michael@0: /*
michael@0: *******************************************************************************
michael@0: *
michael@0: *   Copyright (C) 2003-2012, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *
michael@0: *******************************************************************************
michael@0: *   file name:  gensprep.c
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2003-02-06
michael@0: *   created by: Ram Viswanadha
michael@0: *
michael@0: *   This program reads the Profile.txt files,
michael@0: *   parses them, and extracts the data for StringPrep profile.
michael@0: *   It then preprocesses it and writes a binary file for efficient use
michael@0: *   in various StringPrep conversion processes.
michael@0: */
michael@0: 
michael@0: #define USPREP_TYPE_NAMES_ARRAY 1
michael@0: 
michael@0: #include <stdio.h>
michael@0: #include <stdlib.h>
michael@0: 
michael@0: #include "cmemory.h"
michael@0: #include "cstring.h"
michael@0: #include "unewdata.h"
michael@0: #include "uoptions.h"
michael@0: #include "uparse.h"
michael@0: #include "sprpimpl.h"
michael@0: 
michael@0: #include "unicode/uclean.h"
michael@0: #include "unicode/udata.h"
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/putil.h"
michael@0: 
michael@0: 
michael@0: U_CDECL_BEGIN
michael@0: #include "gensprep.h"
michael@0: U_CDECL_END
michael@0: 
michael@0: UBool beVerbose=FALSE, haveCopyright=TRUE;
michael@0: 
michael@0: #define NORM_CORRECTIONS_FILE_NAME "NormalizationCorrections.txt"
michael@0: 
michael@0: #define NORMALIZE_DIRECTIVE "normalize"
michael@0: #define NORMALIZE_DIRECTIVE_LEN 9
michael@0: #define CHECK_BIDI_DIRECTIVE "check-bidi"
michael@0: #define CHECK_BIDI_DIRECTIVE_LEN 10
michael@0: 
michael@0: /* prototypes --------------------------------------------------------------- */
michael@0: 
michael@0: static void
michael@0: parseMappings(const char *filename, UBool reportError, UErrorCode *pErrorCode);
michael@0: 
michael@0: static void
michael@0: parseNormalizationCorrections(const char *filename, UErrorCode *pErrorCode);
michael@0: 
michael@0: 
michael@0: /* -------------------------------------------------------------------------- */
michael@0: 
michael@0: static UOption options[]={
michael@0:     UOPTION_HELP_H,
michael@0:     UOPTION_HELP_QUESTION_MARK,
michael@0:     UOPTION_VERBOSE,
michael@0:     UOPTION_COPYRIGHT,
michael@0:     UOPTION_DESTDIR,
michael@0:     UOPTION_SOURCEDIR,
michael@0:     UOPTION_ICUDATADIR,
michael@0:     UOPTION_BUNDLE_NAME,
michael@0:     { "normalization", NULL, NULL, NULL, 'n', UOPT_REQUIRES_ARG, 0 },
michael@0:     { "norm-correction", NULL, NULL, NULL, 'm', UOPT_REQUIRES_ARG, 0 },
michael@0:     { "check-bidi", NULL, NULL, NULL,  'k', UOPT_NO_ARG, 0},
michael@0:     { "unicode", NULL, NULL, NULL, 'u', UOPT_REQUIRES_ARG, 0 },
michael@0: };
michael@0: 
michael@0: enum{
michael@0:     HELP,
michael@0:     HELP_QUESTION_MARK,
michael@0:     VERBOSE,
michael@0:     COPYRIGHT,
michael@0:     DESTDIR,
michael@0:     SOURCEDIR,
michael@0:     ICUDATADIR,
michael@0:     BUNDLE_NAME,
michael@0:     NORMALIZE,
michael@0:     NORM_CORRECTION_DIR,
michael@0:     CHECK_BIDI,
michael@0:     UNICODE_VERSION
michael@0: };
michael@0: 
michael@0: static int printHelp(int argc, char* argv[]){
michael@0:     /*
michael@0:      * Broken into chucks because the C89 standard says the minimum
michael@0:      * required supported string length is 509 bytes.
