char *text =
+ "\n"
+ "\n";
+ ExtTest test_data = {
+ "\n"
+ "\n",
+ NULL,
+ NULL
+ };
+
+ XML_SetUserData(parser, &test_data);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test a different form of unknown external entity */
+typedef struct ext_hdlr_data {
+ const char *parse_text;
+ XML_ExternalEntityRefHandler handler;
+} ExtHdlrData;
+
+static int XMLCALL
+external_entity_oneshot_loader(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId))
+{
+ ExtHdlrData *test_data = (ExtHdlrData *)XML_GetUserData(parser);
+ XML_Parser ext_parser;
+
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ fail("Could not create external entity parser.");
+ /* Use the requested entity parser for further externals */
+ XML_SetExternalEntityRefHandler(ext_parser, test_data->handler);
+ if ( _XML_Parse_SINGLE_BYTES(ext_parser,
+ test_data->parse_text,
+ strlen(test_data->parse_text),
+ XML_TRUE) == XML_STATUS_ERROR) {
+ xml_failure(ext_parser);
+ }
+
+ XML_ParserFree(ext_parser);
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_skipped_null_loaded_ext_entity)
+{
+ const char *text =
+ "\n"
+ "";
+ ExtHdlrData test_data = {
+ "\n"
+ "\n"
+ "%pe2;\n",
+ external_entity_null_loader
+ };
+
+ XML_SetUserData(parser, &test_data);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_oneshot_loader);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+START_TEST(test_skipped_unloaded_ext_entity)
+{
+ const char *text =
+ "\n"
+ "";
+ ExtHdlrData test_data = {
+ "\n"
+ "\n"
+ "%pe2;\n",
+ NULL
+ };
+
+ XML_SetUserData(parser, &test_data);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_oneshot_loader);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test that a parameter entity value ending with a carriage return
+ * has it translated internally into a newline.
+ */
+START_TEST(test_param_entity_with_trailing_cr)
+{
+#define PARAM_ENTITY_NAME "pe"
+#define PARAM_ENTITY_CORE_VALUE ""
+ const char *text =
+ "\n"
+ "";
+ ExtTest test_data = {
+ "\n"
+ "%" PARAM_ENTITY_NAME ";\n",
+ NULL,
+ NULL
+ };
+
+ XML_SetUserData(parser, &test_data);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader);
+ XML_SetEntityDeclHandler(parser, param_entity_match_handler);
+ entity_name_to_match = PARAM_ENTITY_NAME;
+ entity_value_to_match = PARAM_ENTITY_CORE_VALUE "\n";
+ entity_match_flag = ENTITY_MATCH_NOT_FOUND;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ if (entity_match_flag == ENTITY_MATCH_FAIL)
+ fail("Parameter entity CR->NEWLINE conversion failed");
+ else if (entity_match_flag == ENTITY_MATCH_NOT_FOUND)
+ fail("Parameter entity not parsed");
+}
+#undef PARAM_ENTITY_NAME
+#undef PARAM_ENTITY_CORE_VALUE
+END_TEST
+
+START_TEST(test_invalid_character_entity)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&entity;";
+
+ expect_failure(text, XML_ERROR_BAD_CHAR_REF,
+ "Out of range character reference not faulted");
+}
+END_TEST
+
+START_TEST(test_invalid_character_entity_2)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&entity;";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Out of range character reference not faulted");
+}
+END_TEST
+
+START_TEST(test_invalid_character_entity_3)
+{
+ const char text[] =
+ /* \n */
+ "\0<\0!\0E\0N\0T\0I\0T\0Y\0 \0e\0n\0t\0i\0t\0y\0 "
+ "\0'\0&\x0e\x04\x0e\x08\0;\0'\0>\0\n"
+ /* ]>\n */
+ "\0]\0>\0\n"
+ /* &entity; */
+ "\0<\0d\0o\0c\0>\0&\0e\0n\0t\0i\0t\0y\0;\0<\0/\0d\0o\0c\0>";
+
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) != XML_STATUS_ERROR)
+ fail("Invalid start of entity name not faulted");
+ if (XML_GetErrorCode(parser) != XML_ERROR_UNDEFINED_ENTITY)
+ xml_failure(parser);
+}
+END_TEST
+
+START_TEST(test_invalid_character_entity_4)
+{
+ const char *text =
+ "\n" /* = � */
+ "]>\n"
+ "&entity;";
+
+ expect_failure(text, XML_ERROR_BAD_CHAR_REF,
+ "Out of range character reference not faulted");
+}
+END_TEST
+
+
+/* Test that processing instructions are picked up by a default handler */
+START_TEST(test_pi_handled_in_default)
+{
+ const char *text = "\n";
+ const XML_Char *expected = "\n";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetDefaultHandler(parser, accumulate_characters);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE)== XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+
+/* Test that comments are picked up by a default handler */
+START_TEST(test_comment_handled_in_default)
+{
+ const char *text = "\n";
+ const XML_Char *expected = "\n";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetDefaultHandler(parser, accumulate_characters);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test PIs that look almost but not quite like XML declarations */
+static void XMLCALL
+accumulate_pi_characters(void *userData,
+ const XML_Char *target,
+ const XML_Char *data)
+{
+ CharData *storage = (CharData *)userData;
+
+ CharData_AppendXMLChars(storage, target, -1);
+ CharData_AppendXMLChars(storage, ": ", 2);
+ CharData_AppendXMLChars(storage, data, -1);
+ CharData_AppendXMLChars(storage, "\n", 1);
+}
+
+START_TEST(test_pi_yml)
+{
+ const char *text = "";
+ const XML_Char *expected = "yml: something like data\n";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetProcessingInstructionHandler(parser, accumulate_pi_characters);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_pi_xnl)
+{
+ const char *text = "";
+ const XML_Char *expected = "xnl: nothing like data\n";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetProcessingInstructionHandler(parser, accumulate_pi_characters);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_pi_xmm)
+{
+ const char *text = "";
+ const XML_Char *expected = "xmm: everything like data\n";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetProcessingInstructionHandler(parser, accumulate_pi_characters);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_utf16_pi)
+{
+ const char text[] =
+ /*
+ * where {KHO KHWAI} = U+0E04
+ * and {CHO CHAN} = U+0E08
+ */
+ "<\0?\0\x04\x0e\x08\x0e?\0>\0"
+ /* */
+ "<\0q\0/\0>\0";
+ const XML_Char *expected = "\xe0\xb8\x84\xe0\xb8\x88: \n";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetProcessingInstructionHandler(parser, accumulate_pi_characters);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_utf16_be_pi)
+{
+ const char text[] =
+ /*
+ * where {KHO KHWAI} = U+0E04
+ * and {CHO CHAN} = U+0E08
+ */
+ "\0<\0?\x0e\x04\x0e\x08\0?\0>"
+ /* */
+ "\0<\0q\0/\0>";
+ const XML_Char *expected = "\xe0\xb8\x84\xe0\xb8\x88: \n";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetProcessingInstructionHandler(parser, accumulate_pi_characters);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test that comments can be picked up and translated */
+static void XMLCALL
+accumulate_comment(void *userData,
+ const XML_Char *data)
+{
+ CharData *storage = (CharData *)userData;
+
+ CharData_AppendXMLChars(storage, data, -1);
+}
+
+START_TEST(test_utf16_be_comment)
+{
+ const char text[] =
+ /* */
+ "\0<\0!\0-\0-\0 \0C\0o\0m\0m\0e\0n\0t\0 \0A\0 \0-\0-\0>\0\n"
+ /* */
+ "\0<\0d\0o\0c\0/\0>";
+ const XML_Char *expected = " Comment A ";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetCommentHandler(parser, accumulate_comment);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_utf16_le_comment)
+{
+ const char text[] =
+ /* */
+ "<\0!\0-\0-\0 \0C\0o\0m\0m\0e\0n\0t\0 \0B\0 \0-\0-\0>\0\n\0"
+ /* */
+ "<\0d\0o\0c\0/\0>\0";
+ const XML_Char *expected = " Comment B ";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetCommentHandler(parser, accumulate_comment);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test that the unknown encoding handler with map entries that expect
+ * conversion but no conversion function is faulted
+ */
+static int XMLCALL
+failing_converter(void *UNUSED_P(data), const char *UNUSED_P(s))
+{
+ /* Always claim to have failed */
+ return -1;
+}
+
+static int XMLCALL
+prefix_converter(void *UNUSED_P(data), const char *s)
+{
+ /* If the first byte is 0xff, raise an error */
+ if (s[0] == -1)
+ return -1;
+ /* Just add the low bits of the first byte to the second */
+ return (s[1] + (s[0] & 0x7f)) & 0x01ff;
+}
+
+static int XMLCALL
+MiscEncodingHandler(void *data,
+ const XML_Char *encoding,
+ XML_Encoding *info)
+{
+ int i;
+ int high_map = -2; /* Assume a 2-byte sequence */
+
+ if (!strcmp(encoding, "invalid-9") ||
+ !strcmp(encoding, "ascii-like") ||
+ !strcmp(encoding, "invalid-len") ||
+ !strcmp(encoding, "invalid-a") ||
+ !strcmp(encoding, "invalid-surrogate") ||
+ !strcmp(encoding, "invalid-high"))
+ high_map = -1;
+
+ for (i = 0; i < 128; ++i)
+ info->map[i] = i;
+ for (; i < 256; ++i)
+ info->map[i] = high_map;
+
+ /* If required, put an invalid value in the ASCII entries */
+ if (!strcmp(encoding, "invalid-9"))
+ info->map[9] = 5;
+ /* If required, have a top-bit set character starts a 5-byte sequence */
+ if (!strcmp(encoding, "invalid-len"))
+ info->map[0x81] = -5;
+ /* If required, make a top-bit set character a valid ASCII character */
+ if (!strcmp(encoding, "invalid-a"))
+ info->map[0x82] = 'a';
+ /* If required, give a top-bit set character a forbidden value,
+ * what would otherwise be the first of a surrogate pair.
+ */
+ if (!strcmp(encoding, "invalid-surrogate"))
+ info->map[0x83] = 0xd801;
+ /* If required, give a top-bit set character too high a value */
+ if (!strcmp(encoding, "invalid-high"))
+ info->map[0x84] = 0x010101;
+
+ info->data = data;
+ info->release = NULL;
+ if (!strcmp(encoding, "failing-conv"))
+ info->convert = failing_converter;
+ else if (!strcmp(encoding, "prefix-conv"))
+ info->convert = prefix_converter;
+ else
+ info->convert = NULL;
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_missing_encoding_conversion_fn)
+{
+ const char *text =
+ "\n"
+ "\x81";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ /* MiscEncodingHandler sets up an encoding with every top-bit-set
+ * character introducing a two-byte sequence. For this, it
+ * requires a convert function. The above function call doesn't
+ * pass one through, so when BadEncodingHandler actually gets
+ * called it should supply an invalid encoding.
+ */
+ expect_failure(text, XML_ERROR_UNKNOWN_ENCODING,
+ "Encoding with missing convert() not faulted");
+}
+END_TEST
+
+START_TEST(test_failing_encoding_conversion_fn)
+{
+ const char *text =
+ "\n"
+ "\x81";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ /* BadEncodingHandler sets up an encoding with every top-bit-set
+ * character introducing a two-byte sequence. For this, it
+ * requires a convert function. The above function call passes
+ * one that insists all possible sequences are invalid anyway.
+ */
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Encoding with failing convert() not faulted");
+}
+END_TEST
+
+/* Test unknown encoding conversions */
+START_TEST(test_unknown_encoding_success)
+{
+ const char *text =
+ "\n"
+ /* Equivalent to Hello, world */
+ "<\x81\x64\x80oc>Hello, world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ run_character_check(text, "Hello, world");
+}
+END_TEST
+
+/* Test bad name character in unknown encoding */
+START_TEST(test_unknown_encoding_bad_name)
+{
+ const char *text =
+ "\n"
+ "<\xff\x64oc>Hello, world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Bad name start in unknown encoding not faulted");
+}
+END_TEST
+
+/* Test bad mid-name character in unknown encoding */
+START_TEST(test_unknown_encoding_bad_name_2)
+{
+ const char *text =
+ "\n"
+ "Hello, world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Bad name in unknown encoding not faulted");
+}
+END_TEST
+
+/* Test element name that is long enough to fill the conversion buffer
+ * in an unknown encoding, finishing with an encoded character.
