intl/icu/source/tools/toolutil/package.cpp

changeset 0
6474c204b198
     1.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.2 +++ b/intl/icu/source/tools/toolutil/package.cpp	Wed Dec 31 06:09:35 2014 +0100
     1.3 @@ -0,0 +1,1313 @@
     1.4 +/*
     1.5 +*******************************************************************************
     1.6 +*
     1.7 +*   Copyright (C) 1999-2013, International Business Machines
     1.8 +*   Corporation and others.  All Rights Reserved.
     1.9 +*
    1.10 +*******************************************************************************
    1.11 +*   file name:  package.cpp
    1.12 +*   encoding:   US-ASCII
    1.13 +*   tab size:   8 (not used)
    1.14 +*   indentation:4
    1.15 +*
    1.16 +*   created on: 2005aug25
    1.17 +*   created by: Markus W. Scherer
    1.18 +*
    1.19 +*   Read, modify, and write ICU .dat data package files.
    1.20 +*   This is an integral part of the icupkg tool, moved to the toolutil library
    1.21 +*   because parts of tool implementations tend to be later shared by
    1.22 +*   other tools.
    1.23 +*   Subsumes functionality and implementation code from
    1.24 +*   gencmn, decmn, and icuswap tools.
    1.25 +*/
    1.26 +
    1.27 +#include "unicode/utypes.h"
    1.28 +#include "unicode/putil.h"
    1.29 +#include "unicode/udata.h"
    1.30 +#include "cstring.h"
    1.31 +#include "uarrsort.h"
    1.32 +#include "ucmndata.h"
    1.33 +#include "udataswp.h"
    1.34 +#include "swapimpl.h"
    1.35 +#include "toolutil.h"
    1.36 +#include "package.h"
    1.37 +#include "cmemory.h"
    1.38 +
    1.39 +#include <stdio.h>
    1.40 +#include <stdlib.h>
    1.41 +#include <string.h>
    1.42 +
    1.43 +
    1.44 +static const int32_t kItemsChunk = 256; /* How much to increase the filesarray by each time */
    1.45 +
    1.46 +// general definitions ----------------------------------------------------- ***
    1.47 +
    1.48 +#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
    1.49 +
    1.50 +/* UDataInfo cf. udata.h */
    1.51 +static const UDataInfo dataInfo={
    1.52 +    (uint16_t)sizeof(UDataInfo),
    1.53 +    0,
    1.54 +
    1.55 +    U_IS_BIG_ENDIAN,
    1.56 +    U_CHARSET_FAMILY,
    1.57 +    (uint8_t)sizeof(UChar),
    1.58 +    0,
    1.59 +
    1.60 +    {0x43, 0x6d, 0x6e, 0x44},     /* dataFormat="CmnD" */
    1.61 +    {1, 0, 0, 0},                 /* formatVersion */
    1.62 +    {3, 0, 0, 0}                  /* dataVersion */
    1.63 +};
    1.64 +
    1.65 +U_CDECL_BEGIN
    1.66 +static void U_CALLCONV
    1.67 +printPackageError(void *context, const char *fmt, va_list args) {
    1.68 +    vfprintf((FILE *)context, fmt, args);
    1.69 +}
    1.70 +U_CDECL_END
    1.71 +
    1.72 +static uint16_t
    1.73 +readSwapUInt16(uint16_t x) {
    1.74 +    return (uint16_t)((x<<8)|(x>>8));
    1.75 +}
    1.76 +
    1.77 +// platform types ---------------------------------------------------------- ***
    1.78 +
    1.79 +static const char *types="lb?e";
    1.80 +
    1.81 +enum { TYPE_L, TYPE_B, TYPE_LE, TYPE_E, TYPE_COUNT };
    1.82 +
    1.83 +static inline int32_t
    1.84 +makeTypeEnum(uint8_t charset, UBool isBigEndian) {
    1.85 +    return 2*(int32_t)charset+isBigEndian;
    1.86 +}
    1.87 +
    1.88 +static inline int32_t
    1.89 +makeTypeEnum(char type) {
    1.90 +    return
    1.91 +        type == 'l' ? TYPE_L :
    1.92 +        type == 'b' ? TYPE_B :
    1.93 +        type == 'e' ? TYPE_E :
    1.94 +               -1;
    1.95 +}
    1.96 +
    1.97 +static inline char
    1.98 +makeTypeLetter(uint8_t charset, UBool isBigEndian) {
    1.99 +    return types[makeTypeEnum(charset, isBigEndian)];
   1.100 +}
   1.101 +
   1.102 +static inline char
   1.103 +makeTypeLetter(int32_t typeEnum) {
   1.104 +    return types[typeEnum];
   1.105 +}
   1.106 +
   1.107 +static void
   1.108 +makeTypeProps(char type, uint8_t &charset, UBool &isBigEndian) {
   1.109 +    int32_t typeEnum=makeTypeEnum(type);
   1.110 +    charset=(uint8_t)(typeEnum>>1);
   1.111 +    isBigEndian=(UBool)(typeEnum&1);
   1.112 +}
   1.113 +
   1.114 +U_CFUNC const UDataInfo *
   1.115 +getDataInfo(const uint8_t *data, int32_t length,
   1.116 +            int32_t &infoLength, int32_t &headerLength,
   1.117 +            UErrorCode *pErrorCode) {
   1.118 +    const DataHeader *pHeader;
   1.119 +    const UDataInfo *pInfo;
   1.120 +
   1.121 +    if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
   1.122 +        return NULL;
   1.123 +    }
   1.124 +    if( data==NULL ||
   1.125 +        (length>=0 && length<(int32_t)sizeof(DataHeader))
   1.126 +    ) {
   1.127 +        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
   1.128 +        return NULL;
   1.129 +    }
   1.130 +
   1.131 +    pHeader=(const DataHeader *)data;
   1.132 +    pInfo=&pHeader->info;
   1.133 +    if( (length>=0 && length<(int32_t)sizeof(DataHeader)) ||
   1.134 +        pHeader->dataHeader.magic1!=0xda ||
   1.135 +        pHeader->dataHeader.magic2!=0x27 ||
   1.136 +        pInfo->sizeofUChar!=2
   1.137 +    ) {
   1.138 +        *pErrorCode=U_UNSUPPORTED_ERROR;
   1.139 +        return NULL;
   1.140 +    }
   1.141 +
   1.142 +    if(pInfo->isBigEndian==U_IS_BIG_ENDIAN) {
   1.143 +        headerLength=pHeader->dataHeader.headerSize;
   1.144 +        infoLength=pInfo->size;
   1.145 +    } else {
   1.146 +        headerLength=readSwapUInt16(pHeader->dataHeader.headerSize);
   1.147 +        infoLength=readSwapUInt16(pInfo->size);
   1.148 +    }
   1.149 +
   1.150 +    if( headerLength<(int32_t)sizeof(DataHeader) ||
   1.151 +        infoLength<(int32_t)sizeof(UDataInfo) ||
   1.152 +        headerLength<(int32_t)(sizeof(pHeader->dataHeader)+infoLength) ||
   1.153 +        (length>=0 && length<headerLength)
   1.154 +    ) {
   1.155 +        *pErrorCode=U_UNSUPPORTED_ERROR;
   1.156 +        return NULL;
   1.157 +    }
   1.158 +
   1.159 +    return pInfo;
   1.160 +}
   1.161 +
   1.162 +static int32_t
   1.163 +getTypeEnumForInputData(const uint8_t *data, int32_t length,
   1.164 +                        UErrorCode *pErrorCode) {
   1.165 +    const UDataInfo *pInfo;
   1.166 +    int32_t infoLength, headerLength;
   1.167 +
   1.168 +    /* getDataInfo() checks for illegal arguments */
   1.169 +    pInfo=getDataInfo(data, length, infoLength, headerLength, pErrorCode);
   1.170 +    if(pInfo==NULL) {
   1.171 +        return -1;
   1.172 +    }
   1.173 +
   1.174 +    return makeTypeEnum(pInfo->charsetFamily, (UBool)pInfo->isBigEndian);
   1.175 +}
   1.176 +
   1.177 +// file handling ----------------------------------------------------------- ***
   1.178 +
   1.179 +static void
   1.180 +extractPackageName(const char *filename,
   1.181 +                   char pkg[], int32_t capacity) {
   1.182 +    const char *basename;
   1.183 +    int32_t len;
   1.184 +
   1.185 +    basename=findBasename(filename);
   1.186 +    len=(int32_t)strlen(basename)-4; /* -4: subtract the length of ".dat" */
   1.187 +
   1.188 +    if(len<=0 || 0!=strcmp(basename+len, ".dat")) {
   1.189 +        fprintf(stderr, "icupkg: \"%s\" is not recognized as a package filename (must end with .dat)\n",
   1.190 +                         basename);
   1.191 +        exit(U_ILLEGAL_ARGUMENT_ERROR);
   1.192 +    }
   1.193 +
   1.194 +    if(len>=capacity) {
   1.195 +        fprintf(stderr, "icupkg: the package name \"%s\" is too long (>=%ld)\n",
   1.196 +                         basename, (long)capacity);
   1.197 +        exit(U_ILLEGAL_ARGUMENT_ERROR);
   1.198 +    }
   1.199 +
   1.200 +    memcpy(pkg, basename, len);
   1.201 +    pkg[len]=0;
   1.202 +}
   1.203 +
   1.204 +static int32_t
   1.205 +getFileLength(FILE *f) {
   1.206 +    int32_t length;
   1.207 +
   1.208 +    fseek(f, 0, SEEK_END);
   1.209 +    length=(int32_t)ftell(f);
   1.210 +    fseek(f, 0, SEEK_SET);
   1.211 +    return length;
   1.212 +}
   1.213 +
   1.214 +/*
   1.215 + * Turn tree separators and alternate file separators into normal file separators.
