Per-stream language-tags extraction in asfdec.

Patch by Cyril Comparon: gmail(name, surname);
Original thread: Suggestion for a centralized language-tag facility in libavformat
Date: 04/10/2009 07:33 PM

Originally committed as revision 18800 to svn://svn.ffmpeg.org/ffmpeg/trunk

4 files changed, 40 insertions, 2 deletions

diff --git a/libavformat/Makefile b/libavformat/Makefile
index b9e5eb7229..c0be725f86 100644
--- a/libavformat/Makefile
+++ b/libavformat/Makefile
@@ -18,7 +18,7 @@ OBJS-$(CONFIG_AMR_DEMUXER)               += amr.o
 OBJS-$(CONFIG_AMR_MUXER)                 += amr.o
 OBJS-$(CONFIG_APC_DEMUXER)               += apc.o
 OBJS-$(CONFIG_APE_DEMUXER)               += ape.o
-OBJS-$(CONFIG_ASF_DEMUXER)               += asfdec.o asf.o asfcrypt.o riff.o
+OBJS-$(CONFIG_ASF_DEMUXER)               += asfdec.o asf.o asfcrypt.o riff.o avlanguage.o
 OBJS-$(CONFIG_ASF_MUXER)                 += asfenc.o asf.o riff.o
 OBJS-$(CONFIG_ASF_STREAM_MUXER)          += asfenc.o asf.o riff.o
 OBJS-$(CONFIG_ASS_DEMUXER)               += assdec.o
diff --git a/libavformat/asf.c b/libavformat/asf.c
index 2e17554f03..79ef61444b 100644
--- a/libavformat/asf.c
+++ b/libavformat/asf.c
@@ -112,6 +112,9 @@ const ff_asf_guid ff_asf_my_guid = {
     0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
 };
 
+const ff_asf_guid ff_asf_language_guid = {
+    0xa9, 0x46, 0x43, 0x7c, 0xe0, 0xef, 0xfc, 0x4b, 0xb2, 0x29, 0x39, 0x3e, 0xde, 0x41, 0x5c, 0x85
+};
 
 const AVMetadataConv ff_asf_metadata_conv[] = {
     { "AlbumArtist", "artist"    },
diff --git a/libavformat/asf.h b/libavformat/asf.h
index 6d6619dcc8..f322ee2a58 100644
--- a/libavformat/asf.h
+++ b/libavformat/asf.h
@@ -42,6 +42,8 @@ typedef struct {
 
     int64_t packet_pos;
 
+    uint16_t stream_language_index;
+
 } ASFStream;
 
 typedef uint8_t ff_asf_guid[16];
@@ -85,6 +87,7 @@ typedef struct {
     int asfid2avid[128];                 ///< conversion table from asf ID 2 AVStream ID
     ASFStream streams[128];              ///< it's max number and it's not that big
     uint32_t stream_bitrates[128];       ///< max number of streams, bitrate for each (for streaming)
+    char stream_languages[128][6];       ///< max number of streams, language for each (RFC1766, e.g. en-US)
     /* non streamed additonnal info */
     uint64_t nb_packets;                 ///< how many packets are there in the file, invalid if broadcasting
     int64_t duration;                    ///< in 100ns units
@@ -157,6 +160,7 @@ extern const ff_asf_guid ff_asf_ext_stream_embed_stream_header;
 extern const ff_asf_guid ff_asf_ext_stream_audio_stream;
 extern const ff_asf_guid ff_asf_metadata_header;
 extern const ff_asf_guid ff_asf_my_guid;
+extern const ff_asf_guid ff_asf_language_guid;
 
 extern const AVMetadataConv ff_asf_metadata_conv[];
 
diff --git a/libavformat/asfdec.c b/libavformat/asfdec.c
index e76f98beb2..3810663c2f 100644
--- a/libavformat/asfdec.c
+++ b/libavformat/asfdec.c
@@ -20,11 +20,13 @@
  */
 
 #include "libavutil/common.h"
+#include "libavutil/avstring.h"
 #include "libavcodec/mpegaudio.h"
 #include "avformat.h"
 #include "riff.h"
 #include "asf.h"
 #include "asfcrypt.h"
+#include "avlanguage.h"
 
 void ff_mms_set_stream_selection(URLContext *h, AVFormatContext *format);
 
