organicmaps/icu
13
Fork 0
mirror of https://github.com/unicode-org/icu.git synced 2025-04-08 23:10:40 +00:00
Code Issues 1 Projects Releases Wiki Activity Actions 1

ICU-1292 add variable range overlap detection

Browse source 
X-SVN-Rev: 6786
This commit is contained in:
Alan Liu 2001-11-12 19:12:08 +00:00
parent b07d31376d
commit 9160747c21
4 changed files with 61 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

33
icu4c/source/i18n/rbt_pars.cpp
View file

@ -354,6 +354,9 @@ int32_t RuleHalf::parseSection(const UnicodeString& rule, int32_t pos, int32_t l
if (escaped == (UChar32) -1) {
return syntaxError(U_MALFORMED_UNICODE_ESCAPE, rule, start);
}
if (!parser.checkVariableRange(escaped)) {
return syntaxError(U_VARIABLE_RANGE_OVERLAP, rule, start);
}
buf.append(escaped);
continue;
}
@ -388,9 +391,20 @@ int32_t RuleHalf::parseSection(const UnicodeString& rule, int32_t pos, int32_t l
}
}
quoteLimit = buf.length();
for (iq=quoteStart; iq<quoteLimit; ++iq) {
if (!parser.checkVariableRange(buf.charAt(iq))) {
return syntaxError(U_VARIABLE_RANGE_OVERLAP, rule, start);
}
}
}
continue;
}
if (!parser.checkVariableRange(c)) {
return syntaxError(U_VARIABLE_RANGE_OVERLAP, rule, start);
}
switch (c) {
//------------------------------------------------------
@ -913,6 +927,15 @@ void TransliteratorParser::setVariableRange(int32_t start, int32_t end) {
variableLimit = (UChar) (end + 1);
}
/**
* Assert that the given character is NOT within the variable range.
* If it is, return FALSE. This is neccesary to ensure that the
* variable range does not overlap characters used in a rule.
*/
UBool TransliteratorParser::checkVariableRange(UChar32 ch) const {
return !(ch >= data->variablesBase && ch < variableLimit);
}
/**
* Set the maximum backup to 'backup', in response to a pragma
* statement.
@ -929,15 +952,15 @@ void TransliteratorParser::pragmaNormalizeRules(UNormalizationMode mode) {
//TODO Finish
}
static const UChar PRAGMA_USE[] = {0x75,0x73,0x65,0x20}; // "use "
static const UChar PRAGMA_USE[] = {0x75,0x73,0x65,0x20,0}; // "use "
static const UChar PRAGMA_VARIABLE_RANGE[] = {0x7E,0x76,0x61,0x72,0x69,0x61,0x62,0x6C,0x65,0x20,0x72,0x61,0x6E,0x67,0x65,0x20,0x23,0x20,0x23,0x7E,0x3B}; // "~variable range # #~;"
static const UChar PRAGMA_VARIABLE_RANGE[] = {0x7E,0x76,0x61,0x72,0x69,0x61,0x62,0x6C,0x65,0x20,0x72,0x61,0x6E,0x67,0x65,0x20,0x23,0x20,0x23,0x7E,0x3B,0}; // "~variable range # #~;"
static const UChar PRAGMA_MAXIMUM_BACKUP[] = {0x7E,0x6D,0x61,0x78,0x69,0x6D,0x75,0x6D,0x20,0x62,0x61,0x63,0x6B,0x75,0x70,0x20,0x23,0x7E,0x3B}; // "~maximum backup #~;"
static const UChar PRAGMA_MAXIMUM_BACKUP[] = {0x7E,0x6D,0x61,0x78,0x69,0x6D,0x75,0x6D,0x20,0x62,0x61,0x63,0x6B,0x75,0x70,0x20,0x23,0x7E,0x3B,0}; // "~maximum backup #~;"
static const UChar PRAGMA_NFD_RULES[] = {0x7E,0x6E,0x66,0x64,0x20,0x72,0x75,0x6C,0x65,0x73,0x7E,0x3B}; // "~nfd rules~;"
static const UChar PRAGMA_NFD_RULES[] = {0x7E,0x6E,0x66,0x64,0x20,0x72,0x75,0x6C,0x65,0x73,0x7E,0x3B,0}; // "~nfd rules~;"
static const UChar PRAGMA_NFC_RULES[] = {0x7E,0x6E,0x66,0x63,0x20,0x72,0x75,0x6C,0x65,0x73,0x7E,0x3B}; // "~nfc rules~;"
static const UChar PRAGMA_NFC_RULES[] = {0x7E,0x6E,0x66,0x63,0x20,0x72,0x75,0x6C,0x65,0x73,0x7E,0x3B,0}; // "~nfc rules~;"
/**
* Return true if the given rule looks like a pragma.

7
icu4c/source/i18n/rbt_pars.h
View file

@ -179,6 +179,13 @@ private:
*/
void setVariableRange(int32_t start, int32_t end);
/**
* Assert that the given character is NOT within the variable range.
* If it is, return FALSE. This is neccesary to ensure that the
* variable range does not overlap characters used in a rule.
*/
UBool checkVariableRange(UChar32 ch) const;
/**
* Set the maximum backup to 'backup', in response to a pragma
* statement.

20
icu4c/source/test/intltest/transtst.cpp
View file

@ -136,6 +136,7 @@ TransliteratorTest::runIndexedTest(int32_t index, UBool exec,
TESTCASE(54,TestLocaleResource);
TESTCASE(55,TestParseError);
TESTCASE(56,TestOutputSet);
TESTCASE(57,TestVariableRange);
default: name = ""; break;
}
}
@ -2645,6 +2646,25 @@ void TransliteratorTest::TestOutputSet() {
errln("FAIL: No syntax error");
}
/**
* Test the use variable range pragma, making sure that use of
* variable range characters is detected and flagged as an error.
*/
void TransliteratorTest::TestVariableRange() {
UnicodeString rule = "use variable range 0x70 0x72; a > A; b > B; q > Q;";
UErrorCode ec = U_ZERO_ERROR;
UParseError pe;
Transliterator *t = Transliterator::createFromRules("ID", rule, UTRANS_FORWARD, pe, ec);
delete t;
if (U_FAILURE(ec)) {
UnicodeString err(pe.preContext);
err.append((UChar)124/*|*/).append(pe.postContext);
logln("Ok: " + err);
return;
}
errln("FAIL: No syntax error");
}
//======================================================================
// icu4c ONLY
// These tests are not mirrored (yet) in icu4j at

6
icu4c/source/test/intltest/transtst.h
View file

@ -266,6 +266,12 @@ class TransliteratorTest : public IntlTest {
*/
void TestOutputSet();
/**
* Test the use variable range pragma, making sure that use of
* variable range characters is detected and flagged as an error.
*/
void TestVariableRange();
//======================================================================
// Support methods
//======================================================================