ICU-59117.0.1.tar.gz

[apple/icu.git] / icuSources / tools / icupkg / icupkg.cpp

diff --git a/icuSources/tools/icupkg/icupkg.cpp b/icuSources/tools/icupkg/icupkg.cpp
index 93c488ddda71f1e46cd8f55f6c2c24f9fb4fcf4a..ea7be4a90923fd7adda726a46c63928892522b7c 100644 (file)
--- a/icuSources/tools/icupkg/icupkg.cpp
+++ b/icuSources/tools/icupkg/icupkg.cpp
@@ -1,12 +1,14 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html

 /*
 *******************************************************************************
 *
 /*
 *******************************************************************************
 *

-*   Copyright (C) 2005-2006, International Business Machines
+*   Copyright (C) 2005-2014, International Business Machines

 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
 *   file name:  icupkg.cpp
 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
 *   file name:  icupkg.cpp

-*   encoding:   US-ASCII
+*   encoding:   UTF-8

 *   tab size:   8 (not used)
 *   indentation:4
 *
 *   tab size:   8 (not used)
 *   indentation:4
 *


@@ -33,145 +35,20 @@
 #include "toolutil.h"
 #include "uoptions.h"
 #include "uparse.h"
 #include "toolutil.h"
 #include "uoptions.h"
 #include "uparse.h"

+#include "filestrm.h"

 #include "package.h"
 #include "package.h"

+#include "pkg_icu.h"

 
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
 
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 

+U_NAMESPACE_USE
+

 // TODO: add --matchmode=regex for using the ICU regex engine for item name pattern matching?
 
 // general definitions ----------------------------------------------------- ***
 
 // TODO: add --matchmode=regex for using the ICU regex engine for item name pattern matching?
 
 // general definitions ----------------------------------------------------- ***
 

-#define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
-
-// read a file list -------------------------------------------------------- ***
-
-static const char *reservedChars="\"%&'()*+,-./:;<=>?_";
-
-static const struct {
-    const char *suffix;
-    int32_t length;
-} listFileSuffixes[]={
-    { ".txt", 4 },
-    { ".lst", 4 },
-    { ".tmp", 4 }
-};
-
-/* check for multiple text file suffixes to see if this list name is a text file name */
-static UBool
-isListTextFile(const char *listname) {
-    const char *listNameEnd=strchr(listname, 0);
-    const char *suffix;
-    int32_t i, length;
-    for(i=0; i<LENGTHOF(listFileSuffixes); ++i) {
-        suffix=listFileSuffixes[i].suffix;
-        length=listFileSuffixes[i].length;
-        if((listNameEnd-listname)>length && 0==memcmp(listNameEnd-length, suffix, length)) {
-            return TRUE;
-        }
-    }
-    return FALSE;
-}
-
-/*
- * Read a file list.
- * If the listname ends with ".txt", then read the list file
- * (in the system/ invariant charset).
- * If the listname ends with ".dat", then read the ICU .dat package file.
- * Otherwise, read the file itself as a single-item list.
- */
-static Package *
-readList(const char *filesPath, const char *listname, UBool readContents) {
-    Package *listPkg;
-    FILE *file;
-    const char *listNameEnd;
-
-    if(listname==NULL || listname[0]==0) {
-        fprintf(stderr, "missing list file\n");
-        return NULL;
-    }
-
-    listPkg=new Package();
-    if(listPkg==NULL) {
-        fprintf(stderr, "icupkg: not enough memory\n");
-        exit(U_MEMORY_ALLOCATION_ERROR);
-    }
-
-    listNameEnd=strchr(listname, 0);
-    if(isListTextFile(listname)) {
-        // read the list file
-        char line[1024];
-        char *end;
-        const char *start;
-
-        file=fopen(listname, "r");
-        if(file==NULL) {
-            fprintf(stderr, "icupkg: unable to open list file \"%s\"\n", listname);
-            delete listPkg;
-            exit(U_FILE_ACCESS_ERROR);
-        }
-
-        while(fgets(line, sizeof(line), file)) {
-            // remove comments
-            end=strchr(line, '#');
-            if(end!=NULL) {
-                *end=0;
-            } else {
-                // remove trailing CR LF
-                end=strchr(line, 0);
-                while(line<end && (*(end-1)=='\r' || *(end-1)=='\n')) {
-                    *--end=0;
-                }
-            }
-
-            // check first non-whitespace character and
-            // skip empty lines and
-            // skip lines starting with reserved characters
-            start=u_skipWhitespace(line);
-            if(*start==0 || NULL!=strchr(reservedChars, *start)) {
-                continue;
-            }
-
-            // take whitespace-separated items from the line
-            for(;;) {
-                // find whitespace after the item or the end of the line
-                for(end=(char *)start; *end!=0 && *end!=' ' && *end!='\t'; ++end) {}
-                if(*end==0) {
-                    // this item is the last one on the line
-                    end=NULL;
-                } else {
-                    // the item is terminated by whitespace, terminate it with NUL
-                    *end=0;
-                }
-                if(readContents) {
-                    listPkg->addFile(filesPath, start);
-                } else {
-                    listPkg->addItem(start);
-                }
-
-                // find the start of the next item or exit the loop
-                if(end==NULL || *(start=u_skipWhitespace(end+1))==0) {
-                    break;
-                }
-            }
-        }
-        fclose(file);
-    } else if((listNameEnd-listname)>4 && 0==memcmp(listNameEnd-4, ".dat", 4)) {
-        // read the ICU .dat package
-        listPkg->readPackage(listname);
-    } else {
-        // list the single file itself
-        if(readContents) {
-            listPkg->addFile(filesPath, listname);
-        } else {
-            listPkg->addItem(listname);
-        }
-    }
-
-    return listPkg;
-}
-

