Merge pull request #913 from notmasteryet/issue-863

Bad pdf with scanned image

src/core.js

@@ -392,12 +392,17 @@ var PDFDocModel = (function pdfDoc() {
         if (find(stream, 'endobj', 1024))
           startXRef = stream.pos + 6;
       } else {
-        // Find startxref at the end of the file.
-        var start = stream.end - 1024;
-        if (start < 0)
-          start = 0;
-        stream.pos = start;
-        if (find(stream, 'startxref', 1024, true)) {
+        // Find startxref by jumping backward from the end of the file.
+        var step = 1024;
+        var found = false, pos = stream.end;
+        while (!found && pos > 0) {
+          pos -= step - 'startxref'.length;
+          if (pos < 0)
+            pos = 0;
+          stream.pos = pos;
+          found = find(stream, 'startxref', step, true);
+        }
+        if (found) {
           stream.skip(9);
           var ch;
           do {

test/pdfs/.gitignore

@@ -17,4 +17,5 @@
 !devicen.pdf
 !cmykjpeg.pdf
 !issue840.pdf
+!scan-bad.pdf
 !freeculture.pdf

test/test.py

@@ -12,6 +12,7 @@ DOC_ROOT = os.path.abspath(os.path.join(os.path.dirname(__file__),".."))
 ANAL = True
 DEFAULT_MANIFEST_FILE = 'test_manifest.json'
 EQLOG_FILE = 'eq.log'
+BROWSERLOG_FILE = 'browser.log'
 REFDIR = 'ref'
 TMPDIR = 'tmp'
 VERBOSE = False
@@ -229,6 +230,7 @@ class BaseBrowserCommand(object):
     def setup(self):
         self.tempDir = tempfile.mkdtemp()
         self.profileDir = os.path.join(self.tempDir, "profile")
+        self.browserLog = open(BROWSERLOG_FILE, "w")
 
     def teardown(self):
         # If the browser is still running, wait up to ten seconds for it to quit
@@ -245,6 +247,8 @@ class BaseBrowserCommand(object):
         if self.tempDir is not None and os.path.exists(self.tempDir):
             shutil.rmtree(self.tempDir)
 
+        self.browserLog.close()
+
     def start(self, url):
         raise Exception("Can't start BaseBrowserCommand")
 
@@ -262,7 +266,7 @@ class FirefoxBrowserCommand(BaseBrowserCommand):
         if platform.system() == "Darwin":
             cmds.append("-foreground")
         cmds.extend(["-no-remote", "-profile", self.profileDir, url])
-        self.process = subprocess.Popen(cmds)
+        self.process = subprocess.Popen(cmds, stdout = self.browserLog, stderr = self.browserLog)
 
 class ChromeBrowserCommand(BaseBrowserCommand):
     def _fixupMacPath(self):
@@ -272,7 +276,7 @@ class ChromeBrowserCommand(BaseBrowserCommand):
         cmds = [self.path]
         cmds.extend(["--user-data-dir=%s" % self.profileDir,
                      "--no-first-run", "--disable-sync", url])
-        self.process = subprocess.Popen(cmds)
+        self.process = subprocess.Popen(cmds, stdout = self.browserLog, stderr = self.browserLog)
 
 def makeBrowserCommand(browser):
     path = browser["path"].lower()

test/test_manifest.json

@@ -227,6 +227,12 @@
        "rounds": 1,
        "type": "load"
     },
+    {  "id": "scan-bad",
+       "file": "pdfs/scan-bad.pdf",
+       "md5": "4cf988f01ab83f61aca57f406dfd6584",
+       "rounds": 1,
+       "type": "load"
+    },
     {  "id": "ibwa-bad",
        "file": "pdfs/ibwa-bad.pdf",
        "md5": "6ca059d32b74ac2688ae06f727fee755",