michael@0: /*
michael@0: *******************************************************************************
michael@0: *
michael@0: *   Copyright (C) 1999-2012, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *
michael@0: *******************************************************************************
michael@0: *   file name:  gencnval.c
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 1999nov05
michael@0: *   created by: Markus W. Scherer
michael@0: *
michael@0: *   This program reads convrtrs.txt and writes a memory-mappable
michael@0: *   converter name alias table to cnvalias.dat .
michael@0: *
michael@0: *   This program currently writes version 2.1 of the data format. See
michael@0: *   ucnv_io.c for more details on the format. Note that version 2.1
michael@0: *   is written in such a way that a 2.0 reader will be able to use it,
michael@0: *   and a 2.1 reader will be able to read 2.0.
michael@0: */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/putil.h"
michael@0: #include "unicode/ucnv.h" /* ucnv_compareNames() */
michael@0: #include "ucnv_io.h"
michael@0: #include "cmemory.h"
michael@0: #include "cstring.h"
michael@0: #include "uinvchar.h"
michael@0: #include "filestrm.h"
michael@0: #include "unicode/uclean.h"
michael@0: #include "unewdata.h"
michael@0: #include "uoptions.h"
michael@0: 
michael@0: #include <stdio.h>
michael@0: #include <stdlib.h>
michael@0: #include <ctype.h>
michael@0: 
michael@0: /* TODO: Need to check alias name length is less than UCNV_MAX_CONVERTER_NAME_LENGTH */
michael@0: 
michael@0: /* STRING_STORE_SIZE + TAG_STORE_SIZE <= ((2^16 - 1) * 2)
michael@0:  That is the maximum size for the string stores combined
michael@0:  because the strings are index at 16-bit boundries by a
michael@0:  16-bit index, and there is only one section for the 
michael@0:  strings.
michael@0:  */
michael@0: #define STRING_STORE_SIZE 0x1FBFE   /* 130046 */
michael@0: #define TAG_STORE_SIZE      0x400   /* 1024 */
michael@0: 
michael@0: /* The combined tag and converter count can affect the number of lists
michael@0:  created.  The size of all lists must be less than (2^17 - 1)
michael@0:  because the lists are indexed as a 16-bit array with a 16-bit index.
michael@0:  */
michael@0: #define MAX_TAG_COUNT 0x3F      /* 63 */
michael@0: #define MAX_CONV_COUNT UCNV_CONVERTER_INDEX_MASK
michael@0: #define MAX_ALIAS_COUNT 0xFFFF  /* 65535 */
michael@0: 
michael@0: /* The maximum number of aliases that a standard tag/converter combination can have.
michael@0:  At this moment 6/18/2002, IANA has 12 names for ASCII. Don't go below 15 for
michael@0:  this value. I don't recommend more than 31 for this value.
michael@0:  */
michael@0: #define MAX_TC_ALIAS_COUNT 0x1F    /* 31 */
michael@0: 
michael@0: #define MAX_LINE_SIZE 0x7FFF    /* 32767 */
michael@0: #define MAX_LIST_SIZE 0xFFFF    /* 65535 */
michael@0: 
michael@0: #define DATA_NAME "cnvalias"
michael@0: #define DATA_TYPE "icu" /* ICU alias table */
michael@0: 
michael@0: #define ALL_TAG_STR "ALL"
michael@0: #define ALL_TAG_NUM 1
michael@0: #define EMPTY_TAG_NUM 0
michael@0: 
michael@0: /* UDataInfo cf. udata.h */
michael@0: static const UDataInfo dataInfo={
michael@0:     sizeof(UDataInfo),
michael@0:     0,
michael@0: 
michael@0:     U_IS_BIG_ENDIAN,
michael@0:     U_CHARSET_FAMILY,
michael@0:     sizeof(UChar),
michael@0:     0,
michael@0: 
michael@0:     {0x43, 0x76, 0x41, 0x6c},     /* dataFormat="CvAl" */
michael@0:     {3, 0, 1, 0},                 /* formatVersion */
michael@0:     {1, 4, 2, 0}                  /* dataVersion */
michael@0: };
michael@0: 
michael@0: typedef struct {
michael@0:     char *store;
michael@0:     uint32_t top;
michael@0:     uint32_t max;
michael@0: } StringBlock;
michael@0: 
michael@0: static char stringStore[STRING_STORE_SIZE];
michael@0: static StringBlock stringBlock = { stringStore, 0, STRING_STORE_SIZE };
michael@0: 
michael@0: typedef struct {
michael@0:     uint16_t    aliasCount;
michael@0:     uint16_t    *aliases;     /* Index into stringStore */
michael@0: } AliasList;
michael@0: 
michael@0: typedef struct {
michael@0:     uint16_t converter;     /* Index into stringStore */
michael@0:     uint16_t totalAliasCount;    /* Total aliases in this column */
michael@0: } Converter;
michael@0: 
michael@0: static Converter converters[MAX_CONV_COUNT];
michael@0: static uint16_t converterCount=0;
michael@0: 
michael@0: static char tagStore[TAG_STORE_SIZE];
michael@0: static StringBlock tagBlock = { tagStore, 0, TAG_STORE_SIZE };
michael@0: 
michael@0: typedef struct {
michael@0:     uint16_t    tag;        /* Index into tagStore */
michael@0:     uint16_t    totalAliasCount; /* Total aliases in this row */
michael@0:     AliasList   aliasList[MAX_CONV_COUNT];
michael@0: } Tag;
michael@0: 
michael@0: /* Think of this as a 3D array. It's tagCount by converterCount by aliasCount */
michael@0: static Tag tags[MAX_TAG_COUNT];
michael@0: static uint16_t tagCount = 0;
michael@0: 
michael@0: /* Used for storing all aliases  */
michael@0: static uint16_t knownAliases[MAX_ALIAS_COUNT];
michael@0: static uint16_t knownAliasesCount = 0;
michael@0: /*static uint16_t duplicateKnownAliasesCount = 0;*/
michael@0: 
michael@0: /* Used for storing the lists section that point to aliases */
michael@0: static uint16_t aliasLists[MAX_LIST_SIZE];
michael@0: static uint16_t aliasListsSize = 0;
michael@0: 
michael@0: /* Were the standard tags declared before the aliases. */
michael@0: static UBool standardTagsUsed = FALSE;
michael@0: static UBool verbose = FALSE;
michael@0: static int lineNum = 1;
michael@0: 
michael@0: static UConverterAliasOptions tableOptions = {
michael@0:     UCNV_IO_STD_NORMALIZED,
michael@0:     1 /* containsCnvOptionInfo */
michael@0: };
michael@0: 
michael@0: 
michael@0: /**
michael@0:  * path to convrtrs.txt
michael@0:  */
michael@0: const char *path;
michael@0: 
michael@0: /* prototypes --------------------------------------------------------------- */
michael@0: 
michael@0: static void
michael@0: parseLine(const char *line);
michael@0: 
michael@0: static void
michael@0: parseFile(FileStream *in);
michael@0: 
michael@0: static int32_t
michael@0: chomp(char *line);
michael@0: 
michael@0: static void
michael@0: addOfficialTaggedStandards(char *line, int32_t lineLen);
michael@0: 
michael@0: static uint16_t
