[libsoup: 9/10] sniffing: Add audio/video sniffing

[Dan Winship <danw src gnome org>]

commit 9530c7b58197d7d45e21aa60a9986735ccf515b1
Author: Gustavo Noronha Silva <gns gnome org>
Date:   Tue Dec 10 19:45:55 2013 +0100

    sniffing: Add audio/video sniffing

 libsoup/soup-content-sniffer.c |  111 ++++++++++++++++++++++++++++++++++++++++
 tests/resources/test.aiff      |  Bin 0 -> 384088 bytes
 tests/resources/test.mp4       |  Bin 0 -> 192844 bytes
 tests/resources/test.ogg       |  Bin 0 -> 16994 bytes
 tests/resources/test.wav       |  Bin 0 -> 384080 bytes
 tests/resources/test.webm      |  Bin 0 -> 149879 bytes
 tests/sniffing-test.c          |   19 +++++++
 tests/soup-tests.gresource.xml |    5 ++
 8 files changed, 135 insertions(+), 0 deletions(-)
---
diff --git a/libsoup/soup-content-sniffer.c b/libsoup/soup-content-sniffer.c
index 5e0b2a7..5659af9 100644
--- a/libsoup/soup-content-sniffer.c
+++ b/libsoup/soup-content-sniffer.c
@@ -177,6 +177,103 @@ sniff_images (SoupContentSniffer *sniffer, SoupBuffer *buffer)
 }
 
 /* This table is based on the MIMESNIFF spec;
+ * See 6.2 Matching an audio or video type pattern
+ */
+static SoupContentSnifferMediaPattern audio_video_types_table[] = {
+       { (const guchar *)"\xFF\xFF\xFF\xFF",
+         (const guchar *)"\x1A\x45\xDF\xA3",
+         4,
+         "video/webm" },
+
+       { (const guchar *)"\xFF\xFF\xFF\xFF",
+         (const guchar *)".snd",
+         4,
+         "audio/basic" },
+
+
+       { (const guchar *)"\xFF\xFF\xFF\xFF\x00\x00\x00\x00\xFF\xFF\xFF\xFF",
+         (const guchar *)"FORM\0\0\0\0AIFF",
+         12,
+         "audio/aiff" },
+
+       { (const guchar *)"\xFF\xFF\xFF",
+         (const guchar *)"ID3",
+         3,
+         "audio/mpeg" },
+
+       { (const guchar *)"\xFF\xFF\xFF\xFF\xFF",
+         (const guchar *)"OggS\0",
+         5,
+         "application/ogg" },
+
+       { (const guchar *)"\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF",
+         (const guchar *)"MThd\x00\x00\x00\x06",
+         8,
+         "audio/midi" },
+
+       { (const guchar *)"\xFF\xFF\xFF\xFF\x00\x00\x00\x00\xFF\xFF\xFF\xFF",
+         (const guchar *)"RIFF\x00\x00\x00\x00AVI ",
+         12,
+         "video/avi" },
+
+       { (const guchar *)"\xFF\xFF\xFF\xFF\x00\x00\x00\x00\xFF\xFF\xFF\xFF",
+         (const guchar *)"RIFF\x00\x00\x00\x00WAVE",
+         12,
+         "audio/wave" },
+};
+
+static gboolean
+sniff_mp4 (SoupContentSniffer *sniffer, SoupBuffer *buffer)
+{
+       const char *resource = (const char *)buffer->data;
+       int resource_length = MIN (512, buffer->length);
+       guint32 box_size = *((guint32*)resource);
+       int i;
+
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+       box_size = ((box_size >> 24) |
+                   ((box_size << 8) & 0x00FF0000) |
+                   ((box_size >> 8) & 0x0000FF00) |
+                   (box_size << 24));
+#endif
+
+       if (resource_length < 12 || resource_length < box_size || box_size % 4 != 0)
+               return FALSE;
+
+       if (!g_str_has_prefix (resource + 4, "ftyp"))
+               return FALSE;
+
+       if (!g_str_has_prefix (resource + 8, "mp4"))
+               return FALSE;
+
+       for (i = 16; i < box_size && i < resource_length; i = i + 4) {
+               if (g_str_has_prefix (resource + i, "mp4"))
+                       return TRUE;
+       }
+
+       return FALSE;
+}
+
+static char*
+sniff_audio_video (SoupContentSniffer *sniffer, SoupBuffer *buffer)
+{
+       char *sniffed_type;
+
+       sniffed_type = sniff_media (sniffer,
+                                   buffer,
+                                   audio_video_types_table,
+                                   G_N_ELEMENTS (audio_video_types_table));
+
+       if (sniffed_type != NULL)
+               return sniffed_type;
+
+       if (sniff_mp4 (sniffer, buffer))
+               return g_strdup ("video/mp4");
+
+       return NULL;
+}
+
+/* This table is based on the MIMESNIFF spec;
  * See 7.1 Identifying a resource with an unknown MIME type
  */
 typedef struct {
@@ -462,6 +559,10 @@ sniff_unknown (SoupContentSniffer *sniffer, SoupBuffer *buffer,
        if (sniffed_type != NULL)
                return sniffed_type;
 
