[GNUnet-SVN] r18218 - Extractor/src/plugins

[gnunet]

Author: grothoff
Date: 2011-11-19 15:35:25 +0100 (Sat, 19 Nov 2011)
New Revision: 18218

Modified:
   Extractor/src/plugins/mkv_extractor.c
Log:
some mkv code cleanup

Modified: Extractor/src/plugins/mkv_extractor.c
===================================================================
--- Extractor/src/plugins/mkv_extractor.c       2011-11-19 14:20:42 UTC (rev 
18217)
+++ Extractor/src/plugins/mkv_extractor.c       2011-11-19 14:35:25 UTC (rev 
18218)
@@ -26,14 +26,28 @@
 
 #include "platform.h"
 #include "extractor.h"
+#include <stdint.h>
 
-
 #define ADD(s,t) do { if (0 != (ret = proc (proc_cls, "mkv", t, 
EXTRACTOR_METAFORMAT_UTF8, "text/plain", s, strlen(s)+1))) goto EXIT; } while 
(0)
 
+/**
+ * FIXME: document
+ */
 #define BUFFER_SIZE 0x4000
+
+/**
+ * FIXME: document
+ */
 #define MAX_STRING_SIZE 1024
+
+/**
+ * FIXME: document
+ */
 #define MAX_STREAMS 9
-typedef long long int64;
+
+/**
+ * FIXME: document
+ */
 enum
 {
   MKV_Track_video = 1,
@@ -41,6 +55,10 @@
   MKV_Track_subtitle = 3,
   MKV_Track_subtitle_orig = 0x11
 };
+
+/**
+ * FIXME: document
+ */
 enum
 {
   MKVID_OutputSamplingFrequency = 0x78B5,
@@ -73,103 +91,113 @@
 };
 
 
-static int
-VINTparse (const unsigned char *buffer, const int start, const int end,
-           int64 * result, const int flag)
+/**
+ * FIXME: document 'flag', should 'temp'/'result' really be signed?
+ *
+ * @return 0 on error, otherwise number of bytes read from buffer
+ */
+static size_t
+VINTparse (const unsigned char *buffer, size_t start, size_t end,
+           int64_t * result, int flag)
 {
-  unsigned const char mask[8] = { 0x80, 0x40, 0x20, 0x10, 0x8, 0x4, 0x2, 0x1 };
-  unsigned const char imask[8] = { 0x7F, 0x3F, 0x1F, 0xF, 0x7, 0x3, 0x1, 00 };
-  int VINT_WIDTH;
-  int c;
-
-  VINT_WIDTH = 0;
-  int64 temp;
+  static const unsigned char mask[8] = { 0x80, 0x40, 0x20, 0x10, 0x8, 0x4, 
0x2, 0x1 };
+  static const unsigned char imask[8] = { 0x7F, 0x3F, 0x1F, 0xF, 0x7, 0x3, 
0x1, 00 };
+  int vint_width;
+  unsigned int c;
+  int64_t temp;
   unsigned char tempc;
-  *result = 0;
 
-  if (end - start < 2)
-  {
-    return 0;                   /*ops */
-  }
+  if (end - start < 2)  
+    return 0;                   /*ops */  
   
+  vint_width = 0;
   for (c = 0; c < 8; c++)
     if (!(buffer[start] & mask[c]))
-      VINT_WIDTH++;
+      vint_width++;
     else
       break;
-  if (VINT_WIDTH >= 8 || VINT_WIDTH + start + 1 >= end)
-  {
+  if ( (vint_width >= 8) || (vint_width + start + 1 >= end) )
     return 0;
-  }
   
