130660a898545416f0fea2d717f16f75640001e38Ted Kremenek/******************************************************************************
230660a898545416f0fea2d717f16f75640001e38Ted Kremenek *   Copyright (C) 2008-2013, International Business Machines
330660a898545416f0fea2d717f16f75640001e38Ted Kremenek *   Corporation and others.  All Rights Reserved.
430660a898545416f0fea2d717f16f75640001e38Ted Kremenek *******************************************************************************
530660a898545416f0fea2d717f16f75640001e38Ted Kremenek */
630660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include "unicode/utypes.h"
730660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include "unicode/putil.h"
830660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include "cstring.h"
930660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include "toolutil.h"
1030660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include "uoptions.h"
1155fc873017f10f6f566b182b70f6fc22aefa3464Chandler Carruth#include "uparse.h"
1230660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include "package.h"
1330660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include "pkg_icu.h"
1430660a898545416f0fea2d717f16f75640001e38Ted Kremenek
1530660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include <stdio.h>
1630660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include <stdlib.h>
1730660a898545416f0fea2d717f16f75640001e38Ted Kremenek#include <string.h>
1830660a898545416f0fea2d717f16f75640001e38Ted Kremenek
1930660a898545416f0fea2d717f16f75640001e38Ted Kremenek
2030660a898545416f0fea2d717f16f75640001e38Ted Kremenek#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
2130660a898545416f0fea2d717f16f75640001e38Ted Kremenek
2230660a898545416f0fea2d717f16f75640001e38Ted Kremenek// read a file list -------------------------------------------------------- ***
2330660a898545416f0fea2d717f16f75640001e38Ted Kremenek
2430660a898545416f0fea2d717f16f75640001e38Ted KremenekU_NAMESPACE_USE
2530660a898545416f0fea2d717f16f75640001e38Ted Kremenek
2630660a898545416f0fea2d717f16f75640001e38Ted Kremenekstatic const struct {
2730660a898545416f0fea2d717f16f75640001e38Ted Kremenek    const char *suffix;
2830660a898545416f0fea2d717f16f75640001e38Ted Kremenek    int32_t length;
2930660a898545416f0fea2d717f16f75640001e38Ted Kremenek} listFileSuffixes[]={
3030660a898545416f0fea2d717f16f75640001e38Ted Kremenek    { ".txt", 4 },
3130660a898545416f0fea2d717f16f75640001e38Ted Kremenek    { ".lst", 4 },
3230660a898545416f0fea2d717f16f75640001e38Ted Kremenek    { ".tmp", 4 }
3330660a898545416f0fea2d717f16f75640001e38Ted Kremenek};
3430660a898545416f0fea2d717f16f75640001e38Ted Kremenek
3530660a898545416f0fea2d717f16f75640001e38Ted Kremenek/* check for multiple text file suffixes to see if this list name is a text file name */
3630660a898545416f0fea2d717f16f75640001e38Ted Kremenekstatic UBool
3730660a898545416f0fea2d717f16f75640001e38Ted KremenekisListTextFile(const char *listname) {
3830660a898545416f0fea2d717f16f75640001e38Ted Kremenek    const char *listNameEnd=strchr(listname, 0);
3930660a898545416f0fea2d717f16f75640001e38Ted Kremenek    const char *suffix;
4030660a898545416f0fea2d717f16f75640001e38Ted Kremenek    int32_t i, length;
4130660a898545416f0fea2d717f16f75640001e38Ted Kremenek    for(i=0; i<LENGTHOF(listFileSuffixes); ++i) {
4230660a898545416f0fea2d717f16f75640001e38Ted Kremenek        suffix=listFileSuffixes[i].suffix;
4330660a898545416f0fea2d717f16f75640001e38Ted Kremenek        length=listFileSuffixes[i].length;
4430660a898545416f0fea2d717f16f75640001e38Ted Kremenek        if((listNameEnd-listname)>length && 0==memcmp(listNameEnd-length, suffix, length)) {
4530660a898545416f0fea2d717f16f75640001e38Ted Kremenek            return TRUE;
4630660a898545416f0fea2d717f16f75640001e38Ted Kremenek        }
4730660a898545416f0fea2d717f16f75640001e38Ted Kremenek    }
4830660a898545416f0fea2d717f16f75640001e38Ted Kremenek    return FALSE;
4930660a898545416f0fea2d717f16f75640001e38Ted Kremenek}
5030660a898545416f0fea2d717f16f75640001e38Ted Kremenek
5130660a898545416f0fea2d717f16f75640001e38Ted Kremenek/*
5230660a898545416f0fea2d717f16f75640001e38Ted Kremenek * Read a file list.
