Add codepage translation for non-utf8 id3v1 tags

admin-root/config.xml

@@ -116,6 +116,13 @@
         </item>
     </section>
     <section name="Scanning" advanced="true">
+        <item id="scanning:mp3_tag_codepage">
+	    <name>MP3 Tag Codepage</name>
+	    <short_description>
+	        What codepage non-utf mp3 tags should be converted from (default: ISO-8859-1)
+            </short_description>
+            <type>text</type>
+	</item>
         <item id="scanning:process_playlists">
             <name>Process Playlists</name>
             <short_description>

src/conf.c

@@ -146,6 +146,7 @@ static CONF_ELEMENTS conf_elements[] = {
     { 0, 0, CONF_T_INT,"scanning","case_sensitive" },
     { 0, 0, CONF_T_INT,"scanning","follow_symlinks" },
     { 0, 0, CONF_T_INT,"scanning","skip_first" },
+    { 0, 0, CONF_T_STRING,"scanning","mp3_tag_codepage" },
     { 0, 0, CONF_T_INT,"scan","correct_order" },
 
     /* remapped values */

src/scan-mp3.c

@@ -303,6 +303,7 @@ int scan_mp3_get_mp3tags(char *file, MP3FILE *pmp3) {
     char *tmp;
     int got_numeric_genre;
     int rating;
+    char *conversion_codepage;
 
     pid3file=id3_file_open(file,ID3_FILE_MODE_READONLY);
     if(!pid3file) {
@@ -322,6 +323,9 @@ int scan_mp3_get_mp3tags(char *file, MP3FILE *pmp3) {
 
     DPRINTF(E_SPAM,L_SCAN,"Starting mp3 tag scan\n");
 
+    conversion_codepage = conf_alloc_string("scanning","mp3_tag_codepage",
+                                            "ISO-8859-1");
+
     index=0;
     while((pid3frame=id3_tag_findframe(pid3tag,"",index))) {
         used=0;
@@ -345,20 +349,15 @@ int scan_mp3_get_mp3tags(char *file, MP3FILE *pmp3) {
             native_text=id3_field_getstrings(&pid3frame->fields[1],0);
 
             if(native_text) {
-                /* FIXME: I didn't understand what was happening here.
-                 * this should really be a switch to evaluate latin1
-                 * tags as native codepage.  Not only is this hackish,
-                 * it's just plain wrong.
-                 */
                 have_utf8=1;
-                if(conf_get_int("general","latin1_tags",0)) {
-                    utf8_text=(char *)id3_ucs4_latin1duplicate(native_text);
-                } else {
-                    utf8_text=(char *)id3_ucs4_utf8duplicate(native_text);
-                }
+                utf8_text = util_xtoutf8_alloc(native_text,strlen(native_text),
+                                               conversion_codepage);
 
-                if(utf8_text)
-                    mem_register(utf8_text,0);
+                if(!utf8_text) {
+                    utf8_text = (char*)id3_ucs4_latin1duplicate(native_text);
+                    if(utf8_text)
+                        mem_register(utf8_text,0);
+                }
 
                 if(!strcmp(pid3frame->id,"TIT2")) { /* Title */
                     used=1;
@@ -509,6 +508,7 @@ int scan_mp3_get_mp3tags(char *file, MP3FILE *pmp3) {
         index++;
     }
 
+    free(conversion_codepage);
     id3_file_close(pid3file);
     DPRINTF(E_DBG,L_SCAN,"Got id3 tag successfully\n");
     return TRUE;

src/util.c

@@ -102,7 +102,7 @@ int util_utf16_byte_len(unsigned char *utf16) {
 
 
 unsigned char *util_utf8toutf16_alloc(unsigned char *utf8) {
-    char *utf16;
+    unsigned char *utf16;
 
     utf16 = calloc(1,strlen((char*)utf8) * 4 + 1);
     if(util_xtoy(utf16,strlen((char*)utf8) * 4 + 1, utf8, strlen((char*)utf8),"UTF-8","UTF-16LE")) {
@@ -114,7 +114,7 @@ unsigned char *util_utf8toutf16_alloc(unsigned char *utf8) {
 }
 
 unsigned char *util_utf16toutf8_alloc(unsigned char *utf16, int slen) {
-    char *utf8;
+    unsigned char *utf8;
 
     utf8=calloc(1, slen * 2 + 1);
     if(util_xtoy(utf8,slen * 2 + 1,utf16,slen,"UTF-16LE","UTF-8")) {
@@ -125,17 +125,33 @@ unsigned char *util_utf16toutf8_alloc(unsigned char *utf16, int slen) {
     return NULL;
 }
 
+unsigned char *util_xtoutf8_alloc(unsigned char *x, int slen, char *from) {
+    unsigned char *utf8;
+
+    utf8 = calloc(1, slen * 4 + 1);
+    if(util_xtoy(utf8,slen * 4 + 1, x, slen, from, "UTF-8")) {
+        return utf8;
+    }
+    free(utf8);
+    return NULL;
+}
+
 int util_xtoy(unsigned char *dbuffer, int dlen, unsigned char *sbuffer, int slen, char *from, char *to) {
     iconv_t iv;
     size_t csize;
 
+    size_t st_dlen = dlen;
+    size_t st_slen = slen;
+
     memset(dbuffer,0,dlen);
+
     iv=iconv_open(to,from);
     if(iv == (iconv_t)-1) {
         DPRINTF(E_LOG,L_MISC,"iconv error: iconv_open failed with %d\n",errno);
     }
 
-    csize = iconv(iv,&sbuffer,&slen,&dbuffer,&dlen);
+    csize = iconv(iv,(const char **)&sbuffer,&st_slen,
+                  (char **)&dbuffer,&st_dlen);
     if(csize == (size_t)-1) {
         switch(errno) {
             case EILSEQ:
@@ -397,6 +413,7 @@ char *util_vasprintf(char *fmt, va_list ap) {
     if(!outbuf)
         DPRINTF(E_FATAL,L_MISC,"Could not allocate buffer in vasprintf\n");
 
+    ap2=ap; /* shut up lint warnings */
     VA_COPY(ap2,ap);
 
     while(1) {

src/util.h

@@ -46,6 +46,7 @@ extern unsigned char *util_utf16touft8_alloc(unsigned char *utf16, int slen);
 extern int util_utf8toutf16(unsigned char *utf16, int dlen, unsigned char *utf8, int slen);
 extern int util_utf16toutf8(unsigned char *utf8, int dlen, unsigned char *utf16, int slen);
 extern int util_xtoy(unsigned char *dbuffer, int dlen, unsigned char *sbuffer, int slen, char *from, char *to);
+extern unsigned char *util_xtoutf8_alloc(unsigned char *x,int slen,char *from);
 extern int util_utf16_byte_len(unsigned char *utf16);
 
 extern void util_hexdump(unsigned char *block, int len);