michael@0:      */
michael@0:     fprintf(stderr,
michael@0:         "Usage: %s [-options] [file_name]\n"
michael@0:         "\n"
michael@0:         "Read the files specified and\n"
michael@0:         "create a binary file [package-name]_[bundle-name]." DATA_TYPE " with the StringPrep profile data\n"
michael@0:         "\n",
michael@0:         argv[0]);
michael@0:     fprintf(stderr,
michael@0:         "Options:\n"
michael@0:         "\t-h or -? or --help       print this usage text\n"
michael@0:         "\t-v or --verbose          verbose output\n"
michael@0:         "\t-c or --copyright        include a copyright notice\n");
michael@0:     fprintf(stderr,
michael@0:         "\t-d or --destdir          destination directory, followed by the path\n"
michael@0:         "\t-s or --sourcedir        source directory of ICU data, followed by the path\n"
michael@0:         "\t-b or --bundle-name      generate the ouput data file with the name specified\n"
michael@0:         "\t-i or --icudatadir       directory for locating any needed intermediate data files,\n"
michael@0:         "\t                         followed by path, defaults to %s\n",
michael@0:         u_getDataDirectory());
michael@0:     fprintf(stderr,
michael@0:         "\t-n or --normalize        turn on the option for normalization and include mappings\n"
michael@0:         "\t                         from NormalizationCorrections.txt from the given path,\n"
michael@0:         "\t                         e.g: /test/icu/source/data/unidata\n");
michael@0:     fprintf(stderr,
michael@0:         "\t-m or --norm-correction  use NormalizationCorrections.txt from the given path\n"
michael@0:         "\t                         when the input file contains a normalization directive.\n"
michael@0:         "\t                         unlike -n/--normalize, this option does not force the\n"
michael@0:         "\t                         normalization.\n");
michael@0:     fprintf(stderr,
michael@0:         "\t-k or --check-bidi       turn on the option for checking for BiDi in the profile\n"
michael@0:         "\t-u or --unicode          version of Unicode to be used with this profile followed by the version\n"
michael@0:         );
michael@0:     return argc<0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
michael@0: }
michael@0: 
michael@0: 
michael@0: extern int
michael@0: main(int argc, char* argv[]) {
michael@0: #if !UCONFIG_NO_IDNA
michael@0:     char* filename = NULL;
michael@0: #endif
michael@0:     const char *srcDir=NULL, *destDir=NULL, *icuUniDataDir=NULL;
michael@0:     const char *bundleName=NULL, *inputFileName = NULL;
michael@0:     char *basename=NULL;
michael@0:     int32_t sprepOptions = 0;
michael@0: 
michael@0:     UErrorCode errorCode=U_ZERO_ERROR;
michael@0: 
michael@0:     U_MAIN_INIT_ARGS(argc, argv);
michael@0: 
michael@0:     /* preset then read command line options */
michael@0:     options[DESTDIR].value=u_getDataDirectory();
michael@0:     options[SOURCEDIR].value="";
michael@0:     options[UNICODE_VERSION].value="0"; /* don't assume the unicode version */
michael@0:     options[BUNDLE_NAME].value = DATA_NAME;
michael@0:     options[NORMALIZE].value = "";
michael@0: 
michael@0:     argc=u_parseArgs(argc, argv, sizeof(options)/sizeof(options[0]), options);
michael@0: 
michael@0:     /* error handling, printing usage message */
michael@0:     if(argc<0) {
michael@0:         fprintf(stderr,
michael@0:             "error in command line argument \"%s\"\n",
michael@0:             argv[-argc]);
michael@0:     }
michael@0:     if(argc<0 || options[HELP].doesOccur || options[HELP_QUESTION_MARK].doesOccur) {
michael@0:         return printHelp(argc, argv);
michael@0:         
michael@0:     }
michael@0: 
michael@0:     /* get the options values */
michael@0:     beVerbose=options[VERBOSE].doesOccur;
michael@0:     haveCopyright=options[COPYRIGHT].doesOccur;
