Created attachment 101828 [details] PDF test case This was reported in GNOME's Bugzilla: https://bugzilla.gnome.org/show_bug.cgi?id=732271 Poppler gets in a infinite loop when rendering the document attached. After interrupting poppler-glib-demo, this is the backtrace: Program received signal SIGINT, Interrupt. 0xb726c4ba in TextBlock::isBeforeByRule1 (this=0x940d698, blk1=0x9081220) at TextOutputDev.cc:1738 1738 (this->ExMin <= blk1->ExMax)); Thread 1 (Thread 0xb55f2ac0 (LWP 31946)): #0 0xb726c4ba in TextBlock::isBeforeByRule1 (this=0x940d698, blk1=0x9081220) at TextOutputDev.cc:1738 #1 0xb726c7c3 in TextBlock::visitDepthFirst (this=0x940d698, blkList=0x847daf0, pos1=2898, sorted=0x99a0f80, sortPos=1494, visited=0x99a7fb0) at TextOutputDev.cc:1856 #2 0xb726c6f4 in TextBlock::visitDepthFirst (this=0x9406cf8, blkList=0x847daf0, pos1=2878, sorted=0x99a0f80, sortPos=1494, visited=0x99a7fb0) at TextOutputDev.cc:1874 #3 0xb726c6f4 in TextBlock::visitDepthFirst (this=0x93c5af8, blkList=0x847daf0, pos1=2681, sorted=0x99a0f80, sortPos=1494, visited=0x99a7fb0) at TextOutputDev.cc:1874 #4 0xb726c6f4 in TextBlock::visitDepthFirst (this=0x93c0678, blkList=0x847daf0, pos1=2665, sorted=0x99a0f80, sortPos=1347, visited=0x99a7fb0) at TextOutputDev.cc:1874 #5 0xb726c6f4 in TextBlock::visitDepthFirst (this=0x93bcc60, blkList=0x847daf0, pos1=2654, sorted=0x99a0f80, sortPos=1347, visited=0x99a7fb0) at TextOutputDev.cc:1874 #6 0xb726c6f4 in TextBlock::visitDepthFirst (this=0x93b7298, blkList=0x847daf0, pos1=2637, sorted=0x99a0f80, sortPos=1347, visited=0x99a7fb0) at TextOutputDev.cc:1874 Acroread works fine and renders the document fast.
This is with poppler from master (and reported as 0.26.1) with cairo backend.
This looks pretty much a duplicate of an existing bug. Also looks pretty much not cairo specific. I'm busy enough to do the work though. Anyone that has time look for something related to very slow extraction of text. I doubt it's an infinite loop, just extremely slow.
This is the same as bug #77087. pdftotext takes almost 8 minutes to complete, and that's mostly spent in TextBlock::visitDepthFirst. Evince will take longer, because it seems to sort the text twice before showing anything. *** This bug has been marked as a duplicate of bug 77087 ***
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.