-  for (c = 0; c < VINT_WIDTH; c++)
+  *result = 0;
+  for (c = 0; c < vint_width; c++)
   {
-    tempc = buffer[start + VINT_WIDTH - c];
+    tempc = buffer[start + vint_width - c];
     temp = tempc << (c * 8);
     *result += temp;
   }
-
   if (flag)
-    *result += (buffer[start] & imask[VINT_WIDTH]) << (VINT_WIDTH * 8);
+    *result += (buffer[start] & imask[vint_width]) << (vint_width * 8);
   else
-    *result += (buffer[start]) << (VINT_WIDTH * 8);
-
-  return VINT_WIDTH + 1;
+    *result += (buffer[start]) << (vint_width * 8);
+  return vint_width + 1;
 }
 
+
+/**
+ * FIXME: 'unsigned int' or 'uint32_t' for ID? 
+ * FIXME: document arguments, return value...
+ */
 static int
-elementRead (const char *buffer, const int start, const int end,
-             unsigned int *ID, int64 * size)
+elementRead (const unsigned char *buffer, size_t start, size_t end,
+             unsigned int *id, int64_t * size)
 {
-  int64 tempID;
-  int64 tempsize;
-  int ID_offset, size_offset;
+  int64_t tempID;
+  int64_t tempsize;
+  size_t id_offset;
+  size_t size_offset;
 
   tempID = 0;
   
-  ID_offset = VINTparse (buffer, start, end, &tempID, 0);
-  if (!ID_offset)
+  id_offset = VINTparse (buffer, start, end, &tempID, 0);
+  if (!id_offset)
     return 0;
-  size_offset = VINTparse (buffer, start + ID_offset, end, &tempsize, 1);
+  size_offset = VINTparse (buffer, start + id_offset, end, &tempsize, 1);
   if (!size_offset)
     return 0;
-  *ID = (int) tempID;           /*id must be <4 and must to feet in uint */
+  *id = (unsigned int) tempID;           /*id must be <4 and must fit in uint 
*/
   *size = tempsize;
-  return ID_offset + size_offset;
+  return id_offset + size_offset;
 }
 
-static int64
-getInt (const char *buffer, const int start, const int size)
+
+/**
+ * FIXME: signed or unsigned return value?
+ */
+static int64_t
+getInt (const unsigned char *buffer, size_t start, size_t size)
 {
-/*return a int [8-64], from buffer, Big Endian*/
-  int64 result = 0;
-  int c;
+  /*return a int [8-64], from buffer, Big Endian*/
+  int64_t result;
+  size_t c;
 
+  result = 0;
   for (c = 1; c <= size; c++)
-  {
-    result += buffer[start + c - 1] << (8 * (size - c));
-  }
+    result += buffer[start + c - 1] << (8 * (size - c));  
   return result;
 }
 
 static float
-getFloat (const char *buffer, const int start, const int size)
+getFloat (const char *buffer, size_t start, size_t size)
 {
-  float result = 0;
+  float result;
   char tmp[4];
 
+  if (size != sizeof (float))
+    return 0.0;
   if (size == sizeof (float))
   {
     tmp[0] = buffer[start + 3];
     tmp[1] = buffer[start + 2];
     tmp[2] = buffer[start + 1];
     tmp[3] = buffer[start];
-    result = *((float *) (tmp));
+    return *((float *) (tmp));
   }
   return result;
 }
 
-const unsigned int MKV_Parse_list[] = { /*Elements, containing requed 
information (sub-elements), see enum in mkv.h for values */
+static const unsigned int MKV_Parse_list[] = { 
+  /*Elements, containing requed information (sub-elements), see enum in mkv.h 
for values */
   MKVID_Segment,
   MKVID_Info,
   MKVID_Video,
@@ -178,26 +206,26 @@
   MKVID_Tracks
 };
 
-const char stream_type_letters[] = "?vat";      /*[0]-no, 
[1]-video,[2]-audio,[3]-text */
+static const char stream_type_letters[] = "?vat";      /*[0]-no, 
[1]-video,[2]-audio,[3]-text */
 
