unzip 6.0 日本語化パッチ(Shift_JISファイル名などに対応)

README.md

unzip 6.0 日本語化パッチ

概要

unzip 6.0 で Shift_JIS などの日本語ファイル名を含む zipファイルを展開する為の patch です。

Ubuntu に入っていた unzip 5.52 用のpatchを移植、改変したものです。オリジナルの patch はこちら:

http://archive.ubuntulinux.jp/ubuntu-ja/intrepid-ja/unzip_5.52-11ubuntu1.1ja1.diff.gz

Linuxでのビルド方法

% tar xzf unzip60.tar.gz
% cd unzip60
% wget http://gist.github.com/raw/573753/unzip-6.0-japanese_charset.patch
% patch -p1 < unzip-6.0-japanese_charset.patch
% make -f unix/Makefile LOCAL_UNZIP="-D_FILE_OFFSET_BITS=64 -DNO_LCHMOD -D_MBCS -DNO_WORKING_ISPRINT" linux_noasm

CentOS6の場合 make -f unix/Makefile ... linux の代わりに generic_gcc じゃないとダメみたい。

unzip-6.0-japanese_charset.patch

# 01_japanese_charset.dpatch by <[email protected]>
# port to unzip 6.0 by Tsukasa Hamano <[email protected]>

diff --git a/crypt.c b/crypt.c
index 784e411..d4a54b7 100644
--- a/crypt.c
+++ b/crypt.c
@@ -192,10 +192,74 @@ void init_keys(__G__ passwd)
     GLOBAL(keys[0]) = 305419896L;
     GLOBAL(keys[1]) = 591751049L;
     GLOBAL(keys[2]) = 878082192L;
+#ifdef _MBCS
+    char *buf, *bufp, *inbufp;
+    size_t inlen, outlen, size, inleft, outleft;
+    int i, updated = 0;
+    iconv_t cd;
+    static int mode = -1;
+    static char *locale_charset = NULL;
+
+    if (mode == -1) {
+      const char *ctype = setlocale(LC_CTYPE, "");
+      size_t ctype_len = 0;
+      if ( ctype != NULL ) ctype_len = strlen(ctype);
+
+      if ( (ctype_len >= 5 ) && (strncmp(ctype + ctype_len -5, "UTF-8", 5) == 0) ) {
+        mode = 1;
+        locale_charset = "UTF-8";
+      } else if ( ( (ctype_len >= 6 ) && (strncmp(ctype + ctype_len - 6, "EUC-JP", 6) == 0) ) ||
+                  ( (ctype_len >= 5 ) && (strncmp(ctype + ctype_len - 5, "eucJP",  5) == 0) ) ||
+                  ( (ctype_len >= 4 ) && (strncmp(ctype + ctype_len - 4, "ujis",   4) == 0) ) ) {
+        mode = 2;
+        locale_charset = "EUC-JP";
+/*
+      } else if ( ( (ctype_len >= 9 ) && (strncmp(ctype + ctype_len - 9, "Shift_JIS", 9) == 0) ) ||
+                  ( (ctype_len >= 4 ) && (strncmp(ctype + ctype_len - 4, "SJIS",      4) == 0) ) ||
+                  ( (ctype_len >= 4 ) && (strncmp(ctype + ctype_len - 4, "sjis",      4) == 0) ) ) {
+        mode = 3;
+        locale_charset = "CP932";
+*/
+      } else {
+        mode = 0;
+      }
+    }
+
+    if ( (mode > 0) && (locale_charset != NULL) ) {
+      inlen = strlen(passwd);
+      size = inlen * 6;
+      buf = malloc(size);
+      inleft = inlen;
+      outleft = size;
+      inbufp = (char *)passwd;
+      bufp = buf;
+      cd = iconv_open("CP932", locale_charset);
+      if (cd != (iconv_t)-1) {
+        iconv(cd, &inbufp, &inleft, &bufp, &outleft);
+        iconv_close(cd);
+        if (inleft == 0) {
+	  outlen = size - outleft;
+	  for(i=0; i<outlen; i++) {
+	    update_keys(__G__ (int)*(buf+i));
+	  }
+	  updated = 1;
+	}
+      }
+      free(buf);
+    }
+
+    if ( updated <= 0 ) {
+      while (*passwd != '\0') {
+        update_keys(__G__ (int)*passwd);
+        passwd++;
+      }
+    }
+#else
     while (*passwd != '\0') {
         update_keys(__G__ (int)*passwd);
         passwd++;
     }
+#endif
 }
 
 
diff --git a/crypt.h b/crypt.h
index 0c533e9..e9321a8 100644
--- a/crypt.h
+++ b/crypt.h
@@ -165,5 +165,9 @@ void init_keys OF((__GPRO__ ZCONST char *passwd));
 