5330660a898545416f0fea2d717f16f75640001e38Ted Kremenek * If the listname ends with ".txt", then read the list file
5430660a898545416f0fea2d717f16f75640001e38Ted Kremenek * (in the system/ invariant charset).
5530660a898545416f0fea2d717f16f75640001e38Ted Kremenek * If the listname ends with ".dat", then read the ICU .dat package file.
5630660a898545416f0fea2d717f16f75640001e38Ted Kremenek * Otherwise, read the file itself as a single-item list.
5730660a898545416f0fea2d717f16f75640001e38Ted Kremenek */
5830660a898545416f0fea2d717f16f75640001e38Ted KremenekU_CAPI Package * U_EXPORT2
5930660a898545416f0fea2d717f16f75640001e38Ted KremenekreadList(const char *filesPath, const char *listname, UBool readContents, Package *listPkgIn) {
6030660a898545416f0fea2d717f16f75640001e38Ted Kremenek    Package *listPkg = listPkgIn;
6130660a898545416f0fea2d717f16f75640001e38Ted Kremenek    FILE *file;
6230660a898545416f0fea2d717f16f75640001e38Ted Kremenek    const char *listNameEnd;
6330660a898545416f0fea2d717f16f75640001e38Ted Kremenek
6430660a898545416f0fea2d717f16f75640001e38Ted Kremenek    if(listname==NULL || listname[0]==0) {
6530660a898545416f0fea2d717f16f75640001e38Ted Kremenek        fprintf(stderr, "missing list file\n");
6630660a898545416f0fea2d717f16f75640001e38Ted Kremenek        return NULL;
6730660a898545416f0fea2d717f16f75640001e38Ted Kremenek    }
6830660a898545416f0fea2d717f16f75640001e38Ted Kremenek
6930660a898545416f0fea2d717f16f75640001e38Ted Kremenek    if (listPkg == NULL) {
7030660a898545416f0fea2d717f16f75640001e38Ted Kremenek        listPkg=new Package();
7130660a898545416f0fea2d717f16f75640001e38Ted Kremenek        if(listPkg==NULL) {
7230660a898545416f0fea2d717f16f75640001e38Ted Kremenek            fprintf(stderr, "icupkg: not enough memory\n");
7330660a898545416f0fea2d717f16f75640001e38Ted Kremenek            exit(U_MEMORY_ALLOCATION_ERROR);
7430660a898545416f0fea2d717f16f75640001e38Ted Kremenek        }
7530660a898545416f0fea2d717f16f75640001e38Ted Kremenek    }
7630660a898545416f0fea2d717f16f75640001e38Ted Kremenek
7730660a898545416f0fea2d717f16f75640001e38Ted Kremenek    listNameEnd=strchr(listname, 0);
7830660a898545416f0fea2d717f16f75640001e38Ted Kremenek    if(isListTextFile(listname)) {
7930660a898545416f0fea2d717f16f75640001e38Ted Kremenek        // read the list file
8030660a898545416f0fea2d717f16f75640001e38Ted Kremenek        char line[1024];
8130660a898545416f0fea2d717f16f75640001e38Ted Kremenek        char *end;
8230660a898545416f0fea2d717f16f75640001e38Ted Kremenek        const char *start;
8330660a898545416f0fea2d717f16f75640001e38Ted Kremenek
8430660a898545416f0fea2d717f16f75640001e38Ted Kremenek        file=fopen(listname, "r");
8530660a898545416f0fea2d717f16f75640001e38Ted Kremenek        if(file==NULL) {
8630660a898545416f0fea2d717f16f75640001e38Ted Kremenek            fprintf(stderr, "icupkg: unable to open list file \"%s\"\n", listname);
8730660a898545416f0fea2d717f16f75640001e38Ted Kremenek            delete listPkg;
8830660a898545416f0fea2d717f16f75640001e38Ted Kremenek            exit(U_FILE_ACCESS_ERROR);
8930660a898545416f0fea2d717f16f75640001e38Ted Kremenek        }
9030660a898545416f0fea2d717f16f75640001e38Ted Kremenek
9130660a898545416f0fea2d717f16f75640001e38Ted Kremenek        while(fgets(line, sizeof(line), file)) {
9230660a898545416f0fea2d717f16f75640001e38Ted Kremenek            // remove comments
9330660a898545416f0fea2d717f16f75640001e38Ted Kremenek            end=strchr(line, '#');
9430660a898545416f0fea2d717f16f75640001e38Ted Kremenek            if(end!=NULL) {
9530660a898545416f0fea2d717f16f75640001e38Ted Kremenek                *end=0;
9630660a898545416f0fea2d717f16f75640001e38Ted Kremenek            } else {
9730660a898545416f0fea2d717f16f75640001e38Ted Kremenek                // remove trailing CR LF