michael@0: addAlias(const char *alias, uint16_t standard, uint16_t converter, UBool defaultName);
michael@0: 
michael@0: static uint16_t
michael@0: addConverter(const char *converter);
michael@0: 
michael@0: static char *
michael@0: allocString(StringBlock *block, const char *s, int32_t length);
michael@0: 
michael@0: static uint16_t
michael@0: addToKnownAliases(const char *alias);
michael@0: 
michael@0: static int
michael@0: compareAliases(const void *alias1, const void *alias2);
michael@0: 
michael@0: static uint16_t
michael@0: getTagNumber(const char *tag, uint16_t tagLen);
michael@0: 
michael@0: /*static void
michael@0: addTaggedAlias(uint16_t tag, const char *alias, uint16_t converter);*/
michael@0: 
michael@0: static void
michael@0: writeAliasTable(UNewDataMemory *out);
michael@0: 
michael@0: /* -------------------------------------------------------------------------- */
michael@0: 
michael@0: /* Presumes that you used allocString() */
michael@0: #define GET_ALIAS_STR(index) (stringStore + ((size_t)(index) << 1))
michael@0: #define GET_TAG_STR(index) (tagStore + ((size_t)(index) << 1))
michael@0: 
michael@0: /* Presumes that you used allocString() */
michael@0: #define GET_ALIAS_NUM(str) ((uint16_t)((str - stringStore) >> 1))
michael@0: #define GET_TAG_NUM(str) ((uint16_t)((str - tagStore) >> 1))
michael@0: 
michael@0: enum
michael@0: {
michael@0:     HELP1,
michael@0:     HELP2,
michael@0:     VERBOSE,
michael@0:     COPYRIGHT,
michael@0:     DESTDIR,
michael@0:     SOURCEDIR
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: };
michael@0: 
michael@0: extern int
michael@0: main(int argc, char* argv[]) {
michael@0:     int i, n;
michael@0:     char pathBuf[512];
michael@0:     FileStream *in;
michael@0:     UNewDataMemory *out;
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=options[SOURCEDIR].value=u_getDataDirectory();
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[HELP1].doesOccur || options[HELP2].doesOccur) {
michael@0:         fprintf(stderr,
michael@0:             "usage: %s [-options] [convrtrs.txt]\n"
michael@0:             "\tread convrtrs.txt and create " U_ICUDATA_NAME "_" DATA_NAME "." DATA_TYPE "\n"
michael@0:             "options:\n"
michael@0:             "\t-h or -? or --help  this usage text\n"
michael@0:             "\t-v or --verbose     prints out extra information about the alias table\n"
michael@0:             "\t-c or --copyright   include a copyright notice\n"
michael@0:             "\t-d or --destdir     destination directory, followed by the path\n"
michael@0:             "\t-s or --sourcedir   source directory, followed by the path\n",
michael@0:             argv[0]);
michael@0:         return argc<0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
michael@0:     }
michael@0: 
michael@0:     if(options[VERBOSE].doesOccur) {
michael@0:         verbose = TRUE;
michael@0:     }
michael@0: 
michael@0:     if(argc>=2) {
michael@0:         path=argv[1];
michael@0:     } else {
michael@0:         path=options[SOURCEDIR].value;
michael@0:         if(path!=NULL && *path!=0) {
michael@0:             char *end;
michael@0: 
michael@0:             uprv_strcpy(pathBuf, path);
michael@0:             end = uprv_strchr(pathBuf, 0);
michael@0:             if(*(end-1)!=U_FILE_SEP_CHAR) {
michael@0:                 *(end++)=U_FILE_SEP_CHAR;
michael@0:             }
michael@0:             uprv_strcpy(end, "convrtrs.txt");
michael@0:             path=pathBuf;
michael@0:         } else {
michael@0:             path = "convrtrs.txt";
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     uprv_memset(stringStore, 0, sizeof(stringStore));
michael@0:     uprv_memset(tagStore, 0, sizeof(tagStore));
michael@0:     uprv_memset(converters, 0, sizeof(converters));
michael@0:     uprv_memset(tags, 0, sizeof(tags));
michael@0:     uprv_memset(aliasLists, 0, sizeof(aliasLists));
michael@0:     uprv_memset(knownAliases, 0, sizeof(aliasLists));
michael@0: 
michael@0: 
michael@0:     in=T_FileStream_open(path, "r");
michael@0:     if(in==NULL) {
michael@0:         fprintf(stderr, "gencnval: unable to open input file %s\n", path);
michael@0:         exit(U_FILE_ACCESS_ERROR);
michael@0:     }
michael@0:     parseFile(in);
michael@0:     T_FileStream_close(in);
michael@0: 
michael@0:     /* create the output file */
michael@0:     out=udata_create(options[DESTDIR].value, DATA_TYPE, DATA_NAME, &dataInfo,
michael@0:                      options[COPYRIGHT].doesOccur ? U_COPYRIGHT_STRING : NULL, &errorCode);
michael@0:     if(U_FAILURE(errorCode)) {
michael@0:         fprintf(stderr, "gencnval: unable to open output file - error %s\n", u_errorName(errorCode));
michael@0:         exit(errorCode);
michael@0:     }
michael@0: 
michael@0:     /* write the table of aliases based on a tag/converter name combination */
michael@0:     writeAliasTable(out);
michael@0: 
michael@0:     /* finish */
michael@0:     udata_finish(out, &errorCode);
michael@0:     if(U_FAILURE(errorCode)) {
michael@0:         fprintf(stderr, "gencnval: error finishing output file - %s\n", u_errorName(errorCode));
michael@0:         exit(errorCode);
michael@0:     }
michael@0: 
michael@0:     /* clean up tags */
michael@0:     for (i = 0; i < MAX_TAG_COUNT; i++) {
michael@0:         for (n = 0; n < MAX_CONV_COUNT; n++) {
michael@0:             if (tags[i].aliasList[n].aliases!=NULL) {
michael@0:                 uprv_free(tags[i].aliasList[n].aliases);
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: static void
michael@0: parseFile(FileStream *in) {
michael@0:     char line[MAX_LINE_SIZE];
michael@0:     char lastLine[MAX_LINE_SIZE];
michael@0:     int32_t lineSize = 0;
michael@0:     int32_t lastLineSize = 0;
michael@0:     UBool validParse = TRUE;
michael@0: 
michael@0:     lineNum = 0;
michael@0: 
michael@0:     /* Add the empty tag, which is for untagged aliases */
michael@0:     getTagNumber("", 0);
michael@0:     getTagNumber(ALL_TAG_STR, 3);
michael@0:     allocString(&stringBlock, "", 0);
michael@0: 
michael@0:     /* read the list of aliases */
michael@0:     while (validParse) {
michael@0:         validParse = FALSE;
michael@0: 
michael@0:         /* Read non-empty lines that don't start with a space character. */
michael@0:         while (T_FileStream_readLine(in, lastLine, MAX_LINE_SIZE) != NULL) {
michael@0:             lastLineSize = chomp(lastLine);
michael@0:             if (lineSize == 0 || (lastLineSize > 0 && isspace((int)*lastLine))) {
