Efficient lookup for an ID3v2 tag in MPEG files with garbage.

This looks for an ID3v2 tag until reaching the first valid MPEG frame in O(n) time.

diff --git a/taglib/mpeg/mpegfile.cpp b/taglib/mpeg/mpegfile.cpp
index af7253fa573ac50ef79aaa7ded36a012a9445e20..c634eeb87e9c1c6dbcfb3f4e2c390b7ac145d9df 100644 (file)
--- a/taglib/mpeg/mpegfile.cpp
+++ b/taglib/mpeg/mpegfile.cpp
@@ -488,28 +488,41 @@ long MPEG::File::findID3v2()
   const ByteVector headerID = ID3v2::Header::fileIdentifier();
 
   seek(0);
+  if(readBlock(headerID.size()) == headerID)
+    return 0;
 
-  const ByteVector data = readBlock(headerID.size());
-  if(data.size() < headerID.size())
+  Header firstHeader(this, 0, true);
+  if(firstHeader.isValid())
     return -1;
 
-  if(data == headerID)
-    return 0;
+  // Look for an ID3v2 tag until reaching the first valid MPEG frame.
 
-  if(firstSyncByte(data[0]) && secondSynchByte(data[1]))
-    return -1;
+  char frameSyncBytes[2] = {};
+  char tagHeaderBytes[4] = {};
+  long position = 0;
 
-  // Look for the entire file, if neither an MEPG frame or ID3v2 tag was found
-  // at the beginning of the file.
-  // We don't care about the inefficiency of the code, since this is a seldom case.
+  while(true) {
+    seek(position);
+    const ByteVector buffer = readBlock(bufferSize());
+    if(buffer.isEmpty())
+      return -1;
 
-  const long tagOffset = find(headerID);
-  if(tagOffset < 0)
-    return -1;
+    for(unsigned int i = 0; i < buffer.size(); ++i) {
+      frameSyncBytes[0] = frameSyncBytes[1];
+      frameSyncBytes[1] = buffer[i];
+      if(firstSyncByte(frameSyncBytes[0]) && secondSynchByte(frameSyncBytes[1])) {
+        Header header(this, position + i - 1, true);
+        if(header.isValid())
+          return -1;
+      }
 
-  const long frameOffset = firstFrameOffset();
-  if(frameOffset < tagOffset)
-    return -1;
+      tagHeaderBytes[0] = tagHeaderBytes[1];
+      tagHeaderBytes[1] = tagHeaderBytes[2];
+      tagHeaderBytes[2] = buffer[i];
+      if(headerID == tagHeaderBytes)
+        return position + i - 2;
+    }
 
-  return tagOffset;
+    position += bufferSize();
+  }
 }