   1.216 + */
   1.217 +#if U_TREE_ENTRY_SEP_CHAR==U_FILE_SEP_CHAR && U_FILE_ALT_SEP_CHAR==U_FILE_SEP_CHAR
   1.218 +#define treeToPath(s)
   1.219 +#else
   1.220 +static void
   1.221 +treeToPath(char *s) {
   1.222 +    char *t;
   1.223 +
   1.224 +    for(t=s; *t!=0; ++t) {
   1.225 +        if(*t==U_TREE_ENTRY_SEP_CHAR || *t==U_FILE_ALT_SEP_CHAR) {
   1.226 +            *t=U_FILE_SEP_CHAR;
   1.227 +        }
   1.228 +    }
   1.229 +}
   1.230 +#endif
   1.231 +
   1.232 +/*
   1.233 + * Turn file separators into tree separators.
   1.234 + */
   1.235 +#if U_TREE_ENTRY_SEP_CHAR==U_FILE_SEP_CHAR && U_FILE_ALT_SEP_CHAR==U_FILE_SEP_CHAR
   1.236 +#define pathToTree(s)
   1.237 +#else
   1.238 +static void
   1.239 +pathToTree(char *s) {
   1.240 +    char *t;
   1.241 +
   1.242 +    for(t=s; *t!=0; ++t) {
   1.243 +        if(*t==U_FILE_SEP_CHAR || *t==U_FILE_ALT_SEP_CHAR) {
   1.244 +            *t=U_TREE_ENTRY_SEP_CHAR;
   1.245 +        }
   1.246 +    }
   1.247 +}
   1.248 +#endif
   1.249 +
   1.250 +/*
   1.251 + * Prepend the path (if any) to the name and run the name through treeToName().
   1.252 + */
   1.253 +static void
   1.254 +makeFullFilename(const char *path, const char *name,
   1.255 +                 char *filename, int32_t capacity) {
   1.256 +    char *s;
   1.257 +
   1.258 +    // prepend the path unless NULL or empty
   1.259 +    if(path!=NULL && path[0]!=0) {
   1.260 +        if((int32_t)(strlen(path)+1)>=capacity) {
   1.261 +            fprintf(stderr, "pathname too long: \"%s\"\n", path);
   1.262 +            exit(U_BUFFER_OVERFLOW_ERROR);
   1.263 +        }
   1.264 +        strcpy(filename, path);
   1.265 +
   1.266 +        // make sure the path ends with a file separator
   1.267 +        s=strchr(filename, 0);
   1.268 +        if(*(s-1)!=U_FILE_SEP_CHAR && *(s-1)!=U_FILE_ALT_SEP_CHAR) {
   1.269 +            *s++=U_FILE_SEP_CHAR;
   1.270 +        }
   1.271 +    } else {
   1.272 +        s=filename;
   1.273 +    }
   1.274 +
   1.275 +    // turn the name into a filename, turn tree separators into file separators
   1.276 +    if((int32_t)((s-filename)+strlen(name))>=capacity) {
   1.277 +        fprintf(stderr, "path/filename too long: \"%s%s\"\n", filename, name);
   1.278 +        exit(U_BUFFER_OVERFLOW_ERROR);
   1.279 +    }
   1.280 +    strcpy(s, name);
   1.281 +    treeToPath(s);
   1.282 +}
   1.283 +
   1.284 +static void
   1.285 +makeFullFilenameAndDirs(const char *path, const char *name,
   1.286 +                        char *filename, int32_t capacity) {
   1.287 +    char *sep;
   1.288 +    UErrorCode errorCode;
   1.289 +
   1.290 +    makeFullFilename(path, name, filename, capacity);
   1.291 +
   1.292 +    // make tree directories
   1.293 +    errorCode=U_ZERO_ERROR;
   1.294 +    sep=strchr(filename, 0)-strlen(name);
   1.295 +    while((sep=strchr(sep, U_FILE_SEP_CHAR))!=NULL) {
   1.296 +        if(sep!=filename) {
   1.297 +            *sep=0;                 // truncate temporarily
   1.298 +            uprv_mkdir(filename, &errorCode);
   1.299 +            if(U_FAILURE(errorCode)) {
   1.300 +                fprintf(stderr, "icupkg: unable to create tree directory \"%s\"\n", filename);
   1.301 +                exit(U_FILE_ACCESS_ERROR);
   1.302 +            }
   1.303 +        }
   1.304 +        *sep++=U_FILE_SEP_CHAR; // restore file separator character
   1.305 +    }
   1.306 +}
   1.307 +
   1.308 +static uint8_t *
   1.309 +readFile(const char *path, const char *name, int32_t &length, char &type) {
   1.310 +    char filename[1024];
   1.311 +    FILE *file;
   1.312 +    uint8_t *data;
   1.313 +    UErrorCode errorCode;
   1.314 +    int32_t fileLength, typeEnum;
   1.315 +
   1.316 +    makeFullFilename(path, name, filename, (int32_t)sizeof(filename));
   1.317 +
   1.318 +    /* open the input file, get its length, allocate memory for it, read the file */
   1.319 +    file=fopen(filename, "rb");
   1.320 +    if(file==NULL) {
   1.321 +        fprintf(stderr, "icupkg: unable to open input file \"%s\"\n", filename);
   1.322 +        exit(U_FILE_ACCESS_ERROR);
   1.323 +    }
   1.324 +
   1.325 +    /* get the file length */
   1.326 +    fileLength=getFileLength(file);
   1.327 +    if(ferror(file) || fileLength<=0) {
   1.328 +        fprintf(stderr, "icupkg: empty input file \"%s\"\n", filename);
   1.329 +        fclose(file);
   1.330 +        exit(U_FILE_ACCESS_ERROR);
   1.331 +    }
   1.332 +
   1.333 +    /* allocate the buffer, pad to multiple of 16 */
   1.334 +    length=(fileLength+0xf)&~0xf;
   1.335 +    data=(uint8_t *)uprv_malloc(length);
   1.336 +    if(data==NULL) {
   1.337 +        fclose(file);
   1.338 +        fprintf(stderr, "icupkg: malloc error allocating %d bytes.\n", (int)length);
   1.339 +        exit(U_MEMORY_ALLOCATION_ERROR);
   1.340 +    }
   1.341 +
   1.342 +    /* read the file */
   1.343 +    if(fileLength!=(int32_t)fread(data, 1, fileLength, file)) {
   1.344 +        fprintf(stderr, "icupkg: error reading \"%s\"\n", filename);
   1.345 +        fclose(file);
   1.346 +        free(data);
   1.347 +        exit(U_FILE_ACCESS_ERROR);
   1.348 +    }
   1.349 +
   1.350 +    /* pad the file to a multiple of 16 using the usual padding byte */
   1.351 +    if(fileLength<length) {
   1.352 +        memset(data+fileLength, 0xaa, length-fileLength);
   1.353 +    }
   1.354 +
   1.355 +    fclose(file);
   1.356 +
   1.357 +    // minimum check for ICU-format data
   1.358 +    errorCode=U_ZERO_ERROR;
   1.359 +    typeEnum=getTypeEnumForInputData(data, length, &errorCode);
