Added scored nodes into candidates

11 years ago · df5cb8c8f6
parent f858f0dbb0
commit df5cb8c8f6
3 changed files with 23 additions and 25 deletions
--- a/readability/scoring.py
+++ b/readability/scoring.py
@ -192,6 +192,10 @@ def score_candidates(nodes):
        logger.debug("Giving parent bonus points: %f", candidates[parent].content_score)
        # the grand node gets half
        candidates[grand].content_score += content_score / 2.0
+
+        if node not in candidates:
+            candidates[node] = ScoredNode(node)
+        candidates[node].content_score += content_score
        logger.debug("Giving grand bonus points: %f", candidates[grand].content_score)

    for candidate in candidates.values():
--- a/tests/test_readable.py
+++ b/tests/test_readable.py
@ -216,20 +216,16 @@ class TestScoringNodes(unittest.TestCase):

    def test_we_get_candidates(self):
        """Processing candidates should get us a list of nodes to try out."""
-        # we'll start out using our first real test document
-        test_nodes = []
-        doc = document_fromstring(load_article('ars.001.html'))
-        for node in doc.iter('p', 'td', 'pre'):
-            test_nodes.append(node)
-
+        doc = document_fromstring(load_article("ars.001.html"))
+        test_nodes = tuple(doc.iter("p", "td", "pre"))
        candidates = score_candidates(test_nodes)

-        # this might change as we tweak our algorithm, but if it does change,
+        # this might change as we tweak our algorithm, but if it does,
        # it signifies we need to look at what we changed.
-        self.assertEqual(len(candidates.keys()), 6)
+        self.assertEqual(len(candidates.keys()), 37)

        # one of these should have a decent score
-        scores = sorted([c.content_score for c in candidates.values()])
+        scores = sorted(c.content_score for c in candidates.values())
        self.assertTrue(scores[-1] > 100)

    def test_bonus_score_per_100_chars_in_p(self):
--- a/tests/test_scoring.py
+++ b/tests/test_scoring.py
@ -248,7 +248,7 @@ class TestScoreCandidates(unittest.TestCase):

    def test_simple_candidate_set(self):
        """Tests a simple case of two candidate nodes"""
-        doc = """
+        html = """
            <html>
            <body>
                <div class="content">
@ -262,18 +262,16 @@ class TestScoreCandidates(unittest.TestCase):
            </body>
            </html>
        """
-        d_elem = document_fromstring(doc)
-        divs = d_elem.findall(".//div")
-        f_elem = divs[0]
-        s_elem = divs[1]
-
-        res = score_candidates([f_elem, s_elem])
-        ordered = sorted([c for c in res.values()],
-                          key=attrgetter('content_score'),
-                          reverse=True)
-
-        # the body element should have a higher score
-        self.assertTrue(ordered[0].node.tag == 'body')
-
-        # the html element is the outer should come in second
-        self.assertTrue(ordered[1].node.tag == 'html')
+        dom = document_fromstring(html)
+        div_nodes = dom.findall(".//div")
+
+        candidates = score_candidates(div_nodes)
+        ordered = sorted((c for c in candidates.values()), reverse=True,
+            key=attrgetter("content_score"))
+
+        self.assertEqual(ordered[0].node.tag, "div")
+        self.assertEqual(ordered[0].node.attrib["class"], "content")
+        self.assertEqual(ordered[1].node.tag, "body")
+        self.assertEqual(ordered[2].node.tag, "html")
+        self.assertEqual(ordered[3].node.tag, "div")
+        self.assertEqual(ordered[3].node.attrib["class"], "footer")