organicmaps/icu
13
Fork 0
mirror of https://github.com/unicode-org/icu.git synced 2025-04-13 17:01:16 +00:00
Code Issues 1 Projects Releases Wiki Activity Actions 3

ICU-1923 upgrade to Unicode 3.2: verify all DerivedCoreProperties.txt against u_hasBinaryProperty() which computes most of them

Browse source 
X-SVN-Rev: 8751
This commit is contained in:
Markus Scherer 2002-05-31 01:21:57 +00:00
parent 802f0c5735
commit 338fed0add
2 changed files with 191 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

183
icu4c/source/test/intltest/ucdtest.cpp
View file

@ -4,12 +4,15 @@
* others. All Rights Reserved.
********************************************************************/
#include "ucdtest.h"
#include "unicode/unicode.h"
#include "unicode/ustring.h"
#include "unicode/uchar.h"
#include "unicode/uniset.h"
#include "cstring.h"
#include "uparse.h"
#include "ucdtest.h"
#define LENGTHOF(array) (sizeof(array)/sizeof(array[0]))
UnicodeTest::UnicodeTest()
{
@ -32,6 +35,7 @@ void UnicodeTest::runIndexedTest( int32_t index, UBool exec, const char* &name,
case 6: name = "TestCharLength"; if(exec) TestCharLength(); break;
case 7: name = "TestIdentifier"; if(exec) TestIdentifier(); break;
case 8: name = "TestScript"; if(exec) TestScript(); break;
case 9: name = "TestAdditionalProperties"; if(exec) TestAdditionalProperties(); break;
default: name = ""; break; //needed to end loop
}
}
@ -170,7 +174,7 @@ void UnicodeTest::TestMisc()
const UChar sampleNonSpaces[] = {0x61, 0x62, 0x63, 0x64, 0x74};
const UChar sampleWhiteSpaces[] = {0x2008, 0x2009, 0x200a, 0x001c, 0x000c};
const UChar sampleNonWhiteSpaces[] = {0x61, 0x62, 0x3c, 0x28, 0x3f};
const UChar sampleUndefined[] = {0xfff1, 0xfff7, 0xfa30};
const UChar sampleUndefined[] = {0xfff1, 0xfff7, 0xfa6b};
const UChar sampleDefined[] = {0x523E, 0x4f88, 0xfffd};
const UChar sampleBase[] = {0x0061, 0x0031, 0x03d2};
const UChar sampleNonBase[] = {0x002B, 0x0020, 0x203B};
@ -724,3 +728,178 @@ void UnicodeTest::TestScript()
errln("ERROR: Diffe rent Thai values in EUnicodeScript and UCharScript");
}
}
// test DerivedCoreProperties.txt -------------------------------------------
// copied from genprops.c
static int32_t
getTokenIndex(const char *const tokens[], int32_t countTokens, const char *s) {
const char *t, *z;
int32_t i, j;
s=u_skipWhitespace(s);
for(i=0; i<countTokens; ++i) {
t=tokens[i];
if(t!=NULL) {
for(j=0;; ++j) {
if(t[j]!=0) {
if(s[j]!=t[j]) {
break;
}
} else {
z=u_skipWhitespace(s+j);
if(*z==';' || *z==0) {
return i;
} else {
break;
}
}
}
}
}
return -1;
}
static const char *const
derivedCorePropsNames[]={
"Math",
"Alphabetic",
"Lowercase",
"Uppercase",
"ID_Start",
"ID_Continue",
"XID_Start",
"XID_Continue",
"Default_Ignorable_Code_Point",
"Grapheme_Extend",
"Grapheme_Base"
};
static const UProperty
derivedCorePropsIndex[]={
UCHAR_MATH,
UCHAR_ALPHABETIC,
UCHAR_LOWERCASE,
UCHAR_UPPERCASE,
UCHAR_ID_START,
UCHAR_ID_CONTINUE,
UCHAR_XID_START,
UCHAR_XID_CONTINUE,
UCHAR_DEFAULT_IGNORABLE_CODE_POINT,
UCHAR_GRAPHEME_EXTEND,
UCHAR_GRAPHEME_BASE
};
void U_CALLCONV
UnicodeTest::derivedCorePropsLineFn(void *context,
char *fields[][2], int32_t fieldCount,
UErrorCode *pErrorCode) {
UnicodeTest *me=(UnicodeTest *)context;
uint32_t start, end;
int32_t i;
u_parseCodePointRange(fields[0][0], &start, &end, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
me->errln("UnicodeTest: syntax error in DerivedCoreProperties.txt field 0 at %s\n", fields[0][0]);
return;
}
/* parse derived binary property name, ignore unknown names */
i=getTokenIndex(derivedCorePropsNames, LENGTHOF(derivedCorePropsNames), fields[1][0]);
if(i<0) {
me->errln("UnicodeTest warning: unknown property name '%s' in \n", fields[1][0]);