+ */
+START_TEST(test_unknown_encoding_long_name_1)
+{
+ const char *text =
+ "\n"
+ ""
+ "Hi"
+ "";
+ const XML_Char *expected = "abcdefghabcdefghabcdefghijklmnop";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ XML_SetStartElementHandler(parser, record_element_start_handler);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test element name that is long enough to fill the conversion buffer
+ * in an unknown encoding, finishing with an simple character.
+ */
+START_TEST(test_unknown_encoding_long_name_2)
+{
+ const char *text =
+ "\n"
+ ""
+ "Hi"
+ "";
+ const XML_Char *expected = "abcdefghabcdefghabcdefghijklmnop";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ XML_SetStartElementHandler(parser, record_element_start_handler);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_invalid_unknown_encoding)
+{
+ const char *text =
+ "\n"
+ "Hello world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_UNKNOWN_ENCODING,
+ "Invalid unknown encoding not faulted");
+}
+END_TEST
+
+START_TEST(test_unknown_ascii_encoding_ok)
+{
+ const char *text =
+ "\n"
+ "Hello, world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ run_character_check(text, "Hello, world");
+}
+END_TEST
+
+START_TEST(test_unknown_ascii_encoding_fail)
+{
+ const char *text =
+ "\n"
+ "Hello, \x80 world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Invalid character not faulted");
+}
+END_TEST
+
+START_TEST(test_unknown_encoding_invalid_length)
+{
+ const char *text =
+ "\n"
+ "Hello, world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_UNKNOWN_ENCODING,
+ "Invalid unknown encoding not faulted");
+}
+END_TEST
+
+START_TEST(test_unknown_encoding_invalid_topbit)
+{
+ const char *text =
+ "\n"
+ "Hello, world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_UNKNOWN_ENCODING,
+ "Invalid unknown encoding not faulted");
+}
+END_TEST
+
+START_TEST(test_unknown_encoding_invalid_surrogate)
+{
+ const char *text =
+ "\n"
+ "Hello, \x82 world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Invalid unknown encoding not faulted");
+}
+END_TEST
+
+START_TEST(test_unknown_encoding_invalid_high)
+{
+ const char *text =
+ "\n"
+ "Hello, world";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_UNKNOWN_ENCODING,
+ "Invalid unknown encoding not faulted");
+}
+END_TEST
+
+START_TEST(test_unknown_encoding_invalid_attr_value)
+{
+ const char *text =
+ "\n"
+ "";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Invalid attribute valid not faulted");
+}
+END_TEST
+
+/* Test an external entity parser set to use latin-1 detects UTF-16
+ * BOMs correctly.
+ */
+enum ee_parse_flags {
+ EE_PARSE_NONE = 0x00,
+ EE_PARSE_FULL_BUFFER = 0x01
+};
+
+typedef struct ExtTest2 {
+ const char *parse_text;
+ int parse_len;
+ const char *encoding;
+ CharData *storage;
+ enum ee_parse_flags flags;
+} ExtTest2;
+
+static int XMLCALL
+external_entity_loader2(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId))
+{
+ ExtTest2 *test_data = (ExtTest2 *)XML_GetUserData(parser);
+ XML_Parser extparser;
+
+ extparser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (extparser == NULL)
+ fail("Coulr not create external entity parser");
+ if (test_data->encoding != NULL) {
+ if (!XML_SetEncoding(extparser, test_data->encoding))
+ fail("XML_SetEncoding() ignored for external entity");
+ }
+ if (test_data->flags & EE_PARSE_FULL_BUFFER) {
+ if (XML_Parse(extparser,
+ test_data->parse_text,
+ test_data->parse_len,
+ XML_TRUE) == XML_STATUS_ERROR) {
+ xml_failure(extparser);
+ }
+ }
+ else if (_XML_Parse_SINGLE_BYTES(extparser,
+ test_data->parse_text,
+ test_data->parse_len,
+ XML_TRUE) == XML_STATUS_ERROR) {
+ xml_failure(extparser);
+ }
+
+ XML_ParserFree(extparser);
+ return XML_STATUS_OK;
+}
+
+/* Test that UTF-16 BOM does not select UTF-16 given explicit encoding */
+static void XMLCALL
+ext2_accumulate_characters(void *userData, const XML_Char *s, int len)
+{
+ ExtTest2 *test_data = (ExtTest2 *)userData;
+ accumulate_characters(test_data->storage, s, len);
+}
+
+START_TEST(test_ext_entity_latin1_utf16le_bom)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&en;";
+ ExtTest2 test_data = {
+ /* If UTF-16, 0xfeff is the BOM and 0x204c is black left bullet */
+ /* If Latin-1, 0xff = Y-diaeresis, 0xfe = lowercase thorn,
+ * 0x4c = L and 0x20 is a space
+ */
+ "\xff\xfe\x4c\x20",
+ 4,
+ "iso-8859-1",
+ NULL,
+ EE_PARSE_NONE
+ };
+ /* In UTF-8, y-diaeresis is 0xc3 0xbf, lowercase thorn is 0xc3 0xbe */
+ const XML_Char *expected = "\xc3\xbf\xc3\xbeL ";
+ CharData storage;
+
+
+ CharData_Init(&storage);
+ test_data.storage = &storage;
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader2);
+ XML_SetUserData(parser, &test_data);
+ XML_SetCharacterDataHandler(parser, ext2_accumulate_characters);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_ext_entity_latin1_utf16be_bom)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&en;";
+ ExtTest2 test_data = {
+ /* If UTF-16, 0xfeff is the BOM and 0x204c is black left bullet */
+ /* If Latin-1, 0xff = Y-diaeresis, 0xfe = lowercase thorn,
+ * 0x4c = L and 0x20 is a space
+ */
+ "\xfe\xff\x20\x4c",
+ 4,
+ "iso-8859-1",
+ NULL,
+ EE_PARSE_NONE
+ };
+ /* In UTF-8, y-diaeresis is 0xc3 0xbf, lowercase thorn is 0xc3 0xbe */
+ const XML_Char *expected = "\xc3\xbe\xc3\xbf L";
+ CharData storage;
+
+
+ CharData_Init(&storage);
+ test_data.storage = &storage;
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader2);
+ XML_SetUserData(parser, &test_data);
+ XML_SetCharacterDataHandler(parser, ext2_accumulate_characters);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+
+/* Parsing the full buffer rather than a byte at a time makes a
+ * difference to the encoding scanning code, so repeat the above tests
+ * without breaking them down by byte.
+ */
+START_TEST(test_ext_entity_latin1_utf16le_bom2)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&en;";
+ ExtTest2 test_data = {
+ /* If UTF-16, 0xfeff is the BOM and 0x204c is black left bullet */
+ /* If Latin-1, 0xff = Y-diaeresis, 0xfe = lowercase thorn,
+ * 0x4c = L and 0x20 is a space
+ */
+ "\xff\xfe\x4c\x20",
+ 4,
+ "iso-8859-1",
+ NULL,
+ EE_PARSE_FULL_BUFFER
+ };
+ /* In UTF-8, y-diaeresis is 0xc3 0xbf, lowercase thorn is 0xc3 0xbe */
+ const XML_Char *expected = "\xc3\xbf\xc3\xbeL ";
+ CharData storage;
+
+
+ CharData_Init(&storage);
+ test_data.storage = &storage;
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader2);
+ XML_SetUserData(parser, &test_data);
+ XML_SetCharacterDataHandler(parser, ext2_accumulate_characters);
+ if (XML_Parse(parser, text, strlen(text), XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_ext_entity_latin1_utf16be_bom2)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&en;";
+ ExtTest2 test_data = {
+ /* If UTF-16, 0xfeff is the BOM and 0x204c is black left bullet */
+ /* If Latin-1, 0xff = Y-diaeresis, 0xfe = lowercase thorn,
+ * 0x4c = L and 0x20 is a space
+ */
+ "\xfe\xff\x20\x4c",
+ 4,
+ "iso-8859-1",
+ NULL,
+ EE_PARSE_FULL_BUFFER
+ };
+ /* In UTF-8, y-diaeresis is 0xc3 0xbf, lowercase thorn is 0xc3 0xbe */
+ const XML_Char *expected = "\xc3\xbe\xc3\xbf L";
+ CharData storage;
+
+
+ CharData_Init(&storage);
+ test_data.storage = &storage;
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader2);
+ XML_SetUserData(parser, &test_data);
+ XML_SetCharacterDataHandler(parser, ext2_accumulate_characters);
+ if (XML_Parse(parser, text, strlen(text), XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test little-endian UTF-16 given an explicit big-endian encoding */
+START_TEST(test_ext_entity_utf16_be)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&en;";
+ ExtTest2 test_data = {
+ "<\0e\0/\0>\0",
+ 8,
+ "utf-16be",
+ NULL,
+ EE_PARSE_NONE
+ };
+ const XML_Char *expected =
+ "\xe3\xb0\x80" /* U+3C00 */
+ "\xe6\x94\x80" /* U+6A00 */
+ "\xe2\xbc\x80" /* U+2F00 */
+ "\xe3\xb8\x80"; /* U+3E00 */
+ CharData storage;
+
+ CharData_Init(&storage);
+ test_data.storage = &storage;
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader2);
+ XML_SetUserData(parser, &test_data);
+ XML_SetCharacterDataHandler(parser, ext2_accumulate_characters);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test big-endian UTF-16 given an explicit little-endian encoding */
+START_TEST(test_ext_entity_utf16_le)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&en;";
+ ExtTest2 test_data = {
+ "\0<\0e\0/\0>",
+ 8,
+ "utf-16le",
+ NULL,
+ EE_PARSE_NONE
+ };
+ const XML_Char *expected =
+ "\xe3\xb0\x80" /* U+3C00 */
+ "\xe6\x94\x80" /* U+6A00 */
+ "\xe2\xbc\x80" /* U+2F00 */
+ "\xe3\xb8\x80"; /* U+3E00 */
+ CharData storage;
+
+ CharData_Init(&storage);
+ test_data.storage = &storage;
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader2);
+ XML_SetUserData(parser, &test_data);
+ XML_SetCharacterDataHandler(parser, ext2_accumulate_characters);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test little-endian UTF-16 given no explicit encoding.
+ * The existing default encoding (UTF-8) is assumed to hold without a
+ * BOM to contradict it, so the entity value will in fact provoke an
+ * error because 0x00 is not a valid XML character. We parse the
+ * whole buffer in one go rather than feeding it in byte by byte to
+ * exercise different code paths in the initial scanning routines.
+ */
+typedef struct ExtFaults2 {
+ const char *parse_text;
+ int parse_len;
+ const char *fail_text;
+ const char *encoding;
+ enum XML_Error error;
+} ExtFaults2;
+
+static int XMLCALL
+external_entity_faulter2(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId))
+{
+ ExtFaults2 *test_data = (ExtFaults2 *)XML_GetUserData(parser);
+ XML_Parser extparser;
+
+ extparser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (extparser == NULL)
+ fail("Could not create external entity parser");
+ if (test_data->encoding != NULL) {
+ if (!XML_SetEncoding(extparser, test_data->encoding))
+ fail("XML_SetEncoding() ignored for external entity");
+ }
+ if (XML_Parse(extparser,
+ test_data->parse_text,
+ test_data->parse_len,
+ XML_TRUE) != XML_STATUS_ERROR)
+ fail(test_data->fail_text);
+ if (XML_GetErrorCode(extparser) != test_data->error)
+ xml_failure(extparser);
+
+ XML_ParserFree(extparser);
+ return XML_STATUS_ERROR;
+}
+
+START_TEST(test_ext_entity_utf16_unknown)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&en;";
+ ExtFaults2 test_data = {
+ "a\0b\0c\0",
+ 6,
+ "Invalid character in entity not faulted",
+ NULL,
+ XML_ERROR_INVALID_TOKEN
+ };
+
+ XML_SetExternalEntityRefHandler(parser, external_entity_faulter2);
+ XML_SetUserData(parser, &test_data);
+ expect_failure(text, XML_ERROR_EXTERNAL_ENTITY_HANDLING,
+ "Invalid character should not have been accepted");
+}
+END_TEST
+
+/* Test not-quite-UTF-8 BOM (0xEF 0xBB 0xBF) */
+START_TEST(test_ext_entity_utf8_non_bom)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&en;";
+ ExtTest2 test_data = {
+ "\xef\xbb\x80", /* Arabic letter DAD medial form, U+FEC0 */
+ 3,
+ NULL,
+ NULL,
+ EE_PARSE_NONE
+ };
+ const XML_Char *expected = "\xef\xbb\x80";
+ CharData storage;
+
+ CharData_Init(&storage);
+ test_data.storage = &storage;
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader2);
+ XML_SetUserData(parser, &test_data);
+ XML_SetCharacterDataHandler(parser, ext2_accumulate_characters);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test that UTF-8 in a CDATA section is correctly passed through */
+START_TEST(test_utf8_in_cdata_section)
+{
+ const char *text = "";
+ const XML_Char *expected = "one \xc3\xa9 two";
+
+ run_character_check(text, expected);
+}
+END_TEST
+
+/* Test that little-endian UTF-16 in a CDATA section is handled */
+START_TEST(test_utf8_in_cdata_section_2)
+{
+ const char *text = "";
+ const XML_Char *expected = "\xc3\xa9]\xc3\xa9two";
+
+ run_character_check(text, expected);
+}
+END_TEST
+
+/* Test trailing spaces in elements are accepted */
+static void XMLCALL
+record_element_end_handler(void *userData,
+ const XML_Char *name)
+{
+ CharData *storage = (CharData *)userData;
+
+ CharData_AppendXMLChars(storage, "/", 1);
+ CharData_AppendXMLChars(storage, name, -1);
+}
+
+START_TEST(test_trailing_spaces_in_elements)
+{
+ const char *text = "Hi";
+ const XML_Char *expected = "doc/doc";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetElementHandler(parser, record_element_start_handler,
+ record_element_end_handler);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_utf16_attribute)
+{
+ const char text[] =
+ /*
+ * where {KHO KHWAI} = U+0E04 = 0xe0 0xb8 0x84 in UTF-8
+ * and {CHO CHAN} = U+0E08 = 0xe0 0xb8 0x88 in UTF-8
+ */
+ "<\0d\0 \0\x04\x0e\x08\x0e=\0'\0a\0'\0/\0>\0";
+ const XML_Char *expected = "a";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetStartElementHandler(parser, accumulate_attribute);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_utf16_second_attr)
+{
+ /*
+ * where {KHO KHWAI} = U+0E04 = 0xe0 0xb8 0x84 in UTF-8
+ * and {CHO CHAN} = U+0E08 = 0xe0 0xb8 0x88 in UTF-8
+ */
+ const char text[] =
+ "<\0d\0 \0a\0=\0'\0\x31\0'\0 \0"
+ "\x04\x0e\x08\x0e=\0'\0\x32\0'\0/\0>\0";
+ const XML_Char *expected = "1";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetStartElementHandler(parser, accumulate_attribute);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_attr_after_solidus)
+{
+ const char *text = "";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Misplaced / not faulted");
+}
+END_TEST
+
+static void XMLCALL
+accumulate_entity_decl(void *userData,
+ const XML_Char *entityName,
+ int UNUSED_P(is_parameter_entity),
+ const XML_Char *value,
+ int value_length,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId),
+ const XML_Char *UNUSED_P(notationName))
+{
+ CharData *storage = (CharData *)userData;
+
+ CharData_AppendXMLChars(storage, entityName, -1);
+ CharData_AppendXMLChars(storage, "=", 1);
+ CharData_AppendXMLChars(storage, value, value_length);
+ CharData_AppendXMLChars(storage, "\n", 1);
+}
+
+
+START_TEST(test_utf16_pe)
+{
+ /* '>
+ * %{KHO KHWAI}{CHO CHAN};
+ * ]>
+ *
+ *
+ * where {KHO KHWAI} = U+0E04 = 0xe0 0xb8 0x84 in UTF-8
+ * and {CHO CHAN} = U+0E08 = 0xe0 0xb8 0x88 in UTF-8
+ */
+ const char text[] =
+ "\0<\0!\0D\0O\0C\0T\0Y\0P\0E\0 \0d\0o\0c\0 \0[\0\n"
+ "\0<\0!\0E\0N\0T\0I\0T\0Y\0 \0%\0 \x0e\x04\x0e\x08\0 "
+ "\0'\0<\0!\0E\0L\0E\0M\0E\0N\0T\0 "
+ "\0d\0o\0c\0 \0(\0#\0P\0C\0D\0A\0T\0A\0)\0>\0'\0>\0\n"
+ "\0%\x0e\x04\x0e\x08\0;\0\n"
+ "\0]\0>\0\n"
+ "\0<\0d\0o\0c\0>\0<\0/\0d\0o\0c\0>";
+ const XML_Char *expected =
+ "\xe0\xb8\x84\xe0\xb8\x88=\n";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetUserData(parser, &storage);
+ XML_SetEntityDeclHandler(parser, accumulate_entity_decl);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test that duff attribute description keywords are rejected */
+START_TEST(test_bad_attr_desc_keyword)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Bad keyword !IMPLIED not faulted");
+}
+END_TEST
+
+/* Test that an invalid attribute description keyword consisting of
+ * UTF-16 characters with their top bytes non-zero are correctly
+ * faulted
+ */
+START_TEST(test_bad_attr_desc_keyword_utf16)
+{
+ /*
+ * ]>
+ *
+ * where {KHO KHWAI} = U+0E04 = 0xe0 0xb8 0x84 in UTF-8
+ * and {CHO CHAN} = U+0E08 = 0xe0 0xb8 0x88 in UTF-8
+ */
+ const char text[] =
+ "\0<\0!\0D\0O\0C\0T\0Y\0P\0E\0 \0d\0 \0[\0\n"
+ "\0<\0!\0A\0T\0T\0L\0I\0S\0T\0 \0d\0 \0a\0 \0C\0D\0A\0T\0A\0 "
+ "\0#\x0e\x04\x0e\x08\0>\0\n"
+ "\0]\0>\0<\0d\0/\0>";
+
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) != XML_STATUS_ERROR)
+ fail("Invalid UTF16 attribute keyword not faulted");
+ if (XML_GetErrorCode(parser) != XML_ERROR_SYNTAX)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test that invalid syntax in a is rejected. Do this
+ * using prefix-encoding (see above) to trigger specific code paths
+ */
+START_TEST(test_bad_doctype)
+{
+ const char *text =
+ "\n"
+ "";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "Invalid bytes in DOCTYPE not faulted");
+}
+END_TEST
+
+START_TEST(test_bad_doctype_utf16)
+{
+ const char text[] =
+ /*
+ *
+ * U+06F2 = EXTENDED ARABIC-INDIC DIGIT TWO, a valid number
+ * (name character) but not a valid letter (name start character)
+ */
+ "\0<\0!\0D\0O\0C\0T\0Y\0P\0E\0 \0d\0o\0c\0 \0[\0 "
+ "\x06\xf2"
+ "\0 \0]\0>\0<\0d\0o\0c\0/\0>";
+
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) != XML_STATUS_ERROR)
+ fail("Invalid bytes in DOCTYPE not faulted");
+ if (XML_GetErrorCode(parser) != XML_ERROR_SYNTAX)
+ xml_failure(parser);
+}
+END_TEST
+
+START_TEST(test_bad_doctype_plus)
+{
+ const char *text =
+ " ]>\n"
+ "<1+>&foo;";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "'+' in document name not faulted");
+}
+END_TEST
+
+START_TEST(test_bad_doctype_star)
+{
+ const char *text =
+ " ]>\n"
+ "<1*>&foo;";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "'*' in document name not faulted");
+}
+END_TEST
+
+START_TEST(test_bad_doctype_query)
+{
+ const char *text =
+ " ]>\n"
+ "<1?>&foo;";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "'?' in document name not faulted");
+}
+END_TEST
+
+START_TEST(test_unknown_encoding_bad_ignore)
+{
+ const char *text =
+ ""
+ ""
+ "&entity;";
+ ExtFaults fault = {
+ "]]>",
+ "Invalid character not faulted",
+ "prefix-conv",
+ XML_ERROR_INVALID_TOKEN
+ };
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_faulter);
+ XML_SetUserData(parser, &fault);
+ expect_failure(text, XML_ERROR_EXTERNAL_ENTITY_HANDLING,
+ "Bad IGNORE section with unknown encoding not failed");
+}
+END_TEST
+
+START_TEST(test_entity_in_utf16_be_attr)
+{
+ const char text[] =
+ /* */
+ "\0<\0e\0 \0a\0=\0'\0&\0#\0\x32\0\x32\0\x38\0;\0 "
+ "\0&\0#\0x\0\x30\0\x30\0E\0\x34\0;\0'\0>\0<\0/\0e\0>";
+ const XML_Char *expected = "\xc3\xa4 \xc3\xa4";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetUserData(parser, &storage);
+ XML_SetStartElementHandler(parser, accumulate_attribute);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_entity_in_utf16_le_attr)
+{
+ const char text[] =
+ /* */
+ "<\0e\0 \0a\0=\0'\0&\0#\0\x32\0\x32\0\x38\0;\0 \0"
+ "&\0#\0x\0\x30\0\x30\0E\0\x34\0;\0'\0>\0<\0/\0e\0>\0";
+ const XML_Char *expected = "\xc3\xa4 \xc3\xa4";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetUserData(parser, &storage);
+ XML_SetStartElementHandler(parser, accumulate_attribute);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_entity_public_utf16_be)
+{
+ const char text[] =
+ /* */
+ "\0<\0!\0E\0N\0T\0I\0T\0Y\0 \0%\0 \0e\0 \0P\0U\0B\0L\0I\0C\0 "
+ "\0'\0f\0o\0o\0'\0 \0'\0b\0a\0r\0.\0e\0n\0t\0'\0>\0\n"
+ /* %e; */
+ "\0%\0e\0;\0\n"
+ /* ]> */
+ "\0]\0>\0\n"
+ /* &j; */
+ "\0<\0d\0>\0&\0j\0;\0<\0/\0d\0>";
+ ExtTest2 test_data = {
+ /* */
+ "\0<\0!\0E\0N\0T\0I\0T\0Y\0 \0j\0 \0'\0b\0a\0z\0'\0>",
+ 34,
+ NULL,
+ NULL,
+ EE_PARSE_NONE
+ };
+ const XML_Char *expected = "baz";
+ CharData storage;
+
+ CharData_Init(&storage);
+ test_data.storage = &storage;
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader2);
+ XML_SetUserData(parser, &test_data);
+ XML_SetCharacterDataHandler(parser, ext2_accumulate_characters);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_entity_public_utf16_le)
+{
+ const char text[] =
+ /* */
+ "<\0!\0E\0N\0T\0I\0T\0Y\0 \0%\0 \0e\0 \0P\0U\0B\0L\0I\0C\0 \0"
+ "'\0f\0o\0o\0'\0 \0'\0b\0a\0r\0.\0e\0n\0t\0'\0>\0\n\0"
+ /* %e; */
+ "%\0e\0;\0\n\0"
+ /* ]> */
+ "]\0>\0\n\0"
+ /* &j; */
+ "<\0d\0>\0&\0j\0;\0<\0/\0d\0>\0";
+ ExtTest2 test_data = {
+ /* */
+ "<\0!\0E\0N\0T\0I\0T\0Y\0 \0j\0 \0'\0b\0a\0z\0'\0>\0",
+ 34,
+ NULL,
+ NULL,
+ EE_PARSE_NONE
+ };
+ const XML_Char *expected = "baz";
+ CharData storage;
+
+ CharData_Init(&storage);
+ test_data.storage = &storage;
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_loader2);
+ XML_SetUserData(parser, &test_data);
+ XML_SetCharacterDataHandler(parser, ext2_accumulate_characters);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+/* Test that a doctype with neither an internal nor external subset is
+ * faulted
+ */
+START_TEST(test_short_doctype)
+{
+ const char *text = "";
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "DOCTYPE without subset not rejected");
+}
+END_TEST
+
+START_TEST(test_short_doctype_2)
+{
+ const char *text = "";
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "DOCTYPE without Public ID not rejected");
+}
+END_TEST
+
+START_TEST(test_short_doctype_3)
+{
+ const char *text = "";
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "DOCTYPE without System ID not rejected");
+}
+END_TEST
+
+START_TEST(test_long_doctype)
+{
+ const char *text = "";
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "DOCTYPE with extra ID not rejected");
+}
+END_TEST
+
+START_TEST(test_bad_entity)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "";
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "ENTITY without Public ID is not rejected");
+}
+END_TEST
+
+/* Test unquoted value is faulted */
+START_TEST(test_bad_entity_2)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "";
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "ENTITY without Public ID is not rejected");
+}
+END_TEST
+
+START_TEST(test_bad_entity_3)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "";
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "Parameter ENTITY without Public ID is not rejected");
+}
+END_TEST
+
+START_TEST(test_bad_entity_4)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "";
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "Parameter ENTITY without Public ID is not rejected");
+}
+END_TEST
+
+START_TEST(test_bad_notation)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "";
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "Notation without System ID is not rejected");
+}
+END_TEST
/*
* Namespaces tests.