   1.360 +    if(typeEnum<0 || U_FAILURE(errorCode)) {
   1.361 +        fprintf(stderr, "icupkg: not an ICU data file: \"%s\"\n", filename);
   1.362 +        free(data);
   1.363 +#if !UCONFIG_NO_LEGACY_CONVERSION
   1.364 +        exit(U_INVALID_FORMAT_ERROR);
   1.365 +#else
   1.366 +        fprintf(stderr, "U_INVALID_FORMAT_ERROR occurred but UCONFIG_NO_LEGACY_CONVERSION is on so this is expected.\n");
   1.367 +        exit(0);
   1.368 +#endif
   1.369 +    }
   1.370 +    type=makeTypeLetter(typeEnum);
   1.371 +
   1.372 +    return data;
   1.373 +}
   1.374 +
   1.375 +// .dat package file representation ---------------------------------------- ***
   1.376 +
   1.377 +U_CDECL_BEGIN
   1.378 +
   1.379 +static int32_t U_CALLCONV
   1.380 +compareItems(const void * /*context*/, const void *left, const void *right) {
   1.381 +    U_NAMESPACE_USE
   1.382 +
   1.383 +    return (int32_t)strcmp(((Item *)left)->name, ((Item *)right)->name);
   1.384 +}
   1.385 +
   1.386 +U_CDECL_END
   1.387 +
   1.388 +U_NAMESPACE_BEGIN
   1.389 +
   1.390 +Package::Package() 
   1.391 +        : doAutoPrefix(FALSE), prefixEndsWithType(FALSE) {
   1.392 +    inPkgName[0]=0;
   1.393 +    pkgPrefix[0]=0;
   1.394 +    inData=NULL;
   1.395 +    inLength=0;
   1.396 +    inCharset=U_CHARSET_FAMILY;
   1.397 +    inIsBigEndian=U_IS_BIG_ENDIAN;
   1.398 +
   1.399 +    itemCount=0;
   1.400 +    itemMax=0;
   1.401 +    items=NULL;
   1.402 +
   1.403 +    inStringTop=outStringTop=0;
   1.404 +
   1.405 +    matchMode=0;
   1.406 +    findPrefix=findSuffix=NULL;
   1.407 +    findPrefixLength=findSuffixLength=0;
   1.408 +    findNextIndex=-1;
   1.409 +
   1.410 +    // create a header for an empty package
   1.411 +    DataHeader *pHeader;
   1.412 +    pHeader=(DataHeader *)header;
   1.413 +    pHeader->dataHeader.magic1=0xda;
   1.414 +    pHeader->dataHeader.magic2=0x27;
   1.415 +    memcpy(&pHeader->info, &dataInfo, sizeof(dataInfo));
   1.416 +    headerLength=(int32_t)(4+sizeof(dataInfo));
   1.417 +    if(headerLength&0xf) {
   1.418 +        /* NUL-pad the header to a multiple of 16 */
   1.419 +        int32_t length=(headerLength+0xf)&~0xf;
   1.420 +        memset(header+headerLength, 0, length-headerLength);
   1.421 +        headerLength=length;
   1.422 +    }
   1.423 +    pHeader->dataHeader.headerSize=(uint16_t)headerLength;
   1.424 +}
   1.425 +
   1.426 +Package::~Package() {
   1.427 +    int32_t idx;
   1.428 +
   1.429 +    free(inData);
   1.430 +
   1.431 +    for(idx=0; idx<itemCount; ++idx) {
   1.432 +        if(items[idx].isDataOwned) {
   1.433 +            free(items[idx].data);
   1.434 +        }
   1.435 +    }
   1.436 +
   1.437 +    uprv_free((void*)items);
   1.438 +}
   1.439 +
   1.440 +void
   1.441 +Package::setPrefix(const char *p) {
   1.442 +    if(strlen(p)>=sizeof(pkgPrefix)) {
   1.443 +        fprintf(stderr, "icupkg: --toc_prefix %s too long\n", p);
   1.444 +        exit(U_ILLEGAL_ARGUMENT_ERROR);
   1.445 +    }
   1.446 +    strcpy(pkgPrefix, p);
   1.447 +}
   1.448 +
   1.449 +void
   1.450 +Package::readPackage(const char *filename) {
   1.451 +    UDataSwapper *ds;
   1.452 +    const UDataInfo *pInfo;
   1.453 +    UErrorCode errorCode;
   1.454 +
   1.455 +    const uint8_t *inBytes;
   1.456 +
   1.457 +    int32_t length, offset, i;
   1.458 +    int32_t itemLength, typeEnum;
   1.459 +    char type;
   1.460 +
   1.461 +    const UDataOffsetTOCEntry *inEntries;
   1.462 +
   1.463 +    extractPackageName(filename, inPkgName, (int32_t)sizeof(inPkgName));
   1.464 +
   1.465 +    /* read the file */
   1.466 +    inData=readFile(NULL, filename, inLength, type);
   1.467 +    length=inLength;
   1.468 +
   1.469 +    /*
   1.470 +     * swap the header - even if the swapping itself is a no-op
   1.471 +     * because it tells us the header length
   1.472 +     */
   1.473 +    errorCode=U_ZERO_ERROR;
   1.474 +    makeTypeProps(type, inCharset, inIsBigEndian);
   1.475 +    ds=udata_openSwapper(inIsBigEndian, inCharset, U_IS_BIG_ENDIAN, U_CHARSET_FAMILY, &errorCode);
   1.476 +    if(U_FAILURE(errorCode)) {
   1.477 +        fprintf(stderr, "icupkg: udata_openSwapper(\"%s\") failed - %s\n",
   1.478 +                filename, u_errorName(errorCode));
   1.479 +        exit(errorCode);
   1.480 +    }
   1.481 +
   1.482 +    ds->printError=printPackageError;
   1.483 +    ds->printErrorContext=stderr;
   1.484 +
   1.485 +    headerLength=sizeof(header);
   1.486 +    if(length<headerLength) {
   1.487 +        headerLength=length;
   1.488 +    }
   1.489 +    headerLength=udata_swapDataHeader(ds, inData, headerLength, header, &errorCode);
   1.490 +    if(U_FAILURE(errorCode)) {
   1.491 +        exit(errorCode);
   1.492 +    }
   1.493 +
   1.494 +    /* check data format and format version */
   1.495 +    pInfo=(const UDataInfo *)((const char *)inData+4);
   1.496 +    if(!(
   1.497 +        pInfo->dataFormat[0]==0x43 &&   /* dataFormat="CmnD" */
   1.498 +        pInfo->dataFormat[1]==0x6d &&
   1.499 +        pInfo->dataFormat[2]==0x6e &&
   1.500 +        pInfo->dataFormat[3]==0x44 &&
   1.501 +        pInfo->formatVersion[0]==1
   1.502 +    )) {
   1.503 +        fprintf(stderr, "icupkg: data format %02x.%02x.%02x.%02x (format version %02x) is not recognized as an ICU .dat package\n",
   1.504 +                pInfo->dataFormat[0], pInfo->dataFormat[1],
   1.505 +                pInfo->dataFormat[2], pInfo->dataFormat[3],
   1.506 +                pInfo->formatVersion[0]);
   1.507 +        exit(U_UNSUPPORTED_ERROR);
   1.508 +    }
   1.509 +    inIsBigEndian=(UBool)pInfo->isBigEndian;