michael@0:                 uprv_strcpy(line + lineSize, lastLine);
michael@0:                 lineSize += lastLineSize;
michael@0:             } else if (lineSize > 0) {
michael@0:                 validParse = TRUE;
michael@0:                 break;
michael@0:             }
michael@0:             lineNum++;
michael@0:         }
michael@0: 
michael@0:         if (validParse || lineSize > 0) {
michael@0:             if (isspace((int)*line)) {
michael@0:                 fprintf(stderr, "%s:%d: error: cannot start an alias with a space\n", path, lineNum-1);
michael@0:                 exit(U_PARSE_ERROR);
michael@0:             } else if (line[0] == '{') {
michael@0:                 if (!standardTagsUsed && line[lineSize - 1] != '}') {
michael@0:                     fprintf(stderr, "%s:%d: error: alias needs to start with a converter name\n", path, lineNum);
michael@0:                     exit(U_PARSE_ERROR);
michael@0:                 }
michael@0:                 addOfficialTaggedStandards(line, lineSize);
michael@0:                 standardTagsUsed = TRUE;
michael@0:             } else {
michael@0:                 if (standardTagsUsed) {
michael@0:                     parseLine(line);
michael@0:                 }
michael@0:                 else {
michael@0:                     fprintf(stderr, "%s:%d: error: alias table needs to start a list of standard tags\n", path, lineNum);
michael@0:                     exit(U_PARSE_ERROR);
michael@0:                 }
michael@0:             }
michael@0:             /* Was the last line consumed */
michael@0:             if (lastLineSize > 0) {
michael@0:                 uprv_strcpy(line, lastLine);
michael@0:                 lineSize = lastLineSize;
michael@0:             }
michael@0:             else {
michael@0:                 lineSize = 0;
michael@0:             }
michael@0:         }
michael@0:         lineNum++;
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* This works almost like the Perl chomp.
michael@0:  It removes the newlines, comments and trailing whitespace (not preceding whitespace).
michael@0: */
michael@0: static int32_t
michael@0: chomp(char *line) {
michael@0:     char *s = line;
michael@0:     char *lastNonSpace = line;
michael@0:     while(*s!=0) {
michael@0:         /* truncate at a newline or a comment */
michael@0:         if(*s == '\r' || *s == '\n' || *s == '#') {
michael@0:             *s = 0;
michael@0:             break;
michael@0:         }
michael@0:         if (!isspace((int)*s)) {
michael@0:             lastNonSpace = s;
michael@0:         }
michael@0:         ++s;
michael@0:     }
michael@0:     if (lastNonSpace++ > line) {
michael@0:         *lastNonSpace = 0;
michael@0:         s = lastNonSpace;
michael@0:     }
michael@0:     return (int32_t)(s - line);
michael@0: }
michael@0: 
michael@0: static void
michael@0: parseLine(const char *line) {
michael@0:     uint16_t pos=0, start, limit, length, cnv;
michael@0:     char *converter, *alias;
michael@0: 
michael@0:     /* skip leading white space */
michael@0:     /* There is no whitespace at the beginning anymore */
michael@0: /*    while(line[pos]!=0 && isspace(line[pos])) {
michael@0:         ++pos;
michael@0:     }
michael@0: */
michael@0: 
michael@0:     /* is there nothing on this line? */
michael@0:     if(line[pos]==0) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /* get the converter name */
michael@0:     start=pos;
michael@0:     while(line[pos]!=0 && !isspace((int)line[pos])) {
michael@0:         ++pos;
michael@0:     }
michael@0:     limit=pos;
michael@0: 
michael@0:     /* store the converter name */
michael@0:     length=(uint16_t)(limit-start);
michael@0:     converter=allocString(&stringBlock, line+start, length);
michael@0: 
michael@0:     /* add the converter to the converter table */
michael@0:     cnv=addConverter(converter);
michael@0: 
michael@0:     /* The name itself may be tagged, so let's added it to the aliases list properly */
michael@0:     pos = start;
michael@0: 
michael@0:     /* get all the real aliases */
michael@0:     for(;;) {
michael@0: 
michael@0:         /* skip white space */
michael@0:         while(line[pos]!=0 && isspace((int)line[pos])) {
michael@0:             ++pos;
michael@0:         }
michael@0: 
michael@0:         /* is there no more alias name on this line? */
michael@0:         if(line[pos]==0) {
michael@0:             break;
michael@0:         }
michael@0: 
michael@0:         /* get an alias name */
michael@0:         start=pos;
michael@0:         while(line[pos]!=0 && line[pos]!='{' && !isspace((int)line[pos])) {
michael@0:             ++pos;
michael@0:         }
michael@0:         limit=pos;
michael@0: 
michael@0:         /* store the alias name */
michael@0:         length=(uint16_t)(limit-start);
michael@0:         if (start == 0) {
michael@0:             /* add the converter as its own alias to the alias table */
michael@0:             alias = converter;
michael@0:             addAlias(alias, ALL_TAG_NUM, cnv, TRUE);
michael@0:         }
michael@0:         else {
michael@0:             alias=allocString(&stringBlock, line+start, length);
michael@0:             addAlias(alias, ALL_TAG_NUM, cnv, FALSE);
michael@0:         }
michael@0:         addToKnownAliases(alias);
michael@0: 
michael@0:         /* add the alias/converter pair to the alias table */
michael@0:         /* addAlias(alias, 0, cnv, FALSE);*/
michael@0: 
michael@0:         /* skip whitespace */
michael@0:         while (line[pos] && isspace((int)line[pos])) {
michael@0:             ++pos;
michael@0:         }
michael@0: 
michael@0:         /* handle tags if they are present */
michael@0:         if (line[pos] == '{') {
michael@0:             ++pos;
michael@0:             do {
michael@0:                 start = pos;
michael@0:                 while (line[pos] && line[pos] != '}' && !isspace((int)line[pos])) {
michael@0:                     ++pos;
michael@0:                 }
michael@0:                 limit = pos;
michael@0: 
michael@0:                 if (start != limit) {
michael@0:                     /* add the tag to the tag table */
michael@0:                     uint16_t tag = getTagNumber(line + start, (uint16_t)(limit - start));
michael@0:                     addAlias(alias, tag, cnv, (UBool)(line[limit-1] == '*'));
michael@0:                 }
michael@0: 
