Prevent an infinite loop in `XRef.readXRef` by keeping track of already parsed tables (bug 1393476)

With this patch, not only is the infinite loop prevented, but we're also able to actually render the file (which e.g. Adobe Reader isn't able to).

Fixes https://bugzilla.mozilla.org/show_bug.cgi?id=1393476.

src/core/obj.js

@@ -1186,11 +1186,22 @@ var XRef = (function XRefClosure() {
 
     readXRef: function XRef_readXRef(recoveryMode) {
       var stream = this.stream;
+      // Keep track of already parsed XRef tables, to prevent an infinite loop
+      // when parsing corrupt PDF files where e.g. the /Prev entries create a
+      // circular dependency between tables (fixes bug1393476.pdf).
+      let startXRefParsedCache = Object.create(null);
 
       try {
         while (this.startXRefQueue.length) {
           var startXRef = this.startXRefQueue[0];
 
+          if (startXRefParsedCache[startXRef]) {
+            warn('readXRef - skipping XRef table since it was already parsed.');
+            this.startXRefQueue.shift();
+            continue;
+          }
+          startXRefParsedCache[startXRef] = true;
+
           stream.pos = startXRef + stream.start;
 
           var parser = new Parser(new Lexer(stream), true, this);

test/pdfs/bug1393476.pdf.link

@@ -0,0 +1 @@
+https://bugzilla.mozilla.org/attachment.cgi?id=8900754

test/test_manifest.json

@@ -820,6 +820,13 @@
        "link": false,
        "type": "eq"
     },
+    {  "id": "bug1393476",
+       "file": "pdfs/bug1393476.pdf",
+       "md5": "163ee8727c77f27ee651eec777bb20a9",
+       "rounds": 1,
+       "link": true,
+       "type": "eq"
+    },
     {  "id": "bug1252420",
        "file": "pdfs/bug1252420.pdf",
        "md5": "f21c911b9b655972b06ef782a1fa6a17",