 
 /* video/mkv */
 int
-EXTRACTOR_mkv_extract (const unsigned char *data, size_t size1,
+EXTRACTOR_mkv_extract (const unsigned char *data, size_t size,
                        EXTRACTOR_MetaDataProcessor proc, void *proc_cls,
                        const char *options)
 {
   int ret;
-  char buffer[128],temp[128];
-  int p;                        /*pointer in buffer */
+  char buffer[128];
+  char temp[128];
+  size_t p;                        /*pointer in buffer */
   int c, c2;                    /*counter in some loops */
-
   unsigned int eID;             /*element ID */
-  int64 eSize;                  /*Size of element content */
+  int64_t eSize;                  /*Size of element content */
   int offs;
-  int64 timescale = 1000000;
-  float Duration = 0;
-  int64 DefaultDuration = 0;
+  int64_t timescale = 1000000;
+  float duration = -1.0;
+  int64_t DefaultDuration = 0;
   int TrackType = 0;
   int pvt_look_flag = 0;
   int curr_c = -1;
@@ -207,34 +235,20 @@
   int value_width = 0;
   int value_height = 0;
   int value = 0;
-  int size;
+  size_t size1;
+  const unsigned char *start;
 
-  ret = 0;
-  p = 0;
- 
-  if (size1 > 16777216)
-  {
-       size = 16777216;
-  }else{
-       size = size1;
-  }
-
-  if (!size)
-  {
+  if (size > 32 * 1024)
+    size1 = 32 * 1024;
+  else
+    size1 = size;
+  start = memchr (data, MKVID_FILE_BEGIN, size1);
+  if (NULL == start)
     return 0;
-  }
-
- 
-  while (data[p] != MKVID_FILE_BEGIN)
-  {
-    p++;
-    if (p >= size)
-    {
-      return 0;
-    }
-  };                            /*skip text while EBML begin */
-
+  p = start - data;
+  
 /*main loop*/
+  ret = 0;
   do
   {
     offs = elementRead (data, p, size, &eID, &eSize);
@@ -264,18 +278,32 @@
         v_c++;
         if (v_c > MAX_STREAMS)
           v_c = MAX_STREAMS;
-        sprintf (buffer, "%u(video)",
-                 (int) (Duration / 1e+9 * (float) timescale));
-        ADD (buffer, EXTRACTOR_METATYPE_DURATION);
+       if (duration != -1.0)
+       {
+         /* FIXME: duration might still be -1 here, defer until
+            end of function & check if duration is != -1 */
+         snprintf (buffer,
+                   sizeof(buffer),
+                   "%u s (video)",
+                   (int) (duration / 1e+9 * (float) timescale));
+         ADD (buffer, EXTRACTOR_METATYPE_DURATION);
+       }
         curr_c = v_c;
         break;
       case MKV_Track_audio:
         a_c++;
         if (a_c > MAX_STREAMS)
           a_c = MAX_STREAMS;
-        sprintf (buffer, "%u(audio)",
-                 (int) (Duration / 1e+9 * (float) timescale));
-        ADD (buffer, EXTRACTOR_METATYPE_DURATION);
+       if (duration != -1.0)
+       {
+         /* FIXME: duration might still be -1 here, defer until
+            end of function & check if duration is != -1 */
+         snprintf (buffer,
+                   sizeof (buffer),
+                   "%u s (audio)",
+                   (unsigned int) (duration / 1e+9 * (float) timescale));
+         ADD (buffer, EXTRACTOR_METATYPE_DURATION);
+       }
         curr_c = a_c;
         break;
       case MKV_Track_subtitle_orig:
@@ -293,7 +321,10 @@
         DefaultDuration = value;
         if (DefaultDuration > 100)
         {
-          sprintf (buffer, "fps: %u", 1000000000 / DefaultDuration);
+         /* FIXME: integrate with codec name/id into 'METATYPE_FORMAT' */
+          snprintf (buffer, 
+                   sizeof (buffer),
+                   "fps: %u", 1000000000 / DefaultDuration);
