1b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru/****************************************************************************** 28393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius * Copyright (C) 2008-2013, International Business Machines 3b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * Corporation and others. All Rights Reserved. 4b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru ******************************************************************************* 5b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru */ 6b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include "unicode/utypes.h" 7b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include "unicode/putil.h" 8b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include "cstring.h" 9b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include "toolutil.h" 10b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include "uoptions.h" 11b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include "uparse.h" 12b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include "package.h" 13b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include "pkg_icu.h" 14b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 15b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include <stdio.h> 16b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include <stdlib.h> 17b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#include <string.h> 18b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 19b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 20b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0])) 21b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 22b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru// read a file list -------------------------------------------------------- *** 23b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 2450294ead5e5d23f5bbfed76e00e6b510bd41eee1clairehoU_NAMESPACE_USE 2550294ead5e5d23f5bbfed76e00e6b510bd41eee1claireho 26b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Querustatic const struct { 27b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru const char *suffix; 28b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru int32_t length; 29b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru} listFileSuffixes[]={ 30b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru { ".txt", 4 }, 31b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru { ".lst", 4 }, 32b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru { ".tmp", 4 } 33b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru}; 34b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 35b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru/* check for multiple text file suffixes to see if this list name is a text file name */ 36b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Querustatic UBool 37b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste QueruisListTextFile(const char *listname) { 38b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru const char *listNameEnd=strchr(listname, 0); 39b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru const char *suffix; 40b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru int32_t i, length; 41b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru for(i=0; i<LENGTHOF(listFileSuffixes); ++i) { 42b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru suffix=listFileSuffixes[i].suffix; 43b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru length=listFileSuffixes[i].length; 44b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if((listNameEnd-listname)>length && 0==memcmp(listNameEnd-length, suffix, length)) { 45b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru return TRUE; 46b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 47b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 48b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru return FALSE; 49b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru} 50b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 51b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru/* 52b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * Read a file list. 53b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * If the listname ends with ".txt", then read the list file 54b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * (in the system/ invariant charset). 55b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * If the listname ends with ".dat", then read the ICU .dat package file. 56b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru * Otherwise, read the file itself as a single-item list. 57b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru */ 58b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste QueruU_CAPI Package * U_EXPORT2 5954dcd9b6a06071f647dac967e9e267abb9410720Craig CorneliusreadList(const char *filesPath, const char *listname, UBool readContents, Package *listPkgIn) { 6054dcd9b6a06071f647dac967e9e267abb9410720Craig Cornelius Package *listPkg = listPkgIn; 61b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru FILE *file; 62b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru const char *listNameEnd; 63b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 64b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(listname==NULL || listname[0]==0) { 65b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru fprintf(stderr, "missing list file\n"); 66b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru return NULL; 67b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 68b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 6954dcd9b6a06071f647dac967e9e267abb9410720Craig Cornelius if (listPkg == NULL) { 7054dcd9b6a06071f647dac967e9e267abb9410720Craig Cornelius listPkg=new Package(); 7154dcd9b6a06071f647dac967e9e267abb9410720Craig Cornelius if(listPkg==NULL) { 7254dcd9b6a06071f647dac967e9e267abb9410720Craig Cornelius fprintf(stderr, "icupkg: not enough memory\n"); 7354dcd9b6a06071f647dac967e9e267abb9410720Craig Cornelius exit(U_MEMORY_ALLOCATION_ERROR); 7454dcd9b6a06071f647dac967e9e267abb9410720Craig Cornelius } 75b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 76b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 77b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru listNameEnd=strchr(listname, 0); 78b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(isListTextFile(listname)) { 79b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // read the list file 80b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru char line[1024]; 81b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru char *end; 82b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru const char *start; 83b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 84b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru file=fopen(listname, "r"); 85b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(file==NULL) { 86b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru fprintf(stderr, "icupkg: unable to open list file \"%s\"\n", listname); 87b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru delete listPkg; 88b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru exit(U_FILE_ACCESS_ERROR); 89b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 90b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 91b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru while(fgets(line, sizeof(line), file)) { 92b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // remove comments 93b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru end=strchr(line, '#'); 94b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(end!=NULL) { 95b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru *end=0; 96b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } else { 97b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // remove trailing CR LF 98b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru end=strchr(line, 0); 99b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru while(line<end && (*(end-1)=='\r' || *(end-1)=='\n')) { 100b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru *--end=0; 101b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 102b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 103b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 104b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // check first non-whitespace character and 105b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // skip empty lines and 106b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // skip lines starting with reserved characters 107b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru start=u_skipWhitespace(line); 108b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(*start==0 || NULL!=strchr(U_PKG_RESERVED_CHARS, *start)) { 109b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru continue; 110b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 111b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 112b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // take whitespace-separated items from the line 113b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru for(;;) { 114b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // find whitespace after the item or the end of the line 115b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru for(end=(char *)start; *end!=0 && *end!=' ' && *end!='\t'; ++end) {} 116b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(*end==0) { 117b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // this item is the last one on the line 118b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru end=NULL; 119b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } else { 120b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // the item is terminated by whitespace, terminate it with NUL 121b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru *end=0; 122b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 123b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(readContents) { 124b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru listPkg->addFile(filesPath, start); 125b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } else { 126b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru listPkg->addItem(start); 127b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 128b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 129b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // find the start of the next item or exit the loop 130b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(end==NULL || *(start=u_skipWhitespace(end+1))==0) { 131b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru break; 132b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 133b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 134b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 135b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru fclose(file); 136b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } else if((listNameEnd-listname)>4 && 0==memcmp(listNameEnd-4, ".dat", 4)) { 137b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // read the ICU .dat package 1388393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius // Accept a .dat file whose name differs from the ToC prefixes. 1398393335b955da7340c9f19b1b4b2d6c0c2c04be7Craig Cornelius listPkg->setAutoPrefix(); 140b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru listPkg->readPackage(listname); 141b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } else { 142b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru // list the single file itself 143b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(readContents) { 144b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru listPkg->addFile(filesPath, listname); 145b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } else { 146b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru listPkg->addItem(listname); 147b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 148b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 149b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 150b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru return listPkg; 151b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru} 152b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 153b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste QueruU_CAPI int U_EXPORT2 154b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste QueruwritePackageDatFile(const char *outFilename, const char *outComment, const char *sourcePath, const char *addList, Package *pkg, char outType) { 155b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru Package *addListPkg = NULL; 156b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru UBool pkgDelete = FALSE; 157b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 158b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if (pkg == NULL) { 159b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru pkg = new Package; 160b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(pkg == NULL) { 161b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru fprintf(stderr, "icupkg: not enough memory\n"); 162b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru return U_MEMORY_ALLOCATION_ERROR; 163b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 164b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 16554dcd9b6a06071f647dac967e9e267abb9410720Craig Cornelius addListPkg = readList(sourcePath, addList, TRUE, NULL); 166b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if(addListPkg != NULL) { 167b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru pkg->addItems(*addListPkg); 168b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } else { 169b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru return U_ILLEGAL_ARGUMENT_ERROR; 170b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 171b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 172b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru pkgDelete = TRUE; 173b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 174b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 175b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru pkg->writePackage(outFilename, outType, outComment); 176b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 177b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru if (pkgDelete) { 178b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru delete pkg; 179b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru delete addListPkg; 180b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru } 181b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 182b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru return 0; 183b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru} 184b0ac937921a2c196d8b9da665135bf6ba01a1ccfJean-Baptiste Queru 185