Merge pull request #1413 from saebekassebil/metadata

Metadata Parsing - Setting proper document title

Makefile

@@ -39,6 +39,7 @@ PDF_JS_FILES = \
   ../external/jpgjs/jpg.js \
   jpx.js \
   bidi.js \
+  metadata.js \
 	$(NULL)
 
 # make server

make.js

@@ -97,7 +97,8 @@ target.bundle = function() {
         'worker.js',
         '../external/jpgjs/jpg.js',
         'jpx.js',
-        'bidi.js'];
+        'bidi.js',
+        'metadata.js'];
 
   if (!exists(BUILD_DIR))
     mkdir(BUILD_DIR);

src/core.js

@@ -587,14 +587,6 @@ var PDFDocModel = (function PDFDocModelClosure() {
                           this.mainXRefEntriesOffset);
       this.xref = xref;
       this.catalog = new Catalog(xref);
-      if (xref.trailer && xref.trailer.has('ID')) {
-        var fileID = '';
-        var id = xref.fetchIfRef(xref.trailer.get('ID'))[0];
-        id.split('').forEach(function(el) {
-          fileID += Number(el.charCodeAt(0)).toString(16);
-        });
-        this.fileID = fileID;
-      }
     },
     get numPages() {
       var linearization = this.linearization;
@@ -602,21 +594,33 @@ var PDFDocModel = (function PDFDocModelClosure() {
       // shadow the prototype getter
       return shadow(this, 'numPages', num);
     },
+    getDocumentInfo: function pdfDocGetDocumentInfo() {
+      var info;
+      if (this.xref.trailer.has('Info'))
+        info = this.xref.fetch(this.xref.trailer.get('Info'));
+
+      return shadow(this, 'getDocumentInfo', info);
+    },
     getFingerprint: function pdfDocGetFingerprint() {
-      if (this.fileID) {
-        return this.fileID;
+      var xref = this.xref, fileID;
+      if (xref.trailer.has('ID')) {
+        fileID = '';
+        var id = xref.fetchIfRef(xref.trailer.get('ID'))[0];
+        id.split('').forEach(function(el) {
+          fileID += Number(el.charCodeAt(0)).toString(16);
+        });
       } else {
         // If we got no fileID, then we generate one,
         // from the first 100 bytes of PDF
         var data = this.stream.bytes.subarray(0, 100);
         var hash = calculateMD5(data, 0, data.length);
-        var strHash = '';
+        fileID = '';
         for (var i = 0, length = hash.length; i < length; i++) {
-          strHash += Number(hash[i]).toString(16);
+          fileID += Number(hash[i]).toString(16);
         }
-
-        return strHash;
       }
+
+      return shadow(this, 'getFingerprint', fileID);
     },
     getPage: function pdfDocGetPage(n) {
       return this.catalog.getPage(n);
@@ -645,6 +649,7 @@ var PDFDoc = (function PDFDocClosure() {
     this.stream = stream;
     this.pdfModel = new PDFDocModel(stream);
     this.fingerprint = this.pdfModel.getFingerprint();
+    this.info = this.pdfModel.getDocumentInfo();
     this.catalog = this.pdfModel.catalog;
     this.objs = new PDFObjects();
 

