[Lame-cvs] SF.net SVN: lame:[6528] trunk/lame/frontend/parse.c

[<ale...@us...>]

Revision: 6528
          http://sourceforge.net/p/lame/svn/6528
Author:   aleidinger
Date:     2023-12-14 09:13:45 +0000 (Thu, 14 Dec 2023)
Log Message:
-----------
Unbreak UTF8 genre tag. Text tags still broken.

Disable UTF8 for now in the frontend.

Modified Paths:
--------------
    trunk/lame/frontend/parse.c

Modified: trunk/lame/frontend/parse.c
===================================================================
--- trunk/lame/frontend/parse.c	2023-12-14 09:12:21 UTC (rev 6527)
+++ trunk/lame/frontend/parse.c	2023-12-14 09:13:45 UTC (rev 6528)
@@ -304,6 +304,7 @@
     size_t w = currCharCodeSize();
     char* dst = 0;
     if (src != 0) {
+        /* XXX: size calculation for UTF-16, a bit too much for UTF-8, but not too small */
         size_t const l = strlenMultiByte(src, w);
         size_t const n = (l+1)*4;
         dst = calloc(n+4, 4);
@@ -310,11 +311,9 @@
         if (dst != 0) {
             char* cur_code = currentCharacterEncoding();
             iconv_t xiconv = iconv_open("UTF-8//TRANSLIT", cur_code);
-            dst[0] = 0xff;
-            dst[1] = 0xfe;
             if (xiconv != (iconv_t)-1) {
                 char* i_ptr = (char*)src;
-                char* o_ptr = &dst[2];
+                char* o_ptr = &dst[0];
                 size_t srcln = l*w;
                 size_t avail = n;
                 iconv(xiconv, &i_ptr, &srcln, &o_ptr, &avail);
@@ -370,7 +369,7 @@
 
 unsigned short* toUtf8(char const* s)
 {
-    return (unsigned short *)s;
+    return local8BitToUtf8(s);
 }
 #endif
 
@@ -404,18 +403,38 @@
 
 #ifdef ID3TAGS_EXTENDED
 static int
-set_id3v2tag(lame_global_flags* gfp, int type, unsigned short const* str)
+set_id3v2tag(lame_global_flags* gfp, TextEncoding enc, int type, unsigned short const* str)
 {
-    switch (type)
+    switch (enc)
     {
-        case 'a': return id3tag_set_textinfo_utf16(gfp, "TPE1", str);
-        case 't': return id3tag_set_textinfo_utf16(gfp, "TIT2", str);
-        case 'l': return id3tag_set_textinfo_utf16(gfp, "TALB", str);
-        case 'g': return id3tag_set_textinfo_utf16(gfp, "TCON", str);
-        case 'c': return id3tag_set_comment_utf16(gfp, 0, 0, str);
-        case 'n': return id3tag_set_textinfo_utf16(gfp, "TRCK", str);
-        case 'y': return id3tag_set_textinfo_utf16(gfp, "TYER", str);
-        case 'v': return id3tag_set_fieldvalue_utf16(gfp, str);
+        case TENC_UTF8:
+            switch (type)
+            {
+                case 'a': return id3tag_set_textinfo_utf8(gfp, "TPE1", str);
+                case 't': return id3tag_set_textinfo_utf8(gfp, "TIT2", str);
+                case 'l': return id3tag_set_textinfo_utf8(gfp, "TALB", str);
+                case 'g': return id3tag_set_textinfo_utf8(gfp, "TCON", str);
+                case 'c': return id3tag_set_comment_ucs2(gfp, 0, 0, str);
+                case 'n': return id3tag_set_textinfo_utf8(gfp, "TRCK", str);
+                case 'y': return id3tag_set_textinfo_utf8(gfp, "TYER", str);
+                case 'v': return id3tag_set_fieldvalue_ucs2(gfp, str);
+            }
+            ;;
+        case TENC_UTF16:
+            switch (type)
+            {
+                case 'a': return id3tag_set_textinfo_utf16(gfp, "TPE1", str);
+                case 't': return id3tag_set_textinfo_utf16(gfp, "TIT2", str);
+                case 'l': return id3tag_set_textinfo_utf16(gfp, "TALB", str);
+                case 'g': return id3tag_set_textinfo_utf16(gfp, "TCON", str);
+                case 'c': return id3tag_set_comment_utf16(gfp, 0, 0, str);
+                case 'n': return id3tag_set_textinfo_utf16(gfp, "TRCK", str);
+                case 'y': return id3tag_set_textinfo_utf16(gfp, "TYER", str);
+                case 'v': return id3tag_set_fieldvalue_utf16(gfp, str);
+            }
+            ;;
+        default:
+            return -3;
     }
     return 0;
 }
@@ -443,7 +462,7 @@
 {
     void* x = 0;
     int result;
-    if (enc == TENC_UTF16 && type != 'v' ) {
+    if ((enc == TENC_UTF16 || enc == TENC_UTF8) && type != 'v' ) {
         id3_tag(gfp, type, TENC_LATIN1, str); /* for id3v1 */
     }
     switch (enc) 
@@ -462,8 +481,8 @@
         default:
 #ifdef ID3TAGS_EXTENDED
         case TENC_LATIN1: result = set_id3tag(gfp, type, x);   break;
-        case TENC_UTF16:  result = set_id3v2tag(gfp, type, x); break;
-        case TENC_UTF8:   result = set_id3v2tag(gfp, type, x); break;
+        case TENC_UTF16:  result = set_id3v2tag(gfp, enc, type, x); break;
+        case TENC_UTF8:   result = set_id3v2tag(gfp, enc, type, x); break;
 #else
         case TENC_RAW:    result = set_id3tag(gfp, type, x);   break;
 #endif
@@ -679,7 +698,9 @@
             "    --id3v1-only    add only a version 1 tag\n"
             "    --id3v2-only    add only a version 2 tag\n"
 #ifdef ID3TAGS_EXTENDED
+#ifdef UTF8BROKEN
             "    --id3v2-utf8    add following options in unicode UTF-8 text encoding\n"
+#endif
             "    --id3v2-utf16   add following options in unicode UTF-16 text encoding\n"
             "    --id3v2-latin1  add following options in latin-1 text encoding\n"
 #endif
@@ -1596,7 +1617,7 @@
     int     id3tag_mode = ID3TAG_MODE_DEFAULT;
     int     ignore_tag_errors = 0;  /* Ignore errors in values passed for tags */
 #ifdef ID3TAGS_EXTENDED
-    enum TextEncoding id3_tenc = TENC_UTF8;
+    enum TextEncoding id3_tenc = TENC_UTF16;
 #else
     enum TextEncoding id3_tenc = TENC_LATIN1;
 #endif
@@ -1809,9 +1830,11 @@
                     id3_tenc = TENC_UTF16;
                     id3tag_add_v2(gfp);
 
+#ifdef UTF8BROKEN
                 T_ELIF("id3v2-utf8")
                     id3_tenc = TENC_UTF8;
                     id3tag_add_v2_4_UTF8(gfp);
+#endif
 
                 T_ELIF("id3v2-latin1")
                     id3_tenc = TENC_LATIN1;