Merge 0e0c84a into 34dc93c

martin-majlis · Mar 24, 2018 · 7750efd · 7750efd
2 parents 34dc93c + 0e0c84a
commit 7750efd
Show file tree

Hide file tree

Showing 3 changed files with 49 additions and 1 deletion.
diff --git a/tests/extract_html_format_test.py b/tests/extract_html_format_test.py
@@ -108,3 +108,19 @@ def test_text(self):
                 "<p>Text for section 5.1\n\n\n</p>"
             )
         )
+
+    def test_with_erroneous_edit(self):
+        page = self.wiki.page('Test_Edit')
+        self.maxDiff = None
+        section = page.section_by_title('Section with Edit')
+        self.assertEqual(section.title, 'Section with Edit')
+        self.assertEqual(
+            page.text,
+            (
+                "<p><b>Summary</b> text\n\n</p>\n\n" +
+                "<h2>Section 1</h2>\n" +
+                "<p>Text for section 1</p>\n\n"
+                "<h3>Section with Edit</h3>\n" +
+                "<p>Text for section with edit\n\n\n</p>"
+            )
+        )
diff --git a/tests/mock_data.py b/tests/mock_data.py
@@ -108,6 +108,36 @@ def wikipedia_api_request(page, params):
             }
         }
     },
+    'en:action=query&prop=extracts&titles=Test_Edit&': {
+        "batchcomplete": "",
+        "warnings": {
+            "extracts": {
+                "*": "\"exlimit\" was too large for a whole article extracts request, lowered to 1."
+            }
+        },
+        "query": {
+            "normalized": [
+                {
+                    "from": "Test_Edit",
+                    "to": "Test Edit"
+                }
+            ],
+            "pages": {
+                "4": {
+                    "pageid": 4,
+                    "ns": 0,
+                    "title": "Test Edit",
+                    "extract": (
+                        "<p><b>Summary</b> text\n\n</p>\n" +
+                        "<h2>Section 1</h2>\n" +
+                        "<p>Text for section 1</p>\n\n\n" +
+                        "<h3><span id=\"s1.Edit\">Section with Edit</span><span>Edit</span></h3>\n" +
+                        "<p>Text for section with edit\n\n\n</p>"
+                    )
+                }
+            }
+        }
+    },
     'en:action=query&inprop=protection|talkid|watched|watchers|visitingwatchers|notificationtimestamp|subjectid|url|readable|preload|displaytitle&prop=info&titles=Test_1&': {
         "batchcomplete": "",
         "query": {

diff --git a/wikipediaapi/wikipedia.py b/wikipediaapi/wikipedia.py
@@ -65,7 +65,9 @@ class Namespace(object):
     ExtractFormat.HTML: re.compile(
         r'\n? *<h(\d)[^>]*?>(<span[^>]*><\/span>)? *' +
         '(<span[^>]*>)? *(<span[^>]*><\/span>)? *(.*?) *' +
-        '(<\/span>)?<\/h\d>\n?'
+        '(<\/span>)?(<span>Edit<\/span>)?<\/h\d>\n?'
+        #                  ^^^^
+        # Example page with 'Edit' erroneous links: https://bit.ly/2ui4FWs
     ),
     # ExtractFormat.PLAIN.value: re.compile(r'\n\n *(===*) (.*?) (===*) *\n'),
 }