   1.510 +    inCharset=pInfo->charsetFamily;
   1.511 +
   1.512 +    inBytes=(const uint8_t *)inData+headerLength;
   1.513 +    inEntries=(const UDataOffsetTOCEntry *)(inBytes+4);
   1.514 +
   1.515 +    /* check that the itemCount fits, then the ToC table, then at least the header of the last item */
   1.516 +    length-=headerLength;
   1.517 +    if(length<4) {
   1.518 +        /* itemCount does not fit */
   1.519 +        offset=0x7fffffff;
   1.520 +    } else {
   1.521 +        itemCount=udata_readInt32(ds, *(const int32_t *)inBytes);
   1.522 +        setItemCapacity(itemCount); /* resize so there's space */
   1.523 +        if(itemCount==0) {
   1.524 +            offset=4;
   1.525 +        } else if(length<(4+8*itemCount)) {
   1.526 +            /* ToC table does not fit */
   1.527 +            offset=0x7fffffff;
   1.528 +        } else {
   1.529 +            /* offset of the last item plus at least 20 bytes for its header */
   1.530 +            offset=20+(int32_t)ds->readUInt32(inEntries[itemCount-1].dataOffset);
   1.531 +        }
   1.532 +    }
   1.533 +    if(length<offset) {
   1.534 +        fprintf(stderr, "icupkg: too few bytes (%ld after header) for a .dat package\n",
   1.535 +                        (long)length);
   1.536 +        exit(U_INDEX_OUTOFBOUNDS_ERROR);
   1.537 +    }
   1.538 +    /* do not modify the package length variable until the last item's length is set */
   1.539 +
   1.540 +    if(itemCount<=0) {
   1.541 +        if(doAutoPrefix) {
   1.542 +            fprintf(stderr, "icupkg: --auto_toc_prefix[_with_type] but the input package is empty\n");
   1.543 +            exit(U_INVALID_FORMAT_ERROR);
   1.544 +        }
   1.545 +    } else {
   1.546 +        char prefix[MAX_PKG_NAME_LENGTH+4];
   1.547 +        char *s, *inItemStrings;
   1.548 +
   1.549 +        if(itemCount>itemMax) {
   1.550 +            fprintf(stderr, "icupkg: too many items, maximum is %d\n", itemMax);
   1.551 +            exit(U_BUFFER_OVERFLOW_ERROR);
   1.552 +        }
   1.553 +
   1.554 +        /* swap the item name strings */
   1.555 +        int32_t stringsOffset=4+8*itemCount;
   1.556 +        itemLength=(int32_t)(ds->readUInt32(inEntries[0].dataOffset))-stringsOffset;
   1.557 +
   1.558 +        // don't include padding bytes at the end of the item names
   1.559 +        while(itemLength>0 && inBytes[stringsOffset+itemLength-1]!=0) {
   1.560 +            --itemLength;
   1.561 +        }
   1.562 +
   1.563 +        if((inStringTop+itemLength)>STRING_STORE_SIZE) {
   1.564 +            fprintf(stderr, "icupkg: total length of item name strings too long\n");
   1.565 +            exit(U_BUFFER_OVERFLOW_ERROR);
   1.566 +        }
   1.567 +
   1.568 +        inItemStrings=inStrings+inStringTop;
   1.569 +        ds->swapInvChars(ds, inBytes+stringsOffset, itemLength, inItemStrings, &errorCode);
   1.570 +        if(U_FAILURE(errorCode)) {
   1.571 +            fprintf(stderr, "icupkg failed to swap the input .dat package item name strings\n");
   1.572 +            exit(U_INVALID_FORMAT_ERROR);
   1.573 +        }
   1.574 +        inStringTop+=itemLength;
   1.575 +
   1.576 +        // reset the Item entries
   1.577 +        memset(items, 0, itemCount*sizeof(Item));
   1.578 +
   1.579 +        /*
   1.580 +         * Get the common prefix of the items.
   1.581 +         * New-style ICU .dat packages use tree separators ('/') between package names,
   1.582 +         * tree names, and item names,
   1.583 +         * while old-style ICU .dat packages (before multi-tree support)
   1.584 +         * use an underscore ('_') between package and item names.
   1.585 +         */
   1.586 +        offset=(int32_t)ds->readUInt32(inEntries[0].nameOffset)-stringsOffset;
   1.587 +        s=inItemStrings+offset;  // name of the first entry
   1.588 +        int32_t prefixLength;
   1.589 +        if(doAutoPrefix) {
   1.590 +            // Use the first entry's prefix. Must be a new-style package.
   1.591 +            const char *prefixLimit=strchr(s, U_TREE_ENTRY_SEP_CHAR);
   1.592 +            if(prefixLimit==NULL) {
   1.593 +                fprintf(stderr,
   1.594 +                        "icupkg: --auto_toc_prefix[_with_type] but "
   1.595 +                        "the first entry \"%s\" does not contain a '%c'\n",
   1.596 +                        s, U_TREE_ENTRY_SEP_CHAR);
   1.597 +                exit(U_INVALID_FORMAT_ERROR);
   1.598 +            }
   1.599 +            prefixLength=(int32_t)(prefixLimit-s);
   1.600 +            if(prefixLength==0 || prefixLength>=LENGTHOF(pkgPrefix)) {
   1.601 +                fprintf(stderr,
   1.602 +                        "icupkg: --auto_toc_prefix[_with_type] but "
   1.603 +                        "the prefix of the first entry \"%s\" is empty or too long\n",
   1.604 +                        s);
   1.605 +                exit(U_INVALID_FORMAT_ERROR);
   1.606 +            }
   1.607 +            if(prefixEndsWithType && s[prefixLength-1]!=type) {
   1.608 +                fprintf(stderr,
   1.609 +                        "icupkg: --auto_toc_prefix_with_type but "
   1.610 +                        "the prefix of the first entry \"%s\" does not end with '%c'\n",
   1.611 +                        s, type);
   1.612 +                exit(U_INVALID_FORMAT_ERROR);
   1.613 +            }
   1.614 +            memcpy(pkgPrefix, s, prefixLength);
   1.615 +            memcpy(prefix, s, ++prefixLength);  // include the /
   1.616 +        } else {
   1.617 +            // Use the package basename as prefix.
