Merge pull request #5508 from Snuffleupagus/jpeg-stream-find-soi

Refactor searching for the SOI marker of inline JPEG image streams
This commit is contained in:
Yury Delendik 2014-12-15 09:36:24 -06:00
commit bc27774cfd
2 changed files with 9 additions and 18 deletions

View File

@ -372,22 +372,6 @@ var Parser = (function ParserClosure() {
return new LZWStream(stream, maybeLength, earlyChange);
}
if (name === 'DCTDecode' || name === 'DCT') {
// According to the specification: for inline images, the ID operator
// shall be followed by a single whitespace character (unless it uses
// ASCII85Decode or ASCIIHexDecode filters).
// In practice this only seems to be followed for inline JPEG images,
// and generally ignoring the first byte of the stream if it is a
// whitespace char can even *cause* issues (e.g. in the CCITTFaxDecode
// filters used in issue2984.pdf).
// Hence when the first byte of the stream of an inline JPEG image is
// a whitespace character, we thus simply skip over it.
if (isCmd(this.buf1, 'ID')) {
var firstByte = stream.peekByte();
if (firstByte === 0x0A /* LF */ || firstByte === 0x0D /* CR */ ||
firstByte === 0x20 /* SPACE */) {
stream.skip();
}
}
xrefStreamStats[StreamType.DCT] = true;
return new JpegStream(stream, maybeLength, stream.dict, this.xref);
}

View File

@ -857,8 +857,15 @@ var PredictorStream = (function PredictorStreamClosure() {
*/
var JpegStream = (function JpegStreamClosure() {
function JpegStream(stream, maybeLength, dict, xref) {
// TODO: per poppler, some images may have 'junk' before that
// need to be removed
// Some images may contain 'junk' before the SOI (start-of-image) marker.
// Note: this seems to mainly affect inline images.
var ch;
while ((ch = stream.getByte()) !== -1) {
if (ch === 0xFF) { // Find the first byte of the SOI marker (0xFFD8).
stream.skip(-1); // Reset the stream position to the SOI.
break;
}
}
this.stream = stream;
this.maybeLength = maybeLength;
this.dict = dict;