 // main() ------------------------------------------------------------------ ***
 
 static void
 // main() ------------------------------------------------------------------ ***
 
 static void


@@ -180,8 +57,9 @@ printUsage(const char *pname, UBool isHelp) {
 
     fprintf(where,
             "%csage: %s [-h|-?|--help ] [-tl|-tb|-te] [-c] [-C comment]\n"
 
     fprintf(where,
             "%csage: %s [-h|-?|--help ] [-tl|-tb|-te] [-c] [-C comment]\n"

-            "\t[-a list] [-r list] [-x list] [-l]\n"
+            "\t[-a list] [-r list] [-x list] [-l [-o outputListFileName]]\n"

             "\t[-s path] [-d path] [-w] [-m mode]\n"
             "\t[-s path] [-d path] [-w] [-m mode]\n"

+            "\t[--auto_toc_prefix] [--auto_toc_prefix_with_type] [--toc_prefix]\n"

             "\tinfilename [outfilename]\n",
             isHelp ? 'U' : 'u', pname);
     if(isHelp) {
             "\tinfilename [outfilename]\n",
             isHelp ? 'U' : 'u', pname);
     if(isHelp) {


@@ -241,6 +119,31 @@ printUsage(const char *pname, UBool isHelp) {
             "\t-m mode or --matchmode mode  set the matching mode for item names with\n"
             "\t                             wildcards\n"
             "\t        noslash: the '*' wildcard does not match the '/' tree separator\n");
             "\t-m mode or --matchmode mode  set the matching mode for item names with\n"
             "\t                             wildcards\n"
             "\t        noslash: the '*' wildcard does not match the '/' tree separator\n");

+        fprintf(where,
+            "\n"
+            "\tIn the .dat package, the Table of Contents (ToC) contains an entry\n"
+            "\tfor each item of the form prefix/tree/itemname .\n"
+            "\tThe prefix normally matches the package basename, and icupkg checks that,\n"
+            "\tbut this is not necessary when ICU need not find and load the package by filename.\n"
+            "\tICU package names end with the platform type letter, and thus differ\n"
+            "\tbetween platform types. This is not required for user data packages.\n");
+        fprintf(where,
+            "\n"
+            "\t--auto_toc_prefix            automatic ToC entries prefix\n"
+            "\t                             Uses the prefix of the first entry of the\n"
+            "\t                             input package, rather than its basename.\n"
+            "\t                             Requires a non-empty input package.\n"
+            "\t--auto_toc_prefix_with_type  auto_toc_prefix + adjust platform type\n"
+            "\t                             Same as auto_toc_prefix but also checks that\n"
+            "\t                             the prefix ends with the input platform\n"
+            "\t                             type letter, and modifies it to the output\n"
+            "\t                             platform type letter.\n"
+            "\t                At most one of the auto_toc_prefix options\n"
+            "\t                can be used at a time.\n"
+            "\t--toc_prefix prefix          ToC prefix to be used in the output package\n"
+            "\t                             Overrides the package basename\n"
+            "\t                             and --auto_toc_prefix.\n"
+            "\t                             Cannot be combined with --auto_toc_prefix_with_type.\n");