   1.618 +            int32_t inPkgNameLength=strlen(inPkgName);
   1.619 +            memcpy(prefix, inPkgName, inPkgNameLength);
   1.620 +            prefixLength=inPkgNameLength;
   1.621 +
   1.622 +            if( (int32_t)strlen(s)>=(inPkgNameLength+2) &&
   1.623 +                0==memcmp(s, inPkgName, inPkgNameLength) &&
   1.624 +                s[inPkgNameLength]=='_'
   1.625 +            ) {
   1.626 +                // old-style .dat package
   1.627 +                prefix[prefixLength++]='_';
   1.628 +            } else {
   1.629 +                // new-style .dat package
   1.630 +                prefix[prefixLength++]=U_TREE_ENTRY_SEP_CHAR;
   1.631 +                // if it turns out to not contain U_TREE_ENTRY_SEP_CHAR
   1.632 +                // then the test in the loop below will fail
   1.633 +            }
   1.634 +        }
   1.635 +        prefix[prefixLength]=0;
   1.636 +
   1.637 +        /* read the ToC table */
   1.638 +        for(i=0; i<itemCount; ++i) {
   1.639 +            // skip the package part of the item name, error if it does not match the actual package name
   1.640 +            // or if nothing follows the package name
   1.641 +            offset=(int32_t)ds->readUInt32(inEntries[i].nameOffset)-stringsOffset;
   1.642 +            s=inItemStrings+offset;
   1.643 +            if(0!=strncmp(s, prefix, prefixLength) || s[prefixLength]==0) {
   1.644 +                fprintf(stderr, "icupkg: input .dat item name \"%s\" does not start with \"%s\"\n",
   1.645 +                        s, prefix);
   1.646 +                exit(U_INVALID_FORMAT_ERROR);
   1.647 +            }
   1.648 +            items[i].name=s+prefixLength;
   1.649 +
   1.650 +            // set the item's data
   1.651 +            items[i].data=(uint8_t *)inBytes+ds->readUInt32(inEntries[i].dataOffset);
   1.652 +            if(i>0) {
   1.653 +                items[i-1].length=(int32_t)(items[i].data-items[i-1].data);
   1.654 +
   1.655 +                // set the previous item's platform type
   1.656 +                typeEnum=getTypeEnumForInputData(items[i-1].data, items[i-1].length, &errorCode);
   1.657 +                if(typeEnum<0 || U_FAILURE(errorCode)) {
   1.658 +                    fprintf(stderr, "icupkg: not an ICU data file: item \"%s\" in \"%s\"\n", items[i-1].name, filename);
   1.659 +                    exit(U_INVALID_FORMAT_ERROR);
   1.660 +                }
   1.661 +                items[i-1].type=makeTypeLetter(typeEnum);
   1.662 +            }
   1.663 +            items[i].isDataOwned=FALSE;
   1.664 +        }
   1.665 +        // set the last item's length
   1.666 +        items[itemCount-1].length=length-ds->readUInt32(inEntries[itemCount-1].dataOffset);
   1.667 +
   1.668 +        // set the last item's platform type
   1.669 +        typeEnum=getTypeEnumForInputData(items[itemCount-1].data, items[itemCount-1].length, &errorCode);
   1.670 +        if(typeEnum<0 || U_FAILURE(errorCode)) {
   1.671 +            fprintf(stderr, "icupkg: not an ICU data file: item \"%s\" in \"%s\"\n", items[i-1].name, filename);
   1.672 +            exit(U_INVALID_FORMAT_ERROR);
   1.673 +        }
   1.674 +        items[itemCount-1].type=makeTypeLetter(typeEnum);
   1.675 +
   1.676 +        if(type!=U_ICUDATA_TYPE_LETTER[0]) {
   1.677 +            // sort the item names for the local charset
   1.678 +            sortItems();
   1.679 +        }
   1.680 +    }
   1.681 +
   1.682 +    udata_closeSwapper(ds);
   1.683 +}
   1.684 +
   1.685 +char
   1.686 +Package::getInType() {
   1.687 +    return makeTypeLetter(inCharset, inIsBigEndian);
   1.688 +}
   1.689 +
   1.690 +void
   1.691 +Package::writePackage(const char *filename, char outType, const char *comment) {
   1.692 +    char prefix[MAX_PKG_NAME_LENGTH+4];
   1.693 +    UDataOffsetTOCEntry entry;
   1.694 +    UDataSwapper *dsLocalToOut, *ds[TYPE_COUNT];
   1.695 +    FILE *file;
   1.696 +    Item *pItem;
   1.697 +    char *name;
   1.698 +    UErrorCode errorCode;
   1.699 +    int32_t i, length, prefixLength, maxItemLength, basenameOffset, offset, outInt32;
   1.700 +    uint8_t outCharset;
   1.701 +    UBool outIsBigEndian;
   1.702 +
   1.703 +    extractPackageName(filename, prefix, MAX_PKG_NAME_LENGTH);
   1.704 +
   1.705 +    // if there is an explicit comment, then use it, else use what's in the current header
   1.706 +    if(comment!=NULL) {
   1.707 +        /* get the header size minus the current comment */
   1.708 +        DataHeader *pHeader;
   1.709 +        int32_t length;
   1.710 +
   1.711 +        pHeader=(DataHeader *)header;
   1.712 +        headerLength=4+pHeader->info.size;
   1.713 +        length=(int32_t)strlen(comment);
   1.714 +        if((int32_t)(headerLength+length)>=(int32_t)sizeof(header)) {
   1.715 +            fprintf(stderr, "icupkg: comment too long\n");
   1.716 +            exit(U_BUFFER_OVERFLOW_ERROR);
   1.717 +        }
   1.718 +        memcpy(header+headerLength, comment, length+1);
   1.719 +        headerLength+=length;
   1.720 +        if(headerLength&0xf) {
   1.721 +            /* NUL-pad the header to a multiple of 16 */
   1.722 +            length=(headerLength+0xf)&~0xf;
   1.723 +            memset(header+headerLength, 0, length-headerLength);
   1.724 +            headerLength=length;
   1.725 +        }
   1.726 +        pHeader->dataHeader.headerSize=(uint16_t)headerLength;
   1.727 +    }
   1.728 +
   1.729 +    makeTypeProps(outType, outCharset, outIsBigEndian);
   1.730 +
   1.731 +    // open (TYPE_COUNT-2) swappers
   1.732 +    // one is a no-op for local type==outType
   1.733 +    // one type (TYPE_LE) is bogus
   1.734 +    errorCode=U_ZERO_ERROR;
   1.735 +    i=makeTypeEnum(outType);
   1.736 +    ds[TYPE_B]= i==TYPE_B ? NULL : udata_openSwapper(TRUE, U_ASCII_FAMILY, outIsBigEndian, outCharset, &errorCode);
   1.737 +    ds[TYPE_L]= i==TYPE_L ? NULL : udata_openSwapper(FALSE, U_ASCII_FAMILY, outIsBigEndian, outCharset, &errorCode);
   1.738 +    ds[TYPE_LE]=NULL;
   1.739 +    ds[TYPE_E]= i==TYPE_E ? NULL : udata_openSwapper(TRUE, U_EBCDIC_FAMILY, outIsBigEndian, outCharset, &errorCode);
   1.740 +    if(U_FAILURE(errorCode)) {
   1.741 +        fprintf(stderr, "icupkg: udata_openSwapper() failed - %s\n", u_errorName(errorCode));
   1.742 +        exit(errorCode);
   1.743 +    }
   1.744 +    for(i=0; i<TYPE_COUNT; ++i) {
   1.745 +        if(ds[i]!=NULL) {
   1.746 +            ds[i]->printError=printPackageError;
   1.747 +            ds[i]->printErrorContext=stderr;
   1.748 +        }
   1.749 +    }
   1.750 +
   1.751 +    dsLocalToOut=ds[makeTypeEnum(U_CHARSET_FAMILY, U_IS_BIG_ENDIAN)];
   1.752 +
   1.753 +    // create the file and write its contents
   1.754 +    file=fopen(filename, "wb");
   1.755 +    if(file==NULL) {
   1.756 +        fprintf(stderr, "icupkg: unable to create file \"%s\"\n", filename);
   1.757 +        exit(U_FILE_ACCESS_ERROR);
   1.758 +    }
   1.759 +
   1.760 +    // swap and write the header
   1.761 +    if(dsLocalToOut!=NULL) {
   1.762 +        udata_swapDataHeader(dsLocalToOut, header, headerLength, header, &errorCode);
