U_CAPI void U_EXPORT2
u_parseDelimitedFile(const char *filename, char delimiter,
char *fields[][2], int32_t fieldCount,
UParseLineFn *lineFn, void *context,
UErrorCode *pErrorCode) {
FileStream *file;
char line[300];
char *start, *limit;
int32_t i, length;
if(U_FAILURE(*pErrorCode)) {
return;
}
if(fields==NULL || lineFn==NULL || fieldCount<=0) {
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
return;
}
if(filename==NULL || *filename==0 || (*filename=='-' && filename[1]==0)) {
filename=NULL;
file=T_FileStream_stdin();
} else {
file=T_FileStream_open(filename, "r");
}
if(file==NULL) {
*pErrorCode=U_FILE_ACCESS_ERROR;
return;
}
while(T_FileStream_readLine(file, line, sizeof(line))!=NULL) {
/* remove trailing newline characters */
length=(int32_t)(u_rtrim(line)-line);
/*
* detect a line with # @missing:
* start parsing after that, or else from the beginning of the line
* set the default warning for @missing lines
*/
start=(char *)getMissingLimit(line);
if(start==line) {
*pErrorCode=U_ZERO_ERROR;
} else {
*pErrorCode=U_USING_DEFAULT_WARNING;
}
/* skip this line if it is empty or a comment */
if(*start==0 || *start=='#') {
continue;
}
/* remove in-line comments */
limit=uprv_strchr(start, '#');
if(limit!=NULL) {
/* get white space before the pound sign */
while(limit>start && U_IS_INV_WHITESPACE(*(limit-1))) {
--limit;
}
/* truncate the line */
*limit=0;
}
/* skip lines with only whitespace */
if(u_skipWhitespace(start)[0]==0) {
continue;
}
/* for each field, call the corresponding field function */
for(i=0; i<fieldCount; ++i) {
/* set the limit pointer of this field */
limit=start;
while(*limit!=delimiter && *limit!=0) {
++limit;
}
/* set the field start and limit in the fields array */
fields[i][0]=start;
fields[i][1]=limit;
/* set start to the beginning of the next field, if any */
start=limit;
if(*start!=0) {
++start;
} else if(i+1<fieldCount) {
*pErrorCode=U_PARSE_ERROR;
limit=line+length;
i=fieldCount;
break;
}
}
/* error in a field function? */
if(U_FAILURE(*pErrorCode)) {
break;
}
/* call the field function */
lineFn(context, fields, fieldCount, pErrorCode);
if(U_FAILURE(*pErrorCode)) {
break;
}
}
if(filename!=NULL) {
T_FileStream_close(file);
}
}
void BiDiConformanceTest::TestBidiTest() {
if(isICUVersionBefore(52, 1)) {
// TODO: Update the ICU BiDi code to implement the additions in the Unicode 6.3 BiDi Algorithm,
// and reenable the BiDi conformance test.
return;
}
IcuTestErrorCode errorCode(*this, "TestBidiTest");
const char *sourceTestDataPath=getSourceTestData(errorCode);
if(errorCode.logIfFailureAndReset("unable to find the source/test/testdata "
"folder (getSourceTestData())")) {
return;
}
char bidiTestPath[400];
strcpy(bidiTestPath, sourceTestDataPath);
strcat(bidiTestPath, "BidiTest.txt");
LocalStdioFilePointer bidiTestFile(fopen(bidiTestPath, "r"));
if(bidiTestFile.isNull()) {
errln("unable to open %s", bidiTestPath);
return;
}
LocalUBiDiPointer ubidi(ubidi_open());
ubidi_setClassCallback(ubidi.getAlias(), biDiConfUBiDiClassCallback, NULL,
NULL, NULL, errorCode);
if(errorCode.logIfFailureAndReset("ubidi_setClassCallback()")) {
return;
}
lineNumber=0;
levelsCount=0;
orderingCount=0;
errorCount=0;
while(errorCount<10 && fgets(line, (int)sizeof(line), bidiTestFile.getAlias())!=NULL) {
++lineNumber;
// Remove trailing comments and whitespace.
char *commentStart=strchr(line, '#');
if(commentStart!=NULL) {
*commentStart=0;
}
u_rtrim(line);
const char *start=u_skipWhitespace(line);
if(*start==0) {
continue; // Skip empty and comment-only lines.
}
if(*start=='@') {
++start;
if(0==strncmp(start, "Levels:", 7)) {
if(!parseLevels(start+7)) {
return;
}
} else if(0==strncmp(start, "Reorder:", 8)) {
if(!parseOrdering(start+8)) {
return;
}
}
// Skip unknown @Xyz: ...
} else {
if(!parseInputStringFromBiDiClasses(start)) {
return;
}
start=u_skipWhitespace(start);
if(*start!=';') {
errln("missing ; separator on input line %s", line);
return;
}
start=u_skipWhitespace(start+1);
char *end;
uint32_t bitset=(uint32_t)strtoul(start, &end, 16);
if(end<=start || (!U_IS_INV_WHITESPACE(*end) && *end!=';' && *end!=0)) {
errln("input bitset parse error at %s", start);
return;
}
// Loop over the bitset.
static const UBiDiLevel paraLevels[]={ UBIDI_DEFAULT_LTR, 0, 1, UBIDI_DEFAULT_RTL };
static const char *const paraLevelNames[]={ "auto/LTR", "LTR", "RTL", "auto/RTL" };
for(int i=0; i<=3; ++i) {
if(bitset&(1<<i)) {
ubidi_setPara(ubidi.getAlias(), inputString.getBuffer(), inputString.length(),
paraLevels[i], NULL, errorCode);
const UBiDiLevel *actualLevels=ubidi_getLevels(ubidi.getAlias(), errorCode);
if(errorCode.logIfFailureAndReset("ubidi_setPara() or ubidi_getLevels()")) {
errln("Input line %d: %s", (int)lineNumber, line);
return;
}
if(!checkLevels(actualLevels, ubidi_getProcessedLength(ubidi.getAlias()),
paraLevelNames[i])) {
// continue outerLoop; does not exist in C++
// so just break out of the inner loop.
break;
}
if(!checkOrdering(ubidi.getAlias(), paraLevelNames[i])) {
// continue outerLoop; does not exist in C++
// so just break out of the inner loop.
break;
}
}
}
}
}
}
/*
*******************************************************************************
*
* created on: 2013jul01
* created by: Matitiahu Allouche
This function performs a conformance test for implementations of the
Unicode Bidirectional Algorithm, specified in UAX #9: Unicode
Bidirectional Algorithm, at http://www.unicode.org/unicode/reports/tr9/
Each test case is represented in a single line which is read from a file
named BidiCharacter.txt. Empty, blank and comment lines may also appear
in this file.
The format of the test data is specified below. Note that each test
case constitutes a single line of text; reordering is applied within a
single line and independently of a rendering engine, and rules L3 and L4
are out of scope.
The number sign '#' is the comment character: everything is ignored from
the occurrence of '#' until the end of the line,
Empty lines and lines containing only spaces and/or comments are ignored.
Lines which represent test cases consist of 4 or 5 fields separated by a
semicolon. Each field consists of tokens separated by whitespace (space
or Tab). Whitespace before and after semicolons is optional.
Field 0: A sequence of hexadecimal code point values separated by space
Field 1: A value representing the paragraph direction, as follows:
- 0 represents left-to-right
- 1 represents right-to-left
- 2 represents auto-LTR according to rules P2 and P3 of the algorithm
- 3 represents auto-RTL according to rules P2 and P3 of the algorithm
- a negative number whose absolute value is taken as paragraph level;
this may be useful to test cases where the embedding level approaches
or exceeds the maximum embedding level.
Field 2: The resolved paragraph embedding level. If the input (field 0)
includes more than one paragraph, this field represents the
resolved level of the first paragraph.
Field 3: An ordered list of resulting levels for each token in field 0
(each token represents one source character).
The UBA does not assign levels to certain characters (e.g. LRO);
characters removed in rule X9 are indicated with an 'x'.
Field 4: An ordered list of indices showing the resulting visual ordering
from left to right; characters with a resolved level of 'x' are
skipped. The number are zero-based. Each index corresponds to
a character in the reordered (visual) string. It represents the
index of the source character in the input (field 0).
This field is optional. When it is absent, the visual ordering
is not verified.