michael@0:                 while (line[pos] && isspace((int)line[pos])) {
michael@0:                     ++pos;
michael@0:                 }
michael@0:             } while (line[pos] && line[pos] != '}');
michael@0: 
michael@0:             if (line[pos] == '}') {
michael@0:                 ++pos;
michael@0:             } else {
michael@0:                 fprintf(stderr, "%s:%d: Unterminated tag list\n", path, lineNum);
michael@0:                 exit(U_UNMATCHED_BRACES);
michael@0:             }
michael@0:         } else {
michael@0:             addAlias(alias, EMPTY_TAG_NUM, cnv, (UBool)(tags[0].aliasList[cnv].aliasCount == 0));
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: static uint16_t
michael@0: getTagNumber(const char *tag, uint16_t tagLen) {
michael@0:     char *atag;
michael@0:     uint16_t t;
michael@0:     UBool preferredName = ((tagLen > 0) ? (tag[tagLen - 1] == '*') : (FALSE));
michael@0: 
michael@0:     if (tagCount >= MAX_TAG_COUNT) {
michael@0:         fprintf(stderr, "%s:%d: too many tags\n", path, lineNum);
michael@0:         exit(U_BUFFER_OVERFLOW_ERROR);
michael@0:     }
michael@0: 
michael@0:     if (preferredName) {
michael@0: /*        puts(tag);*/
michael@0:         tagLen--;
michael@0:     }
michael@0: 
michael@0:     for (t = 0; t < tagCount; ++t) {
michael@0:         const char *currTag = GET_TAG_STR(tags[t].tag);
michael@0:         if (uprv_strlen(currTag) == tagLen && !uprv_strnicmp(currTag, tag, tagLen)) {
michael@0:             return t;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* we need to add this tag */
michael@0:     if (tagCount >= MAX_TAG_COUNT) {
michael@0:         fprintf(stderr, "%s:%d: error: too many tags\n", path, lineNum);
michael@0:         exit(U_BUFFER_OVERFLOW_ERROR);
michael@0:     }
michael@0: 
michael@0:     /* allocate a new entry in the tag table */
michael@0:     atag = allocString(&tagBlock, tag, tagLen);
michael@0: 
michael@0:     if (standardTagsUsed) {
michael@0:         fprintf(stderr, "%s:%d: error: Tag \"%s\" is not declared at the beginning of the alias table.\n",
michael@0:             path, lineNum, atag);
michael@0:         exit(1);
michael@0:     }
michael@0:     else if (tagLen > 0 && strcmp(tag, ALL_TAG_STR) != 0) {
michael@0:         fprintf(stderr, "%s:%d: warning: Tag \"%s\" was added to the list of standards because it was not declared at beginning of the alias table.\n",
michael@0:             path, lineNum, atag);
michael@0:     }
michael@0: 
michael@0:     /* add the tag to the tag table */
michael@0:     tags[tagCount].tag = GET_TAG_NUM(atag);
michael@0:     /* The aliasList should be set to 0's already */
michael@0: 
michael@0:     return tagCount++;
michael@0: }
michael@0: 
michael@0: /*static void
michael@0: addTaggedAlias(uint16_t tag, const char *alias, uint16_t converter) {
michael@0:     tags[tag].aliases[converter] = alias;
michael@0: }
michael@0: */
michael@0: 
michael@0: static void
michael@0: addOfficialTaggedStandards(char *line, int32_t lineLen) {
michael@0:     char *atag;
michael@0:     char *endTagExp;
michael@0:     char *tag;
michael@0:     static const char WHITESPACE[] = " \t";
michael@0: 
michael@0:     if (tagCount > UCNV_NUM_RESERVED_TAGS) {
michael@0:         fprintf(stderr, "%s:%d: error: official tags already added\n", path, lineNum);
michael@0:         exit(U_BUFFER_OVERFLOW_ERROR);
michael@0:     }
michael@0:     tag = strchr(line, '{');
michael@0:     if (tag == NULL) {
michael@0:         /* Why were we called? */
michael@0:         fprintf(stderr, "%s:%d: error: Missing start of tag group\n", path, lineNum);
michael@0:         exit(U_PARSE_ERROR);
michael@0:     }
michael@0:     tag++;
michael@0:     endTagExp = strchr(tag, '}');
michael@0:     if (endTagExp == NULL) {
michael@0:         fprintf(stderr, "%s:%d: error: Missing end of tag group\n", path, lineNum);
michael@0:         exit(U_PARSE_ERROR);
michael@0:     }
michael@0:     endTagExp[0] = 0;
michael@0: 
michael@0:     tag = strtok(tag, WHITESPACE);
michael@0:     while (tag != NULL) {
michael@0: /*        printf("Adding original tag \"%s\"\n", tag);*/
michael@0: 
michael@0:         /* allocate a new entry in the tag table */
michael@0:         atag = allocString(&tagBlock, tag, -1);
michael@0: 
michael@0:         /* add the tag to the tag table */
michael@0:         tags[tagCount++].tag = (uint16_t)((atag - tagStore) >> 1);
michael@0: 
michael@0:         /* The aliasList should already be set to 0's */
michael@0: 
michael@0:         /* Get next tag */
michael@0:         tag = strtok(NULL, WHITESPACE);
michael@0:     }
michael@0: }
michael@0: 
michael@0: static uint16_t
michael@0: addToKnownAliases(const char *alias) {
michael@0: /*    uint32_t idx; */
michael@0:     /* strict matching */
michael@0: /*    for (idx = 0; idx < knownAliasesCount; idx++) {
michael@0:         uint16_t num = GET_ALIAS_NUM(alias);
michael@0:         if (knownAliases[idx] != num
michael@0:             && uprv_strcmp(alias, GET_ALIAS_STR(knownAliases[idx])) == 0)
michael@0:         {
michael@0:             fprintf(stderr, "%s:%d: warning: duplicate alias %s and %s found\n", path, 
michael@0:                 lineNum, alias, GET_ALIAS_STR(knownAliases[idx]));
michael@0:             duplicateKnownAliasesCount++;
michael@0:             break;
michael@0:         }
michael@0:         else if (knownAliases[idx] != num
michael@0:             && ucnv_compareNames(alias, GET_ALIAS_STR(knownAliases[idx])) == 0)
michael@0:         {
michael@0:             if (verbose) {
michael@0:                 fprintf(stderr, "%s:%d: information: duplicate alias %s and %s found\n", path, 
michael@0:                     lineNum, alias, GET_ALIAS_STR(knownAliases[idx]));
michael@0:             }
michael@0:             duplicateKnownAliasesCount++;
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0: */
michael@0:     if (knownAliasesCount >= MAX_ALIAS_COUNT) {
michael@0:         fprintf(stderr, "%s:%d: warning: Too many aliases defined for all converters\n",
michael@0:             path, lineNum);
michael@0:         exit(U_BUFFER_OVERFLOW_ERROR);
michael@0:     }
michael@0:     /* TODO: We could try to unlist exact duplicates. */
michael@0:     return knownAliases[knownAliasesCount++] = GET_ALIAS_NUM(alias);
michael@0: }
michael@0: 
michael@0: /*
michael@0: @param standard When standard is 0, then it's the "empty" tag.