src/metadata.js

@@ -0,0 +1,66 @@
+/* -*- Mode: Java; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
+/* vim: set shiftwidth=2 tabstop=2 autoindent cindent expandtab: */
+
+'use strict';
+
+var Metadata = PDFJS.Metadata = (function MetadataClosure() {
+  function Metadata(meta) {
+    if (typeof meta === 'string') {
+      var parser = new DOMParser();
+      meta = parser.parseFromString(meta, 'application/xml');
+    } else if (!(meta instanceof Document)) {
+      error('Metadata: Invalid metadata object');
+    }
+
+    this.metaDocument = meta;
+    this.metadata = {};
+    this.parse();
+  }
+
+  Metadata.prototype = {
+    parse: function() {
+      var doc = this.metaDocument;
+      var rdf = doc.documentElement;
+
+      if (rdf.nodeName.toLowerCase() !== 'rdf:rdf') { // Wrapped in <xmpmeta>
+        rdf = rdf.firstChild;
+        while (rdf && rdf.nodeName.toLowerCase() !== 'rdf:rdf')
+          rdf = rdf.nextSibling;
+      }
+
+      var nodeName = (rdf) ? rdf.nodeName.toLowerCase() : null;
+      if (!rdf || nodeName !== 'rdf:rdf' || !rdf.hasChildNodes())
+        return;
+
+      var childNodes = rdf.childNodes, desc, namespace, entries, entry;
+
+      for (var i = 0, length = childNodes.length; i < length; i++) {
+        desc = childNodes[i];
+        if (desc.nodeName.toLowerCase() !== 'rdf:description')
+          continue;
+
+        entries = [];
+        for (var ii = 0, iLength = desc.childNodes.length; ii < iLength; ii++) {
+          if (desc.childNodes[ii].nodeName.toLowerCase() !== '#text')
+            entries.push(desc.childNodes[ii]);
+        }
+
+        for (ii = 0, iLength = entries.length; ii < iLength; ii++) {
+          var entry = entries[ii];
+          var name = entry.nodeName.toLowerCase();
+          this.metadata[name] = entry.textContent.trim();
+        }
+      }
+    },
+
+    get: function(name) {
+      return this.metadata[name] || null;
+    },
+
+    has: function(name) {
+      return typeof this.metadata[name] !== 'undefined';
+    }
+  };
+
+  return Metadata;
+})();

src/obj.js

@@ -111,6 +111,22 @@ var Catalog = (function CatalogClosure() {
   }
 
   Catalog.prototype = {
+    get metadata() {
+      var ref = this.catDict.get('Metadata');
+      var stream = this.xref.fetchIfRef(ref);
+      var metadata;
+      if (stream && isDict(stream.dict)) {
+        var type = stream.dict.get('Type');
+        var subtype = stream.dict.get('Subtype');
+
+        if (isName(type) && isName(subtype) &&
+            type.name === 'Metadata' && subtype.name === 'XML') {
+          metadata = stringToPDFString(bytesToString(stream.getBytes()));
+        }
+      }
+
+      return shadow(this, 'metadata', metadata);
+    },
     get toplevelPagesDict() {
       var pagesObj = this.catDict.get('Pages');
       assertWellFormed(isRef(pagesObj), 'invalid top-level pages reference');

web/viewer.html

@@ -11,6 +11,7 @@
         <!-- PDFJSSCRIPT_INCLUDE_BUILD -->
         <script type="text/javascript" src="../src/core.js"></script> <!-- PDFJSSCRIPT_REMOVE_CORE -->
         <script type="text/javascript" src="../src/util.js"></script>  <!-- PDFJSSCRIPT_REMOVE_CORE -->
+        <script type="text/javascript" src="../src/metadata.js"></script>  <!-- PDFJSSCRIPT_REMOVE_CORE -->
         <script type="text/javascript" src="../src/canvas.js"></script>  <!-- PDFJSSCRIPT_REMOVE_CORE -->
         <script type="text/javascript" src="../src/obj.js"></script>  <!-- PDFJSSCRIPT_REMOVE_CORE -->
         <script type="text/javascript" src="../src/function.js"></script> <!-- PDFJSSCRIPT_REMOVE_CORE -->

web/viewer.js

@@ -499,6 +499,24 @@ var PDFView = {
       // Setting the default one.
       this.parseScale(kDefaultScale, true);
     }
+
+    this.metadata = null;
+    var metadata = pdf.catalog.metadata;
+    var info = this.documentInfo = pdf.info;
+    var pdfTitle;
+
+    if (metadata) {
+      this.metadata = metadata = new PDFJS.Metadata(metadata);
+
+      if (metadata.has('dc:title'))
+        pdfTitle = metadata.get('dc:title');
+    }
+
+    if (!pdfTitle && info && info.has('Title'))
+      pdfTitle = info.get('Title');
+
+    if (pdfTitle)
+      document.title = pdfTitle;
   },
 
   setHash: function pdfViewSetHash(hash) {