[iso] fix UDF UNICODE support

* libcdio's unicode16_decode() was not decoding labels at all,
  but simply stripping the MSB of the UDF USC-16 sequences
* use an actual conversion to UTF-8
* also update the rest of udf_fs.c to latest libcdio version
* closes #285

src/libcdio/driver/utf8.c

@@ -113,7 +113,6 @@ FILE* fopen_utf8(const char* filename, const char* mode)
 }
 #endif
 
-#ifdef HAVE_JOLIET
 #ifdef HAVE_ICONV
 #include <iconv.h>
 struct cdio_charset_coverter_s
@@ -344,5 +343,3 @@ bool cdio_charset_to_utf8(const char *src, size_t src_len, cdio_utf8_t **dst,
   return (*dst != NULL);
 }
 #endif /* HAVE_ICONV */
-
-#endif /* HAVE_JOLIET */

src/libcdio/udf/udf_fs.c

@@ -1,5 +1,5 @@
 /*
-  Copyright (C) 2005, 2006, 2008, 2011 Rocky Bernstein <rocky@gnu.org>
+  Copyright (C) 2005-2006, 2008, 2011, 2013 Rocky Bernstein <rocky@gnu.org>
 
   This program is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -70,6 +70,9 @@ const char VSD_STD_ID_NSR03[] = {'N', 'S', 'R', '0', '3'};
 const char VSD_STD_ID_TEA01[] = {'T', 'E', 'A', '0', '1'};
 
 #include <cdio/bytesex.h>
+#include <cdio/utf8.h>
+#include <cdio/util.h>
+
 #include "udf_private.h"
 #include "udf_fs.h"
 #include "cdio_assert.h"
@@ -134,6 +137,10 @@ udf_checktag(const udf_tag_t *p_tag, udf_Uint16_t tag_id)
 
   itag = (uint8_t *)p_tag;
 
+#ifdef WORDS_BIGENDIAN
+  tag_id = UINT16_SWAP_LE_BE(tag_id);
+#endif
+
   if (p_tag->id != tag_id)
     return -1;
 
@@ -270,23 +277,31 @@ udf_fopen(udf_dirent_t *p_udf_root, const char *psz_name)
   return p_udf_file;
 }
 
-/* Convert unicode16 to 8-bit char by dripping MSB. 
+/* Convert unicode16 to UTF-8.
-   Wonder if iconv can be used here
+   The returned string is allocated and must be freed by the caller
 */
-static int 
+static char*
-unicode16_decode( const uint8_t *data, int i_len, char *target ) 
+unicode16_decode(const uint8_t *data, int i_len)
 {
-  int p = 1, i = 0;
+  int i;
+  char* r = NULL;
 
-  if( ( data[ 0 ] == 8 ) || ( data[ 0 ] == 16 ) ) do {
+  switch (data[0])
-    if( data[ 0 ] == 16 ) p++;  /* Ignore MSB of unicode16 */
+  {
-    if( p < i_len ) {
+  case 8:
-      target[ i++ ] = data[ p++ ];
+    r = (char*)malloc(i_len);
+    if (r == NULL)
+      return r;
+    for (i=0; i<i_len-1; i++)
+      r[i] = data[i+1];
+    r[i] = 0;
+    return r;
+  case 16:
+    cdio_charset_to_utf8((char*)&data[1], i_len-1, &r, "UCS-2BE");
+    return r;
+  default:
+    return NULL;
   }
-  } while( p < i_len );
-  
-  target[ i ] = '\0';
-  return 0;
 }
 
 
@@ -389,7 +404,8 @@ udf_open (const char *psz_path)
    * Then try to find a reference to a Primary Volume Descriptor.
    */
   {
-    const anchor_vol_desc_ptr_t *p_avdp = &p_udf->anchor_vol_desc_ptr;
+    anchor_vol_desc_ptr_t *p_avdp = &p_udf->anchor_vol_desc_ptr;
+
     const uint32_t mvds_start =
       uint32_from_le(p_avdp->main_vol_desc_seq_ext.loc);
     const uint32_t mvds_end   = mvds_start +
@@ -398,7 +414,6 @@ udf_open (const char *psz_path)
     uint32_t i_lba;
 
     for (i_lba = mvds_start; i_lba < mvds_end; i_lba++) {
-
       udf_pvd_t *p_pvd = (udf_pvd_t *) &data;
 
       if (DRIVER_OP_SUCCESS != udf_read_sectors (p_udf, p_pvd, i_lba, 1) )
@@ -427,18 +442,24 @@ udf_open (const char *psz_path)
 }
 