   1.763 +        if(U_FAILURE(errorCode)) {
   1.764 +            fprintf(stderr, "icupkg: udata_swapDataHeader(local to out) failed - %s\n", u_errorName(errorCode));
   1.765 +            exit(errorCode);
   1.766 +        }
   1.767 +    }
   1.768 +    length=(int32_t)fwrite(header, 1, headerLength, file);
   1.769 +    if(length!=headerLength) {
   1.770 +        fprintf(stderr, "icupkg: unable to write complete header to file \"%s\"\n", filename);
   1.771 +        exit(U_FILE_ACCESS_ERROR);
   1.772 +    }
   1.773 +
   1.774 +    // prepare and swap the package name with a tree separator
   1.775 +    // for prepending to item names
   1.776 +    if(pkgPrefix[0]==0) {
   1.777 +        prefixLength=(int32_t)strlen(prefix);
   1.778 +    } else {
   1.779 +        prefixLength=(int32_t)strlen(pkgPrefix);
   1.780 +        memcpy(prefix, pkgPrefix, prefixLength);
   1.781 +        if(prefixEndsWithType) {
   1.782 +            prefix[prefixLength-1]=outType;
   1.783 +        }
   1.784 +    }
   1.785 +    prefix[prefixLength++]=U_TREE_ENTRY_SEP_CHAR;
   1.786 +    prefix[prefixLength]=0;
   1.787 +    if(dsLocalToOut!=NULL) {
   1.788 +        dsLocalToOut->swapInvChars(dsLocalToOut, prefix, prefixLength, prefix, &errorCode);
   1.789 +        if(U_FAILURE(errorCode)) {
   1.790 +            fprintf(stderr, "icupkg: swapInvChars(output package name) failed - %s\n", u_errorName(errorCode));
   1.791 +            exit(errorCode);
   1.792 +        }
   1.793 +
   1.794 +        // swap and sort the item names (sorting needs to be done in the output charset)
   1.795 +        dsLocalToOut->swapInvChars(dsLocalToOut, inStrings, inStringTop, inStrings, &errorCode);
   1.796 +        if(U_FAILURE(errorCode)) {
   1.797 +            fprintf(stderr, "icupkg: swapInvChars(item names) failed - %s\n", u_errorName(errorCode));
   1.798 +            exit(errorCode);
   1.799 +        }
   1.800 +        sortItems();
   1.801 +    }
   1.802 +
   1.803 +    // create the output item names in sorted order, with the package name prepended to each
   1.804 +    for(i=0; i<itemCount; ++i) {
   1.805 +        length=(int32_t)strlen(items[i].name);
   1.806 +        name=allocString(FALSE, length+prefixLength);
   1.807 +        memcpy(name, prefix, prefixLength);
   1.808 +        memcpy(name+prefixLength, items[i].name, length+1);
   1.809 +        items[i].name=name;
   1.810 +    }
   1.811 +
   1.812 +    // calculate offsets for item names and items, pad to 16-align items
   1.813 +    // align only the first item; each item's length is a multiple of 16
   1.814 +    basenameOffset=4+8*itemCount;
   1.815 +    offset=basenameOffset+outStringTop;
   1.816 +    if((length=(offset&15))!=0) {
   1.817 +        length=16-length;
   1.818 +        memset(allocString(FALSE, length-1), 0xaa, length);
   1.819 +        offset+=length;
   1.820 +    }
   1.821 +
   1.822 +    // write the table of contents
   1.823 +    // first the itemCount
   1.824 +    outInt32=itemCount;
   1.825 +    if(dsLocalToOut!=NULL) {
   1.826 +        dsLocalToOut->swapArray32(dsLocalToOut, &outInt32, 4, &outInt32, &errorCode);
   1.827 +        if(U_FAILURE(errorCode)) {
   1.828 +            fprintf(stderr, "icupkg: swapArray32(item count) failed - %s\n", u_errorName(errorCode));
   1.829 +            exit(errorCode);
   1.830 +        }
   1.831 +    }
   1.832 +    length=(int32_t)fwrite(&outInt32, 1, 4, file);
   1.833 +    if(length!=4) {
   1.834 +        fprintf(stderr, "icupkg: unable to write complete item count to file \"%s\"\n", filename);
   1.835 +        exit(U_FILE_ACCESS_ERROR);
   1.836 +    }
   1.837 +
   1.838 +    // then write the item entries (and collect the maxItemLength)
   1.839 +    maxItemLength=0;
   1.840 +    for(i=0; i<itemCount; ++i) {
   1.841 +        entry.nameOffset=(uint32_t)(basenameOffset+(items[i].name-outStrings));
   1.842 +        entry.dataOffset=(uint32_t)offset;
   1.843 +        if(dsLocalToOut!=NULL) {
   1.844 +            dsLocalToOut->swapArray32(dsLocalToOut, &entry, 8, &entry, &errorCode);
   1.845 +            if(U_FAILURE(errorCode)) {
   1.846 +                fprintf(stderr, "icupkg: swapArray32(item entry %ld) failed - %s\n", (long)i, u_errorName(errorCode));
   1.847 +                exit(errorCode);
   1.848 +            }
   1.849 +        }
   1.850 +        length=(int32_t)fwrite(&entry, 1, 8, file);
   1.851 +        if(length!=8) {
   1.852 +            fprintf(stderr, "icupkg: unable to write complete item entry %ld to file \"%s\"\n", (long)i, filename);
   1.853 +            exit(U_FILE_ACCESS_ERROR);
   1.854 +        }
   1.855 +
   1.856 +        length=items[i].length;
   1.857 +        if(length>maxItemLength) {
   1.858 +            maxItemLength=length;
   1.859 +        }
   1.860 +        offset+=length;
   1.861 +    }
   1.862 +
   1.863 +    // write the item names
   1.864 +    length=(int32_t)fwrite(outStrings, 1, outStringTop, file);
   1.865 +    if(length!=outStringTop) {
   1.866 +        fprintf(stderr, "icupkg: unable to write complete item names to file \"%s\"\n", filename);
   1.867 +        exit(U_FILE_ACCESS_ERROR);
   1.868 +    }
   1.869 +
   1.870 +    // write the items
   1.871 +    for(pItem=items, i=0; i<itemCount; ++pItem, ++i) {
   1.872 +        int32_t type=makeTypeEnum(pItem->type);
   1.873 +        if(ds[type]!=NULL) {
   1.874 +            // swap each item from its platform properties to the desired ones
   1.875 +            udata_swap(
   1.876 +                ds[type],
   1.877 +                pItem->data, pItem->length, pItem->data,
   1.878 +                &errorCode);
   1.879 +            if(U_FAILURE(errorCode)) {
   1.880 +                fprintf(stderr, "icupkg: udata_swap(item %ld) failed - %s\n", (long)i, u_errorName(errorCode));
   1.881 +                exit(errorCode);
   1.882 +            }
   1.883 +        }
   1.884 +        length=(int32_t)fwrite(pItem->data, 1, pItem->length, file);
   1.885 +        if(length!=pItem->length) {
   1.886 +            fprintf(stderr, "icupkg: unable to write complete item %ld to file \"%s\"\n", (long)i, filename);
   1.887 +            exit(U_FILE_ACCESS_ERROR);
   1.888 +        }
   1.889 +    }
   1.890 +
   1.891 +    if(ferror(file)) {
   1.892 +        fprintf(stderr, "icupkg: unable to write complete file \"%s\"\n", filename);
   1.893 +        exit(U_FILE_ACCESS_ERROR);
   1.894 +    }
   1.895 +
   1.896 +    fclose(file);
   1.897 +    for(i=0; i<TYPE_COUNT; ++i) {
   1.898 +        udata_closeSwapper(ds[i]);
   1.899 +    }
   1.900 +}
   1.901 +
   1.902 +int32_t
   1.903 +Package::findItem(const char *name, int32_t length) const {
   1.904 +    int32_t i, start, limit;
   1.905 +    int result;
   1.906 +
   1.907 +    /* do a binary search for the string */
   1.908 +    start=0;
   1.909 +    limit=itemCount;
   1.910 +    while(start<limit) {
   1.911 +        i=(start+limit)/2;
   1.912 +        if(length>=0) {
   1.913 +            result=strncmp(name, items[i].name, length);
   1.914 +        } else {
   1.915 +            result=strcmp(name, items[i].name);
   1.916 +        }
   1.917 +
   1.918 +        if(result==0) {