@@ -3658,9 +6883,6 @@ START_TEST(test_return_ns_triplet)
dummy_end_namespace_decl_handler);
triplet_start_flag = XML_FALSE;
triplet_end_flag = XML_FALSE;
- XML_SetNamespaceDeclHandler(parser,
- dummy_start_namespace_decl_handler,
- dummy_end_namespace_decl_handler);
dummy_handler_flags = 0;
if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
XML_FALSE) == XML_STATUS_ERROR)
@@ -4128,6 +7350,234 @@ START_TEST(test_ns_reserved_attributes_2)
}
END_TEST
+/* Test string pool handling of namespace names of 2048 characters */
+/* Exercises a particular string pool growth path */
+START_TEST(test_ns_extremely_long_prefix)
+{
+ const char *text =
+ ""
+ "";
+
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_FALSE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+}
+END_TEST
+
+/* Test unknown encoding handlers in namespace setup */
+START_TEST(test_ns_unknown_encoding_success)
+{
+ const char *text =
+ "\n"
+ "Hi";
+
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ run_character_check(text, "Hi");
+}
+END_TEST
+
+/* Test that too many colons are rejected */
+START_TEST(test_ns_double_colon)
+{
+ const char *text =
+ "";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Double colon in attribute name not faulted");
+}
+END_TEST
+
+START_TEST(test_ns_double_colon_element)
+{
+ const char *text =
+ "";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Double colon in element name not faulted");
+}
+END_TEST
+
+/* Test that non-name characters after a colon are rejected */
+START_TEST(test_ns_bad_attr_leafname)
+{
+ const char *text =
+ "";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Invalid character in leafname not faulted");
+}
+END_TEST
+
+START_TEST(test_ns_bad_element_leafname)
+{
+ const char *text =
+ "";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Invalid character in element leafname not faulted");
+}
+END_TEST
+
+/* Test high-byte-set UTF-16 characters are valid in a leafname */
+START_TEST(test_ns_utf16_leafname)
+{
+ const char text[] =
+ /*
+ * where {KHO KHWAI} = U+0E04 = 0xe0 0xb8 0x84 in UTF-8
+ */
+ "<\0n\0:\0e\0 \0x\0m\0l\0n\0s\0:\0n\0=\0'\0U\0R\0I\0'\0 \0"
+ "n\0:\0\x04\x0e=\0'\0a\0'\0 \0/\0>\0";
+ const XML_Char *expected = "a";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetStartElementHandler(parser, accumulate_attribute);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_ns_utf16_element_leafname)
+{
+ const char text[] =
+ /*
+ * where {KHO KHWAI} = U+0E04 = 0xe0 0xb8 0x84 in UTF-8
+ */
+ "\0<\0n\0:\x0e\x04\0 \0x\0m\0l\0n\0s\0:\0n\0=\0'\0U\0R\0I\0'\0/\0>";
+ const XML_Char *expected = "URI \xe0\xb8\x84";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetStartElementHandler(parser, start_element_event_handler);
+ XML_SetUserData(parser, &storage);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_ns_utf16_doctype)
+{
+ const char text[] =
+ /* ]>\n
+ * where {KHO KHWAI} = U+0E04 = 0xe0 0xb8 0x84 in UTF-8
+ */
+ "\0<\0!\0D\0O\0C\0T\0Y\0P\0E\0 \0f\0o\0o\0:\x0e\x04\0 "
+ "\0[\0 \0<\0!\0E\0N\0T\0I\0T\0Y\0 \0b\0a\0r\0 \0'\0b\0a\0z\0'\0>\0 "
+ "\0]\0>\0\n"
+ /* &bar; */
+ "\0<\0f\0o\0o\0:\x0e\x04\0 "
+ "\0x\0m\0l\0n\0s\0:\0f\0o\0o\0=\0'\0U\0R\0I\0'\0>"
+ "\0&\0b\0a\0r\0;"
+ "\0<\0/\0f\0o\0o\0:\x0e\x04\0>";
+ const XML_Char *expected = "URI \xe0\xb8\x84";
+ CharData storage;
+
+ CharData_Init(&storage);
+ XML_SetUserData(parser, &storage);
+ XML_SetStartElementHandler(parser, start_element_event_handler);
+ XML_SetUnknownEncodingHandler(parser, MiscEncodingHandler, NULL);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
+START_TEST(test_ns_invalid_doctype)
+{
+ const char *text =
+ "\n"
+ "&bar;";
+
+ expect_failure(text, XML_ERROR_INVALID_TOKEN,
+ "Invalid character in document local name not faulted");
+}
+END_TEST
+
+START_TEST(test_ns_double_colon_doctype)
+{
+ const char *text =
+ "\n"
+ "&bar;";
+
+ expect_failure(text, XML_ERROR_SYNTAX,
+ "Double colon in document name not faulted");
+}
+END_TEST
+
/* Control variable; the number of times duff_allocator() will successfully allocate */
#define ALLOC_ALWAYS_SUCCEED (-1)
#define REALLOC_ALWAYS_SUCCEED (-1)
@@ -4216,38 +7666,74 @@ START_TEST(test_misc_error_string)
END_TEST
/* Test the version information is consistent */
+
+/* Since we are working in XML_LChars (potentially 16-bits), we
+ * can't use the standard C library functions for character
+ * manipulation and have to roll our own.
+ */
+static int
+parse_version(const XML_LChar *version_text,
+ XML_Expat_Version *version_struct)
+{
+ while (*version_text != 0x00) {
+ if (*version_text >= ASCII_0 && *version_text <= ASCII_9)
+ break;
+ version_text++;
+ }
+ if (*version_text == 0x00)
+ return XML_FALSE;
+
+ /* version_struct->major = strtoul(version_text, 10, &version_text) */
+ version_struct->major = 0;
+ while (*version_text >= ASCII_0 && *version_text <= ASCII_9) {
+ version_struct->major =
+ 10 * version_struct->major + (*version_text++ - ASCII_0);
+ }
+ if (*version_text++ != ASCII_PERIOD)
+ return XML_FALSE;
+
+ /* Now for the minor version number */
+ version_struct->minor = 0;
+ while (*version_text >= ASCII_0 && *version_text <= ASCII_9) {
+ version_struct->minor =
+ 10 * version_struct->minor + (*version_text++ - ASCII_0);
+ }
+ if (*version_text++ != ASCII_PERIOD)
+ return XML_FALSE;
+
+ /* Finally the micro version number */
+ version_struct->micro = 0;
+ while (*version_text >= ASCII_0 && *version_text <= ASCII_9) {
+ version_struct->micro =
+ 10 * version_struct->micro + (*version_text++ - ASCII_0);
+ }
+ if (*version_text != 0x00)
+ return XML_FALSE;
+ return XML_TRUE;
+}
+
+static int
+versions_equal(const XML_Expat_Version *first,
+ const XML_Expat_Version *second)
+{
+ return (first->major == second->major &&
+ first->minor == second->minor &&
+ first->micro == second->micro);
+}
+
START_TEST(test_misc_version)
{
- XML_Expat_Version version_struct = XML_ExpatVersionInfo();
+ XML_Expat_Version read_version = XML_ExpatVersionInfo();
+ /* Silence compiler warning with the following assignment */
+ XML_Expat_Version parsed_version = { 0, 0, 0 };
const XML_LChar *version_text = XML_ExpatVersion();
- long value;
- const char *p;
- char *endp;
if (version_text == NULL)
fail("Could not obtain version text");
- for (p = version_text; *p != '\0'; p++)
- if (isdigit(*p))
- break;
- if (*p == '\0')
- fail("No numbers in version text");
- value = strtoul(p, &endp, 10);
- if (*endp != '.')
- fail("Major version conversion from text failed");
- if (value != version_struct.major)
- fail("Major version mismatch");
- p = endp + 1;
- value = strtoul(p, &endp, 10);
- if (*endp != '.')
- fail("Minor version conversion from text failed");
- if (value != version_struct.minor)
- fail("Minor version mismatch");
- p = endp + 1;
- value = strtoul(p, &endp, 10);
- if (*endp != '\0')
- fail("Micro version conversion from text failed");
- if (value != version_struct.micro)
- fail("Micro version mismatch");
+ if (!parse_version(version_text, &parsed_version))
+ fail("Unable to parse version text");
+ if (!versions_equal(&read_version, &parsed_version))
+ fail("Version mismatch");
}
END_TEST
@@ -4303,6 +7789,31 @@ START_TEST(test_misc_attribute_leak)
}
END_TEST
+/* Test parser created for UTF-16LE is successful */
+START_TEST(test_misc_utf16le)
+{
+ const char text[] =
+ /* Hi
*/
+ "<\0?\0x\0m\0l\0 \0"
+ "v\0e\0r\0s\0i\0o\0n\0=\0'\0\x31\0.\0\x30\0'\0?\0>\0"
+ "<\0q\0>\0H\0i\0<\0/\0q\0>\0";
+ const XML_Char *expected = "Hi";
+ CharData storage;
+
+ parser = XML_ParserCreate("UTF-16LE");
+ if (parser == NULL)
+ fail("Parser not created");
+
+ CharData_Init(&storage);
+ XML_SetUserData(parser, &storage);
+ XML_SetCharacterDataHandler(parser, accumulate_characters);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, sizeof(text)-1,
+ XML_TRUE) == XML_STATUS_ERROR)
+ xml_failure(parser);
+ CharData_CheckXMLChars(&storage, expected);
+}
+END_TEST
+
static void
alloc_setup(void)
@@ -4328,6 +7839,113 @@ alloc_teardown(void)
}
+/* Test the effects of allocation failures on xml declaration processing */
+START_TEST(test_alloc_parse_xdecl)
+{
+ const char *text =
+ "\n"
+ "Hello, world";
+ int i;
+ int repeat = 0;
+#define MAX_ALLOC_COUNT 10
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some (most) counts to defeat cached allocations */
+ if (i == 2 && repeat != 1) {
+ i--;
+ repeat++;
+ }
+ else if (i == 3) {
+ i -= 2;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetXmlDeclHandler(parser, dummy_xdecl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed with max allocations");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+/* As above, but with an encoding big enough to cause storing the
+ * version information to expand the string pool being used.
