From ca0212ba19b64488b9e8459a762c11ecd6e7d0bd Mon Sep 17 00:00:00 2001

From: Petr Stodulka <pstodulk@redhat.com>

Date: Tue, 24 Nov 2015 17:56:11 +0100

Subject: [PATCH] print correctly non-ascii filenames

---

 extract.c | 289 ++++++++++++++++++++++++++++++++++++++++++++++++--------------

 unzpriv.h |   7 ++

 2 files changed, 233 insertions(+), 63 deletions(-)

diff --git a/extract.c b/extract.c

index 0ee4e93..741b7e0 100644

--- a/extract.c

+++ b/extract.c

@@ -2648,8 +2648,21 @@ static void set_deferred_symlink(__G__ slnk_entry)

 } /* end function set_deferred_symlink() */

 #endif /* SYMLINKS */

+/*

+ * If Unicode is supported, assume we have what we need to do this

+ * check using wide characters, avoiding MBCS issues.

+ */

-

+#ifndef UZ_FNFILTER_REPLACECHAR

+        /* A convenient choice for the replacement of unprintable char codes is

+         * the "single char wildcard", as this character is quite unlikely to

+         * appear in filenames by itself.  The following default definition

+         * sets the replacement char to a question mark as the most common

+         * "single char wildcard"; this setting should be overridden in the

+         * appropiate system-specific configuration header when needed.

+         */

+# define UZ_FNFILTER_REPLACECHAR      '?'

+#endif

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

 /*  Function fnfilter()  */        /* here instead of in list.c for SFX */

@@ -2661,48 +2674,168 @@ char *fnfilter(raw, space, size)   /* convert name to safely printable form */

     extent size;

 {

 #ifndef NATIVE   /* ASCII:  filter ANSI escape codes, etc. */

-    ZCONST uch *r=(ZCONST uch *)raw;

+    ZCONST uch *r; // =(ZCONST uch *)raw;

     uch *s=space;

     uch *slim=NULL;

     uch *se=NULL;

     int have_overflow = FALSE;

-    if (size > 0) {

-        slim = space + size

-#ifdef _MBCS

-                     - (MB_CUR_MAX - 1)

-#endif

-                     - 4;

+# if defined( UNICODE_SUPPORT) && defined( _MBCS)

+/* If Unicode support is enabled, and we have multi-byte characters,

+ * then do the isprint() checks by first converting to wide characters

+ * and checking those.  This avoids our having to parse multi-byte

+ * characters for ourselves.  After the wide-char replacements have been

+ * made, the wide string is converted back to the local character set.

+ */

+    wchar_t *wstring;    /* wchar_t version of raw */

+    size_t wslen;        /* length of wstring */

+    wchar_t *wostring;   /* wchar_t version of output string */

+    size_t woslen;       /* length of wostring */

+    char *newraw;        /* new raw */

+

+    /* 2012-11-06 SMS.

+     * Changed to check the value returned by mbstowcs(), and bypass the

+     * Unicode processing if it fails.  This seems to fix a problem

+     * reported in the SourceForge forum, but it's not clear that we

+     * should be doing any Unicode processing without some evidence that

+     * the name actually is Unicode.  (Check bit 11 in the flags before

+     * coming here?)

+     * http://sourceforge.net/p/infozip/bugs/40/

+     */

+

+    if (MB_CUR_MAX <= 1)

+    {

+        /* There's no point to converting multi-byte chars if there are

+         * no multi-byte chars.

+         */

+        wslen = (size_t)-1;

     }

-    while (*r) {

-        if (size > 0 && s >= slim && se == NULL) {

-            se = s;

+    else

+    {

+        /* Get Unicode wide character count (for storage allocation). */

+        wslen = mbstowcs( NULL, raw, 0);

+    }

+

+    if (wslen != (size_t)-1)

+    {

+        /* Apparently valid Unicode.  Allocate wide-char storage. */

+        wstring = (wchar_t *)malloc((wslen + 1) * sizeof(wchar_t));

+        if (wstring == NULL) {

+            strcpy( (char *)space, raw);

+            return (char *)space;

         }

-#ifdef QDOS

-        if (qlflag & 2) {

-            if (*r == '/' || *r == '.') {

+        wostring = (wchar_t *)malloc(2 * (wslen + 1) * sizeof(wchar_t));

+        if (wostring == NULL) {

+            free(wstring);

+            strcpy( (char *)space, raw);

+            return (char *)space;