+       sniffed_type = sniff_audio_video (sniffer, buffer);
+
+       if (sniffed_type != NULL)
+               return sniffed_type;
 
        for (i = 0; i < resource_length; i++) {
                if (byte_looks_binary[resource[i]])
@@ -714,6 +815,16 @@ soup_content_sniffer_real_sniff (SoupContentSniffer *sniffer, SoupMessage *msg,
                return g_strdup (content_type);
        }
 
+       /* 7. Audio and video types. */
+       if (!g_ascii_strncasecmp (content_type, "audio/", 6) ||
+           !g_ascii_strncasecmp (content_type, "video/", 6) ||
+           !g_ascii_strcasecmp (content_type, "application/ogg")) {
+               sniffed_type = sniff_audio_video (sniffer, buffer);
+               if (sniffed_type != NULL)
+                       return sniffed_type;
+               return g_strdup (content_type);
+        }
+
        /* If we got text/plain, use text_or_binary */
        if (g_str_equal (content_type, "text/plain")) {
                return sniff_text_or_binary (sniffer, buffer);
diff --git a/tests/resources/test.aiff b/tests/resources/test.aiff
new file mode 100644
index 0000000..9a1ecbb
Binary files /dev/null and b/tests/resources/test.aiff differ
diff --git a/tests/resources/test.mp4 b/tests/resources/test.mp4
new file mode 100644
index 0000000..d278c8a
Binary files /dev/null and b/tests/resources/test.mp4 differ
diff --git a/tests/resources/test.ogg b/tests/resources/test.ogg
new file mode 100644
index 0000000..e8f49ac
Binary files /dev/null and b/tests/resources/test.ogg differ
diff --git a/tests/resources/test.wav b/tests/resources/test.wav
new file mode 100644
index 0000000..11660b2
Binary files /dev/null and b/tests/resources/test.wav differ
diff --git a/tests/resources/test.webm b/tests/resources/test.webm
new file mode 100644
index 0000000..7e53d0b
Binary files /dev/null and b/tests/resources/test.webm differ
diff --git a/tests/sniffing-test.c b/tests/sniffing-test.c
index 868b7c5..5b0e6ee 100644
--- a/tests/sniffing-test.c
+++ b/tests/sniffing-test.c
@@ -567,6 +567,25 @@ main (int argc, char **argv)
                              "type/image_png/tux.webp => image/webp",
                              do_sniffing_test);
 
+       /* Test audio and video sniffing path */
+       g_test_add_data_func ("/sniffing/type/audio/wav",
+                             "type/audio_mpeg/test.wav => audio/wave",
+                             do_sniffing_test);
+       g_test_add_data_func ("/sniffing/type/audio/aiff",
+                             "type/audio_mpeg/test.aiff => audio/aiff",
+                             do_sniffing_test);
+       g_test_add_data_func ("/sniffing/type/audio/ogg",
+                             "type/audio_mpeg/test.ogg => application/ogg",
+                             do_sniffing_test);
+       g_test_add_data_func ("/sniffing/type/video/webm",
+                             "type/video_theora/test.webm => video/webm",
+                             do_sniffing_test);
+
+       /* Test the MP4 sniffing path */
+       g_test_add_data_func ("/sniffing/type/video/mp4",
+                             "unknown/test.mp4 => video/mp4",
+                             do_sniffing_test);
+
        /* The spec tells us to only use the last Content-Type header */
        g_test_add_data_func ("/sniffing/multiple-headers",
                              "multiple_headers/home.gif => image/gif",
diff --git a/tests/soup-tests.gresource.xml b/tests/soup-tests.gresource.xml
index 444ed58..b24a729 100644
--- a/tests/soup-tests.gresource.xml
+++ b/tests/soup-tests.gresource.xml
@@ -15,7 +15,12 @@
     <file>resources/mbox.zlib</file>
     <file>resources/ps_binary.ps</file>
     <file>resources/rss20.xml</file>
+    <file>resources/test.aiff</file>
     <file>resources/test.html</file>
+    <file>resources/test.mp4</file>
+    <file>resources/test.ogg</file>
+    <file>resources/test.wav</file>
+    <file>resources/test.webm</file>
     <file>resources/text.txt</file>
     <file>resources/text_binary.txt</file>
     <file>resources/tux.webp</file>