Extract srcset values for use in crawling.

brozzler/browser.py

@@ -470,6 +470,7 @@ class Browser:
                     outlinks = []
                 else:
                     outlinks = self.extract_outlinks()
+                    outlinks = outlinks.union(self.extract_tertiary_assets())
                 if not skip_visit_hashtags:
                     self.visit_hashtags(self.url(), hashtags, outlinks)
                 final_page_url = self.url()
@@ -565,6 +566,30 @@ class Browser:
                     'problem extracting outlinks, result message: %s', message)
             return frozenset()
 
+    def extract_tertiary_assets(self, timeout=60):
+        self.logger.info('extracting tertiary assets')
+        self.websock_thread.expect_result(self._command_id.peek())
+        js = brozzler.jinja2_environment().get_template(
+                'extract-tertiary-assets.js').render()
+        msg_id = self.send_to_chrome(
+                method='Runtime.evaluate', params={'expression': js})
+        self._wait_for(
+                lambda: self.websock_thread.received_result(msg_id),
+                timeout=timeout)
+        message = self.websock_thread.pop_result(msg_id)
+        if ('result' in message and 'result' in message['result']
+                and 'value' in message['result']['result']):
+            if message['result']['result']['value']:
+                return frozenset(
+                        message['result']['result']['value'].split('\n'))
+            else:
+                # no links found
+                return frozenset()
+        else:
+            self.logger.error(
+                    'problem extracting tertiary assets, result message: %s', message)
+            return frozenset()
+
     def screenshot(self, timeout=45):
         self.logger.info('taking screenshot')
         self.websock_thread.expect_result(self._command_id.peek())

brozzler/js-templates/extract-tertiary-assets.js

@@ -0,0 +1,36 @@
+// we have problems if the page has changed the definition of Set or Array
+// http://www.polyvore.com/ does this for example
+var __brzl_framesDone = new Set();
+var __brzl_compileAssets = function(frame) {
+    __brzl_framesDone.add(frame);
+    if (frame && frame.document) {
+        var elem = frame.document.querySelectorAll('[srcset]');
+        var srcset_list = new Array();
+        var base = frame.document.baseURI.substring(0,
+            frame.document.baseURI.lastIndexOf("/"));
+
+        for (var i = 0; i < elem.length; i++) {
+            var srcs = elem[i].srcset.match(/(?:[^\s]+\/[^\s]+)/g);
+
+            for (var i = 0; i < srcs.length; i++) {
+                if ( /https?:/.test(srcs[i]) ) {
+                    srcset_list = srcset_list.concat(srcs[i]);
+                } else {
+                    srcset_list = srcset_list.concat(base + srcs[i]);
+                }
+            }
+        }
+
+        var assets = Array.prototype.slice.call(srcset_list);
+
+        for (var i = 0; i < frame.frames.length; i++) {
+            if (frame.frames[i] && !__brzl_framesDone.has(frame.frames[i])) {
+                assets = assets.concat(
+                            __brzl_compileAssets(frame.frames[i]));
+            }
+        }
+    }
+
+    return assets;
+}
+__brzl_compileAssets(window).join('\n');