michael@0: */
michael@0: static uint16_t
michael@0: addAlias(const char *alias, uint16_t standard, uint16_t converter, UBool defaultName) {
michael@0:     uint32_t idx, idx2;
michael@0:     UBool startEmptyWithoutDefault = FALSE;
michael@0:     AliasList *aliasList;
michael@0: 
michael@0:     if(standard>=MAX_TAG_COUNT) {
michael@0:         fprintf(stderr, "%s:%d: error: too many standard tags\n", path, lineNum);
michael@0:         exit(U_BUFFER_OVERFLOW_ERROR);
michael@0:     }
michael@0:     if(converter>=MAX_CONV_COUNT) {
michael@0:         fprintf(stderr, "%s:%d: error: too many converter names\n", path, lineNum);
michael@0:         exit(U_BUFFER_OVERFLOW_ERROR);
michael@0:     }
michael@0:     aliasList = &tags[standard].aliasList[converter];
michael@0: 
michael@0:     if (strchr(alias, '}')) {
michael@0:         fprintf(stderr, "%s:%d: error: unmatched } found\n", path, 
michael@0:             lineNum);
michael@0:     }
michael@0: 
michael@0:     if(aliasList->aliasCount + 1 >= MAX_TC_ALIAS_COUNT) {
michael@0:         fprintf(stderr, "%s:%d: error: too many aliases for alias %s and converter %s\n", path, 
michael@0:             lineNum, alias, GET_ALIAS_STR(converters[converter].converter));
michael@0:         exit(U_BUFFER_OVERFLOW_ERROR);
michael@0:     }
michael@0: 
michael@0:     /* Show this warning only once. All aliases are added to the "ALL" tag. */
michael@0:     if (standard == ALL_TAG_NUM && GET_ALIAS_STR(converters[converter].converter) != alias) {
michael@0:         /* Normally these option values are parsed at runtime, and they can
michael@0:            be discarded when the alias is a default converter. Options should
michael@0:            only be on a converter and not an alias. */
michael@0:         if (uprv_strchr(alias, UCNV_OPTION_SEP_CHAR) != 0)
michael@0:         {
michael@0:             fprintf(stderr, "warning(line %d): alias %s contains a \""UCNV_OPTION_SEP_STRING"\". Options are parsed at run-time and do not need to be in the alias table.\n",
michael@0:                 lineNum, alias);
michael@0:         }
michael@0:         if (uprv_strchr(alias, UCNV_VALUE_SEP_CHAR) != 0)
michael@0:         {
michael@0:             fprintf(stderr, "warning(line %d): alias %s contains an \""UCNV_VALUE_SEP_STRING"\". Options are parsed at run-time and do not need to be in the alias table.\n",
michael@0:                 lineNum, alias);
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     if (standard != ALL_TAG_NUM) {
michael@0:         /* Check for duplicate aliases for this tag on all converters */
michael@0:         for (idx = 0; idx < converterCount; idx++) {
michael@0:             for (idx2 = 0; idx2 < tags[standard].aliasList[idx].aliasCount; idx2++) {
michael@0:                 uint16_t aliasNum = tags[standard].aliasList[idx].aliases[idx2];
michael@0:                 if (aliasNum
michael@0:                     && ucnv_compareNames(alias, GET_ALIAS_STR(aliasNum)) == 0)
michael@0:                 {
michael@0:                     if (idx == converter) {
michael@0:                         /*
michael@0:                          * (alias, standard) duplicates are harmless if they map to the same converter.
michael@0:                          * Only print a warning in verbose mode, or if the alias is a precise duplicate,
michael@0:                          * not just a lenient-match duplicate.
michael@0:                          */
michael@0:                         if (verbose || 0 == uprv_strcmp(alias, GET_ALIAS_STR(aliasNum))) {
michael@0:                             fprintf(stderr, "%s:%d: warning: duplicate aliases %s and %s found for standard %s and converter %s\n", path, 
michael@0:                                 lineNum, alias, GET_ALIAS_STR(aliasNum),
michael@0:                                 GET_TAG_STR(tags[standard].tag),
michael@0:                                 GET_ALIAS_STR(converters[converter].converter));
michael@0:                         }
michael@0:                     } else {
michael@0:                         fprintf(stderr, "%s:%d: warning: duplicate aliases %s and %s found for standard tag %s between converter %s and converter %s\n", path, 
michael@0:                             lineNum, alias, GET_ALIAS_STR(aliasNum),
michael@0:                             GET_TAG_STR(tags[standard].tag),
michael@0:                             GET_ALIAS_STR(converters[converter].converter),
michael@0:                             GET_ALIAS_STR(converters[idx].converter));
michael@0:                     }
michael@0:                     break;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         /* Check for duplicate default aliases for this converter on all tags */
michael@0:         /* It's okay to have multiple standards prefer the same name */
michael@0: /*        if (verbose && !dupFound) {
michael@0:             for (idx = 0; idx < tagCount; idx++) {
michael@0:                 if (tags[idx].aliasList[converter].aliases) {
michael@0:                     uint16_t aliasNum = tags[idx].aliasList[converter].aliases[0];
michael@0:                     if (aliasNum
michael@0:                         && ucnv_compareNames(alias, GET_ALIAS_STR(aliasNum)) == 0)
michael@0:                     {
michael@0:                         fprintf(stderr, "%s:%d: warning: duplicate alias %s found for converter %s and standard tag %s\n", path, 
michael@0:                             lineNum, alias, GET_ALIAS_STR(converters[converter].converter), GET_TAG_STR(tags[standard].tag));
michael@0:                         break;
michael@0:                     }
michael@0:                 }
michael@0:             }
michael@0:         }*/
michael@0:     }
michael@0: 
michael@0:     if (aliasList->aliasCount <= 0) {
michael@0:         aliasList->aliasCount++;
michael@0:         startEmptyWithoutDefault = TRUE;
michael@0:     }
michael@0:     aliasList->aliases = (uint16_t *)uprv_realloc(aliasList->aliases, (aliasList->aliasCount + 1) * sizeof(aliasList->aliases[0]));
michael@0:     if (startEmptyWithoutDefault) {
michael@0:         aliasList->aliases[0] = 0;
michael@0:     }
michael@0:     if (defaultName) {
michael@0:         if (aliasList->aliases[0] != 0) {
michael@0:             fprintf(stderr, "%s:%d: error: Alias %s and %s cannot both be the default alias for standard tag %s and converter %s\n", path, 
michael@0:                 lineNum,
michael@0:                 alias,
michael@0:                 GET_ALIAS_STR(aliasList->aliases[0]),
michael@0:                 GET_TAG_STR(tags[standard].tag),
michael@0:                 GET_ALIAS_STR(converters[converter].converter));
michael@0:             exit(U_PARSE_ERROR);
michael@0:         }
michael@0:         aliasList->aliases[0] = GET_ALIAS_NUM(alias);
michael@0:     } else {
michael@0:         aliasList->aliases[aliasList->aliasCount++] = GET_ALIAS_NUM(alias);
michael@0:     }
michael@0: /*    aliasList->converter = converter;*/
michael@0: 
michael@0:     converters[converter].totalAliasCount++; /* One more to the column */
michael@0:     tags[standard].totalAliasCount++; /* One more to the row */
michael@0: 
michael@0:     return aliasList->aliasCount;
michael@0: }
michael@0: 
michael@0: static uint16_t
michael@0: addConverter(const char *converter) {
michael@0:     uint32_t idx;
michael@0:     if(converterCount>=MAX_CONV_COUNT) {
michael@0:         fprintf(stderr, "%s:%d: error: too many converters\n", path, lineNum);
michael@0:         exit(U_BUFFER_OVERFLOW_ERROR);
michael@0:     }
michael@0: 
michael@0:     for (idx = 0; idx < converterCount; idx++) {
michael@0:         if (ucnv_compareNames(converter, GET_ALIAS_STR(converters[idx].converter)) == 0) {
michael@0:             fprintf(stderr, "%s:%d: error: duplicate converter %s found!\n", path, lineNum, converter);
michael@0:             exit(U_PARSE_ERROR);
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     converters[converterCount].converter = GET_ALIAS_NUM(converter);
michael@0:     converters[converterCount].totalAliasCount = 0;
michael@0: 
michael@0:     return converterCount++;
michael@0: }
michael@0: 
michael@0: /* resolve this alias based on the prioritization of the standard tags. */
michael@0: static void
michael@0: resolveAliasToConverter(uint16_t alias, uint16_t *tagNum, uint16_t *converterNum) {
michael@0:     uint16_t idx, idx2, idx3;
michael@0: 
michael@0:     for (idx = UCNV_NUM_RESERVED_TAGS; idx < tagCount; idx++) {
michael@0:         for (idx2 = 0; idx2 < converterCount; idx2++) {
michael@0:             for (idx3 = 0; idx3 < tags[idx].aliasList[idx2].aliasCount; idx3++) {
michael@0:                 uint16_t aliasNum = tags[idx].aliasList[idx2].aliases[idx3];
michael@0:                 if (aliasNum == alias) {
michael@0:                     *tagNum = idx;
michael@0:                     *converterNum = idx2;
michael@0:                     return;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:     }
michael@0:     /* Do the leftovers last, just in case */
michael@0:     /* There is no need to do the ALL tag */
michael@0:     idx = 0;
michael@0:     for (idx2 = 0; idx2 < converterCount; idx2++) {
michael@0:         for (idx3 = 0; idx3 < tags[idx].aliasList[idx2].aliasCount; idx3++) {
michael@0:             uint16_t aliasNum = tags[idx].aliasList[idx2].aliases[idx3];
michael@0:             if (aliasNum == alias) {
michael@0:                 *tagNum = idx;
michael@0:                 *converterNum = idx2;
michael@0:                 return;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0:     *tagNum = UINT16_MAX;
michael@0:     *converterNum = UINT16_MAX;
michael@0:     fprintf(stderr, "%s: warning: alias %s not found\n",
michael@0:         path,
michael@0:         GET_ALIAS_STR(alias));
michael@0:     return;
michael@0: }
michael@0: 
michael@0: /* The knownAliases should be sorted before calling this function */
michael@0: static uint32_t
michael@0: resolveAliases(uint16_t *uniqueAliasArr, uint16_t *uniqueAliasToConverterArr, uint16_t aliasOffset) {
michael@0:     uint32_t uniqueAliasIdx = 0;
michael@0:     uint32_t idx;
michael@0:     uint16_t currTagNum, oldTagNum;
michael@0:     uint16_t currConvNum, oldConvNum;
michael@0:     const char *lastName;
michael@0: 
michael@0:     resolveAliasToConverter(knownAliases[0], &oldTagNum, &currConvNum);
michael@0:     uniqueAliasToConverterArr[uniqueAliasIdx] = currConvNum;
michael@0:     oldConvNum = currConvNum;
michael@0:     uniqueAliasArr[uniqueAliasIdx] = knownAliases[0] + aliasOffset;
michael@0:     uniqueAliasIdx++;
michael@0:     lastName = GET_ALIAS_STR(knownAliases[0]);
michael@0: 
michael@0:     for (idx = 1; idx < knownAliasesCount; idx++) {
michael@0:         resolveAliasToConverter(knownAliases[idx], &currTagNum, &currConvNum);
michael@0:         if (ucnv_compareNames(lastName, GET_ALIAS_STR(knownAliases[idx])) == 0) {
michael@0:             /* duplicate found */
michael@0:             if ((currTagNum < oldTagNum && currTagNum >= UCNV_NUM_RESERVED_TAGS)
michael@0:                 || oldTagNum == 0) {
michael@0:                 oldTagNum = currTagNum;
michael@0:                 uniqueAliasToConverterArr[uniqueAliasIdx - 1] = currConvNum;
michael@0:                 uniqueAliasArr[uniqueAliasIdx - 1] = knownAliases[idx] + aliasOffset;
michael@0:                 if (verbose) {
michael@0:                     printf("using %s instead of %s -> %s", 
michael@0:                         GET_ALIAS_STR(knownAliases[idx]),
michael@0:                         lastName,
michael@0:                         GET_ALIAS_STR(converters[currConvNum].converter));
michael@0:                     if (oldConvNum != currConvNum) {
michael@0:                         printf(" (alias conflict)");
michael@0:                     }
michael@0:                     puts("");
michael@0:                 }
michael@0:             }
michael@0:             else {
michael@0:                 /* else ignore it */
michael@0:                 if (verbose) {
michael@0:                     printf("folding %s into %s -> %s",
michael@0:                         GET_ALIAS_STR(knownAliases[idx]),
michael@0:                         lastName,