+ */
+static int XMLCALL
+long_encoding_handler(void *UNUSED_P(userData),
+ const XML_Char *UNUSED_P(encoding),
+ XML_Encoding *info)
+{
+ int i;
+
+ for (i = 0; i < 256; i++)
+ info->map[i] = i;
+ info->data = NULL;
+ info->convert = NULL;
+ info->release = NULL;
+ return XML_STATUS_OK;
+}
+
+START_TEST(test_alloc_parse_xdecl_2)
+{
+ const char *text =
+ ""
+ "Hello, world";
+ int i;
+ int repeat = 0;
+#define MAX_ALLOC_COUNT 10
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat counts to defeat cached allocations */
+ if (i == 4 && repeat == 3) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 3) ||
+ (i == 3 && repeat > 3)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetXmlDeclHandler(parser, dummy_xdecl_handler);
+ XML_SetUnknownEncodingHandler(parser, long_encoding_handler, NULL);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed with max allocations");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
/* Test the effects of allocation failures on a straightforward parse */
START_TEST(test_alloc_parse_pi)
{
@@ -4403,6 +8021,48 @@ START_TEST(test_alloc_parse_pi_2)
}
END_TEST
+START_TEST(test_alloc_parse_pi_3)
+{
+ const char *text =
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Every allocation bar the last is cached */
+ allocation_count = (i == 0) ? 0 : 1;
+ XML_SetProcessingInstructionHandler(parser, dummy_pi_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed with max allocations");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
START_TEST(test_alloc_parse_comment)
{
const char *text =
@@ -4530,16 +8190,6 @@ START_TEST(test_alloc_create_external_parser)
}
END_TEST
-static int XMLCALL
-external_entity_null_loader(XML_Parser UNUSED_P(parser),
- const XML_Char *UNUSED_P(context),
- const XML_Char *UNUSED_P(base),
- const XML_Char *UNUSED_P(systemId),
- const XML_Char *UNUSED_P(publicId))
-{
- return XML_STATUS_OK;
-}
-
/* More external parser memory allocation testing */
START_TEST(test_alloc_run_external_parser)
{
@@ -4750,7 +8400,7 @@ external_entity_alloc_set_encoding(XML_Parser parser,
const XML_Char *UNUSED_P(systemId),
const XML_Char *UNUSED_P(publicId))
{
- /* As for external_entity_loader_set_encoding() */
+ /* As for external_entity_loader() */
const char *text =
""
"\xC3\xA9";
@@ -5105,6 +8755,1649 @@ START_TEST(test_alloc_realloc_many_attributes)
}
END_TEST
+/* Test handling of a public entity with failing allocator */
+START_TEST(test_alloc_public_entity_value)
+{
+ const char *text =
+ "\n"
+ "\n";
+ char dtd_text[] =
+ "\n"
+ "\n"
+ "\n"
+ "%e1;\n";
+ int i;
+#define MAX_ALLOC_COUNT 25
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat certain counts to defeat cached allocations */
+ if ((i == 2 && repeat < 2) ||
+ (i == 3 && repeat == 2) ||
+ (i == 8 && repeat == 3) ||
+ (i == 9 && repeat == 4) ||
+ (i == 18 && repeat == 5) ||
+ (i == 19 && repeat == 6)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ dummy_handler_flags = 0;
+ XML_SetUserData(parser, dtd_text);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_public);
+ /* Provoke a particular code path */
+ XML_SetEntityDeclHandler(parser, dummy_entity_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocation");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed at max allocation count");
+ if (dummy_handler_flags != DUMMY_ENTITY_DECL_HANDLER_FLAG)
+ fail("Entity declaration handler not called");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_realloc_subst_public_entity_value)
+{
+ const char *text =
+ "\n"
+ "\n";
+ char dtd_text[] =
+ "\n"
+ "\n"
+ "%ThisIsAStupidlyLongParameterNameIntendedToTriggerPoolGrowth12345"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP;";
+ int i;
+#define MAX_REALLOC_COUNT 10
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ XML_SetUserData(parser, dtd_text);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_public);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing reallocation");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parsing failed at max reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_parse_public_doctype)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat certain counts to defeat cached allocations */
+ if (i == 4 && repeat == 5) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 3) ||
+ (i == 3 && repeat < 5)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ dummy_handler_flags = 0;
+ XML_SetDoctypeDeclHandler(parser,
+ dummy_start_doctype_decl_handler,
+ dummy_end_doctype_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+ if (dummy_handler_flags != (DUMMY_START_DOCTYPE_DECL_HANDLER_FLAG |
+ DUMMY_END_DOCTYPE_DECL_HANDLER_FLAG))
+ fail("Doctype handler functions not called");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_parse_public_doctype_long_name)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if (i == 4 && repeat == 6) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 3) ||
+ (i == 3 && repeat < 6)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetDoctypeDeclHandler(parser,
+ dummy_start_doctype_decl_handler,
+ dummy_end_doctype_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+static int XMLCALL
+external_entity_alloc(XML_Parser parser,
+ const XML_Char *context,
+ const XML_Char *UNUSED_P(base),
+ const XML_Char *UNUSED_P(systemId),
+ const XML_Char *UNUSED_P(publicId))
+{
+ const char *text = (const char *)XML_GetUserData(parser);
+ XML_Parser ext_parser;
+ int parse_res;
+
+ ext_parser = XML_ExternalEntityParserCreate(parser, context, NULL);
+ if (ext_parser == NULL)
+ return XML_STATUS_ERROR;
+ parse_res = _XML_Parse_SINGLE_BYTES(ext_parser, text, strlen(text),
+ XML_TRUE);
+ XML_ParserFree(ext_parser);
+ return parse_res;
+}
+
+/* Test foreign DTD handling */
+START_TEST(test_alloc_set_foreign_dtd)
+{
+ const char *text1 =
+ "\n"
+ "&entity;";
+ char text2[] = "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to deal with cached allocations */
+ if (i == 9 && repeat == 2) {
+ i -= 2;
+ repeat++;
+ }
+ else if (i == 2 && repeat < 2) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetUserData(parser, &text2);
+ XML_SetExternalEntityRefHandler(parser, external_entity_alloc);
+ if (XML_UseForeignDTD(parser, XML_TRUE) != XML_ERROR_NONE)
+ fail("Could not set foreign DTD");
+ if (_XML_Parse_SINGLE_BYTES(parser, text1, strlen(text1),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+/* Test based on ibm/valid/P32/ibm32v04.xml */
+START_TEST(test_alloc_attribute_enum_value)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "This is a \n \n\nyellow tiger";
+ char dtd_text[] =
+ "\n"
+ "\n"
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 20
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to allow for cached allocations */
+ if (i == 13 && repeat == 5) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 2) ||
+ (i == 3 && repeat == 2) ||
+ (i == 8 && repeat == 3) ||
+ (i == 9 && repeat == 4)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetExternalEntityRefHandler(parser, external_entity_alloc);
+ XML_SetUserData(parser, dtd_text);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ /* An attribute list handler provokes a different code path */
+ XML_SetAttlistDeclHandler(parser, dummy_attlist_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+/* Test attribute enums sufficient to overflow the string pool */
+START_TEST(test_alloc_realloc_attribute_enum_value)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "This is a yellow tiger";
+ /* We wish to define a collection of attribute enums that will
+ * cause the string pool storing them to have to expand. This
+ * means more than 1024 bytes, including the parentheses and
+ * separator bars.
+ */
+ char dtd_text[] =
+ "\n"
+ "";
+ int i;
+#define MAX_REALLOC_COUNT 10
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ XML_SetExternalEntityRefHandler(parser, external_entity_alloc);
+ XML_SetUserData(parser, dtd_text);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ /* An attribute list handler provokes a different code path */
+ XML_SetAttlistDeclHandler(parser, dummy_attlist_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+/* Test attribute enums in a #IMPLIED attribute forcing pool growth */
+START_TEST(test_alloc_realloc_implied_attribute)
+{
+ /* Forcing this particular code path is a balancing act. The
+ * addition of the closing parenthesis and terminal NUL must be
+ * what pushes the string of enums over the 1024-byte limit,
+ * otherwise a different code path will pick up the realloc.
+ */
+ const char *text =
+ "\n"
+ "\n"
+ "]>";
+ int i;
+#define MAX_REALLOC_COUNT 10
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ XML_SetAttlistDeclHandler(parser, dummy_attlist_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+/* Test attribute enums in a defaulted attribute forcing pool growth */
+START_TEST(test_alloc_realloc_default_attribute)
+{
+ /* Forcing this particular code path is a balancing act. The
+ * addition of the closing parenthesis and terminal NUL must be
+ * what pushes the string of enums over the 1024-byte limit,
+ * otherwise a different code path will pick up the realloc.
+ */
+ const char *text =
+ "\n"
+ "\n]>";
+ int i;
+#define MAX_REALLOC_COUNT 10
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ XML_SetAttlistDeclHandler(parser, dummy_attlist_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+/* Test long notation name with dodgy allocator */
+START_TEST(test_alloc_notation)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "\n"
+ "]>\n";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to allow for cached allocations */
+ if ((i == 2 && repeat < 4) || (i == 3)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ dummy_handler_flags = 0;
+ XML_SetNotationDeclHandler(parser, dummy_notation_decl_handler);
+ XML_SetEntityDeclHandler(parser, dummy_entity_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite allocation failures");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+ if (dummy_handler_flags != (DUMMY_ENTITY_DECL_HANDLER_FLAG |
+ DUMMY_NOTATION_DECL_HANDLER_FLAG))
+ fail("Entity declaration handler not called");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+/* Test public notation with dodgy allocator */
+START_TEST(test_alloc_public_notation)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "\n"
+ "]>\n";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to allow for cached allocations */
+ if ((i == 2 && repeat < 5) ||
+ (i == 3 && repeat == 5)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ dummy_handler_flags = 0;
+ XML_SetNotationDeclHandler(parser, dummy_notation_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite allocation failures");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+ if (dummy_handler_flags != DUMMY_NOTATION_DECL_HANDLER_FLAG)
+ fail("Notation handler not called");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+/* Test public notation with dodgy allocator */
+START_TEST(test_alloc_system_notation)
+{
+ const char *text =
+ "\n"
+ "\n"
+ "\n"
+ "]>\n";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to allow for cached allocations */
+ if ((i == 2 && repeat < 5) ||
+ (i == 3 && repeat == 5)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ dummy_handler_flags = 0;
+ XML_SetNotationDeclHandler(parser, dummy_notation_decl_handler);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite allocation failures");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+ if (dummy_handler_flags != DUMMY_NOTATION_DECL_HANDLER_FLAG)
+ fail("Notation handler not called");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_nested_groups)
+{
+ const char *text =
+ "\n"
+ ""
+ "]>\n"
+ "";
+ CharData storage;
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to allow for cached allocations */
+ if ((i == 2 && repeat < 3) || (i == 3 && repeat == 3)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ CharData_Init(&storage);
+ XML_SetElementDeclHandler(parser, dummy_element_decl_handler);
+ XML_SetStartElementHandler(parser, record_element_start_handler);
+ XML_SetUserData(parser, &storage);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+ CharData_CheckString(&storage, "doce");
+ if (dummy_handler_flags != DUMMY_ELEMENT_DECL_HANDLER_FLAG)
+ fail("Element handler not fired");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_realloc_nested_groups)
+{
+ const char *text =
+ "\n"
+ ""
+ "]>\n"
+ "";
+ CharData storage;
+ int i;
+#define MAX_REALLOC_COUNT 10
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ CharData_Init(&storage);
+ XML_SetElementDeclHandler(parser, dummy_element_decl_handler);
+ XML_SetStartElementHandler(parser, record_element_start_handler);
+ XML_SetUserData(parser, &storage);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+ CharData_CheckString(&storage, "doce");
+ if (dummy_handler_flags != DUMMY_ELEMENT_DECL_HANDLER_FLAG)
+ fail("Element handler not fired");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_large_group)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "\n"
+ "\n";
+ int i;
+#define MAX_ALLOC_COUNT 45
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if ((i == 2 && repeat < 3) ||
+ (i == 3 && repeat == 3) ||
+ (i == 37 && repeat == 4)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetElementDeclHandler(parser, dummy_element_decl_handler);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+ if (dummy_handler_flags != DUMMY_ELEMENT_DECL_HANDLER_FLAG)
+ fail("Element handler flag not raised");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_realloc_group_choice)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "\n"
+ "This is a foo\n"
+ "\n"
+ "\n";
+ int i;
+#define MAX_REALLOC_COUNT 10
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ XML_SetElementDeclHandler(parser, dummy_element_decl_handler);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+ if (dummy_handler_flags != DUMMY_ELEMENT_DECL_HANDLER_FLAG)
+ fail("Element handler flag not raised");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_pi_in_epilog)
+{
+ const char *text =
+ "\n"
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat certain counts to allow for cached allocations */
+ if (i == 3 && repeat == 1) {
+ i -= 2;
+ repeat++;
+ }
+ else if (i == 2 && repeat < 4 && repeat != 1) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetProcessingInstructionHandler(parser, dummy_pi_handler);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse completed despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+ if (dummy_handler_flags != DUMMY_PI_HANDLER_FLAG)
+ fail("Processing instruction handler not invoked");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_comment_in_epilog)
+{
+ const char *text =
+ "\n"
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat certain counts to allow for cached allocations */
+ if (i == 3 && repeat == 1) {
+ i -= 2;
+ repeat++;
+ }
+ else if (i == 2 && repeat < 4 && repeat != 1) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetCommentHandler(parser, dummy_comment_handler);
+ dummy_handler_flags = 0;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse completed despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+ if (dummy_handler_flags != DUMMY_COMMENT_HANDLER_FLAG)
+ fail("Processing instruction handler not invoked");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_realloc_long_attribute_value)
+{
+ const char *text =
+ "]>\n"
+ "";
+ int i;
+#define MAX_REALLOC_COUNT 10
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_attribute_whitespace)
+{
+ const char *text = "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if (i == 3 && repeat == 1) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 &&
+ (repeat == 0 || repeat == 2 || repeat == 3)) ||
+ (i == 3 && repeat == 4)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_attribute_predefined_entity)
+{
+ const char *text = "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if (i == 3 && repeat == 1) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 &&
+ (repeat == 0 || repeat == 2 || repeat == 3)) ||
+ (i == 3 && repeat == 4)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+/* Test that a character reference at the end of a suitably long
+ * default value for an attribute can trigger pool growth, and recovers
+ * if the allocator fails on it.