@@ -76,6 +78,7 @@ static void print_guid(const ff_asf_guid *g)
     else PRINT_IF_GUID(g, ff_asf_ext_stream_audio_stream);
     else PRINT_IF_GUID(g, ff_asf_metadata_header);
     else PRINT_IF_GUID(g, stream_bitrate_guid);
+    else PRINT_IF_GUID(g, ff_asf_language_guid);
     else
         dprintf(NULL, "(GUID: unknown) ");
     for(i=0;i<16;i++)
@@ -238,6 +241,8 @@ static int asf_read_header(AVFormatContext *s, AVFormatParameters *ap)
             st->priv_data = asf_st;
             start_time = asf->hdr.preroll;
 
+            asf_st->stream_language_index = 128; // invalid stream index means no language info
+
             if(!(asf->hdr.flags & 0x01)) { // if we aren't streaming...
                 st->duration = asf->hdr.send_time /
                     (10000000 / 1000) - start_time;
@@ -403,6 +408,16 @@ static int asf_read_header(AVFormatContext *s, AVFormatParameters *ap)
 //                av_log(s, AV_LOG_ERROR, "flags: 0x%x stream id %d, bitrate %d\n", flags, stream_id, bitrate);
                 asf->stream_bitrates[stream_id]= bitrate;
             }
+        } else if (!guidcmp(&g, &ff_asf_language_guid)) {
+            int j;
+            int stream_count = get_le16(pb);
+            for(j = 0; j < stream_count; j++) {
+                char lang[6];
+                unsigned int lang_len = get_byte(pb);
+                get_str16_nolen(pb, lang_len, lang, sizeof(lang));
+                if (j < 128)
+                    av_strlcpy(asf->stream_languages[j], lang, sizeof(*asf->stream_languages));
+            }
         } else if (!guidcmp(&g, &ff_asf_extended_content_header)) {
             int desc_count, i;
 
@@ -443,6 +458,7 @@ static int asf_read_header(AVFormatContext *s, AVFormatParameters *ap)
         } else if (!guidcmp(&g, &ff_asf_ext_stream_header)) {
             int ext_len, payload_ext_ct, stream_ct;
             uint32_t ext_d, leak_rate, stream_num;
+            unsigned int stream_languageid_index;
 
             get_le64(pb); // starttime
             get_le64(pb); // endtime
@@ -455,7 +471,11 @@ static int asf_read_header(AVFormatContext *s, AVFormatParameters *ap)
             get_le32(pb); // max-object-size
             get_le32(pb); // flags (reliable,seekable,no_cleanpoints?,resend-live-cleanpoints, rest of bits reserved)
             stream_num = get_le16(pb); // stream-num
-            get_le16(pb); // stream-language-id-index
+
+            stream_languageid_index = get_le16(pb); // stream-language-id-index
+            if (stream_num < 128)
+                asf->streams[stream_num].stream_language_index = stream_languageid_index;
+
             get_le64(pb); // avg frametime in 100ns units
             stream_ct = get_le16(pb); //stream-name-count
             payload_ext_ct = get_le16(pb); //payload-extension-system-count
@@ -535,6 +555,17 @@ static int asf_read_header(AVFormatContext *s, AVFormatParameters *ap)
                           &st->sample_aspect_ratio.den,
                           dar[i].num, dar[i].den, INT_MAX);
 //av_log(s, AV_LOG_ERROR, "dar %d:%d sar=%d:%d\n", dar[i].num, dar[i].den, st->sample_aspect_ratio.num, st->sample_aspect_ratio.den);
+
+            // copy and convert language codes to the frontend
+            if (asf->streams[i].stream_language_index < 128) {
+                const char *rfc1766 = asf->stream_languages[asf->streams[i].stream_language_index];
+                if (rfc1766 && strlen(rfc1766) > 1) {
+                    const char primary_tag[3] = { rfc1766[0], rfc1766[1], '\0' }; // ignore country code if any
+                    const char *iso6392 = av_convert_lang_to(primary_tag, AV_LANG_ISO639_2_BIBL);
+                    if (iso6392)
+                        av_metadata_set(&st->metadata, "language", iso6392);
+                }
+            }
         }
     }