   1.919 +            /* found */
   1.920 +            if(length>=0) {
   1.921 +                /*
   1.922 +                 * if we compared just prefixes, then we may need to back up
   1.923 +                 * to the first item with this prefix
   1.924 +                 */
   1.925 +                while(i>0 && 0==strncmp(name, items[i-1].name, length)) {
   1.926 +                    --i;
   1.927 +                }
   1.928 +            }
   1.929 +            return i;
   1.930 +        } else if(result<0) {
   1.931 +            limit=i;
   1.932 +        } else /* result>0 */ {
   1.933 +            start=i+1;
   1.934 +        }
   1.935 +    }
   1.936 +
   1.937 +    return ~start; /* not found, return binary-not of the insertion point */
   1.938 +}
   1.939 +
   1.940 +void
   1.941 +Package::findItems(const char *pattern) {
   1.942 +    const char *wild;
   1.943 +
   1.944 +    if(pattern==NULL || *pattern==0) {
   1.945 +        findNextIndex=-1;
   1.946 +        return;
   1.947 +    }
   1.948 +
   1.949 +    findPrefix=pattern;
   1.950 +    findSuffix=NULL;
   1.951 +    findSuffixLength=0;
   1.952 +
   1.953 +    wild=strchr(pattern, '*');
   1.954 +    if(wild==NULL) {
   1.955 +        // no wildcard
   1.956 +        findPrefixLength=(int32_t)strlen(pattern);
   1.957 +    } else {
   1.958 +        // one wildcard
   1.959 +        findPrefixLength=(int32_t)(wild-pattern);
   1.960 +        findSuffix=wild+1;
   1.961 +        findSuffixLength=(int32_t)strlen(findSuffix);
   1.962 +        if(NULL!=strchr(findSuffix, '*')) {
   1.963 +            // two or more wildcards
   1.964 +            fprintf(stderr, "icupkg: syntax error (more than one '*') in item pattern \"%s\"\n", pattern);
   1.965 +            exit(U_PARSE_ERROR);
   1.966 +        }
   1.967 +    }
   1.968 +
   1.969 +    if(findPrefixLength==0) {
   1.970 +        findNextIndex=0;
   1.971 +    } else {
   1.972 +        findNextIndex=findItem(findPrefix, findPrefixLength);
   1.973 +    }
   1.974 +}
   1.975 +
   1.976 +int32_t
   1.977 +Package::findNextItem() {
   1.978 +    const char *name, *middle, *treeSep;
   1.979 +    int32_t idx, nameLength, middleLength;
   1.980 +
   1.981 +    if(findNextIndex<0) {
   1.982 +        return -1;
   1.983 +    }
   1.984 +
   1.985 +    while(findNextIndex<itemCount) {
   1.986 +        idx=findNextIndex++;
   1.987 +        name=items[idx].name;
   1.988 +        nameLength=(int32_t)strlen(name);
   1.989 +        if(nameLength<(findPrefixLength+findSuffixLength)) {
   1.990 +            // item name too short for prefix & suffix
   1.991 +            continue;
   1.992 +        }
   1.993 +        if(findPrefixLength>0 && 0!=memcmp(findPrefix, name, findPrefixLength)) {
   1.994 +            // left the range of names with this prefix
   1.995 +            break;
   1.996 +        }
   1.997 +        middle=name+findPrefixLength;
   1.998 +        middleLength=nameLength-findPrefixLength-findSuffixLength;
   1.999 +        if(findSuffixLength>0 && 0!=memcmp(findSuffix, name+(nameLength-findSuffixLength), findSuffixLength)) {
  1.1000 +            // suffix does not match
  1.1001 +            continue;
  1.1002 +        }
  1.1003 +        // prefix & suffix match
  1.1004 +
  1.1005 +        if(matchMode&MATCH_NOSLASH) {
  1.1006 +            treeSep=strchr(middle, U_TREE_ENTRY_SEP_CHAR);
  1.1007 +            if(treeSep!=NULL && (treeSep-middle)<middleLength) {
  1.1008 +                // the middle (matching the * wildcard) contains a tree separator /
  1.1009 +                continue;
  1.1010 +            }
  1.1011 +        }
  1.1012 +
  1.1013 +        // found a matching item
  1.1014 +        return idx;
  1.1015 +    }
  1.1016 +
  1.1017 +    // no more items
  1.1018 +    findNextIndex=-1;
  1.1019 +    return -1;
  1.1020 +}
  1.1021 +
  1.1022 +void
  1.1023 +Package::setMatchMode(uint32_t mode) {
  1.1024 +    matchMode=mode;
  1.1025 +}
  1.1026 +
  1.1027 +void
  1.1028 +Package::addItem(const char *name) {
  1.1029 +    addItem(name, NULL, 0, FALSE, U_ICUDATA_TYPE_LETTER[0]);
  1.1030 +}
  1.1031 +
  1.1032 +void
  1.1033 +Package::addItem(const char *name, uint8_t *data, int32_t length, UBool isDataOwned, char type) {
  1.1034 +    int32_t idx;
  1.1035 +
  1.1036 +    idx=findItem(name);
  1.1037 +    if(idx<0) {
  1.1038 +        // new item, make space at the insertion point
  1.1039 +        ensureItemCapacity();
  1.1040 +        // move the following items down
  1.1041 +        idx=~idx;
  1.1042 +        if(idx<itemCount) {
  1.1043 +            memmove(items+idx+1, items+idx, (itemCount-idx)*sizeof(Item));
  1.1044 +        }
  1.1045 +        ++itemCount;
  1.1046 +
  1.1047 +        // reset this Item entry
  1.1048 +        memset(items+idx, 0, sizeof(Item));
  1.1049 +
  1.1050 +        // copy the item's name
  1.1051 +        items[idx].name=allocString(TRUE, strlen(name));
  1.1052 +        strcpy(items[idx].name, name);
  1.1053 +        pathToTree(items[idx].name);
  1.1054 +    } else {
  1.1055 +        // same-name item found, replace it
  1.1056 +        if(items[idx].isDataOwned) {
  1.1057 +            free(items[idx].data);
  1.1058 +        }
  1.1059 +
  1.1060 +        // keep the item's name since it is the same
  1.1061 +    }
  1.1062 +
  1.1063 +    // set the item's data
  1.1064 +    items[idx].data=data;
  1.1065 +    items[idx].length=length;
  1.1066 +    items[idx].isDataOwned=isDataOwned;
  1.1067 +    items[idx].type=type;
  1.1068 +}
  1.1069 +
  1.1070 +void
  1.1071 +Package::addFile(const char *filesPath, const char *name) {
  1.1072 +    uint8_t *data;
  1.1073 +    int32_t length;
  1.1074 +    char type;
  1.1075 +
  1.1076 +    data=readFile(filesPath, name, length, type);
  1.1077 +    // readFile() exits the tool if it fails
  1.1078 +    addItem(name, data, length, TRUE, type);
  1.1079 +}
  1.1080 +
  1.1081 +void
  1.1082 +Package::addItems(const Package &listPkg) {
  1.1083 +    const Item *pItem;
  1.1084 +    int32_t i;
  1.1085 +
  1.1086 +    for(pItem=listPkg.items, i=0; i<listPkg.itemCount; ++pItem, ++i) {
  1.1087 +        addItem(pItem->name, pItem->data, pItem->length, FALSE, pItem->type);
  1.1088 +    }
  1.1089 +}
  1.1090 +
  1.1091 +void
  1.1092 +Package::removeItem(int32_t idx) {
  1.1093 +    if(idx>=0) {
  1.1094 +        // remove the item
  1.1095 +        if(items[idx].isDataOwned) {
  1.1096 +            free(items[idx].data);
  1.1097 +        }
  1.1098 +
  1.1099 +        // move the following items up
  1.1100 +        if((idx+1)<itemCount) {
  1.1101 +            memmove(items+idx, items+idx+1, (itemCount-(idx+1))*sizeof(Item));
  1.1102 +        }
  1.1103 +        --itemCount;
  1.1104 +
  1.1105 +        if(idx<=findNextIndex) {
  1.1106 +            --findNextIndex;
  1.1107 +        }
  1.1108 +    }
  1.1109 +}
  1.1110 +
  1.1111 +void
  1.1112 +Package::removeItems(const char *pattern) {
  1.1113 +    int32_t idx;
  1.1114 +
  1.1115 +    findItems(pattern);
  1.1116 +    while((idx=findNextItem())>=0) {
  1.1117 +        removeItem(idx);
  1.1118 +    }
  1.1119 +}
  1.1120 +
  1.1121 +void
  1.1122 +Package::removeItems(const Package &listPkg) {
  1.1123 +    const Item *pItem;
  1.1124 +    int32_t i;
  1.1125 +
  1.1126 +    for(pItem=listPkg.items, i=0; i<listPkg.itemCount; ++pItem, ++i) {
  1.1127 +        removeItems(pItem->name);
  1.1128 +    }
  1.1129 +}
  1.1130 +
  1.1131 +void
  1.1132 +Package::extractItem(const char *filesPath, const char *outName, int32_t idx, char outType) {
  1.1133 +    char filename[1024];
  1.1134 +    UDataSwapper *ds;
  1.1135 +    FILE *file;
  1.1136 +    Item *pItem;
  1.1137 +    int32_t fileLength;
  1.1138 +    uint8_t itemCharset, outCharset;
  1.1139 +    UBool itemIsBigEndian, outIsBigEndian;
  1.1140 +
  1.1141 +    if(idx<0 || itemCount<=idx) {
  1.1142 +        return;
  1.1143 +    }
  1.1144 +    pItem=items+idx;
  1.1145 +
  1.1146 +    // swap the data to the outType
  1.1147 +    // outType==0: don't swap
  1.1148 +    if(outType!=0 && pItem->type!=outType) {
  1.1149 +        // open the swapper
  1.1150 +        UErrorCode errorCode=U_ZERO_ERROR;
  1.1151 +        makeTypeProps(pItem->type, itemCharset, itemIsBigEndian);
  1.1152 +        makeTypeProps(outType, outCharset, outIsBigEndian);
  1.1153 +        ds=udata_openSwapper(itemIsBigEndian, itemCharset, outIsBigEndian, outCharset, &errorCode);
  1.1154 +        if(U_FAILURE(errorCode)) {
  1.1155 +            fprintf(stderr, "icupkg: udata_openSwapper(item %ld) failed - %s\n",
  1.1156 +                    (long)idx, u_errorName(errorCode));
  1.1157 +            exit(errorCode);
  1.1158 +        }
  1.1159 +
  1.1160 +        ds->printError=printPackageError;
  1.1161 +        ds->printErrorContext=stderr;
  1.1162 +
  1.1163 +        // swap the item from its platform properties to the desired ones
  1.1164 +        udata_swap(ds, pItem->data, pItem->length, pItem->data, &errorCode);
  1.1165 +        if(U_FAILURE(errorCode)) {
  1.1166 +            fprintf(stderr, "icupkg: udata_swap(item %ld) failed - %s\n", (long)idx, u_errorName(errorCode));
  1.1167 +            exit(errorCode);
  1.1168 +        }
  1.1169 +        udata_closeSwapper(ds);
  1.1170 +        pItem->type=outType;
  1.1171 +    }
  1.1172 +
  1.1173 +    // create the file and write its contents
  1.1174 +    makeFullFilenameAndDirs(filesPath, outName, filename, (int32_t)sizeof(filename));
  1.1175 +    file=fopen(filename, "wb");
  1.1176 +    if(file==NULL) {
  1.1177 +        fprintf(stderr, "icupkg: unable to create file \"%s\"\n", filename);
  1.1178 +        exit(U_FILE_ACCESS_ERROR);
  1.1179 +    }
  1.1180 +    fileLength=(int32_t)fwrite(pItem->data, 1, pItem->length, file);
  1.1181 +
  1.1182 +    if(ferror(file) || fileLength!=pItem->length) {
  1.1183 +        fprintf(stderr, "icupkg: unable to write complete file \"%s\"\n", filename);
  1.1184 +        exit(U_FILE_ACCESS_ERROR);
  1.1185 +    }
  1.1186 +    fclose(file);
  1.1187 +}
  1.1188 +
  1.1189 +void
  1.1190 +Package::extractItem(const char *filesPath, int32_t idx, char outType) {
  1.1191 +    extractItem(filesPath, items[idx].name, idx, outType);
  1.1192 +}
  1.1193 +
  1.1194 +void
  1.1195 +Package::extractItems(const char *filesPath, const char *pattern, char outType) {
  1.1196 +    int32_t idx;
  1.1197 +
  1.1198 +    findItems(pattern);
  1.1199 +    while((idx=findNextItem())>=0) {
  1.1200 +        extractItem(filesPath, idx, outType);
  1.1201 +    }
  1.1202 +}
  1.1203 +
  1.1204 +void
  1.1205 +Package::extractItems(const char *filesPath, const Package &listPkg, char outType) {
  1.1206 +    const Item *pItem;
  1.1207 +    int32_t i;
  1.1208 +
  1.1209 +    for(pItem=listPkg.items, i=0; i<listPkg.itemCount; ++pItem, ++i) {
  1.1210 +        extractItems(filesPath, pItem->name, outType);
  1.1211 +    }
  1.1212 +}
  1.1213 +
  1.1214 +int32_t
  1.1215 +Package::getItemCount() const {
  1.1216 +    return itemCount;
  1.1217 +}
  1.1218 +
  1.1219 +const Item *
  1.1220 +Package::getItem(int32_t idx) const {
  1.1221 +    if (0 <= idx && idx < itemCount) {
  1.1222 +        return &items[idx];
  1.1223 +    }
  1.1224 +    return NULL;
  1.1225 +}
  1.1226 +
  1.1227 +void
  1.1228 +Package::checkDependency(void *context, const char *itemName, const char *targetName) {
  1.1229 +    // check dependency: make sure the target item is in the package
  1.1230 +    Package *me=(Package *)context;
  1.1231 +    if(me->findItem(targetName)<0) {
  1.1232 +        me->isMissingItems=TRUE;
  1.1233 +        fprintf(stderr, "Item %s depends on missing item %s\n", itemName, targetName);
  1.1234 +    }
  1.1235 +}
  1.1236 +
  1.1237 +UBool
  1.1238 +Package::checkDependencies() {
  1.1239 +    isMissingItems=FALSE;
  1.1240 +    enumDependencies(this, checkDependency);
  1.1241 +    return (UBool)!isMissingItems;
  1.1242 +}
  1.1243 +
  1.1244 +void
  1.1245 +Package::enumDependencies(void *context, CheckDependency check) {
  1.1246 +    int32_t i;
  1.1247 +
  1.1248 +    for(i=0; i<itemCount; ++i) {
  1.1249 +        enumDependencies(items+i, context, check);
  1.1250 +    }
  1.1251 +}
  1.1252 +
  1.1253 +char *
  1.1254 +Package::allocString(UBool in, int32_t length) {
  1.1255 +    char *p;
  1.1256 +    int32_t top;
  1.1257 +
  1.1258 +    if(in) {
  1.1259 +        top=inStringTop;
  1.1260 +        p=inStrings+top;
  1.1261 +    } else {
  1.1262 +        top=outStringTop;
  1.1263 +        p=outStrings+top;
  1.1264 +    }
  1.1265 +    top+=length+1;
  1.1266 +
  1.1267 +    if(top>STRING_STORE_SIZE) {
  1.1268 +        fprintf(stderr, "icupkg: string storage overflow\n");
  1.1269 +        exit(U_BUFFER_OVERFLOW_ERROR);
  1.1270 +    }
  1.1271 +    if(in) {
  1.1272 +        inStringTop=top;
  1.1273 +    } else {
  1.1274 +        outStringTop=top;
  1.1275 +    }
  1.1276 +    return p;
  1.1277 +}
  1.1278 +
  1.1279 +void
  1.1280 +Package::sortItems() {
  1.1281 +    UErrorCode errorCode=U_ZERO_ERROR;
  1.1282 +    uprv_sortArray(items, itemCount, (int32_t)sizeof(Item), compareItems, NULL, FALSE, &errorCode);
  1.1283 +    if(U_FAILURE(errorCode)) {
  1.1284 +        fprintf(stderr, "icupkg: sorting item names failed - %s\n", u_errorName(errorCode));
  1.1285 +        exit(errorCode);
  1.1286 +    }
  1.1287 +}
  1.1288 +
  1.1289 +void Package::setItemCapacity(int32_t max) 
  1.1290 +{
  1.1291 +  if(max<=itemMax) {
  1.1292 +    return;
  1.1293 +  }
  1.1294 +  Item *newItems = (Item*)uprv_malloc(max * sizeof(items[0]));
  1.1295 +  Item *oldItems = items;
  1.1296 +  if(newItems == NULL) {
  1.1297 +    fprintf(stderr, "icupkg: Out of memory trying to allocate %lu bytes for %d items\n", 
  1.1298 +        (unsigned long)max*sizeof(items[0]), max);
  1.1299 +    exit(U_MEMORY_ALLOCATION_ERROR);
  1.1300 +  }
  1.1301 +  if(items && itemCount>0) {
  1.1302 +    uprv_memcpy(newItems, items, itemCount*sizeof(items[0]));
  1.1303 +  }
  1.1304 +  itemMax = max;
  1.1305 +  items = newItems;
  1.1306 +  uprv_free(oldItems);
  1.1307 +}
  1.1308 +
  1.1309 +void Package::ensureItemCapacity()
  1.1310 +{
  1.1311 +  if((itemCount+1)>itemMax) {
  1.1312 +    setItemCapacity(itemCount+kItemsChunk);
  1.1313 +  }
  1.1314 +}
  1.1315 +
  1.1316 +U_NAMESPACE_END

mercurial