 /**
- * Gets the Volume Identifier string, in 8bit unicode (latin-1)
+ * Gets the Volume Identifier, as an UTF-8 string
  * psz_volid, place to put the string
  * i_volid, size of the buffer psz_volid points to
  * returns the size of buffer needed for all data
+ * Note: this call accepts a NULL psz_volid, to retrieve the length required.
  */
 int 
 udf_get_volume_id(udf_t *p_udf, /*out*/ char *psz_volid, unsigned int i_volid)
 {
   uint8_t data[UDF_BLOCKSIZE];
   const udf_pvd_t *p_pvd = (udf_pvd_t *) &data;
+  char* r;
   unsigned int volid_len;
 
+  /* clear the output to empty string */
+  if (psz_volid != NULL)
+    psz_volid[0] = 0;
+
   /* get primary volume descriptor */
   if ( DRIVER_OP_SUCCESS != udf_read_sectors(p_udf, &data, p_udf->pvd_lba, 1) )
     return 0;
@@ -448,10 +469,17 @@ udf_get_volume_id(udf_t *p_udf, /*out*/ char *psz_volid,  unsigned int i_volid)
     /* this field is only UDF_VOLID_SIZE bytes something is wrong */
     volid_len = UDF_VOLID_SIZE-1;
   }
-  if(i_volid > volid_len) {
+
-    i_volid = volid_len;
+  r = unicode16_decode((uint8_t *) p_pvd->vol_ident, volid_len);
+  if (r == NULL)
+    return 0;
+
+  volid_len = strlen(r)+1;	/* +1 for NUL terminator */
+  if (psz_volid != NULL) {
+    strncpy(psz_volid, r, MIN(volid_len, i_volid));
+    psz_volid[i_volid-1] = 0;	/* strncpy does not always terminate the dest */
   }
-  unicode16_decode((uint8_t *) p_pvd->vol_ident, i_volid, psz_volid);
+  free(r);
 
   return volid_len;
 }
@@ -486,27 +514,39 @@ udf_get_volumeset_id(udf_t *p_udf, /*out*/ uint8_t *volsetid,
 }
 
 /**
- * Gets the Logical Volume Identifier string, in 8bit unicode (latin-1)
+ * Gets the Logical Volume Identifier string, as an UTF-8 string
  * psz_logvolid, place to put the string (should be at least 64 bytes)
  * i_logvolid, size of the buffer psz_logvolid points to
- * returns the size of buffer needed for all data
+ * returns the size of buffer needed for all data, including NUL terminator
  * A call to udf_get_root() should have been issued before this call
+ * Note: this call accepts a NULL psz_volid, to retrieve the length required.
  */
 int
 udf_get_logical_volume_id(udf_t *p_udf, /*out*/ char *psz_logvolid, unsigned int i_logvolid)
 {
   uint8_t data[UDF_BLOCKSIZE];
   logical_vol_desc_t *p_logvol = (logical_vol_desc_t *) &data;
+  char* r;
   int logvolid_len;
 
+  /* clear the output to empty string */
+  if (psz_logvolid != NULL)
+    psz_logvolid[0] = 0;
+
   if (DRIVER_OP_SUCCESS != udf_read_sectors (p_udf, p_logvol, p_udf->lvd_lba, 1) ) 
     return 0;
 
-  logvolid_len = (p_logvol->logvol_id[127]+1)/2;
+  r = unicode16_decode((uint8_t *) p_logvol->logvol_id, p_logvol->logvol_id[127]);
-  if (i_logvolid > logvolid_len)
+  if (r == NULL)
-    i_logvolid = logvolid_len;
+    return 0;
+
+  logvolid_len = strlen(r)+1;	/* +1 for NUL terminator */
+  if (psz_logvolid != NULL) {
+    strncpy(psz_logvolid, r, MIN(logvolid_len, i_logvolid));
+    psz_logvolid[i_logvolid-1] = 0;	/* strncpy does not always terminate the dest */
+  }
+  free(r);
 