Examples:
# This is a comment line.
L L ON R ; 0 ; 0 ; 0 0 0 1 ; 0 1 2 3
L L ON R;0;0;0 0 0 1;0 1 2 3
# Note: in the next line, 'B' represents a block separator, not the letter 'B'.
LRE A B C PDF;2;0;x 2 0 0 x;1 2 3
# Note: in the next line, 'b' represents the letter 'b', not a block separator.
a b c 05d0 05d1 x ; 0 ; 0 ; 0 0 0 1 1 0 ; 0 1 2 4 3 5
a R R x ; 1 ; 1 ; 2 1 1 2
L L R R R B R R L L L B ON ON ; 3 ; 0 ; 0 0 1 1 1 0 1 1 2 2 2 1 1 1
*
*******************************************************************************
*/
void BiDiConformanceTest::TestBidiCharacterTest() {
IcuTestErrorCode errorCode(*this, "TestBidiCharacterTest");
const char *sourceTestDataPath=getSourceTestData(errorCode);
if(errorCode.logIfFailureAndReset("unable to find the source/test/testdata "
"folder (getSourceTestData())")) {
return;
}
char bidiTestPath[400];
strcpy(bidiTestPath, sourceTestDataPath);
strcat(bidiTestPath, "BidiCharacterTest.txt");
LocalStdioFilePointer bidiTestFile(fopen(bidiTestPath, "r"));
if(bidiTestFile.isNull()) {
errln("unable to open %s", bidiTestPath);
return;
}
LocalUBiDiPointer ubidi(ubidi_open());
lineNumber=0;
levelsCount=0;
orderingCount=0;
errorCount=0;
while(errorCount<20 && fgets(line, (int)sizeof(line), bidiTestFile.getAlias())!=NULL) {
++lineNumber;
paraLevelName="N/A";
inputString="N/A";
// Remove trailing comments and whitespace.
char *commentStart=strchr(line, '#');
if(commentStart!=NULL) {
*commentStart=0;
}
u_rtrim(line);
const char *start=u_skipWhitespace(line);
if(*start==0) {
continue; // Skip empty and comment-only lines.
}
// Parse the code point string in field 0.
UChar *buffer=inputString.getBuffer(200);
int32_t length=u_parseString(start, buffer, inputString.getCapacity(), NULL, errorCode);
if(errorCode.logIfFailureAndReset("Invalid string in field 0")) {
errln("Input line %d: %s", (int)lineNumber, line);
inputString.remove();
continue;
}
inputString.releaseBuffer(length);
start=strchr(start, ';');
if(start==NULL) {
errorCount++;
errln("\nError on line %d: Missing ; separator on line: %s", (int)lineNumber, line);
continue;
}
start=u_skipWhitespace(start+1);
char *end;
int32_t paraDirection=(int32_t)strtol(start, &end, 10);
UBiDiLevel paraLevel=UBIDI_MAX_EXPLICIT_LEVEL+2;
if(paraDirection==0) {
paraLevel=0;
paraLevelName="LTR";
}
else if(paraDirection==1) {
paraLevel=1;
paraLevelName="RTL";
}
else if(paraDirection==2) {
paraLevel=UBIDI_DEFAULT_LTR;
paraLevelName="Auto/LTR";
}
else if(paraDirection==3) {
paraLevel=UBIDI_DEFAULT_RTL;
paraLevelName="Auto/RTL";
}
else if(paraDirection<0 && -paraDirection<=(UBIDI_MAX_EXPLICIT_LEVEL+1)) {
paraLevel=(UBiDiLevel)(-paraDirection);
sprintf(levelNameString, "%d", (int)paraLevel);
paraLevelName=levelNameString;
}
if(end<=start || (!U_IS_INV_WHITESPACE(*end) && *end!=';' && *end!=0) ||
paraLevel==(UBIDI_MAX_EXPLICIT_LEVEL+2)) {
errln("\nError on line %d: Input paragraph direction incorrect at %s", (int)lineNumber, start);
printErrorLine();
continue;
}
start=u_skipWhitespace(end);
if(*start!=';') {
errorCount++;
errln("\nError on line %d: Missing ; separator on line: %s", (int)lineNumber, line);
continue;
}
start++;
uint32_t resolvedParaLevel=(uint32_t)strtoul(start, &end, 10);
if(end<=start || (!U_IS_INV_WHITESPACE(*end) && *end!=';' && *end!=0) ||
resolvedParaLevel>1) {
errln("\nError on line %d: Resolved paragraph level incorrect at %s", (int)lineNumber, start);
printErrorLine();
continue;
}
start=u_skipWhitespace(end);
if(*start!=';') {
errorCount++;
errln("\nError on line %d: Missing ; separator on line: %s", (int)lineNumber, line);
return;
}
start++;
if(!parseLevels(start)) {
continue;
}
start=u_skipWhitespace(start);
if(*start==';') {
if(!parseOrdering(start+1)) {
continue;
}
}
else
orderingCount=-1;
ubidi_setPara(ubidi.getAlias(), inputString.getBuffer(), inputString.length(),
paraLevel, NULL, errorCode);
const UBiDiLevel *actualLevels=ubidi_getLevels(ubidi.getAlias(), errorCode);
if(errorCode.logIfFailureAndReset("ubidi_setPara() or ubidi_getLevels()")) {
errln("Input line %d: %s", (int)lineNumber, line);
continue;
}
UBiDiLevel actualLevel;
if((actualLevel=ubidi_getParaLevel(ubidi.getAlias()))!=resolvedParaLevel) {
printErrorLine();
errln("\nError on line %d: Wrong resolved paragraph level; expected %d actual %d",
(int)lineNumber, resolvedParaLevel, actualLevel);
continue;
}
if(!checkLevels(actualLevels, ubidi_getProcessedLength(ubidi.getAlias()))) {
continue;
}
if(orderingCount>=0 && !checkOrdering(ubidi.getAlias())) {
continue;
}
}
}
void parseFile(FILE *f, Normalizer2DataBuilder &builder) {
IcuToolErrorCode errorCode("gennorm2/parseFile()");
char line[300];
uint32_t startCP, endCP;
while(NULL!=fgets(line, (int)sizeof(line), f)) {
char *comment=(char *)strchr(line, '#');
if(comment!=NULL) {
*comment=0;
}
u_rtrim(line);
if(line[0]==0) {
continue; // skip empty and comment-only lines
}
if(line[0]=='*') {
const char *s=u_skipWhitespace(line+1);
if(0==strncmp(s, "Unicode", 7)) {
s=u_skipWhitespace(s+7);
builder.setUnicodeVersion(s);
}
continue; // reserved syntax
}
const char *delimiter;
int32_t rangeLength=
u_parseCodePointRangeAnyTerminator(line, &startCP, &endCP, &delimiter, errorCode);
if(errorCode.isFailure()) {
fprintf(stderr, "gennorm2 error: parsing code point range from %s\n", line);
exit(errorCode.reset());
}
delimiter=u_skipWhitespace(delimiter);
if(*delimiter==':') {
const char *s=u_skipWhitespace(delimiter+1);
char *end;
unsigned long value=strtoul(s, &end, 10);
if(end<=s || *u_skipWhitespace(end)!=0 || value>=0xff) {
fprintf(stderr, "gennorm2 error: parsing ccc from %s\n", line);
exit(U_PARSE_ERROR);
}
for(UChar32 c=(UChar32)startCP; c<=(UChar32)endCP; ++c) {
builder.setCC(c, (uint8_t)value);
}
continue;
}
if(*delimiter=='-') {
if(*u_skipWhitespace(delimiter+1)!=0) {
fprintf(stderr, "gennorm2 error: parsing remove-mapping %s\n", line);
exit(U_PARSE_ERROR);
}
for(UChar32 c=(UChar32)startCP; c<=(UChar32)endCP; ++c) {
builder.removeMapping(c);
}
continue;
}
if(*delimiter=='=' || *delimiter=='>') {
UChar uchars[Normalizer2Impl::MAPPING_LENGTH_MASK];
int32_t length=u_parseString(delimiter+1, uchars, LENGTHOF(uchars), NULL, errorCode);
if(errorCode.isFailure()) {
fprintf(stderr, "gennorm2 error: parsing mapping string from %s\n", line);
exit(errorCode.reset());
}
UnicodeString mapping(FALSE, uchars, length);
if(*delimiter=='=') {
if(rangeLength!=1) {
fprintf(stderr,
"gennorm2 error: round-trip mapping for more than 1 code point on %s\n",
line);
exit(U_PARSE_ERROR);