michael@0:                         GET_ALIAS_STR(converters[oldConvNum].converter));
michael@0:                     if (oldConvNum != currConvNum) {
michael@0:                         printf(" (alias conflict)");
michael@0:                     }
michael@0:                     puts("");
michael@0:                 }
michael@0:             }
michael@0:             if (oldConvNum != currConvNum) {
michael@0:                 uniqueAliasToConverterArr[uniqueAliasIdx - 1] |= UCNV_AMBIGUOUS_ALIAS_MAP_BIT;
michael@0:             }
michael@0:         }
michael@0:         else {
michael@0:             uniqueAliasToConverterArr[uniqueAliasIdx] = currConvNum;
michael@0:             oldConvNum = currConvNum;
michael@0:             uniqueAliasArr[uniqueAliasIdx] = knownAliases[idx] + aliasOffset;
michael@0:             uniqueAliasIdx++;
michael@0:             lastName = GET_ALIAS_STR(knownAliases[idx]);
michael@0:             oldTagNum = currTagNum;
michael@0:             /*printf("%s -> %s\n", GET_ALIAS_STR(knownAliases[idx]), GET_ALIAS_STR(converters[currConvNum].converter));*/
michael@0:         }
michael@0:         if (uprv_strchr(GET_ALIAS_STR(converters[currConvNum].converter), UCNV_OPTION_SEP_CHAR) != NULL) {
michael@0:             uniqueAliasToConverterArr[uniqueAliasIdx-1] |= UCNV_CONTAINS_OPTION_BIT;
michael@0:         }
michael@0:     }
michael@0:     return uniqueAliasIdx;
michael@0: }
michael@0: 
michael@0: static void
michael@0: createOneAliasList(uint16_t *aliasArrLists, uint32_t tag, uint32_t converter, uint16_t offset) {
michael@0:     uint32_t aliasNum;
michael@0:     AliasList *aliasList = &tags[tag].aliasList[converter];
michael@0: 
michael@0:     if (aliasList->aliasCount == 0) {
michael@0:         aliasArrLists[tag*converterCount + converter] = 0;
michael@0:     }
michael@0:     else {
michael@0:         aliasLists[aliasListsSize++] = aliasList->aliasCount;
michael@0: 
michael@0:         /* write into the array area a 1's based index. */
michael@0:         aliasArrLists[tag*converterCount + converter] = aliasListsSize;
michael@0: 
michael@0: /*        printf("tag %s converter %s\n",
michael@0:             GET_TAG_STR(tags[tag].tag),
michael@0:             GET_ALIAS_STR(converters[converter].converter));*/
michael@0:         for (aliasNum = 0; aliasNum < aliasList->aliasCount; aliasNum++) {
michael@0:             uint16_t value;
michael@0: /*            printf("   %s\n",
michael@0:                 GET_ALIAS_STR(aliasList->aliases[aliasNum]));*/
michael@0:             if (aliasList->aliases[aliasNum]) {
michael@0:                 value = aliasList->aliases[aliasNum] + offset;
michael@0:             } else {
michael@0:                 value = 0;
michael@0:                 if (tag != 0) { /* Only show the warning when it's not the leftover tag. */
michael@0:                     fprintf(stderr, "%s: warning: tag %s does not have a default alias for %s\n",
michael@0:                             path,
michael@0:                             GET_TAG_STR(tags[tag].tag),
michael@0:                             GET_ALIAS_STR(converters[converter].converter));
michael@0:                 }
michael@0:             }
michael@0:             aliasLists[aliasListsSize++] = value;
michael@0:             if (aliasListsSize >= MAX_LIST_SIZE) {
michael@0:                 fprintf(stderr, "%s: error: Too many alias lists\n", path);
michael@0:                 exit(U_BUFFER_OVERFLOW_ERROR);
michael@0:             }
michael@0: 
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: static void
michael@0: createNormalizedAliasStrings(char *normalizedStrings, const char *origStringBlock, int32_t stringBlockLength) {
michael@0:     int32_t currStrLen;
michael@0:     uprv_memcpy(normalizedStrings, origStringBlock, stringBlockLength);
michael@0:     while ((currStrLen = (int32_t)uprv_strlen(origStringBlock)) < stringBlockLength) {
michael@0:         int32_t currStrSize = currStrLen + 1;
michael@0:         if (currStrLen > 0) {
michael@0:             int32_t normStrLen;
michael@0:             ucnv_io_stripForCompare(normalizedStrings, origStringBlock);
michael@0:             normStrLen = uprv_strlen(normalizedStrings);
michael@0:             if (normStrLen > 0) {
michael@0:                 uprv_memset(normalizedStrings + normStrLen, 0, currStrSize - normStrLen);
michael@0:             }
michael@0:         }
michael@0:         stringBlockLength -= currStrSize;
michael@0:         normalizedStrings += currStrSize;
michael@0:         origStringBlock += currStrSize;
michael@0:     }
michael@0: }
michael@0: 
michael@0: static void
michael@0: writeAliasTable(UNewDataMemory *out) {
michael@0:     uint32_t i, j;
michael@0:     uint32_t uniqueAliasesSize;
michael@0:     uint16_t aliasOffset = (uint16_t)(tagBlock.top/sizeof(uint16_t));
michael@0:     uint16_t *aliasArrLists = (uint16_t *)uprv_malloc(tagCount * converterCount * sizeof(uint16_t));
michael@0:     uint16_t *uniqueAliases = (uint16_t *)uprv_malloc(knownAliasesCount * sizeof(uint16_t));
michael@0:     uint16_t *uniqueAliasesToConverter = (uint16_t *)uprv_malloc(knownAliasesCount * sizeof(uint16_t));
michael@0: 
michael@0:     qsort(knownAliases, knownAliasesCount, sizeof(knownAliases[0]), compareAliases);
michael@0:     uniqueAliasesSize = resolveAliases(uniqueAliases, uniqueAliasesToConverter, aliasOffset);
michael@0: 
michael@0:     /* Array index starts at 1. aliasLists[0] is the size of the lists section. */
michael@0:     aliasListsSize = 0;
michael@0: 
michael@0:     /* write the offsets of all the aliases lists in a 2D array, and create the lists. */
michael@0:     for (i = 0; i < tagCount; ++i) {
michael@0:         for (j = 0; j < converterCount; ++j) {
michael@0:             createOneAliasList(aliasArrLists, i, j, aliasOffset);
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* Write the size of the TOC */
michael@0:     if (tableOptions.stringNormalizationType == UCNV_IO_UNNORMALIZED) {
michael@0:         udata_write32(out, 8);
michael@0:     }
michael@0:     else {
michael@0:         udata_write32(out, 9);
michael@0:     }
michael@0: 
michael@0:     /* Write the sizes of each section */