         /*
          * Usage text columns, starting after the initial TAB.
          *      1         2         3         4         5         6         7         8
         /*
          * Usage text columns, starting after the initial TAB.
          *      1         2         3         4         5         6         7         8


@@ -253,7 +156,7 @@ printUsage(const char *pname, UBool isHelp) {
             "\tComments begin with # and are ignored. Empty lines are ignored.\n"
             "\tLines where the first non-whitespace character is one of %s\n"
             "\tare also ignored, to reserve for future syntax.\n",
             "\tComments begin with # and are ignored. Empty lines are ignored.\n"
             "\tLines where the first non-whitespace character is one of %s\n"
             "\tare also ignored, to reserve for future syntax.\n",

-            reservedChars);
+            U_PKG_RESERVED_CHARS);

         fprintf(where,
             "\tItems for removal or extraction may contain a single '*' wildcard\n"
             "\tcharacter. The '*' matches zero or more characters.\n"
         fprintf(where,
             "\tItems for removal or extraction may contain a single '*' wildcard\n"
             "\tcharacter. The '*' matches zero or more characters.\n"


@@ -273,8 +176,10 @@ printUsage(const char *pname, UBool isHelp) {
             "\t-s path or --sourcedir path  directory for the --add items\n"
             "\t-d path or --destdir path    directory for the --extract items\n"
             "\n"
             "\t-s path or --sourcedir path  directory for the --add items\n"
             "\t-d path or --destdir path    directory for the --extract items\n"
             "\n"

-            "\t-l or --list                 list the package items to stdout\n"
-            "\t                             (after modifying the package)\n");
+            "\t-l or --list                 list the package items\n"
+            "\t                             (after modifying the package)\n"
+            "\t                             to stdout or to output list file\n"
+            "\t-o path or --outlist path    path/filename for the --list output\n");

     }
 }
 
     }
 }
 


@@ -297,7 +202,12 @@ static UOption options[]={
     UOPTION_DEF("remove", 'r', UOPT_REQUIRES_ARG),
     UOPTION_DEF("extract", 'x', UOPT_REQUIRES_ARG),
 
     UOPTION_DEF("remove", 'r', UOPT_REQUIRES_ARG),
     UOPTION_DEF("extract", 'x', UOPT_REQUIRES_ARG),
 

-    UOPTION_DEF("list", 'l', UOPT_NO_ARG)
+    UOPTION_DEF("list", 'l', UOPT_NO_ARG),
+    UOPTION_DEF("outlist", 'o', UOPT_REQUIRES_ARG),
+
+    UOPTION_DEF("auto_toc_prefix", '\1', UOPT_NO_ARG),
+    UOPTION_DEF("auto_toc_prefix_with_type", '\1', UOPT_NO_ARG),
+    UOPTION_DEF("toc_prefix", '\1', UOPT_REQUIRES_ARG)

 };
 
 enum {
 };
 
 enum {


@@ -320,6 +230,11 @@ enum {
     OPT_EXTRACT_LIST,
 
     OPT_LIST_ITEMS,
     OPT_EXTRACT_LIST,
 
     OPT_LIST_ITEMS,

+    OPT_LIST_FILE,
+
+    OPT_AUTO_TOC_PREFIX,
+    OPT_AUTO_TOC_PREFIX_WITH_TYPE,
+    OPT_TOC_PREFIX,

 
     OPT_COUNT
 };
 
     OPT_COUNT
 };


@@ -331,30 +246,32 @@ isPackageName(const char *filename) {
     len=(int32_t)strlen(filename)-4; /* -4: subtract the length of ".dat" */
     return (UBool)(len>0 && 0==strcmp(filename+len, ".dat"));
 }
     len=(int32_t)strlen(filename)-4; /* -4: subtract the length of ".dat" */
     return (UBool)(len>0 && 0==strcmp(filename+len, ".dat"));
 }

+/*
+This line is required by MinGW because it incorrectly globs the arguments.
+So when \* is used, it turns into a list of files instead of a literal "*"
+*/
+int _CRT_glob = 0;

 
 extern int
 main(int argc, char *argv[]) {
     const char *pname, *sourcePath, *destPath, *inFilename, *outFilename, *outComment;
     char outType;
     UBool isHelp, isModified, isPackage;
 
 extern int
 main(int argc, char *argv[]) {
     const char *pname, *sourcePath, *destPath, *inFilename, *outFilename, *outComment;
     char outType;
     UBool isHelp, isModified, isPackage;

+    int result = 0;

 
 