-  unicode16_decode((uint8_t *) p_logvol->logvol_id, 2*i_logvolid, psz_logvolid);
   return logvolid_len;
 }
 
@@ -647,6 +687,7 @@ udf_dirent_t *
 udf_readdir(udf_dirent_t *p_udf_dirent)
 {
   udf_t *p_udf;
+  uint8_t* p;
 
   if (p_udf_dirent->dir_left <= 0) {
     udf_dirent_free(p_udf_dirent);
@@ -706,13 +747,9 @@ udf_readdir(udf_dirent_t *p_udf_dirent)
 		return NULL;
 	}
 
-	if (strlen(p_udf_dirent->psz_name) < i_len) 
+	free_and_null(p_udf_dirent->psz_name);
-	  p_udf_dirent->psz_name = (char *)
+	p = (uint8_t*)p_udf_dirent->fid->u.imp_use.data + p_udf_dirent->fid->u.i_imp_use;
-	    realloc(p_udf_dirent->psz_name, sizeof(char)*i_len+1);
+	p_udf_dirent->psz_name = unicode16_decode(p, i_len);
-	
-	unicode16_decode(p_udf_dirent->fid->u.imp_use.data 
-			 + p_udf_dirent->fid->u.i_imp_use, 
-			 i_len, p_udf_dirent->psz_name);
       }
       return p_udf_dirent;
     }

src/rufus.rc

@@ -32,7 +32,7 @@ LANGUAGE LANG_NEUTRAL, SUBLANG_NEUTRAL
 
 IDD_DIALOG DIALOGEX 12, 12, 206, 329
 STYLE DS_SETFONT | DS_MODALFRAME | DS_FIXEDSYS | DS_CENTER | WS_POPUP | WS_CAPTION | WS_SYSMENU
-CAPTION "Rufus 1.4.4.412"
+CAPTION "Rufus 1.4.4.413"
 FONT 8, "MS Shell Dlg", 400, 0, 0x1
 BEGIN
     DEFPUSHBUTTON   "Start",IDC_START,94,291,50,14
@@ -165,7 +165,7 @@ END
 RTL_IDD_DIALOG DIALOGEX 12, 12, 206, 329
 STYLE DS_SETFONT | DS_MODALFRAME | DS_FIXEDSYS | DS_CENTER | WS_POPUP | WS_CAPTION | WS_SYSMENU
 EXSTYLE WS_EX_RTLREADING | WS_EX_APPWINDOW | WS_EX_LAYOUTRTL
-CAPTION "Rufus 1.4.4.412"
+CAPTION "Rufus 1.4.4.413"
 FONT 8, "MS Shell Dlg", 400, 0, 0x1
 BEGIN
     DEFPUSHBUTTON   "Start",IDC_START,94,291,50,14
@@ -427,8 +427,8 @@ END
 //
 
 VS_VERSION_INFO VERSIONINFO
- FILEVERSION 1,4,4,412
+ FILEVERSION 1,4,4,413
- PRODUCTVERSION 1,4,4,412
+ PRODUCTVERSION 1,4,4,413
  FILEFLAGSMASK 0x3fL
 #ifdef _DEBUG
  FILEFLAGS 0x1L
@@ -445,13 +445,13 @@ BEGIN
         BEGIN
             VALUE "CompanyName", "Akeo Consulting (http://akeo.ie)"
             VALUE "FileDescription", "Rufus"
-            VALUE "FileVersion", "1.4.4.412"
+            VALUE "FileVersion", "1.4.4.413"
             VALUE "InternalName", "Rufus"
             VALUE "LegalCopyright", "© 2011-2014 Pete Batard (GPL v3)"
             VALUE "LegalTrademarks", "http://www.gnu.org/copyleft/gpl.html"
             VALUE "OriginalFilename", "rufus.exe"
             VALUE "ProductName", "Rufus"
-            VALUE "ProductVersion", "1.4.4.412"
+            VALUE "ProductVersion", "1.4.4.413"
         END
     END
     BLOCK "VarFileInfo"