+        }

+

+        /* Convert the multi-byte Unicode to wide chars. */

+        wslen = mbstowcs(wstring, raw, wslen + 1);

+

+        /* Filter the wide-character string. */

+        fnfilterw( wstring, wostring, (2 * (wslen + 1) * sizeof(wchar_t)));

+

+        /* Convert filtered wide chars back to multi-byte. */

+        woslen = wcstombs( NULL, wostring, 0);

+        if ((newraw = malloc(woslen + 1)) == NULL) {

+            free(wstring);

+            free(wostring);

+            strcpy( (char *)space, raw);

+            return (char *)space;

+        }

+        woslen = wcstombs( newraw, wostring, (woslen * MB_CUR_MAX) + 1);

+

+        if (size > 0) {

+            slim = space + size - 4;

+        }

+        r = (ZCONST uch *)newraw;

+        while (*r) {

+            if (size > 0 && s >= slim && se == NULL) {

+                se = s;

+            }

+#  ifdef QDOS

+            if (qlflag & 2) {

+                if (*r == '/' || *r == '.') {

+                    if (se != NULL && (s > (space + (size-3)))) {

+                        have_overflow = TRUE;

+                        break;

+                    }

+                    ++r;

+                    *s++ = '_';

+                    continue;

+                }

+            } else

+#  endif

+            {

                 if (se != NULL && (s > (space + (size-3)))) {

                     have_overflow = TRUE;

                     break;

                 }

-                ++r;

-                *s++ = '_';

-                continue;

+                *s++ = *r++;

             }

-        } else

+        }

+        if (have_overflow) {

+            strcpy((char *)se, "...");

+        } else {

+            *s = '\0';

+        }

+

+        free(wstring);

+        free(wostring);

+        free(newraw);

+    }

+    else

+# endif /* defined( UNICODE_SUPPORT) && defined( _MBCS) */

+    {

+        /* No Unicode support, or apparently invalid Unicode. */

+        r = (ZCONST uch *)raw;

+

+        if (size > 0) {

+            slim = space + size

+#ifdef _MBCS

+                         - (MB_CUR_MAX - 1)

+#endif

+                         - 4;

+        }

+        while (*r) {

+            if (size > 0 && s >= slim && se == NULL) {

+                se = s;

+            }

+#ifdef QDOS

+            if (qlflag & 2) {

+                if (*r == '/' || *r == '.') {

+                    if (se != NULL && (s > (space + (size-3)))) {

+                        have_overflow = TRUE;

+                        break;

+                    }

+                    ++r;

+                    *s++ = '_';

+                    continue;

+                }

+            } else

 #endif

 #ifdef HAVE_WORKING_ISPRINT

-# ifndef UZ_FNFILTER_REPLACECHAR

-    /* A convenient choice for the replacement of unprintable char codes is

-     * the "single char wildcard", as this character is quite unlikely to

-     * appear in filenames by itself.  The following default definition

-     * sets the replacement char to a question mark as the most common

-     * "single char wildcard"; this setting should be overridden in the

-     * appropiate system-specific configuration header when needed.

-     */

-#   define UZ_FNFILTER_REPLACECHAR      '?'

-# endif

-        if (!isprint(*r)) {

+            if (!isprint(*r)) {

+                if (*r < 32) {

+                    /* ASCII control codes are escaped as "^{letter}". */

+                    if (se != NULL && (s > (space + (size-4)))) {

+                        have_overflow = TRUE;

+                        break;

+                    }

+                    *s++ = '^', *s++ = (uch)(64 + *r++);

+                } else {

+                    /* Other unprintable codes are replaced by the

+                     * placeholder character. */

+                    if (se != NULL && (s > (space + (size-3)))) {

+                        have_overflow = TRUE;

+                        break;

+                    }

+                    *s++ = UZ_FNFILTER_REPLACECHAR;

+                    INCSTR(r);

+                }

+#else /* !HAVE_WORKING_ISPRINT */

             if (*r < 32) {

                 /* ASCII control codes are escaped as "^{letter}". */

                 if (se != NULL && (s > (space + (size-4)))) {

@@ -2710,47 +2843,30 @@ char *fnfilter(raw, space, size)   /* convert name to safely printable form */

                     break;

                 }

                 *s++ = '^', *s++ = (uch)(64 + *r++);

+#endif /* ?HAVE_WORKING_ISPRINT */

             } else {

-                /* Other unprintable codes are replaced by the

-                 * placeholder character. */

+#ifdef _MBCS

+                unsigned i = CLEN(r);