michael@0:     /* All sizes are the number of uint16_t units, not bytes */
michael@0:     udata_write32(out, converterCount);
michael@0:     udata_write32(out, tagCount);
michael@0:     udata_write32(out, uniqueAliasesSize);  /* list of aliases */
michael@0:     udata_write32(out, uniqueAliasesSize);  /* The preresolved form of mapping an untagged the alias to a converter */
michael@0:     udata_write32(out, tagCount * converterCount);
michael@0:     udata_write32(out, aliasListsSize + 1);
michael@0:     udata_write32(out, sizeof(tableOptions) / sizeof(uint16_t));
michael@0:     udata_write32(out, (tagBlock.top + stringBlock.top) / sizeof(uint16_t));
michael@0:     if (tableOptions.stringNormalizationType != UCNV_IO_UNNORMALIZED) {
michael@0:         udata_write32(out, (tagBlock.top + stringBlock.top) / sizeof(uint16_t));
michael@0:     }
michael@0: 
michael@0:     /* write the table of converters */
michael@0:     /* Think of this as the column headers */
michael@0:     for(i=0; i<converterCount; ++i) {
michael@0:         udata_write16(out, (uint16_t)(converters[i].converter + aliasOffset));
michael@0:     }
michael@0: 
michael@0:     /* write the table of tags */
michael@0:     /* Think of this as the row headers */
michael@0:     for(i=UCNV_NUM_RESERVED_TAGS; i<tagCount; ++i) {
michael@0:         udata_write16(out, tags[i].tag);
michael@0:     }
michael@0:     /* The empty tag is considered the leftover list, and put that at the end of the priority list. */
michael@0:     udata_write16(out, tags[EMPTY_TAG_NUM].tag);
michael@0:     udata_write16(out, tags[ALL_TAG_NUM].tag);
michael@0: 
michael@0:     /* Write the unique list of aliases */
michael@0:     udata_writeBlock(out, uniqueAliases, uniqueAliasesSize * sizeof(uint16_t));
michael@0: 
michael@0:     /* Write the unique list of aliases */
michael@0:     udata_writeBlock(out, uniqueAliasesToConverter, uniqueAliasesSize * sizeof(uint16_t));
michael@0: 
michael@0:     /* Write the array to the lists */
michael@0:     udata_writeBlock(out, (const void *)(aliasArrLists + (2*converterCount)), (((tagCount - 2) * converterCount) * sizeof(uint16_t)));
michael@0:     /* Now write the leftover part of the array for the EMPTY and ALL lists */
michael@0:     udata_writeBlock(out, (const void *)aliasArrLists, (2 * converterCount * sizeof(uint16_t)));
michael@0: 
michael@0:     /* Offset the next array to make the index start at 1. */
michael@0:     udata_write16(out, 0xDEAD);
michael@0: 
michael@0:     /* Write the lists */
michael@0:     udata_writeBlock(out, (const void *)aliasLists, aliasListsSize * sizeof(uint16_t));
michael@0: 
michael@0:     /* Write any options for the alias table. */
michael@0:     udata_writeBlock(out, (const void *)&tableOptions, sizeof(tableOptions));
michael@0: 
michael@0:     /* write the tags strings */
michael@0:     udata_writeString(out, tagBlock.store, tagBlock.top);
michael@0: 
michael@0:     /* write the aliases strings */
michael@0:     udata_writeString(out, stringBlock.store, stringBlock.top);
michael@0: 
michael@0:     /* write the normalized aliases strings */
michael@0:     if (tableOptions.stringNormalizationType != UCNV_IO_UNNORMALIZED) {
michael@0:         char *normalizedStrings = (char *)uprv_malloc(tagBlock.top + stringBlock.top);
michael@0:         createNormalizedAliasStrings(normalizedStrings, tagBlock.store, tagBlock.top);
michael@0:         createNormalizedAliasStrings(normalizedStrings + tagBlock.top, stringBlock.store, stringBlock.top);
michael@0: 
michael@0:         /* Write out the complete normalized array. */
michael@0:         udata_writeString(out, normalizedStrings, tagBlock.top + stringBlock.top);
michael@0:         uprv_free(normalizedStrings);
michael@0:     }
michael@0: 
michael@0:     uprv_free(uniqueAliasesToConverter);
michael@0:     uprv_free(uniqueAliases);
michael@0:     uprv_free(aliasArrLists);
michael@0: }
michael@0: 
michael@0: static char *
michael@0: allocString(StringBlock *block, const char *s, int32_t length) {
michael@0:     uint32_t top;
michael@0:     char *p;
michael@0: 
michael@0:     if(length<0) {
michael@0:         length=(int32_t)uprv_strlen(s);
michael@0:     }
michael@0: 
michael@0:     /*
michael@0:      * add 1 for the terminating NUL
michael@0:      * and round up (+1 &~1)
michael@0:      * to keep the addresses on a 16-bit boundary
michael@0:      */
michael@0:     top=block->top + (uint32_t)((length + 1 + 1) & ~1);
michael@0: 
michael@0:     if(top >= block->max) {
michael@0:         fprintf(stderr, "%s:%d: error: out of memory\n", path, lineNum);
michael@0:         exit(U_MEMORY_ALLOCATION_ERROR);
michael@0:     }
michael@0: 
michael@0:     /* get the pointer and copy the string */
michael@0:     p = block->store + block->top;
michael@0:     uprv_memcpy(p, s, length);
michael@0:     p[length] = 0; /* NUL-terminate it */
michael@0:     if((length & 1) == 0) {
michael@0:         p[length + 1] = 0; /* set the padding byte */
michael@0:     }
michael@0: 
michael@0:     /* check for invariant characters now that we have a NUL-terminated string for easy output */
michael@0:     if(!uprv_isInvariantString(p, length)) {
michael@0:         fprintf(stderr, "%s:%d: error: the name %s contains not just invariant characters\n", path, lineNum, p);
michael@0:         exit(U_INVALID_TABLE_FORMAT);
michael@0:     }
michael@0: 
michael@0:     block->top = top;
michael@0:     return p;
michael@0: }
michael@0: 
michael@0: static int
michael@0: compareAliases(const void *alias1, const void *alias2) {
michael@0:     /* Names like IBM850 and ibm-850 need to be sorted together */
michael@0:     int result = ucnv_compareNames(GET_ALIAS_STR(*(uint16_t*)alias1), GET_ALIAS_STR(*(uint16_t*)alias2));
michael@0:     if (!result) {
michael@0:         /* Sort the shortest first */
michael@0:         return (int)uprv_strlen(GET_ALIAS_STR(*(uint16_t*)alias1)) - (int)uprv_strlen(GET_ALIAS_STR(*(uint16_t*)alias2));
michael@0:     }
michael@0:     return result;
michael@0: }
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:  */
michael@0: