#include "unicode/utypes.h"
#include "unicode/uchar.h"
#include "unicode/ucnv.h"
#include "unicode/uniset.h"
#include "unicode/unistr.h"
#include "unicode/uclean.h"
#include "unicode/udata.h"
#include "unicode/putil.h"
#include "uoptions.h"
#include "unewdata.h"
#include "ucmndata.h"
#include "rbbidata.h"
#include "triedict.h"
#include "cmemory.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
U_NAMESPACE_USE
static char *progName;
static UOption options[]={
UOPTION_HELP_H,
UOPTION_HELP_QUESTION_MARK,
UOPTION_VERBOSE,
{ "out", NULL, NULL, NULL, 'o', UOPT_REQUIRES_ARG, 0 },
UOPTION_ICUDATADIR,
UOPTION_DESTDIR,
UOPTION_COPYRIGHT,
};
void usageAndDie(int retCode) {
printf("Usage: %s [-v] [-options] -o output-file dictionary-file\n", progName);
printf("\tRead in word list and write out compact trie dictionary\n"
"options:\n"
"\t-h or -? or --help this usage text\n"
"\t-V or --version show a version message\n"
"\t-c or --copyright include a copyright notice\n"
"\t-v or --verbose turn on verbose output\n"
"\t-i or --icudatadir directory for locating any needed intermediate data files,\n"
"\t followed by path, defaults to %s\n"
"\t-d or --destdir destination directory, followed by the path\n",
u_getDataDirectory());
exit (retCode);
}
#if UCONFIG_NO_BREAK_ITERATION || UCONFIG_NO_FILE_IO
static UDataInfo dummyDataInfo = {
sizeof(UDataInfo),
0,
U_IS_BIG_ENDIAN,
U_CHARSET_FAMILY,
U_SIZEOF_UCHAR,
0,
{ 0, 0, 0, 0 },
{ 0, 0, 0, 0 },
{ 0, 0, 0, 0 }
};
#else
DataHeader dh ={
{sizeof(DataHeader), 0xda,
0x27},
{ sizeof(UDataInfo), 0, U_IS_BIG_ENDIAN,
U_CHARSET_FAMILY,
U_SIZEOF_UCHAR,
0,
{ 0x54, 0x72, 0x44, 0x63 }, { 1, 0, 0, 0 }, { 0, 0, 0, 0 }, }};
#endif
int main(int argc, char **argv) {
UErrorCode status = U_ZERO_ERROR;
const char *wordFileName;
const char *outFileName;
const char *outDir = NULL;
const char *copyright = NULL;
U_MAIN_INIT_ARGS(argc, argv);
progName = argv[0];
argc=u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
if(argc<0) {
fprintf(stderr, "error in command line argument \"%s\"\n", argv[-argc]);
usageAndDie(U_ILLEGAL_ARGUMENT_ERROR);
}
if(options[0].doesOccur || options[1].doesOccur) {
usageAndDie(0);
}
if (!options[3].doesOccur || argc < 2) {
fprintf(stderr, "input and output file must both be specified.\n");
usageAndDie(U_ILLEGAL_ARGUMENT_ERROR);
}
outFileName = options[3].value;
wordFileName = argv[1];
if (options[4].doesOccur) {
u_setDataDirectory(options[4].value);
}
status = U_ZERO_ERROR;
if(options[5].doesOccur) {
outDir = options[5].value;
}
if (options[6].doesOccur) {
copyright = U_COPYRIGHT_STRING;
}
#if UCONFIG_NO_BREAK_ITERATION || UCONFIG_NO_FILE_IO
UNewDataMemory *pData;
char msg[1024];
sprintf(msg, "genctd writes dummy %s because of UCONFIG_NO_BREAK_ITERATION and/or UCONFIG_NO_FILE_IO, see uconfig.h", outFileName);
fprintf(stderr, "%s\n", msg);
pData = udata_create(outDir, NULL, outFileName, &dummyDataInfo, NULL, &status);
udata_writeBlock(pData, msg, strlen(msg));
udata_finish(pData, &status);
return (int)status;
#else
u_init(&status);
if (U_FAILURE(status)) {
fprintf(stderr, "%s: can not initialize ICU. status = %s\n",
argv[0], u_errorName(status));
exit(1);
}
status = U_ZERO_ERROR;
long result;
long wordFileSize;
FILE *file;
char *wordBufferC;
file = fopen(wordFileName, "rb");
if( file == 0 ) {
fprintf(stderr, "Could not open file \"%s\"\n", wordFileName);
exit(-1);
}
fseek(file, 0, SEEK_END);
wordFileSize = ftell(file);
fseek(file, 0, SEEK_SET);
wordBufferC = new char[wordFileSize+10];
result = (long)fread(wordBufferC, 1, wordFileSize, file);
if (result != wordFileSize) {
fprintf(stderr, "Error reading file \"%s\"\n", wordFileName);
exit (-1);
}
wordBufferC[wordFileSize]=0;
fclose(file);
int32_t signatureLength;
const char * wordSourceC = wordBufferC;
const char* encoding = ucnv_detectUnicodeSignature(
wordSourceC, wordFileSize, &signatureLength, &status);
if (U_FAILURE(status)) {
exit(status);
}
if(encoding!=NULL ){
wordSourceC += signatureLength;
wordFileSize -= signatureLength;
}
UConverter* conv;
conv = ucnv_open(encoding, &status);
if (U_FAILURE(status)) {
fprintf(stderr, "ucnv_open: ICU Error \"%s\"\n", u_errorName(status));
exit(status);
}
uint32_t destCap = ucnv_toUChars(conv,
NULL, 0, wordSourceC,
wordFileSize,
&status);
if (status != U_BUFFER_OVERFLOW_ERROR) {
fprintf(stderr, "ucnv_toUChars: ICU Error \"%s\"\n", u_errorName(status));
exit(status);
};
status = U_ZERO_ERROR;
UChar *wordSourceU = new UChar[destCap+1];
ucnv_toUChars(conv,
wordSourceU, destCap+1,
wordSourceC,
wordFileSize,
&status);
if (U_FAILURE(status)) {
fprintf(stderr, "ucnv_toUChars: ICU Error \"%s\"\n", u_errorName(status));
exit(status);
};
ucnv_close(conv);
delete[] wordBufferC;
UChar *current = wordSourceU + (destCap/2);
UChar uc = *current++;
UnicodeSet breaks;
breaks.add(0x000A); breaks.add(0x000D); breaks.add(0x2028); breaks.add(0x2029);
do {
while (uc && !breaks.contains(uc)) {
uc = *current++;
}
while (uc && breaks.contains(uc)) {
uc = *current++;
}
}
while (uc && (breaks.contains(uc) || u_isspace(uc)));
MutableTrieDictionary *mtd = new MutableTrieDictionary(uc, status);
if (U_FAILURE(status)) {
fprintf(stderr, "new MutableTrieDictionary: ICU Error \"%s\"\n", u_errorName(status));
exit(status);
}
current = wordSourceU;
UChar *candidate = current;
uc = *current++;
int32_t length = 0;
while (uc) {
while (uc && !u_isspace(uc)) {
++length;
uc = *current++;
}
if (length > 0) {
mtd->addWord(candidate, length, status);
if (U_FAILURE(status)) {
fprintf(stderr, "MutableTrieDictionary::addWord: ICU Error \"%s\"\n",
u_errorName(status));
exit(status);
}
}
while (uc && !breaks.contains(uc)) {
uc = *current++;
}
while (uc && breaks.contains(uc)) {
uc = *current++;
}
candidate = current-1;
length = 0;
}
delete[] wordSourceU;
CompactTrieDictionary *ctd = new CompactTrieDictionary(*mtd, status);
if (U_FAILURE(status)) {
fprintf(stderr, "new CompactTrieDictionary: ICU Error \"%s\"\n", u_errorName(status));
exit(status);
}
delete mtd;
uint32_t outDataSize = ctd->dataSize();
const uint8_t *outData = (const uint8_t *)ctd->data();
size_t bytesWritten;
UNewDataMemory *pData;
pData = udata_create(outDir, NULL, outFileName, &(dh.info), copyright, &status);
if(U_FAILURE(status)) {
fprintf(stderr, "genctd: Could not open output file \"%s\", \"%s\"\n",
outFileName, u_errorName(status));
exit(status);
}
udata_writeBlock(pData, outData, outDataSize);
bytesWritten = udata_finish(pData, &status);
if(U_FAILURE(status)) {
fprintf(stderr, "genctd: error \"%s\" writing the output file\n", u_errorName(status));
exit(status);
}
if (bytesWritten != outDataSize) {
fprintf(stderr, "Error writing to output file \"%s\"\n", outFileName);
exit(-1);
}
delete ctd;
u_cleanup();
printf("genctd: tool completed successfully.\n");
return 0;
#endif
}