+ */
+START_TEST(test_alloc_long_attr_default_with_char_ref)
+{
+ const char *text =
+ "]>\n"
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if ((i == 2 && repeat < 3) ||
+ (i == 3 && repeat == 3) ||
+ (i == 4 && repeat == 4)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+/* Test that a long character reference substitution triggers a pool
+ * expansion correctly for an attribute value.
+ */
+START_TEST(test_alloc_long_attr_value)
+{
+ const char *text =
+ "]>\n"
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if ((i == 2 && repeat < 3) ||
+ (i == 3 && repeat < 6) ||
+ (i == 4 && repeat == 6) ||
+ (i == 5 && repeat == 7)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing allocator");
+ if (i == MAX_ALLOC_COUNT)
+ fail("Parse failed at maximum allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+/* Test that an error in a nested parameter entity substitution is
+ * handled correctly. It seems unlikely that the code path being
+ * exercised can be reached purely by carefully crafted XML, but an
+ * allocation error in the right place will definitely do it.
+ */
+START_TEST(test_alloc_nested_entities)
+{
+ const char *text =
+ "\n"
+ "";
+ ExtFaults test_data = {
+ "\n"
+ "\n"
+ "%pe2;",
+ "Memory Fail not faulted",
+ NULL,
+ XML_ERROR_NO_MEMORY
+ };
+
+ /* Causes an allocation error in a nested storeEntityValue() */
+ allocation_count = 12;
+ XML_SetUserData(parser, &test_data);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_faulter);
+ expect_failure(text, XML_ERROR_EXTERNAL_ENTITY_HANDLING,
+ "Entity allocation failure not noted");
+}
+END_TEST
+
+START_TEST(test_alloc_realloc_param_entity_newline)
+{
+ const char *text =
+ "\n"
+ "";
+ char dtd_text[] =
+ "\n'>"
+ "%pe;\n";
+ int i;
+#define MAX_REALLOC_COUNT 5
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ XML_SetUserData(parser, dtd_text);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_alloc);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_realloc_ce_extends_pe)
+{
+ const char *text =
+ "\n"
+ "";
+ char dtd_text[] =
+ "\n'>"
+ "%pe;\n";
+ int i;
+#define MAX_REALLOC_COUNT 5
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ XML_SetUserData(parser, dtd_text);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_alloc);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_realloc_attributes)
+{
+ const char *text =
+ "]>\n"
+ "wombat\n";
+ int i;
+#define MAX_REALLOC_COUNT 5
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+
+ if (i == 0)
+ fail("Parse succeeded despite failing reallocator");
+ if (i == MAX_REALLOC_COUNT)
+ fail("Parse failed at maximum reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_long_doc_name)
+{
+ const char *text =
+ /* 64 characters per line */
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat certain counts to defeat cached allocations */
+ if ((i == 2 && repeat < 4) ||
+ (i == 3 && repeat < 6)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing reallocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max reallocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_long_base)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&e;";
+ char entity_text[] = "Hello world";
+ const char *base =
+ /* 64 characters per line */
+ "LongBaseURI/that/will/overflow/an/internal/buffer/and/cause/it/t"
+ "o/have/to/grow/PQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789A/";
+ int i;
+ #define MAX_ALLOC_COUNT 20
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if (i == 4 && repeat == 4) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 4) ||
+ (i == 3 && repeat == 5) ||
+ (i == 4 && repeat == 6)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetUserData(parser, entity_text);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_alloc);
+ if (XML_SetBase(parser, base) == XML_STATUS_ERROR) {
+ XML_ParserReset(parser, NULL);
+ continue;
+ }
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_long_public_id)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&e;";
+ char entity_text[] = "Hello world";
+ int i;
+#define MAX_ALLOC_COUNT 20
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat certain counts to defeat allocation caching */
+ if (i == 4 && repeat == 4) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 3) ||
+ (i == 3 && (repeat == 3 || repeat == 5)) ||
+ (i == 4 && repeat == 6)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetUserData(parser, entity_text);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_alloc);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_long_entity_value)
+{
+ const char *text =
+ "\n"
+ " \n"
+ "]>\n"
+ "&e2;";
+ char entity_text[] = "Hello world";
+ int i;
+#define MAX_ALLOC_COUNT 20
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat certain counts to defeat cached allocations */
+ if (i == 5 && repeat == 4) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 3) ||
+ (i == 3 && repeat == 3) ||
+ (i == 4 && repeat == 5) ||
+ (i == 5 && repeat == 6)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetUserData(parser, entity_text);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_alloc);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_alloc_long_notation)
+{
+ const char *text =
+ "\n"
+ " \n"
+ " \n"
+ "]>\n"
+ "&e2;";
+ ExtOption options[] = {
+ { "foo", "Entity Foo" },
+ { "bar", "Entity Bar" },
+ { NULL, NULL }
+ };
+ int i;
+#define MAX_ALLOC_COUNT 20
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if (i == 5 && repeat == 5) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 3) ||
+ (i == 3 && (repeat == 3 || repeat == 4)) ||
+ (i == 4 && repeat == 6) ||
+ (i == 5 && repeat == 7)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetUserData(parser, options);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_optioner);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+
+ XML_ParserFree(parser);
+ parser = XML_ParserCreate(NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
static void
nsalloc_setup(void)
@@ -5687,6 +10980,1106 @@ START_TEST(test_nsalloc_realloc_binding_uri)
#undef MAX_REALLOC_COUNT
END_TEST
+/* Check handling of long prefix names (pool growth) */
+START_TEST(test_nsalloc_realloc_long_prefix)
+{
+ const char *text =
+ "<"
+ /* 64 characters per line */
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ ":foo xmlns:"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "='http://example.org/'>"
+ "";
+ int i;
+#define MAX_REALLOC_COUNT 12
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ /* Every reallocation bar the last is cached */
+ reallocation_count = (i == 0) ? 0 : 1;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing reallocations");
+ else if (i == MAX_REALLOC_COUNT)
+ fail("Parsing failed even at max reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+/* Check handling of even long prefix names (different code path) */
+START_TEST(test_nsalloc_realloc_longer_prefix)
+{
+ const char *text =
+ "<"
+ /* 64 characters per line */
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "Q:foo xmlns:"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "Q='http://example.org/'>"
+ "";
+ int i;
+#define MAX_REALLOC_COUNT 12
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ /* Every reallocation bar the last is cached */
+ reallocation_count = (i == 0) ? 0 : 1;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing reallocations");
+ else if (i == MAX_REALLOC_COUNT)
+ fail("Parsing failed even at max reallocation count");
+}
+#undef MAX_REALLOC_COUNT
+END_TEST
+
+START_TEST(test_nsalloc_long_namespace)
+{
+ const char *text =
+ "<"
+ /* 64 characters per line */
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ ":e xmlns:"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "='http://example.org/'>\n"
+ "<"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ ":f "
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ ":attr='foo'/>\n"
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if ((i == 4 && (repeat == 3 || repeat == 6)) ||
+ (i == 7 && repeat == 11)) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 2) ||
+ (i == 3 && (repeat == 2 ||
+ repeat == 4 ||
+ repeat == 5 ||
+ repeat == 7 ||
+ repeat == 8)) ||
+ (i == 5 && (repeat == 9 ||
+ repeat == 10)) ||
+ (i == 6 && repeat == 12) ||
+ (i == 7 && repeat == 13)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+/* Using a slightly shorter namespace name provokes allocations in
+ * slightly different places in the code.
+ */
+START_TEST(test_nsalloc_less_long_namespace)
+{
+ const char *text =
+ "<"
+ /* 64 characters per line */
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz012345678"
+ ":e xmlns:"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz012345678"
+ "='http://example.org/'>\n"
+ "<"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz012345678"
+ ":f "
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789AZ"
+ "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz012345678"
+ ":att='foo'/>\n"
+ "";
+ int i;
+#define MAX_ALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if ((i == 4 && (repeat == 3 || repeat == 5)) ||
+ (i == 7 && repeat == 10)) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 2) ||
+ (i == 3 && (repeat == 2 ||
+ repeat == 4 ||
+ repeat == 6)) ||
+ (i == 4 && repeat == 7) ||
+ (i == 5 && (repeat == 8 ||
+ repeat == 9)) ||
+ (i == 6 && (repeat == 11 || repeat == 12)) ||
+ (i == 7 && repeat == 13)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
+
+START_TEST(test_nsalloc_long_context)
+{
+ const char *text =
+ "\n"
+ " \n"
+ "]>\n"
+ "\n"
+ "&en;"
+ "";
+ ExtOption options[] = {
+ { "foo", ""},
+ { "bar", "" },
+ { NULL, NULL }
+ };
+ int i;
+#define MAX_ALLOC_COUNT 40
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat allocation caching */
+ if ((i == 4 && repeat == 3) ||
+ (i == 13 && repeat == 9) ||
+ (i == 14 && repeat == 10)) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 2) ||
+ (i == 3 && (repeat == 2 || repeat == 4 || repeat == 5)) ||
+ (i == 4 && repeat == 6) ||
+ (i == 5 && repeat == 7) ||
+ (i == 7 && repeat == 8) ||
+ (i == 13 && repeat == 11)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetUserData(parser, options);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_optioner);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+
+ XML_ParserFree(parser);
+ parser = XML_ParserCreate(NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max allocation count");
+#undef MAX_ALLOC_COUNT
+}
+END_TEST
+
+/* This function is void; it will throw a fail() on error, so if it
+ * returns normally it must have succeeded.
+ */
+static void
+context_realloc_test(XML_Parser parser, const char *text)
+{
+ ExtOption options[] = {
+ { "foo", ""},
+ { "bar", "" },
+ { NULL, NULL }
+ };
+ int i;
+#define MAX_REALLOC_COUNT 5
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ reallocation_count = i;
+ XML_SetUserData(parser, options);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_optioner);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing reallocations");
+ else if (i == MAX_REALLOC_COUNT)
+ fail("Parsing failed even at max reallocation count");
+#undef MAX_REALLOC_COUNT
+}
+
+START_TEST(test_nsalloc_realloc_long_context)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "&en;"
+ "";
+
+ context_realloc_test(parser, text);
+}
+END_TEST
+
+START_TEST(test_nsalloc_realloc_long_context_2)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "&en;"
+ "";
+
+ context_realloc_test(parser, text);
+}
+END_TEST
+
+START_TEST(test_nsalloc_realloc_long_context_3)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "&en;"
+ "";
+
+ context_realloc_test(parser, text);
+}
+END_TEST
+
+START_TEST(test_nsalloc_realloc_long_context_4)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "&en;"
+ "";
+
+ context_realloc_test(parser, text);
+}
+END_TEST
+
+START_TEST(test_nsalloc_realloc_long_context_5)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "&en;"
+ "";
+
+ context_realloc_test(parser, text);
+}
+END_TEST
+
+START_TEST(test_nsalloc_realloc_long_context_6)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "&en;"
+ "";
+
+ context_realloc_test(parser, text);
+}
+END_TEST
+
+START_TEST(test_nsalloc_realloc_long_context_7)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "&en;"
+ "";
+
+ context_realloc_test(parser, text);
+}
+END_TEST
+
+START_TEST(test_nsalloc_realloc_long_ge_name)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "\n"
+ "&"
+ /* 64 characters per line */
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ ";"
+ "";
+ ExtOption options[] = {
+ { "foo", "" },
+ { "bar", "" },
+ { NULL, NULL }
+ };
+ int i;
+#define MAX_REALLOC_COUNT 5
+ int repeat = 0;
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat caching */
+ if (i == 2 && repeat < 3) {
+ i--;
+ repeat++;
+ }
+ reallocation_count = i;
+ XML_SetUserData(parser, options);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_optioner);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing reallocations");
+ else if (i == MAX_REALLOC_COUNT)
+ fail("Parsing failed even at max reallocation count");
+#undef MAX_REALLOC_COUNT
+}
+END_TEST
+
+/* Test that when a namespace is passed through the context mechanism
+ * to an external entity parser, the parsers handle reallocation
+ * failures correctly. The prefix is exactly the right length to
+ * provoke particular uncommon code paths.
+ */
+START_TEST(test_nsalloc_realloc_long_context_in_dtd)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "<"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ ":doc xmlns:"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+ "='foo/Second'>&First;";
+ ExtOption options[] = {
+ { "foo/First", "Hello world" },
+ { NULL, NULL }
+ };
+ int i;
+#define MAX_REALLOC_COUNT 10
+ int repeat = 0;
+
+ for (i = 0; i < MAX_REALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat allocation caching */
+ if (i == 2 && repeat < 11) {
+ i--;
+ repeat++;
+ }
+ reallocation_count = i;
+ XML_SetUserData(parser, options);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_optioner);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+ XML_ParserReset(parser, NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing reallocations");
+ else if (i == MAX_REALLOC_COUNT)
+ fail("Parsing failed even at max reallocation count");
+#undef MAX_REALLOC_COUNT
+}
+END_TEST
+
+START_TEST(test_nsalloc_long_default_in_ext)
+{
+ const char *text =
+ "\n"
+ " \n"
+ "]>\n"
+ "&x;";
+ ExtOption options[] = {
+ { "foo", ""},
+ { NULL, NULL }
+ };
+ int i;
+#define MAX_ALLOC_COUNT 30
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat allocation caching */
+ if ((i == 4 && repeat == 3) || (i == 8 && repeat == 9)) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 2) ||
+ (i == 3 && (repeat == 2 || repeat == 4 || repeat == 5)) ||
+ (i == 4 && repeat == 6) ||
+ (i == 5 && repeat == 7) ||
+ (i == 6 && repeat == 8)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetUserData(parser, options);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_optioner);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+
+ XML_ParserFree(parser);
+ parser = XML_ParserCreate(NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max allocation count");
+#undef MAX_ALLOC_COUNT
+}
+END_TEST
+
+START_TEST(test_nsalloc_long_systemid_in_ext)
+{
+ const char *text =
+ "\n"
+ "]>\n"
+ "&en;";
+ ExtOption options[] = {
+ { "foo", "" },
+ {
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/"
+ "ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/ABCDEFGHIJKLMNO/",
+ ""
+ },
+ { NULL, NULL }
+ };
+ int i;
+#define MAX_ALLOC_COUNT 30
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to allow for cached allocations */
+ if ((i == 4 && repeat == 3) || (i == 10 && repeat == 8)) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 2) ||
+ (i == 3 && (repeat == 2 || repeat == 4 || repeat == 5)) ||
+ (i == 4 && repeat == 6) ||
+ (i == 5 && repeat == 7) ||
+ (i == 9 && repeat == 8)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetUserData(parser, options);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_optioner);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+
+ XML_ParserFree(parser);
+ parser = XML_ParserCreate(NULL);
+ }
+ if (i == 0)
+ fail("Parsing worked despite failing allocations");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Parsing failed even at max allocation count");
+#undef MAX_ALLOC_COUNT
+}
+END_TEST
+
+/* Test the effects of allocation failure on parsing an element in a
+ * namespace. Based on test_nsalloc_long_context.
+ */
+START_TEST(test_nsalloc_prefixed_element)
+{
+ const char *text =
+ "\n"
+ " \n"
+ "]>\n"
+ "\n"
+ "&en;"
+ "";
+ ExtOption options[] = {
+ { "foo", "" },
+ { "bar", "" },
+ { NULL, NULL }
+ };
+ int i;
+#define MAX_ALLOC_COUNT 40
+ int repeat = 0;
+
+ for (i = 0; i < MAX_ALLOC_COUNT; i++) {
+ /* Repeat some counts to defeat cached allocations */
+ if ((i == 4 && repeat == 3) ||
+ (i == 14 && repeat == 9) ||
+ (i == 15 && repeat == 10)) {
+ i -= 2;
+ repeat++;
+ }
+ else if ((i == 2 && repeat < 2) ||
+ (i == 3 && (repeat == 2 || repeat == 4 || repeat == 5)) ||
+ (i == 4 && repeat == 6) ||
+ (i == 6 && repeat == 7) ||
+ (i == 8 && repeat == 8)) {
+ i--;
+ repeat++;
+ }
+ allocation_count = i;
+ XML_SetUserData(parser, options);
+ XML_SetParamEntityParsing(parser, XML_PARAM_ENTITY_PARSING_ALWAYS);
+ XML_SetExternalEntityRefHandler(parser, external_entity_optioner);
+ if (_XML_Parse_SINGLE_BYTES(parser, text, strlen(text),
+ XML_TRUE) != XML_STATUS_ERROR)
+ break;
+
+ XML_ParserFree(parser);
+ parser = XML_ParserCreate(NULL);
+ }
+ if (i == 0)
+ fail("Success despite failing allocator");
+ else if (i == MAX_ALLOC_COUNT)
+ fail("Failed even at full allocation count");
+}
+#undef MAX_ALLOC_COUNT
+END_TEST
static Suite *
make_suite(void)
@@ -5707,11 +12100,17 @@ make_suite(void)
tcase_add_test(tc_basic, test_bom_utf8);
tcase_add_test(tc_basic, test_bom_utf16_be);
tcase_add_test(tc_basic, test_bom_utf16_le);
+ tcase_add_test(tc_basic, test_nobom_utf16_le);
tcase_add_test(tc_basic, test_illegal_utf8);
tcase_add_test(tc_basic, test_utf8_auto_align);
tcase_add_test(tc_basic, test_utf16);
tcase_add_test(tc_basic, test_utf16_le_epilog_newline);
+ tcase_add_test(tc_basic, test_not_utf16);
+ tcase_add_test(tc_basic, test_bad_encoding);
tcase_add_test(tc_basic, test_latin1_umlauts);
+ tcase_add_test(tc_basic, test_long_utf8_character);
+ tcase_add_test(tc_basic, test_long_latin1_attribute);
+ tcase_add_test(tc_basic, test_long_ascii_attribute);
/* Regression test for SF bug #491986. */
tcase_add_test(tc_basic, test_danish_latin1);
/* Regression test for SF bug #514281. */
@@ -5730,6 +12129,9 @@ make_suite(void)
tcase_add_test(tc_basic, test_end_element_events);
tcase_add_test(tc_basic, test_attr_whitespace_normalization);
tcase_add_test(tc_basic, test_xmldecl_misplaced);
+ tcase_add_test(tc_basic, test_xmldecl_invalid);
+ tcase_add_test(tc_basic, test_xmldecl_missing_attr);
+ tcase_add_test(tc_basic, test_xmldecl_missing_value);
tcase_add_test(tc_basic, test_unknown_encoding_internal_entity);
tcase_add_test(tc_basic, test_unrecognised_encoding_internal_entity);
tcase_add_test(tc_basic,
@@ -5741,14 +12143,18 @@ make_suite(void)
tcase_add_test(tc_basic, test_not_standalone_handler_accept);
tcase_add_test(tc_basic,
test_wfc_undeclared_entity_with_external_subset_standalone);
+ tcase_add_test(tc_basic,
+ test_entity_with_external_subset_unless_standalone);
tcase_add_test(tc_basic, test_wfc_no_recursive_entity_refs);
tcase_add_test(tc_basic, test_ext_entity_set_encoding);
tcase_add_test(tc_basic, test_ext_entity_no_handler);
tcase_add_test(tc_basic, test_ext_entity_set_bom);
tcase_add_test(tc_basic, test_ext_entity_bad_encoding);
+ tcase_add_test(tc_basic, test_ext_entity_bad_encoding_2);
tcase_add_test(tc_basic, test_ext_entity_invalid_parse);
tcase_add_test(tc_basic, test_ext_entity_invalid_suspended_parse);
tcase_add_test(tc_basic, test_dtd_default_handling);
+ tcase_add_test(tc_basic, test_dtd_attr_handling);
tcase_add_test(tc_basic, test_empty_ns_without_namespaces);
tcase_add_test(tc_basic, test_ns_in_attribute_default_without_namespaces);
tcase_add_test(tc_basic, test_stop_parser_between_char_data_calls);
@@ -5756,8 +12162,10 @@ make_suite(void)
tcase_add_test(tc_basic, test_repeated_stop_parser_between_char_data_calls);
tcase_add_test(tc_basic, test_good_cdata_ascii);
tcase_add_test(tc_basic, test_good_cdata_utf16);
+ tcase_add_test(tc_basic, test_good_cdata_utf16_le);
tcase_add_test(tc_basic, test_long_cdata_utf16);
tcase_add_test(tc_basic, test_multichar_cdata_utf16);
+ tcase_add_test(tc_basic, test_utf16_bad_surrogate_pair);
tcase_add_test(tc_basic, test_bad_cdata);
tcase_add_test(tc_basic, test_bad_cdata_utf16);
tcase_add_test(tc_basic, test_stop_parser_between_cdata_calls);
@@ -5766,6 +12174,11 @@ make_suite(void)
tcase_add_test(tc_basic, test_default_current);
tcase_add_test(tc_basic, test_dtd_elements);
tcase_add_test(tc_basic, test_set_foreign_dtd);
+ tcase_add_test(tc_basic, test_foreign_dtd_not_standalone);
+ tcase_add_test(tc_basic, test_invalid_foreign_dtd);
+ tcase_add_test(tc_basic, test_foreign_dtd_with_doctype);
+ tcase_add_test(tc_basic, test_foreign_dtd_without_external_subset);
+ tcase_add_test(tc_basic, test_empty_foreign_dtd);
tcase_add_test(tc_basic, test_set_base);
tcase_add_test(tc_basic, test_attributes);
tcase_add_test(tc_basic, test_reset_in_entity);
@@ -5792,8 +12205,108 @@ make_suite(void)
tcase_add_test(tc_basic, test_byte_info_at_cdata);
tcase_add_test(tc_basic, test_predefined_entities);
tcase_add_test(tc_basic, test_invalid_tag_in_dtd);
+ tcase_add_test(tc_basic, test_not_predefined_entities);
tcase_add_test(tc_basic, test_ignore_section);
+ tcase_add_test(tc_basic, test_ignore_section_utf16);
+ tcase_add_test(tc_basic, test_ignore_section_utf16_be);
tcase_add_test(tc_basic, test_bad_ignore_section);
+ tcase_add_test(tc_basic, test_external_entity_values);
+ tcase_add_test(tc_basic, test_ext_entity_not_standalone);
+ tcase_add_test(tc_basic, test_ext_entity_value_abort);
+ tcase_add_test(tc_basic, test_bad_public_doctype);
+ tcase_add_test(tc_basic, test_attribute_enum_value);
+ tcase_add_test(tc_basic, test_predefined_entity_redefinition);
+ tcase_add_test(tc_basic, test_dtd_stop_processing);
+ tcase_add_test(tc_basic, test_public_notation_no_sysid);
+ tcase_add_test(tc_basic, test_nested_groups);
+ tcase_add_test(tc_basic, test_group_choice);
+ tcase_add_test(tc_basic, test_standalone_parameter_entity);
+ tcase_add_test(tc_basic, test_skipped_parameter_entity);
+ tcase_add_test(tc_basic, test_recursive_external_parameter_entity);
+ tcase_add_test(tc_basic, test_undefined_ext_entity_in_external_dtd);
+ tcase_add_test(tc_basic, test_suspend_xdecl);
+ tcase_add_test(tc_basic, test_abort_epilog);
+ tcase_add_test(tc_basic, test_abort_epilog_2);
+ tcase_add_test(tc_basic, test_suspend_epilog);
+ tcase_add_test(tc_basic, test_unfinished_epilog);
+ tcase_add_test(tc_basic, test_partial_char_in_epilog);
+ tcase_add_test(tc_basic, test_hash_collision);
+ tcase_add_test(tc_basic, test_suspend_resume_internal_entity);
+ tcase_add_test(tc_basic, test_resume_entity_with_syntax_error);
+ tcase_add_test(tc_basic, test_suspend_resume_parameter_entity);
+ tcase_add_test(tc_basic, test_restart_on_error);
+ tcase_add_test(tc_basic, test_reject_lt_in_attribute_value);
+ tcase_add_test(tc_basic, test_reject_unfinished_param_in_att_value);
+ tcase_add_test(tc_basic, test_trailing_cr_in_att_value);
+ tcase_add_test(tc_basic, test_standalone_internal_entity);
+ tcase_add_test(tc_basic, test_skipped_external_entity);
+ tcase_add_test(tc_basic, test_skipped_null_loaded_ext_entity);
+ tcase_add_test(tc_basic, test_skipped_unloaded_ext_entity);
+ tcase_add_test(tc_basic, test_param_entity_with_trailing_cr);
+ tcase_add_test(tc_basic, test_invalid_character_entity);
+ tcase_add_test(tc_basic, test_invalid_character_entity_2);
+ tcase_add_test(tc_basic, test_invalid_character_entity_3);
+ tcase_add_test(tc_basic, test_invalid_character_entity_4);
+ tcase_add_test(tc_basic, test_pi_handled_in_default);
+ tcase_add_test(tc_basic, test_comment_handled_in_default);
+ tcase_add_test(tc_basic, test_pi_yml);
+ tcase_add_test(tc_basic, test_pi_xnl);
+ tcase_add_test(tc_basic, test_pi_xmm);
+ tcase_add_test(tc_basic, test_utf16_pi);
+ tcase_add_test(tc_basic, test_utf16_be_pi);
+ tcase_add_test(tc_basic, test_utf16_be_comment);
+ tcase_add_test(tc_basic, test_utf16_le_comment);
+ tcase_add_test(tc_basic, test_missing_encoding_conversion_fn);
+ tcase_add_test(tc_basic, test_failing_encoding_conversion_fn);
+ tcase_add_test(tc_basic, test_unknown_encoding_success);
+ tcase_add_test(tc_basic, test_unknown_encoding_bad_name);
+ tcase_add_test(tc_basic, test_unknown_encoding_bad_name_2);
+ tcase_add_test(tc_basic, test_unknown_encoding_long_name_1);
+ tcase_add_test(tc_basic, test_unknown_encoding_long_name_2);
+ tcase_add_test(tc_basic, test_invalid_unknown_encoding);
+ tcase_add_test(tc_basic, test_unknown_ascii_encoding_ok);
+ tcase_add_test(tc_basic, test_unknown_ascii_encoding_fail);
+ tcase_add_test(tc_basic, test_unknown_encoding_invalid_length);
+ tcase_add_test(tc_basic, test_unknown_encoding_invalid_topbit);
+ tcase_add_test(tc_basic, test_unknown_encoding_invalid_surrogate);
+ tcase_add_test(tc_basic, test_unknown_encoding_invalid_high);
+ tcase_add_test(tc_basic, test_unknown_encoding_invalid_attr_value);
+ tcase_add_test(tc_basic, test_ext_entity_latin1_utf16le_bom);
+ tcase_add_test(tc_basic, test_ext_entity_latin1_utf16be_bom);
+ tcase_add_test(tc_basic, test_ext_entity_latin1_utf16le_bom2);
+ tcase_add_test(tc_basic, test_ext_entity_latin1_utf16be_bom2);
+ tcase_add_test(tc_basic, test_ext_entity_utf16_be);
+ tcase_add_test(tc_basic, test_ext_entity_utf16_le);
+ tcase_add_test(tc_basic, test_ext_entity_utf16_unknown);
+ tcase_add_test(tc_basic, test_ext_entity_utf8_non_bom);
+ tcase_add_test(tc_basic, test_utf8_in_cdata_section);
+ tcase_add_test(tc_basic, test_utf8_in_cdata_section_2);
+ tcase_add_test(tc_basic, test_trailing_spaces_in_elements);
+ tcase_add_test(tc_basic, test_utf16_attribute);
+ tcase_add_test(tc_basic, test_utf16_second_attr);
+ tcase_add_test(tc_basic, test_attr_after_solidus);
+ tcase_add_test(tc_basic, test_utf16_pe);
+ tcase_add_test(tc_basic, test_bad_attr_desc_keyword);
+ tcase_add_test(tc_basic, test_bad_attr_desc_keyword_utf16);
+ tcase_add_test(tc_basic, test_bad_doctype);
+ tcase_add_test(tc_basic, test_bad_doctype_utf16);
+ tcase_add_test(tc_basic, test_bad_doctype_plus);
+ tcase_add_test(tc_basic, test_bad_doctype_star);
+ tcase_add_test(tc_basic, test_bad_doctype_query);
+ tcase_add_test(tc_basic, test_unknown_encoding_bad_ignore);
+ tcase_add_test(tc_basic, test_entity_in_utf16_be_attr);
+ tcase_add_test(tc_basic, test_entity_in_utf16_le_attr);
+ tcase_add_test(tc_basic, test_entity_public_utf16_be);
+ tcase_add_test(tc_basic, test_entity_public_utf16_le);
+ tcase_add_test(tc_basic, test_short_doctype);
+ tcase_add_test(tc_basic, test_short_doctype_2);
+ tcase_add_test(tc_basic, test_short_doctype_3);
+ tcase_add_test(tc_basic, test_long_doctype);
+ tcase_add_test(tc_basic, test_bad_entity);
+ tcase_add_test(tc_basic, test_bad_entity_2);
+ tcase_add_test(tc_basic, test_bad_entity_3);
+ tcase_add_test(tc_basic, test_bad_entity_4);
+ tcase_add_test(tc_basic, test_bad_notation);
suite_add_tcase(s, tc_namespace);
tcase_add_checked_fixture(tc_namespace,
@@ -5819,6 +12332,17 @@ make_suite(void)
tcase_add_test(tc_namespace, test_ns_extend_uri_buffer);
tcase_add_test(tc_namespace, test_ns_reserved_attributes);
tcase_add_test(tc_namespace, test_ns_reserved_attributes_2);
+ tcase_add_test(tc_namespace, test_ns_extremely_long_prefix);
+ tcase_add_test(tc_namespace, test_ns_unknown_encoding_success);
+ tcase_add_test(tc_namespace, test_ns_double_colon);
+ tcase_add_test(tc_namespace, test_ns_double_colon_element);
+ tcase_add_test(tc_namespace, test_ns_bad_attr_leafname);
+ tcase_add_test(tc_namespace, test_ns_bad_element_leafname);
+ tcase_add_test(tc_namespace, test_ns_utf16_leafname);
+ tcase_add_test(tc_namespace, test_ns_utf16_element_leafname);
+ tcase_add_test(tc_namespace, test_ns_utf16_doctype);
+ tcase_add_test(tc_namespace, test_ns_invalid_doctype);
+ tcase_add_test(tc_namespace, test_ns_double_colon_doctype);
suite_add_tcase(s, tc_misc);
tcase_add_checked_fixture(tc_misc, NULL, basic_teardown);
@@ -5829,11 +12353,15 @@ make_suite(void)
tcase_add_test(tc_misc, test_misc_version);
tcase_add_test(tc_misc, test_misc_features);
tcase_add_test(tc_misc, test_misc_attribute_leak);
+ tcase_add_test(tc_misc, test_misc_utf16le);
suite_add_tcase(s, tc_alloc);
tcase_add_checked_fixture(tc_alloc, alloc_setup, alloc_teardown);
+ tcase_add_test(tc_alloc, test_alloc_parse_xdecl);
+ tcase_add_test(tc_alloc, test_alloc_parse_xdecl_2);
tcase_add_test(tc_alloc, test_alloc_parse_pi);
tcase_add_test(tc_alloc, test_alloc_parse_pi_2);
+ tcase_add_test(tc_alloc, test_alloc_parse_pi_3);
tcase_add_test(tc_alloc, test_alloc_parse_comment);
tcase_add_test(tc_alloc, test_alloc_parse_comment_2);
tcase_add_test(tc_alloc, test_alloc_create_external_parser);
@@ -5848,6 +12376,38 @@ make_suite(void)
tcase_add_test(tc_alloc, test_alloc_realloc_buffer);
tcase_add_test(tc_alloc, test_alloc_ext_entity_realloc_buffer);
tcase_add_test(tc_alloc, test_alloc_realloc_many_attributes);
+ tcase_add_test(tc_alloc, test_alloc_public_entity_value);
+ tcase_add_test(tc_alloc, test_alloc_realloc_subst_public_entity_value);
+ tcase_add_test(tc_alloc, test_alloc_parse_public_doctype);
+ tcase_add_test(tc_alloc, test_alloc_parse_public_doctype_long_name);
+ tcase_add_test(tc_alloc, test_alloc_set_foreign_dtd);
+ tcase_add_test(tc_alloc, test_alloc_attribute_enum_value);
+ tcase_add_test(tc_alloc, test_alloc_realloc_attribute_enum_value);
+ tcase_add_test(tc_alloc, test_alloc_realloc_implied_attribute);
+ tcase_add_test(tc_alloc, test_alloc_realloc_default_attribute);
+ tcase_add_test(tc_alloc, test_alloc_notation);
+ tcase_add_test(tc_alloc, test_alloc_public_notation);
+ tcase_add_test(tc_alloc, test_alloc_system_notation);
+ tcase_add_test(tc_alloc, test_alloc_nested_groups);
+ tcase_add_test(tc_alloc, test_alloc_realloc_nested_groups);
+ tcase_add_test(tc_alloc, test_alloc_large_group);
+ tcase_add_test(tc_alloc, test_alloc_realloc_group_choice);
+ tcase_add_test(tc_alloc, test_alloc_pi_in_epilog);
+ tcase_add_test(tc_alloc, test_alloc_comment_in_epilog);
+ tcase_add_test(tc_alloc, test_alloc_realloc_long_attribute_value);
+ tcase_add_test(tc_alloc, test_alloc_attribute_whitespace);
+ tcase_add_test(tc_alloc, test_alloc_attribute_predefined_entity);
+ tcase_add_test(tc_alloc, test_alloc_long_attr_default_with_char_ref);
+ tcase_add_test(tc_alloc, test_alloc_long_attr_value);
+ tcase_add_test(tc_alloc, test_alloc_nested_entities);
+ tcase_add_test(tc_alloc, test_alloc_realloc_param_entity_newline);
+ tcase_add_test(tc_alloc, test_alloc_realloc_ce_extends_pe);
+ tcase_add_test(tc_alloc, test_alloc_realloc_attributes);
+ tcase_add_test(tc_alloc, test_alloc_long_doc_name);
+ tcase_add_test(tc_alloc, test_alloc_long_base);
+ tcase_add_test(tc_alloc, test_alloc_long_public_id);
+ tcase_add_test(tc_alloc, test_alloc_long_entity_value);
+ tcase_add_test(tc_alloc, test_alloc_long_notation);
suite_add_tcase(s, tc_nsalloc);
tcase_add_checked_fixture(tc_nsalloc, nsalloc_setup, nsalloc_teardown);
@@ -5860,6 +12420,23 @@ make_suite(void)
tcase_add_test(tc_nsalloc, test_nsalloc_realloc_attributes);
tcase_add_test(tc_nsalloc, test_nsalloc_long_element);
tcase_add_test(tc_nsalloc, test_nsalloc_realloc_binding_uri);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_prefix);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_longer_prefix);
+ tcase_add_test(tc_nsalloc, test_nsalloc_long_namespace);
+ tcase_add_test(tc_nsalloc, test_nsalloc_less_long_namespace);
+ tcase_add_test(tc_nsalloc, test_nsalloc_long_context);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_context);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_context_2);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_context_3);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_context_4);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_context_5);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_context_6);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_context_7);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_ge_name);
+ tcase_add_test(tc_nsalloc, test_nsalloc_realloc_long_context_in_dtd);
+ tcase_add_test(tc_nsalloc, test_nsalloc_long_default_in_ext);
+ tcase_add_test(tc_nsalloc, test_nsalloc_long_systemid_in_ext);
+ tcase_add_test(tc_nsalloc, test_nsalloc_prefixed_element);
return s;
}