           ADD (buffer, EXTRACTOR_METATYPE_UNKNOWN);
         }
       }
@@ -301,9 +332,12 @@
     case MKVID_Language:       /*stream language */
       if (curr_c >= 0 && TrackType < 4 && eSize < MAX_STRING_SIZE)
          {
-                 strncpy(temp,data+p,eSize);
-                 temp[eSize] = '\0';
-                 sprintf (buffer, "%s", temp);
+                 /* FIXME: why first temp, then buffer? */
+                 snprintf (buffer, 
+                           sizeof (buffer),
+                           "%.*s",
+                           (int) eSize,
+                           data + p);
                  ADD (buffer, EXTRACTOR_METATYPE_LANGUAGE);
          }
       break;
@@ -311,12 +345,15 @@
     case MKVID_CodecID:        /*codec detection (if V_MS/VFW/FOURCC - set a 
fourcc code, else fill a vcodecs value) */
       if (curr_c >= 0 && TrackType < 4 && eSize < MAX_STRING_SIZE)
       {
-                 strncpy(temp,data+p,eSize);
-                 temp[eSize] = '\0';
-        if (!strcmp (temp, "V_MS/VFW/FOURCC"))
+       if (!strncmp (data + p, "V_MS/VFW/FOURCC", eSize))
           pvt_look_flag = 1;
-        sprintf (buffer, "codec: %s", temp);
-        ADD (buffer, EXTRACTOR_METATYPE_UNKNOWN);
+        snprintf (buffer,
+                 sizeof (buffer),
+                 "codec: %.*s", 
+                 (int) eSize,
+                 data + p);
+        ADD (buffer, 
+            EXTRACTOR_METATYPE_FORMAT);
       }
       break;
     case MKVID_CodecPrivate:
@@ -344,25 +381,36 @@
       break;
     case MKVID_TimeCodeScale:
       timescale = getInt (data, p, eSize);
-      sprintf (buffer, "TimeScale: %u", timescale);
+      snprintf (buffer, 
+               sizeof (buffer),
+               "TimeScale: %u", timescale);
       ADD (buffer, EXTRACTOR_METATYPE_UNKNOWN);
       break;
     case MKVID_Duration:
-      Duration = getFloat (data, p, eSize);
-      sprintf (buffer, "duration: %u s", Duration);
+      duration = getFloat (data, p, eSize);
+      snprintf (buffer, 
+               sizeof (buffer),
+               "%f ms", duration);
       ADD (buffer, EXTRACTOR_METATYPE_DURATION);
       break;
     case MKVID_Channels:
-      sprintf (buffer, "channels: %u", value);
+      snprintf (buffer,
+               sizeof (buffer),
+               "channels: %u", value);
       ADD (buffer, EXTRACTOR_METATYPE_UNKNOWN);
       break;
     case MKVID_BitDepth:
-      sprintf (buffer, "BitDepth: %u", value);
+      snprintf (buffer,
+               sizeof (buffer),
+               "BitDepth: %u", value);
       ADD (buffer, EXTRACTOR_METATYPE_UNKNOWN);
       break;
     case MKVID_OutputSamplingFrequency:        /*pasthough */
     case MKVID_SamplingFrequency:
-      sprintf (buffer, "Sampling Frequency: %u", value);
+      /* FIXME: the resulting value seems wrong... Unit? */
+      snprintf (buffer,
+               sizeof (buffer),
+               "Sampling Frequency: %u", value);
       ADD (buffer, EXTRACTOR_METATYPE_UNKNOWN);
       break;
       break;
@@ -378,10 +426,13 @@
     p += eSize;                 /*skip unknown or uninteresting */
   }
   while (1);
+  if ( (value_width != 0) && (value_height != 0) )    
+    snprintf (buffer,
+             sizeof(buffer),
+             "%ux%u", 
+             value_width, value_height);
+  ADD (buffer, EXTRACTOR_METATYPE_IMAGE_DIMENSIONS);
   
-  sprintf (buffer, "Image dimensions: %u X %u", value_width, value_height);
-  ADD (buffer, EXTRACTOR_METATYPE_UNKNOWN);
-  
 EXIT:
 
   return ret;