9830660a898545416f0fea2d717f16f75640001e38Ted Kremenek                end=strchr(line, 0);
9930660a898545416f0fea2d717f16f75640001e38Ted Kremenek                while(line<end && (*(end-1)=='\r' || *(end-1)=='\n')) {
10030660a898545416f0fea2d717f16f75640001e38Ted Kremenek                    *--end=0;
10130660a898545416f0fea2d717f16f75640001e38Ted Kremenek                }
10230660a898545416f0fea2d717f16f75640001e38Ted Kremenek            }
103055b395294d190a432e9d87bb665634636a1418aArgyrios Kyrtzidis
104055b395294d190a432e9d87bb665634636a1418aArgyrios Kyrtzidis            // check first non-whitespace character and
105055b395294d190a432e9d87bb665634636a1418aArgyrios Kyrtzidis            // skip empty lines and
10630660a898545416f0fea2d717f16f75640001e38Ted Kremenek            // skip lines starting with reserved characters
107055b395294d190a432e9d87bb665634636a1418aArgyrios Kyrtzidis            start=u_skipWhitespace(line);
10830660a898545416f0fea2d717f16f75640001e38Ted Kremenek            if(*start==0 || NULL!=strchr(U_PKG_RESERVED_CHARS, *start)) {
10930660a898545416f0fea2d717f16f75640001e38Ted Kremenek                continue;
11030660a898545416f0fea2d717f16f75640001e38Ted Kremenek            }
11130660a898545416f0fea2d717f16f75640001e38Ted Kremenek
11230660a898545416f0fea2d717f16f75640001e38Ted Kremenek            // take whitespace-separated items from the line
11330660a898545416f0fea2d717f16f75640001e38Ted Kremenek            for(;;) {
11430660a898545416f0fea2d717f16f75640001e38Ted Kremenek                // find whitespace after the item or the end of the line
11530660a898545416f0fea2d717f16f75640001e38Ted Kremenek                for(end=(char *)start; *end!=0 && *end!=' ' && *end!='\t'; ++end) {}
11630660a898545416f0fea2d717f16f75640001e38Ted Kremenek                if(*end==0) {
11730660a898545416f0fea2d717f16f75640001e38Ted Kremenek                    // this item is the last one on the line
11830660a898545416f0fea2d717f16f75640001e38Ted Kremenek                    end=NULL;
11930660a898545416f0fea2d717f16f75640001e38Ted Kremenek                } else {
12030660a898545416f0fea2d717f16f75640001e38Ted Kremenek                    // the item is terminated by whitespace, terminate it with NUL
12130660a898545416f0fea2d717f16f75640001e38Ted Kremenek                    *end=0;
12230660a898545416f0fea2d717f16f75640001e38Ted Kremenek                }
12330660a898545416f0fea2d717f16f75640001e38Ted Kremenek                if(readContents) {
12430660a898545416f0fea2d717f16f75640001e38Ted Kremenek                    listPkg->addFile(filesPath, start);
12530660a898545416f0fea2d717f16f75640001e38Ted Kremenek                } else {
12630660a898545416f0fea2d717f16f75640001e38Ted Kremenek                    listPkg->addItem(start);
12730660a898545416f0fea2d717f16f75640001e38Ted Kremenek                }
12830660a898545416f0fea2d717f16f75640001e38Ted Kremenek
12930660a898545416f0fea2d717f16f75640001e38Ted Kremenek                // find the start of the next item or exit the loop
13030660a898545416f0fea2d717f16f75640001e38Ted Kremenek                if(end==NULL || *(start=u_skipWhitespace(end+1))==0) {
13130660a898545416f0fea2d717f16f75640001e38Ted Kremenek                    break;
13230660a898545416f0fea2d717f16f75640001e38Ted Kremenek                }
13330660a898545416f0fea2d717f16f75640001e38Ted Kremenek            }
13430660a898545416f0fea2d717f16f75640001e38Ted Kremenek        }
13530660a898545416f0fea2d717f16f75640001e38Ted Kremenek        fclose(file);
13630660a898545416f0fea2d717f16f75640001e38Ted Kremenek    } else if((listNameEnd-listname)>4 && 0==memcmp(listNameEnd-4, ".dat", 4)) {
13730660a898545416f0fea2d717f16f75640001e38Ted Kremenek        // read the ICU .dat package
13830660a898545416f0fea2d717f16f75640001e38Ted Kremenek        // Accept a .dat file whose name differs from the ToC prefixes.
