Extend id3v2 handling by adding (very simplistic) handling for unicode

tags, and also make it deal with gaps after the header or junk before the
actual start of mp3 data.
This makes it able to properly extract info from all mp3 files I have.

1 files changed, 142 insertions, 6 deletions

diff --git a/audio/mserv/patches/patch-ae b/audio/mserv/patches/patch-ae
index 901d5664b02..57a54bdea19 100644
--- a/audio/mserv/patches/patch-ae
+++ b/audio/mserv/patches/patch-ae
@@ -1,16 +1,26 @@
-$NetBSD: patch-ae,v 1.8 2004/02/20 00:41:16 abs Exp $
+$NetBSD: patch-ae,v 1.9 2009/09/26 21:16:58 martin Exp $
 
---- mserv/mp3info.c.orig	Thu Feb 19 17:46:03 2004
-+++ mserv/mp3info.c
-@@ -27,6 +27,7 @@
+--- mserv/mp3info.c.orig	2003-07-29 02:17:48.000000000 +0200
++++ mserv/mp3info.c	2009-09-26 22:11:43.000000000 +0200
+@@ -26,7 +26,17 @@
+ #define h_id(val)             ((val>>19)&1)
  #define h_thing(val)          ((val>>20)&0xfff)
  
++#ifdef STANDALONE
++#define VERBOSE
++#endif
++#ifdef VERBOSE
++#define LOGF(ARG) printf ARG
++#else
++#define LOGF(ARG)
++#endif
++
  #define ID3V2HEADERLEN		10
 +#define min(x,y)		((x)<(y)?(x):(y))
  
  /* mp3 bit rate and sampling frequency tables */
  
-@@ -45,12 +46,14 @@ const int sampling_freq_table[2][3] =
+@@ -45,12 +55,14 @@ const int sampling_freq_table[2][3] =
  /* structure of id3 tag in mp3 file */
  
  typedef struct id3tag_disc_str
@@ -31,7 +41,100 @@ $NetBSD: patch-ae,v 1.8 2004/02/20 00:41:16 abs Exp $
    unsigned char genre;
  } id3tag_disc;
  
-@@ -312,38 +315,44 @@ int mserv_mp3info_readlen(const char *fn
+@@ -228,11 +240,28 @@ static int read_id3v2_frame(FILE *f, id3
+   if (fread(frame->data, 1, frame->datalen, f) != frame->datalen)
+     return -1;
+   if (frame->frameid[0] == 'T' && memcmp(frame->frameid + 1, "XXX", 3)) {
+-    frame->data[frame->datalen] = 0;
+-    if (frame->data[0] == 0) /* Only handle non unicode */
++    if (frame->data[0] == 0) {
++      /* ISO8859-1 */
++      frame->data[frame->datalen] = 0;
+       strcpy(frame->data, frame->data + 1);
+-    else
+-      frame->data[0] = 0;
++    } else if (frame->data[0] == 1 && frame->datalen >= 5) {
++      /* unicode, convert as long as it is 8 bit only */
++      int d = 0, i = 1, off0 = 0, off1 = 1;
++      /* check BOM vs. endianess */
++      if ((unsigned char)frame->data[i] == 0xff && (unsigned char)frame->data[i+1] == 0xfe) {
++	off0 = 1; off1 = 0;
++      } else if ((unsigned char)frame->data[i] == 0xfe && (unsigned char)frame->data[i+1] == 0xff) {
++	off1 = 1; off0 = 0;
++      } else {
++	return -1;
++      }
++      for (i=3; (i+1) < frame->datalen /* && frame->data[i] == 0 */; i+=2) {
++        frame->data[d++] = frame->data[i+off1];
++	if (frame->data[i] == 0 && frame->data[i+1] == 0)
++	  break;
++      }
++      frame->data[d] = 0;
++    }
+   }
+   return 0;
+ }
+@@ -264,12 +293,12 @@ int mserv_mp3info_readlen(const char *fn
+ {
+   FILE              *f;
+   int               bitrate, fs, length, headerlen;
+-  long int          filelen;
++  long int          filelen, skipped;
+   float             mean_frame_size;
+   char              tag[3];
+   unsigned char     data[4];
+   unsigned long int flags;
+-  int               errnok;
++  int               errnok, found;
+   char              *e;
+   
+   if (id3tag)
+@@ -281,15 +310,39 @@ int mserv_mp3info_readlen(const char *fn
+   if ((headerlen = mp3_id3v2head(f)) == -1)
+     goto error;
+ 
+-  if (fseek(f, headerlen, SEEK_SET) == -1 || fread(&data, 4, 1, f) != 1)
++  if (fseek(f, headerlen, SEEK_SET) == -1 || fread(&data[0], 1, 1, f) != 1) {
++    LOGF(("%s: can't read after headerlen %d\n", fname, headerlen));
+     goto error;
++  }
+ 
+-  /* endian independent to 32-bit flags */
+-  flags = (data[0]<<24)+(data[1]<<16)+(data[2]<<8)+data[3];
+-  if (!is_mp3(flags)) {
+-    errno = EDOM;
++  found = skipped = 0;
++  for (;;) {
++    while (data[0] != 0xff) {
++      /* skip junk at beginning or after header - happens quite often */
++      if (fread(&data[0], 1, 1, f) != 1)
++        goto error;
++      skipped++;
++      if (headerlen < 1 && skipped > 4096)
++        break; /* we are not sure this even is a mp3 file */
++    }
++    if (fread(&data[1], 1, 3, f) != 3)
++      goto error;
++
++    /* endian independent to 32-bit flags */
++    flags = (data[0]<<24)+(data[1]<<16)+(data[2]<<8)+data[3];
++    if (is_mp3(flags)) {
++      found = 1;
++      break;
++    }
++    if (headerlen < 1 && skipped > 4096)
++      break; /* we are not sure this even is a mp3 file */
++  }
++  if (!found) {
++    LOGF(("%s: no header found within 4k\n", fname));
+     goto error;
+   }
++  if (skipped)
++    LOGF(("%s: skipped %ld bytes\n", fname, skipped));
+ 
+   bitrate = bitrate_table[h_id(flags)][3-h_layer(flags)]
+     [h_bitrate_index(flags)];
+@@ -312,38 +365,44 @@ int mserv_mp3info_readlen(const char *fn
      if (id3tag)
      {
        id3tag_disc tag_disc;
@@ -86,3 +189,36 @@ $NetBSD: patch-ae,v 1.8 2004/02/20 00:41:16 abs Exp $
        e = id3tag->comment + strlen(id3tag->comment);
        while (e > id3tag->comment && *(e-1) == ' ')
  	*--e = '\0';
+@@ -395,3 +454,32 @@ int mserv_mp3info_readlen(const char *fn
+   errno = errnok;
+   return -1;
+ }
++
++#ifdef STANDALONE
++int main(int argc, char **argv)
++{
++	t_id3tag tag;
++	int bitrate, len;
++
++	if (argc < 2) {
++		printf("usage: minfo (filename)\n");
++		return 1;
++	}
++	len = mserv_mp3info_readlen(argv[1], &bitrate, &tag);
++	if (len < 0) {
++		printf("no mp3 found\n");
++	} else {
++		if (bitrate)
++			printf("bitrate %d kbit/s, estimated length "
++				"%d:%02d.%02d s\n",
++				bitrate, len/6000, (len%6000)/100, len%100);
++		if (tag.present) {
++			printf("title:  %s\n", tag.title);
++			printf("album:  %s\n", tag.album);
++			printf("artist: %s\n", tag.artist);
++			printf("year:   %s\n", tag.year);
++		}
++	}
++	return 0;
++}
++#endif