 #define zfwrite  fwrite
 
+#ifdef _MBCS
+#include <iconv.h>
+#endif
+
 #endif /* ?CRYPT */
 #endif /* !__crypt_h */
diff --git a/fileio.c b/fileio.c
index ba0a1d0..00321ab 100644
--- a/fileio.c
+++ b/fileio.c
@@ -2126,9 +2126,14 @@ int do_string(__G__ length, option)   /* return PK-type error code */
                 /* translate the text coded in the entry's host-dependent
                    "extended ASCII" charset into the compiler's (system's)
                    internal text code page */
+#ifdef _MBCS
+	      if (G.pInfo->hostnum != UNIX_ )
+		buf_to_locale((char *)G.outbuf);
+#else
                 Ext_ASCII_TO_Native((char *)G.outbuf, G.pInfo->hostnum,
                                     G.pInfo->hostver, G.pInfo->HasUxAtt,
                                     FALSE);
+#endif
 #ifdef WINDLL
                 /* translate to ANSI (RTL internal codepage may be OEM) */
                 INTERN_TO_ISO((char *)G.outbuf, (char *)G.outbuf);
@@ -2140,7 +2145,12 @@ int do_string(__G__ length, option)   /* return PK-type error code */
 #endif /* (WIN32 && !_WIN32_WCE) */
 #endif /* ?WINDLL */
             } else {
+#ifdef _MBCS
+	      if (G.pInfo->hostnum != UNIX_ )
+		buf_to_locale((char *)G.outbuf);
+#else
                 A_TO_N(G.outbuf);   /* translate string to native */
+#endif
             }
 
 #ifdef WINDLL
@@ -2238,10 +2248,15 @@ int do_string(__G__ length, option)   /* return PK-type error code */
         G.filename[length] = '\0';      /* terminate w/zero:  ASCIIZ */
 #endif /* ?UNICODE_SUPPORT */
 
+#ifdef _MBCS
+	if (G.pInfo->hostnum != UNIX_ )
+	  buf_to_locale(G.filename);
+#else
         /* translate the Zip entry filename coded in host-dependent "extended
            ASCII" into the compiler's (system's) internal text code page */
         Ext_ASCII_TO_Native(G.filename, G.pInfo->hostnum, G.pInfo->hostver,
                             G.pInfo->HasUxAtt, (option == DS_FN_L));
+#endif
 
         if (G.pInfo->lcflag)      /* replace with lowercase filename */
             STRLOWER(G.filename, G.filename);
@@ -2533,8 +2548,79 @@ char *fzofft(__G__ val, pre, post)
     return G.fzofft_buf[G.fzofft_index];
 }
 
-
-
+void buf_to_locale(char *ptr)
+{
+	char *buf, *bufp, *inbufp;
+	size_t inlen, outlen, size, inleft, outleft;
+	int i;
+	iconv_t cd;
+	static int mode = -1;
+	static char *locale_charset = NULL;
+	static const char* charsets[] = {
+		"UTF-8",
+		"CP932",
+		"EUC-JP",
+		"ISO-2022-JP",
+		NULL
+	};
+
+	if (mode == 0) return;
+
+	if (mode == -1) {
+		const char *ctype = setlocale(LC_CTYPE, "");
+		if ( ctype == NULL ) {
+			mode = 0;
+			return;
+		}
+		size_t ctype_len = strlen(ctype);
+
+		if((ctype_len >= 5 ) &&
+		   (strncasecmp(ctype + ctype_len -5, "UTF-8", 5) == 0)){
+			mode = 1;
+			locale_charset = "UTF-8";
+		}else if(((ctype_len >= 6 ) &&
+				  (strncasecmp(ctype + ctype_len - 6, "EUC-JP", 6) == 0)) ||
+				 ((ctype_len >= 5 ) &&
+				  (strncasecmp(ctype + ctype_len - 5, "eucjp",  5) == 0)) ||
+				 ((ctype_len >= 4 ) &&
+				  (strncasecmp(ctype + ctype_len - 4, "ujis",   4) == 0))){
+			mode = 2;
+			locale_charset = "EUC-JP";
+		} else if (((ctype_len >= 9 ) &&
+					(strncasecmp(ctype + ctype_len - 9, "Shift_JIS", 9) == 0) ) ||
+					((ctype_len >= 4 ) &&
+					 (strncasecmp(ctype + ctype_len - 4, "sjis",      4) == 0) ) ) {
+			mode = 3;
+			locale_charset = "CP932";
+		} else {
+			mode = 0;
+			return;
+		}
+	}
+	if (locale_charset == NULL) return;
+
+	inlen = strlen(ptr);
+	size = inlen * 6;
+	buf = malloc(size);
+
+	for (i=0; charsets[i]; i++) {
+		inleft = inlen;
+		outleft = size;
+		inbufp = ptr;
+		bufp = buf;
+		if (strcmp(locale_charset, charsets[i]) == 0) continue;
+		cd = iconv_open(locale_charset, charsets[i]);
+		if (cd == (iconv_t)-1) continue;
+		iconv(cd, &inbufp, &inleft, &bufp, &outleft);
+		iconv_close(cd);
+		if (inleft != 0) continue;
+		outlen = size - outleft;
+		memcpy(ptr, buf, outlen);
+		ptr[outlen] = '\0';
+		break;
+	}
+	free(buf);
+}
 
 #if CRYPT
 
@@ -2563,6 +2649,13 @@ char *str2iso(dst, src)
 }
 #endif /* NEED_STR2ISO */
 
+int uzmblen(ZCONST unsigned char *ptr) {
+  int i;
+  i = mblen((const char*) ptr, MB_CUR_MAX);
+  if (i >= 0) return i;
+  /* fprintf(stderr, " uzmblen: multi byte strings error (%02x)\n", *ptr); */
+  return 1;
+}
 
 #ifdef NEED_STR2OEM
 /**********************/
diff --git a/unix/unix.c b/unix/unix.c
index efa97fc..c072528 100644
--- a/unix/unix.c
+++ b/unix/unix.c
@@ -214,7 +214,7 @@ char *do_wild(__G__ wildspec)
         }
 
         /* break the wildspec into a directory part and a wildcard filename */
-        if ((G.wildname = (ZCONST char *)strrchr(wildspec, '/')) == NULL) {
+        if ((G.wildname = (ZCONST char *)MBSRCHR(wildspec, '/')) == NULL) {
             G.dirname = ".";
             G.dirnamelen = 1;
             G.have_dirname = FALSE;
@@ -542,7 +542,7 @@ int mapname(__G__ renamed)
     *pathcomp = '\0';           /* initialize translation buffer */
     pp = pathcomp;              /* point to translation buffer */
     if (uO.jflag)               /* junking directories */
-        cp = (char *)strrchr(G.filename, '/');
+        cp = (char *)MBSRCHR(G.filename, '/');
     if (cp == (char *)NULL)     /* no '/' or not junking dirs */
         cp = G.filename;        /* point to internal zipfile-member pathname */
     else
diff --git a/unzpriv.h b/unzpriv.h
index dc9eff5..204a28a 100644
--- a/unzpriv.h
+++ b/unzpriv.h
@@ -1236,6 +1236,8 @@
 #  ifndef PREINCSTR
 #    define PREINCSTR(ptr) (ptr += CLEN(ptr))
 #  endif
+#  include <iconv.h>
+   void buf_to_locale(char *ptr);
 #  define POSTINCSTR(ptr) (___TMP_PTR=(char *)(ptr), PREINCSTR(ptr),___TMP_PTR)
    char *plastchar OF((ZCONST char *ptr, extent len));
 #  define lastchar(ptr, len) ((int)(unsigned)*plastchar(ptr, len))