-    Package *pkg, *listPkg;
+    Package *pkg, *listPkg, *addListPkg;

 
     U_MAIN_INIT_ARGS(argc, argv);
 
     /* get the program basename */
     pname=findBasename(argv[0]);
 
 
     U_MAIN_INIT_ARGS(argc, argv);
 
     /* get the program basename */
     pname=findBasename(argv[0]);
 

-    argc=u_parseArgs(argc, argv, LENGTHOF(options), options);
+    argc=u_parseArgs(argc, argv, UPRV_LENGTHOF(options), options);

     isHelp=options[OPT_HELP_H].doesOccur || options[OPT_HELP_QUESTION_MARK].doesOccur;
     if(isHelp) {
         printUsage(pname, TRUE);
         return U_ZERO_ERROR;
     }
     isHelp=options[OPT_HELP_H].doesOccur || options[OPT_HELP_QUESTION_MARK].doesOccur;
     if(isHelp) {
         printUsage(pname, TRUE);
         return U_ZERO_ERROR;
     }

-    if(argc<2 || 3<argc) {
-        printUsage(pname, FALSE);
-        return U_ILLEGAL_ARGUMENT_ERROR;
-    }

 
     pkg=new Package;
     if(pkg==NULL) {
 
     pkg=new Package;
     if(pkg==NULL) {


@@ -363,6 +280,25 @@ main(int argc, char *argv[]) {
     }
     isModified=FALSE;
 
     }
     isModified=FALSE;
 

+    int autoPrefix=0;
+    if(options[OPT_AUTO_TOC_PREFIX].doesOccur) {
+        pkg->setAutoPrefix();
+        ++autoPrefix;
+    }
+    if(options[OPT_AUTO_TOC_PREFIX_WITH_TYPE].doesOccur) {
+        if(options[OPT_TOC_PREFIX].doesOccur) {
+            fprintf(stderr, "icupkg: --auto_toc_prefix_with_type and also --toc_prefix\n");
+            printUsage(pname, FALSE);
+            return U_ILLEGAL_ARGUMENT_ERROR;
+        }
+        pkg->setAutoPrefixWithType();
+        ++autoPrefix;
+    }
+    if(argc<2 || 3<argc || autoPrefix>1) {
+        printUsage(pname, FALSE);
+        return U_ILLEGAL_ARGUMENT_ERROR;
+    }
+

     if(options[OPT_SOURCEDIR].doesOccur) {
         sourcePath=options[OPT_SOURCEDIR].value;
     } else {
     if(options[OPT_SOURCEDIR].doesOccur) {
         sourcePath=options[OPT_SOURCEDIR].value;
     } else {


@@ -377,6 +313,11 @@ main(int argc, char *argv[]) {
     }
 
     if(0==strcmp(argv[1], "new")) {
     }
 
     if(0==strcmp(argv[1], "new")) {

+        if(autoPrefix) {
+            fprintf(stderr, "icupkg: --auto_toc_prefix[_with_type] but no input package\n");
+            printUsage(pname, FALSE);
+            return U_ILLEGAL_ARGUMENT_ERROR;
+        }

         inFilename=NULL;
         isPackage=TRUE;
     } else {
         inFilename=NULL;
         isPackage=TRUE;
     } else {


@@ -428,7 +369,7 @@ main(int argc, char *argv[]) {
          * If we swap a single file, just assume that we are modifying it.
          * The Package class does not give us access to the item and its type.
          */
          * If we swap a single file, just assume that we are modifying it.
          * The Package class does not give us access to the item and its type.
          */

-        isModified=(UBool)(!isPackage || outType!=pkg->getInType());
+        isModified|=(UBool)(!isPackage || outType!=pkg->getInType());

     } else if(isPackage) {
         outType=pkg->getInType(); // default to input type
     } else /* !isPackage: swap single file */ {
     } else if(isPackage) {
         outType=pkg->getInType(); // default to input type
     } else /* !isPackage: swap single file */ {


@@ -460,7 +401,7 @@ main(int argc, char *argv[]) {
         }
 
         delete pkg;
         }
 
         delete pkg;

-        return 0;
+        return result;