michael@0:     srcDir=options[SOURCEDIR].value;
michael@0:     destDir=options[DESTDIR].value;
michael@0:     bundleName = options[BUNDLE_NAME].value;
michael@0:     if(options[NORMALIZE].doesOccur) {
michael@0:         icuUniDataDir = options[NORMALIZE].value;
michael@0:     } else {
michael@0:         icuUniDataDir = options[NORM_CORRECTION_DIR].value;
michael@0:     }
michael@0: 
michael@0:     if(argc<2) {
michael@0:         /* print the help message */
michael@0:         return printHelp(argc, argv);
michael@0:     } else {
michael@0:         inputFileName = argv[1];
michael@0:     }
michael@0:     if(!options[UNICODE_VERSION].doesOccur){
michael@0:         return printHelp(argc, argv);
michael@0:     }
michael@0:     if(options[ICUDATADIR].doesOccur) {
michael@0:         u_setDataDirectory(options[ICUDATADIR].value);
michael@0:     }
michael@0: #if UCONFIG_NO_IDNA
michael@0: 
michael@0:     fprintf(stderr,
michael@0:         "gensprep writes dummy " U_ICUDATA_NAME "_" DATA_NAME "." DATA_TYPE
michael@0:         " because UCONFIG_NO_IDNA is set, \n"
michael@0:         "see icu/source/common/unicode/uconfig.h\n");
michael@0:     generateData(destDir, bundleName);
michael@0: 
michael@0: #else
michael@0: 
michael@0:     setUnicodeVersion(options[UNICODE_VERSION].value);
michael@0:     filename = (char* ) uprv_malloc(uprv_strlen(srcDir) + 300); /* hopefully this should be enough */
michael@0:    
michael@0:     /* prepare the filename beginning with the source dir */
michael@0:     if(uprv_strchr(srcDir,U_FILE_SEP_CHAR) == NULL && uprv_strchr(srcDir,U_FILE_ALT_SEP_CHAR) == NULL){
michael@0:         filename[0] = '.';
michael@0:         filename[1] = U_FILE_SEP_CHAR;
michael@0:         uprv_strcpy(filename+2,srcDir);
michael@0:     }else{
michael@0:         uprv_strcpy(filename, srcDir);
michael@0:     }
michael@0:     
michael@0:     basename=filename+uprv_strlen(filename);
michael@0:     if(basename>filename && *(basename-1)!=U_FILE_SEP_CHAR) {
michael@0:         *basename++=U_FILE_SEP_CHAR;
michael@0:     }
michael@0:     
michael@0:     /* initialize */
michael@0:     init();
michael@0: 
michael@0:     /* process the file */
michael@0:     uprv_strcpy(basename,inputFileName);
michael@0:     parseMappings(filename,FALSE, &errorCode);
michael@0:     if(U_FAILURE(errorCode)) {
michael@0:         fprintf(stderr, "Could not open file %s for reading. Error: %s \n", filename, u_errorName(errorCode));
michael@0:         return errorCode;
michael@0:     }
michael@0:     
michael@0:     if(options[NORMALIZE].doesOccur){ /* this option might be set by @normalize;; in the source file */
michael@0:         /* set up directory for NormalizationCorrections.txt */
michael@0:         uprv_strcpy(filename,icuUniDataDir);
michael@0:         basename=filename+uprv_strlen(filename);
michael@0:         if(basename>filename && *(basename-1)!=U_FILE_SEP_CHAR) {
michael@0:             *basename++=U_FILE_SEP_CHAR;
michael@0:         }
michael@0: 
michael@0:         *basename++=U_FILE_SEP_CHAR;
michael@0:         uprv_strcpy(basename,NORM_CORRECTIONS_FILE_NAME);
michael@0:     
michael@0:         parseNormalizationCorrections(filename,&errorCode);
michael@0:         if(U_FAILURE(errorCode)){
michael@0:             fprintf(stderr,"Could not open file %s for reading \n", filename);
michael@0:             return errorCode;
michael@0:         }
michael@0:         sprepOptions |= _SPREP_NORMALIZATION_ON;
michael@0:     }
michael@0:     
michael@0:     if(options[CHECK_BIDI].doesOccur){ /* this option might be set by @check-bidi;; in the source file */
michael@0:         sprepOptions |= _SPREP_CHECK_BIDI_ON;
michael@0:     }
michael@0: 
michael@0:     setOptions(sprepOptions);
michael@0: 
michael@0:     /* process parsed data */
michael@0:     if(U_SUCCESS(errorCode)) {
michael@0:         /* write the data file */
michael@0:         generateData(destDir, bundleName);
michael@0: 
