Merge pull request #270 from anteverse/feature/more-dates-in-extractor

Add 4 date metadata patterns in date extraction

newsplease/pipeline/extractor/extractors/date_extractor.py

@@ -124,11 +124,17 @@ def _extract_from_meta(self, html):
                 date = meta['content'].strip()
                 break
 
+            # <meta itemprop="dc.date" content="2015-11-26T11:53:00.000Z" />
+            if 'dc.date' == item_prop:
+                date = meta['content'].strip()
+                break
+
             # <meta property="article:published_time"  content="2015-11-25" />
             if 'article:published_time' == meta_property:
                 date = meta['content'].strip()
                 break
-                # <meta name="Date" content="2015-11-26" />
+
+            # <meta name="Date" content="2015-11-26" />
             if 'date' == meta_name:
                 date = meta['content'].strip()
                 break
@@ -182,6 +188,21 @@ def _extract_from_meta(self, html):
                 date = meta['content'].strip()
                 break
 
+            # <meta itemprop="dcterms.date" content="2015-11-26T11:53:00.000Z" />
+            if 'dcterms.date' == item_prop:
+                date = meta['content'].strip()
+                break
+
+            # <meta itemprop="dcterms.created" content="2015-11-26T11:53:00.000Z" />
+            if 'dcterms.created' == item_prop:
+                date = meta['content'].strip()
+                break
+
+            # <meta itemprop="dcterms.created" content="2015-11-26T11:53:00.000Z" />
+            if 'og:published_time datetime' == item_prop:
+                date = meta['content'].strip()
+                break
+
             # <meta property="og:image" content="http://www.dailytimes.com.pk/digital
             # _images/400/2015-11-26/norway-return-number-of-asylum-seekers-to-pakistan-1448538771-7363.jpg"/>
             if 'og:image' == meta_property or "image" == item_prop: