for implied name: replace img with alt but not with src. update tests

Author: Kartik Prabhu

mf2py/dom_helpers.py

@@ -69,7 +69,7 @@ def get_descendents(node):
         if isinstance(desc, bs4.Tag):
             yield desc
 
-def get_textContent(el, replace_img=False, base_url=''):
+def get_textContent(el, replace_img=False, img_to_src=True, base_url=''):
     """ Get the text content of an element, replacing images by alt or src
     """
 
@@ -78,7 +78,7 @@ def get_textContent(el, replace_img=False, base_url=''):
     P_BREAK_BEFORE = 1
     P_BREAK_AFTER = 0
 
-    def text_collection(el, replace_img=False, base_url=''):
+    def text_collection(el, replace_img=False, img_to_src=True, base_url=''):
         # returns array of strings or integers
 
         items = []
@@ -103,7 +103,7 @@ def text_collection(el, replace_img=False, base_url=''):
 
         elif el.name == 'img' and replace_img:
             value = el.get('alt')
-            if value is None:
+            if value is None and img_to_src:
                 value = el.get('src')
                 if value is not None:
                     value = urljoin(base_url, value)
@@ -117,7 +117,7 @@ def text_collection(el, replace_img=False, base_url=''):
         else:
             for child in el.children:
 
-                child_items = text_collection(child, replace_img, base_url)
+                child_items = text_collection(child, replace_img, img_to_src, base_url)
                 items.extend(child_items)
 
             if el.name == 'p':
@@ -127,7 +127,7 @@ def text_collection(el, replace_img=False, base_url=''):
 
         return items
 
-    results = [t for t in text_collection(el, replace_img, base_url) if t is not '']
+    results = [t for t in text_collection(el, replace_img, img_to_src, base_url) if t is not '']
 
     if results:
         # remove <space> if it is first and last or if it is preceded by a <space> or <int> or followed by a <int>

mf2py/implied_properties.py

@@ -70,8 +70,9 @@ def non_empty(val):
             return text_type(prop_value)
 
     # use text if all else fails
-    # don't replace images in implied name (https://github.com/microformats/microformats2-parsing/issues/35)
-    return get_textContent(el, base_url=base_url)
+    # replace images with alt but not with src in implied name
+    # proposal: https://github.com/microformats/microformats2-parsing/issues/35#issuecomment-393615508
+    return get_textContent(el, replace_img=True, img_to_src=False, base_url=base_url)
 
 
 def photo(el, dict_class, img_with_alt, base_url=''):

test/examples/implied_properties/implied_properties.html

@@ -11,6 +11,8 @@
 
 	<a class="h-card" href="http://tommorris.org/"><img src="http://tommorris.org/photo.png" alt="" />Tom Morris</a>
 
+	<a class="h-card" href="http://tommorris.org/"><img src="http://tommorris.org/photo.png"/>Tom Morris</a>
+
 	<a class="h-card" href="http://tommorris.org/"><img src="http://tommorris.org/photo.png" alt="Tom Morris" /></a>
 
 	<img class="h-card" src="http://tommorris.org/photo.png" alt="Tom Morris" />

test/test_parser.py

@@ -431,7 +431,7 @@ def test_nested_values():
 def test_implied_name():
     result = parse_fixture("implied_properties/implied_properties.html")
 
-    for i in range(6):
+    for i in range(7):
         assert_equal(result["items"][i]["properties"]["name"][0], "Tom Morris")
 
 
@@ -474,8 +474,12 @@ def test_implied_nested_photo():
     result = parse_fixture("implied_properties/implied_properties.html", url="http://bar.org")
     assert_equal(result["items"][2]["properties"]["photo"][0],
                  "http://tommorris.org/photo.png")
+    assert_equal(result["items"][3]["properties"]["photo"][0],
+                 "http://tommorris.org/photo.png")
+    assert_equal(result["items"][4]["properties"]["photo"][0],
+                 "http://tommorris.org/photo.png")
     # src="" is relative to the base url
-    assert_equal(result["items"][5]["properties"]["photo"][0],
+    assert_equal(result["items"][6]["properties"]["photo"][0],
                  "http://bar.org")