+                if (se != NULL && (s > (space + (size-i-2)))) {

+                    have_overflow = TRUE;

+                    break;

+                }

+                for (; i > 0; i--)

+                    *s++ = *r++;

+#else

                 if (se != NULL && (s > (space + (size-3)))) {

                     have_overflow = TRUE;

                     break;

                 }

-                *s++ = UZ_FNFILTER_REPLACECHAR;

-                INCSTR(r);

-            }

-#else /* !HAVE_WORKING_ISPRINT */

-        if (*r < 32) {

-            /* ASCII control codes are escaped as "^{letter}". */

-            if (se != NULL && (s > (space + (size-4)))) {

-                have_overflow = TRUE;

-                break;

-            }

-            *s++ = '^', *s++ = (uch)(64 + *r++);

-#endif /* ?HAVE_WORKING_ISPRINT */

-        } else {

-#ifdef _MBCS

-            unsigned i = CLEN(r);

-            if (se != NULL && (s > (space + (size-i-2)))) {

-                have_overflow = TRUE;

-                break;

-            }

-            for (; i > 0; i--)

                 *s++ = *r++;

-#else

-            if (se != NULL && (s > (space + (size-3)))) {

-                have_overflow = TRUE;

-                break;

-            }

-            *s++ = *r++;

 #endif

-         }

-    }

-    if (have_overflow) {

-        strcpy((char *)se, "...");

-    } else {

-        *s = '\0';

+             }

+        }

+        if (have_overflow) {

+            strcpy((char *)se, "...");

+        } else {

+            *s = '\0';

+        }

     }

 #ifdef WINDLL

@@ -2772,6 +2888,53 @@ char *fnfilter(raw, space, size)   /* convert name to safely printable form */

 } /* end function fnfilter() */

+#if defined( UNICODE_SUPPORT) && defined( _MBCS)

+

+/****************************/

+/*  Function fnfilter[w]()  */  /* (Here instead of in list.c for SFX.) */

+/****************************/

+

+/* fnfilterw() - Convert wide name to safely printable form. */

+

+/* fnfilterw() - Convert wide-character name to safely printable form. */

+

+wchar_t *fnfilterw( src, dst, siz)

+    ZCONST wchar_t *src;        /* Pointer to source char (string). */

+    wchar_t *dst;               /* Pointer to destination char (string). */

+    extent siz;                 /* Not used (!). */

+{

+    wchar_t *dsx = dst;

+

+    /* Filter the wide chars. */

+    while (*src)

+    {

+        if (iswprint( *src))

+        {

+            /* Printable code.  Copy it. */

+            *dst++ = *src;

+        }

+        else

+        {

+            /* Unprintable code.  Substitute something printable for it. */

+            if (*src < 32)

+            {

+                /* Replace ASCII control code with "^{letter}". */

+                *dst++ = (wchar_t)'^';

+                *dst++ = (wchar_t)(64 + *src);

+            }

+            else

+            {

+                /* Replace other unprintable code with the placeholder. */

+                *dst++ = (wchar_t)UZ_FNFILTER_REPLACECHAR;

+            }

+        }

+        src++;

+    }

+    *dst = (wchar_t)0;  /* NUL-terminate the destination string. */

+    return dsx;

+} /* fnfilterw(). */

+

+#endif /* defined( UNICODE_SUPPORT) && defined( _MBCS) */

 #ifdef SET_DIR_ATTRIB

diff --git a/unzpriv.h b/unzpriv.h

index 22d3923..e48a652 100644

--- a/unzpriv.h

+++ b/unzpriv.h

@@ -1212,6 +1212,7 @@

 # ifdef UNICODE_WCHAR

 #  if !(defined(_WIN32_WCE) || defined(POCKET_UNZIP))

 #   include <wchar.h>

+#   include <wctype.h>

 #  endif

 # endif

 # ifndef _MBCS  /* no need to include <locale.h> twice, see below */

@@ -2410,6 +2411,12 @@ int    memflush                  OF((__GPRO__ ZCONST uch *rawbuf, ulg size));

 char  *fnfilter                  OF((ZCONST char *raw, uch *space,

                                      extent size));

+# if defined( UNICODE_SUPPORT) && defined( _MBCS)

+wchar_t *fnfilterw               OF((ZCONST wchar_t *src, wchar_t *dst,

+                                     extent siz));

+#endif

+

+

 /*---------------------------------------------------------------------------

     Decompression functions:

   ---------------------------------------------------------------------------*/

-- 

2.4.3