     }
 
     /* Work with a package. */
     }
 
     /* Work with a package. */


@@ -484,8 +425,12 @@ main(int argc, char *argv[]) {
 
     /* remove items */
     if(options[OPT_REMOVE_LIST].doesOccur) {
 
     /* remove items */
     if(options[OPT_REMOVE_LIST].doesOccur) {

-        listPkg=readList(NULL, options[OPT_REMOVE_LIST].value, FALSE);
-        if(listPkg!=NULL) {
+        listPkg=new Package();
+        if(listPkg==NULL) {
+            fprintf(stderr, "icupkg: not enough memory\n");
+            exit(U_MEMORY_ALLOCATION_ERROR);
+        }
+        if(readList(NULL, options[OPT_REMOVE_LIST].value, FALSE, listPkg)) {

             pkg->removeItems(*listPkg);
             delete listPkg;
             isModified=TRUE;
             pkg->removeItems(*listPkg);
             delete listPkg;
             isModified=TRUE;


@@ -500,11 +445,16 @@ main(int argc, char *argv[]) {
      * use a separate Package so that its memory and items stay around
      * as long as the main Package
      */
      * use a separate Package so that its memory and items stay around
      * as long as the main Package
      */

+    addListPkg=NULL;

     if(options[OPT_ADD_LIST].doesOccur) {
     if(options[OPT_ADD_LIST].doesOccur) {

-        listPkg=readList(sourcePath, options[OPT_ADD_LIST].value, TRUE);
-        if(listPkg!=NULL) {
-            pkg->addItems(*listPkg);
-            delete listPkg;
+        addListPkg=new Package();
+        if(addListPkg==NULL) {
+            fprintf(stderr, "icupkg: not enough memory\n");
+            exit(U_MEMORY_ALLOCATION_ERROR);
+        }
+        if(readList(sourcePath, options[OPT_ADD_LIST].value, TRUE, addListPkg)) {
+            pkg->addItems(*addListPkg);
+            // delete addListPkg; deferred until after writePackage()

             isModified=TRUE;
         } else {
             printUsage(pname, FALSE);
             isModified=TRUE;
         } else {
             printUsage(pname, FALSE);


@@ -514,8 +464,12 @@ main(int argc, char *argv[]) {
 
     /* extract items */
     if(options[OPT_EXTRACT_LIST].doesOccur) {
 
     /* extract items */
     if(options[OPT_EXTRACT_LIST].doesOccur) {

-        listPkg=readList(NULL, options[OPT_EXTRACT_LIST].value, FALSE);
-        if(listPkg!=NULL) {
+        listPkg=new Package();
+        if(listPkg==NULL) {
+            fprintf(stderr, "icupkg: not enough memory\n");
+            exit(U_MEMORY_ALLOCATION_ERROR);
+        }
+        if(readList(NULL, options[OPT_EXTRACT_LIST].value, FALSE, listPkg)) {

             pkg->extractItems(destPath, *listPkg, outType);
             delete listPkg;
         } else {
             pkg->extractItems(destPath, *listPkg, outType);
             delete listPkg;
         } else {


@@ -526,7 +480,24 @@ main(int argc, char *argv[]) {
 
     /* list items */
     if(options[OPT_LIST_ITEMS].doesOccur) {
 
     /* list items */
     if(options[OPT_LIST_ITEMS].doesOccur) {

-        pkg->listItems(stdout);
+        int32_t i;
+        if (options[OPT_LIST_FILE].doesOccur) {
+            FileStream *out;
+            out = T_FileStream_open(options[OPT_LIST_FILE].value, "w");
+            if (out != NULL) {
+                for(i=0; i<pkg->getItemCount(); ++i) {
+                    T_FileStream_writeLine(out, pkg->getItem(i)->name);
+                    T_FileStream_writeLine(out, "\n");
+                }
+                T_FileStream_close(out);
+            } else {
+                return U_ILLEGAL_ARGUMENT_ERROR;
+            }
+        } else {
+            for(i=0; i<pkg->getItemCount(); ++i) {
+                fprintf(stdout, "%s\n", pkg->getItem(i)->name);
+            }
+        }

     }
 
     /* check dependencies between items */
     }
 
     /* check dependencies between items */


@@ -562,11 +533,15 @@ main(int argc, char *argv[]) {
             }
             outFilename=outFilenameBuffer;
         }
             }
             outFilename=outFilenameBuffer;
         }

-        pkg->writePackage(outFilename, outType, outComment);
+        if(options[OPT_TOC_PREFIX].doesOccur) {
+            pkg->setPrefix(options[OPT_TOC_PREFIX].value);
+        }
+        result = writePackageDatFile(outFilename, outComment, NULL, NULL, pkg, outType);

     }
 
     }
 

+    delete addListPkg;

     delete pkg;
     delete pkg;

-    return 0;
+    return result;

 }
 
 /*
 }
 
 /*