13930660a898545416f0fea2d717f16f75640001e38Ted Kremenek        listPkg->setAutoPrefix();
14030660a898545416f0fea2d717f16f75640001e38Ted Kremenek        listPkg->readPackage(listname);
14130660a898545416f0fea2d717f16f75640001e38Ted Kremenek    } else {
14230660a898545416f0fea2d717f16f75640001e38Ted Kremenek        // list the single file itself
14330660a898545416f0fea2d717f16f75640001e38Ted Kremenek        if(readContents) {
14430660a898545416f0fea2d717f16f75640001e38Ted Kremenek            listPkg->addFile(filesPath, listname);
14530660a898545416f0fea2d717f16f75640001e38Ted Kremenek        } else {
14630660a898545416f0fea2d717f16f75640001e38Ted Kremenek            listPkg->addItem(listname);
14730660a898545416f0fea2d717f16f75640001e38Ted Kremenek        }
14830660a898545416f0fea2d717f16f75640001e38Ted Kremenek    }
14930660a898545416f0fea2d717f16f75640001e38Ted Kremenek
15030660a898545416f0fea2d717f16f75640001e38Ted Kremenek    return listPkg;
15130660a898545416f0fea2d717f16f75640001e38Ted Kremenek}
15230660a898545416f0fea2d717f16f75640001e38Ted Kremenek
15330660a898545416f0fea2d717f16f75640001e38Ted KremenekU_CAPI int U_EXPORT2
15430660a898545416f0fea2d717f16f75640001e38Ted KremenekwritePackageDatFile(const char *outFilename, const char *outComment, const char *sourcePath, const char *addList, Package *pkg, char outType) {
15530660a898545416f0fea2d717f16f75640001e38Ted Kremenek    Package *addListPkg = NULL;
15630660a898545416f0fea2d717f16f75640001e38Ted Kremenek    UBool pkgDelete = FALSE;
15730660a898545416f0fea2d717f16f75640001e38Ted Kremenek
15830660a898545416f0fea2d717f16f75640001e38Ted Kremenek    if (pkg == NULL) {
15930660a898545416f0fea2d717f16f75640001e38Ted Kremenek        pkg = new Package;
16030660a898545416f0fea2d717f16f75640001e38Ted Kremenek        if(pkg == NULL) {
16130660a898545416f0fea2d717f16f75640001e38Ted Kremenek            fprintf(stderr, "icupkg: not enough memory\n");
16230660a898545416f0fea2d717f16f75640001e38Ted Kremenek            return U_MEMORY_ALLOCATION_ERROR;
16330660a898545416f0fea2d717f16f75640001e38Ted Kremenek        }
16430660a898545416f0fea2d717f16f75640001e38Ted Kremenek
16530660a898545416f0fea2d717f16f75640001e38Ted Kremenek        addListPkg = readList(sourcePath, addList, TRUE, NULL);
16630660a898545416f0fea2d717f16f75640001e38Ted Kremenek        if(addListPkg != NULL) {
16730660a898545416f0fea2d717f16f75640001e38Ted Kremenek            pkg->addItems(*addListPkg);
16830660a898545416f0fea2d717f16f75640001e38Ted Kremenek        } else {
16930660a898545416f0fea2d717f16f75640001e38Ted Kremenek            return U_ILLEGAL_ARGUMENT_ERROR;
17030660a898545416f0fea2d717f16f75640001e38Ted Kremenek        }
17130660a898545416f0fea2d717f16f75640001e38Ted Kremenek
17230660a898545416f0fea2d717f16f75640001e38Ted Kremenek        pkgDelete = TRUE;
17330660a898545416f0fea2d717f16f75640001e38Ted Kremenek    }
17430660a898545416f0fea2d717f16f75640001e38Ted Kremenek
17530660a898545416f0fea2d717f16f75640001e38Ted Kremenek    pkg->writePackage(outFilename, outType, outComment);
17630660a898545416f0fea2d717f16f75640001e38Ted Kremenek
17730660a898545416f0fea2d717f16f75640001e38Ted Kremenek    if (pkgDelete) {
17830660a898545416f0fea2d717f16f75640001e38Ted Kremenek        delete pkg;
17930660a898545416f0fea2d717f16f75640001e38Ted Kremenek        delete addListPkg;
18030660a898545416f0fea2d717f16f75640001e38Ted Kremenek    }
18130660a898545416f0fea2d717f16f75640001e38Ted Kremenek
18230660a898545416f0fea2d717f16f75640001e38Ted Kremenek    return 0;
18330660a898545416f0fea2d717f16f75640001e38Ted Kremenek}
18430660a898545416f0fea2d717f16f75640001e38Ted Kremenek
18530660a898545416f0fea2d717f16f75640001e38Ted Kremenek