}
builder.setRoundTripMapping((UChar32)startCP, mapping);
} else {
for(UChar32 c=(UChar32)startCP; c<=(UChar32)endCP; ++c) {
builder.setOneWayMapping(c, mapping);
}
}
continue;
}
fprintf(stderr, "gennorm2 error: unrecognized data line %s\n", line);
exit(U_PARSE_ERROR);
}
}
void UCAConformanceTest::testConformance(const Collator *coll)
{
if(testFile == 0) {
return;
}
uint32_t skipFlags = 0;
if(coll->getAttribute(UCOL_ALTERNATE_HANDLING, status) == UCOL_SHIFTED) {
skipFlags |= IS_SHIFTED;
}
if(coll == rbUCA) {
skipFlags |= FROM_RULES;
}
logln("-prop:ucaconfnosortkeys=1 turns off getSortKey() in UCAConformanceTest");
UBool withSortKeys = getProperty("ucaconfnosortkeys") == NULL;
int32_t line = 0;
UChar b1[1024], b2[1024];
UChar *buffer = b1, *oldB = NULL;
char lineB1[1024], lineB2[1024];
char *lineB = lineB1, *oldLineB = lineB2;
uint8_t sk1[1024], sk2[1024];
uint8_t *oldSk = NULL, *newSk = sk1;
int32_t oldLen = 0;
int32_t oldBlen = 0;
uint32_t first = 0;
while (fgets(lineB, 1024, testFile) != NULL) {
// remove trailing whitespace
u_rtrim(lineB);
line++;
if(*lineB == 0 || lineB[0] == '#') {
continue;
}
int32_t buflen = u_parseString(lineB, buffer, 1024, &first, &status);
if(U_FAILURE(status)) {
errln("Error parsing line %ld (%s): %s\n",
(long)line, u_errorName(status), lineB);
status = U_ZERO_ERROR;
}
buffer[buflen] = 0;
if(skipLineBecauseOfBug(buffer, buflen, skipFlags)) {
logln("Skipping line %i because of a known bug", line);
continue;
}
int32_t resLen = withSortKeys ? coll->getSortKey(buffer, buflen, newSk, 1024) : 0;
if(oldSk != NULL) {
UBool ok=TRUE;
int32_t skres = withSortKeys ? strcmp((char *)oldSk, (char *)newSk) : 0;
int32_t cmpres = coll->compare(oldB, oldBlen, buffer, buflen, status);
int32_t cmpres2 = coll->compare(buffer, buflen, oldB, oldBlen, status);
if(cmpres != -cmpres2) {
errln("Compare result not symmetrical on line %i: "
"previous vs. current (%d) / current vs. previous (%d)",
line, cmpres, cmpres2);
ok = FALSE;
}
// TODO: Compare with normalization turned off if the input passes the FCD test.
if(withSortKeys && cmpres != normalizeResult(skres)) {
errln("Difference between coll->compare (%d) and sortkey compare (%d) on line %i",
cmpres, skres, line);
ok = FALSE;
}
int32_t res = cmpres;
if(res == 0 && !isAtLeastUCA62) {
// Up to UCA 6.1, the collation test files use a custom tie-breaker,
// comparing the raw input strings.
res = u_strcmpCodePointOrder(oldB, buffer);
// Starting with UCA 6.2, the collation test files use the standard UCA tie-breaker,
// comparing the NFD versions of the input strings,
// which we do via setting strength=identical.
}
if(res > 0) {
errln("Line %i is not greater or equal than previous line", line);
ok = FALSE;
}
if(!ok) {
errln(" Previous data line %s", oldLineB);
errln(" Current data line %s", lineB);
if(withSortKeys) {
UnicodeString oldS, newS;
prettify(CollationKey(oldSk, oldLen), oldS);
prettify(CollationKey(newSk, resLen), newS);
errln(" Previous key: "+oldS);
errln(" Current key: "+newS);
}
}
}
// swap buffers
oldLineB = lineB;
oldB = buffer;
oldSk = newSk;
if(lineB == lineB1) {
lineB = lineB2;
buffer = b2;
newSk = sk2;
} else {
lineB = lineB1;
buffer = b1;
newSk = sk1;
}
oldLen = resLen;
oldBlen = buflen;
}
}