return;
}
me->derivedCoreProps[i].add(start, end);
}
void UnicodeTest::TestAdditionalProperties() {
// test DerivedCoreProperties.txt
if(LENGTHOF(derivedCoreProps)<LENGTHOF(derivedCorePropsNames)) {
errln("error: UnicodeTest::derivedCoreProps[] too short, need at least %d UnicodeSets\n",
LENGTHOF(derivedCorePropsNames));
return;
}
if(LENGTHOF(derivedCorePropsIndex)!=LENGTHOF(derivedCorePropsNames)) {
errln("error in ucdtest.cpp: LENGTHOF(derivedCorePropsIndex)!=LENGTHOF(derivedCorePropsNames)\n");
return;
}
char newPath[256];
char backupPath[256];
char *fields[2][2];
int32_t length;
UErrorCode errorCode=U_ZERO_ERROR;
/* Look inside ICU_DATA first */
strcpy(newPath, u_getDataDirectory());
// remove trailing "out/"
length=uprv_strlen(newPath);
if(length>=4 && uprv_strcmp(newPath+length-4, "out" U_FILE_SEP_STRING)==0) {
newPath[length-4]=0;
}
strcat(newPath, "unidata" U_FILE_SEP_STRING "DerivedCoreProperties.txt");
// As a fallback, try to guess where the source data was located
// at the time ICU was built, and look there.
# ifdef U_TOPSRCDIR
strcpy(backupPath, U_TOPSRCDIR U_FILE_SEP_STRING "data");
# else
strcpy(backupPath, loadTestData(errorCode));
strcat(backupPath, U_FILE_SEP_STRING ".." U_FILE_SEP_STRING ".." U_FILE_SEP_STRING ".." U_FILE_SEP_STRING ".." U_FILE_SEP_STRING "data");
# endif
strcat(backupPath, U_FILE_SEP_STRING);
strcat(backupPath, "unidata" U_FILE_SEP_STRING "DerivedCoreProperties.txt");
u_parseDelimitedFile(newPath, ';', fields, 2, derivedCorePropsLineFn, this, &errorCode);
if(errorCode==U_FILE_ACCESS_ERROR) {
errorCode=U_ZERO_ERROR;
u_parseDelimitedFile(backupPath, ';', fields, 2, derivedCorePropsLineFn, this, &errorCode);
}
if(U_FAILURE(errorCode)) {
errln("error parsing DerivedCoreProperties.txt: %s\n", u_errorName(errorCode));
return;
}
// now we have all derived core properties in the UnicodeSets
// run them all through the API
int32_t i, rangeCount, range;
UChar32 start, end;
// test all TRUE properties
for(i=0; i<LENGTHOF(derivedCorePropsNames); ++i) {
rangeCount=derivedCoreProps[i].getRangeCount();
for(range=0; range<rangeCount; ++range) {
start=derivedCoreProps[i].getRangeStart(range);
end=derivedCoreProps[i].getRangeEnd(range);
for(; start<=end; ++start) {
if(!u_hasBinaryProperty(start, derivedCorePropsIndex[i])) {
errln("UnicodeTest error: u_hasBinaryProperty(U+%04lx, %s)==FALSE is wrong\n", start, derivedCorePropsNames[i]);
}
}
}
}
// invert all properties
for(i=0; i<LENGTHOF(derivedCorePropsNames); ++i) {
derivedCoreProps[i].complement();
}
// test all FALSE properties
for(i=0; i<LENGTHOF(derivedCorePropsNames); ++i) {
rangeCount=derivedCoreProps[i].getRangeCount();
for(range=0; range<rangeCount; ++range) {
start=derivedCoreProps[i].getRangeStart(range);
end=derivedCoreProps[i].getRangeEnd(range);
for(; start<=end; ++start) {
if(u_hasBinaryProperty(start, derivedCorePropsIndex[i])) {
errln("UnicodeTest error: u_hasBinaryProperty(U+%04lx, %s)==TRUE is wrong\n", start, derivedCorePropsNames[i]);
}
}
}
}
}

10
icu4c/source/test/intltest/ucdtest.h
View file

@ -4,6 +4,7 @@
* others. All Rights Reserved.
********************************************************************/
#include "unicode/uniset.h"
#include "intltest.h"
/** Helper function for TestUnicodeData */
@ -65,6 +66,8 @@ public:
**/
void TestScript();
void TestAdditionalProperties();
private:
/**
* internal utility used by TestUnicodeData
@ -78,5 +81,12 @@ private:
friend void U_CALLCONV unicodeDataLineFn(void *context,
char *fields[][2], int32_t fieldCount,
UErrorCode *pErrorCode);
static void U_CALLCONV
derivedCorePropsLineFn(void *context,
char *fields[][2], int32_t fieldCount,
UErrorCode *pErrorCode);
UnicodeSet derivedCoreProps[30];
};