michael@0:         cleanUpData();
michael@0:     }
michael@0: 
michael@0:     uprv_free(filename);
michael@0: 
michael@0:     u_cleanup();
michael@0: 
michael@0: #endif
michael@0: 
michael@0:     return errorCode;
michael@0: }
michael@0: 
michael@0: #if !UCONFIG_NO_IDNA
michael@0: 
michael@0: static void U_CALLCONV
michael@0: normalizationCorrectionsLineFn(void *context,
michael@0:                     char *fields[][2], int32_t fieldCount,
michael@0:                     UErrorCode *pErrorCode) {
michael@0:     uint32_t mapping[40];
michael@0:     char *end, *s;
michael@0:     uint32_t code;
michael@0:     int32_t length;
michael@0:     UVersionInfo version;
michael@0:     UVersionInfo thisVersion;
michael@0: 
michael@0:     /* get the character code, field 0 */
michael@0:     code=(uint32_t)uprv_strtoul(fields[0][0], &end, 16);
michael@0:     if(U_FAILURE(*pErrorCode)) {
michael@0:         fprintf(stderr, "gensprep: error parsing NormalizationCorrections.txt mapping at %s\n", fields[0][0]);
michael@0:         exit(*pErrorCode);
michael@0:     }
michael@0:     /* Original (erroneous) decomposition */
michael@0:     s = fields[1][0];
michael@0: 
michael@0:     /* parse the mapping string */
michael@0:     length=u_parseCodePoints(s, mapping, sizeof(mapping)/4, pErrorCode);
michael@0: 
michael@0:     /* ignore corrected decomposition */
michael@0: 
michael@0:     u_versionFromString(version,fields[3][0] );
michael@0:     u_versionFromString(thisVersion, "3.2.0");
michael@0: 
michael@0: 
michael@0: 
michael@0:     if(U_FAILURE(*pErrorCode)) {
michael@0:         fprintf(stderr, "gensprep error parsing NormalizationCorrections.txt of U+%04lx - %s\n",
michael@0:                 (long)code, u_errorName(*pErrorCode));
michael@0:         exit(*pErrorCode);
michael@0:     }
michael@0: 
michael@0:     /* store the mapping */
michael@0:     if( version[0] > thisVersion[0] || 
michael@0:         ((version[0]==thisVersion[0]) && (version[1] > thisVersion[1]))
michael@0:         ){
michael@0:         storeMapping(code,mapping, length, USPREP_MAP, pErrorCode);
michael@0:     }
michael@0:     setUnicodeVersionNC(version);
michael@0: }
michael@0: 
michael@0: static void
michael@0: parseNormalizationCorrections(const char *filename, UErrorCode *pErrorCode) {
michael@0:     char *fields[4][2];
michael@0: 
michael@0:     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     u_parseDelimitedFile(filename, ';', fields, 4, normalizationCorrectionsLineFn, NULL, pErrorCode);
michael@0: 
michael@0:     /* fprintf(stdout,"Number of code points that have NormalizationCorrections mapping with length >1 : %i\n",len); */
michael@0: 
michael@0:     if(U_FAILURE(*pErrorCode) && ( *pErrorCode!=U_FILE_ACCESS_ERROR)) {
michael@0:         fprintf(stderr, "gensprep error: u_parseDelimitedFile(\"%s\") failed - %s\n", filename, u_errorName(*pErrorCode));
michael@0:         exit(*pErrorCode);
michael@0:     }
michael@0: }
michael@0: 
michael@0: static void U_CALLCONV
michael@0: strprepProfileLineFn(void *context,
michael@0:               char *fields[][2], int32_t fieldCount,
michael@0:               UErrorCode *pErrorCode) {
michael@0:     uint32_t mapping[40];
michael@0:     char *end, *map;
michael@0:     uint32_t code;
michael@0:     int32_t length;
michael@0:    /*UBool* mapWithNorm = (UBool*) context;*/
michael@0:     const char* typeName;
michael@0:     uint32_t rangeStart=0,rangeEnd =0;
michael@0:     const char* filename = (const char*) context;
michael@0:     const char *s;
michael@0: 
michael@0:     s = u_skipWhitespace(fields[0][0]);
michael@0:     if (*s == '@') {
michael@0:         /* special directive */
michael@0:         s++;
michael@0:         length = fields[0][1] - s;
michael@0:         if (length >= NORMALIZE_DIRECTIVE_LEN
michael@0:             && uprv_strncmp(s, NORMALIZE_DIRECTIVE, NORMALIZE_DIRECTIVE_LEN) == 0) {
michael@0:             options[NORMALIZE].doesOccur = TRUE;
michael@0:             return;
michael@0:         }
michael@0:         else if (length >= CHECK_BIDI_DIRECTIVE_LEN
michael@0:             && uprv_strncmp(s, CHECK_BIDI_DIRECTIVE, CHECK_BIDI_DIRECTIVE_LEN) == 0) {
michael@0:             options[CHECK_BIDI].doesOccur = TRUE;
michael@0:             return;
michael@0:         }
michael@0:         else {
michael@0:             fprintf(stderr, "gensprep error parsing a directive %s.", fields[0][0]);
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     typeName = fields[2][0];
michael@0:     map = fields[1][0];
michael@0: 
michael@0:     if(uprv_strstr(typeName, usprepTypeNames[USPREP_UNASSIGNED])!=NULL){
michael@0: 
michael@0:         u_parseCodePointRange(s, &rangeStart,&rangeEnd, pErrorCode);
michael@0:         if(U_FAILURE(*pErrorCode)){
michael@0:             fprintf(stderr, "Could not parse code point range. Error: %s\n",u_errorName(*pErrorCode));
michael@0:             return;
michael@0:         }
michael@0: 
michael@0:         /* store the range */
michael@0:         storeRange(rangeStart,rangeEnd,USPREP_UNASSIGNED, pErrorCode);
michael@0: 
michael@0:     }else if(uprv_strstr(typeName, usprepTypeNames[USPREP_PROHIBITED])!=NULL){
michael@0: 
michael@0:         u_parseCodePointRange(s, &rangeStart,&rangeEnd, pErrorCode);
michael@0:         if(U_FAILURE(*pErrorCode)){
michael@0:             fprintf(stderr, "Could not parse code point range. Error: %s\n",u_errorName(*pErrorCode));
michael@0:             return;
michael@0:         }
michael@0: 
michael@0:         /* store the range */
michael@0:         storeRange(rangeStart,rangeEnd,USPREP_PROHIBITED, pErrorCode);
michael@0: 
michael@0:     }else if(uprv_strstr(typeName, usprepTypeNames[USPREP_MAP])!=NULL){
michael@0: 
michael@0:         /* get the character code, field 0 */
michael@0:         code=(uint32_t)uprv_strtoul(s, &end, 16);
michael@0:         if(end<=s || end!=fields[0][1]) {
michael@0:             fprintf(stderr, "gensprep: syntax error in field 0 at %s\n", fields[0][0]);
michael@0:             *pErrorCode=U_PARSE_ERROR;
michael@0:             exit(U_PARSE_ERROR);
michael@0:         }
michael@0: 
michael@0:         /* parse the mapping string */
michael@0:         length=u_parseCodePoints(map, mapping, sizeof(mapping)/4, pErrorCode);
michael@0:         
michael@0:         /* store the mapping */
michael@0:         storeMapping(code,mapping, length,USPREP_MAP, pErrorCode);
michael@0: 
michael@0:     }else{
michael@0:         *pErrorCode = U_INVALID_FORMAT_ERROR;
michael@0:     }
michael@0:     
michael@0:     if(U_FAILURE(*pErrorCode)) {
michael@0:         fprintf(stderr, "gensprep error parsing  %s line %s at %s. Error: %s\n",filename,
michael@0:                fields[0][0],fields[2][0],u_errorName(*pErrorCode));
michael@0:         exit(*pErrorCode);
michael@0:     }
michael@0: 
michael@0: }
michael@0: 
michael@0: static void
michael@0: parseMappings(const char *filename, UBool reportError, UErrorCode *pErrorCode) {
michael@0:     char *fields[3][2];
michael@0: 
michael@0:     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     u_parseDelimitedFile(filename, ';', fields, 3, strprepProfileLineFn, (void*)filename, pErrorCode);
michael@0: 
michael@0:     /*fprintf(stdout,"Number of code points that have mappings with length >1 : %i\n",len);*/
michael@0: 
michael@0:     if(U_FAILURE(*pErrorCode) && (reportError || *pErrorCode!=U_FILE_ACCESS_ERROR)) {
michael@0:         fprintf(stderr, "gensprep error: u_parseDelimitedFile(\"%s\") failed - %s\n", filename, u_errorName(*pErrorCode));
michael@0:         exit(*pErrorCode);
michael@0:     }
michael@0: }
michael@0: 
michael@0: 
michael@0: #endif /* #if !UCONFIG_NO_IDNA */
michael@0: 
michael@0: /*
michael@0:  * Hey, Emacs, please set the following:
michael@0:  *
michael@0:  * Local Variables:
michael@0:  * indent-tabs-mode: nil
michael@0:  * End:
michael@0:  *
michael@0:  */