[htmlhead] Increase test coverage, invoke reset after parsing

lovett · Oct 8, 2018 · b7adea0 · b7adea0
1 parent ccb9365
commit b7adea0
Show file tree

Hide file tree

Showing 2 changed files with 59 additions and 8 deletions.
diff --git a/parsers/htmlhead.py b/parsers/htmlhead.py
@@ -1,7 +1,6 @@
 """Extract the tags in the head section of an HTML document."""
 
 from html.parser import HTMLParser
-from html.entities import name2codepoint
 from collections import deque
 
 
@@ -20,7 +19,18 @@ class Parser(HTMLParser):
     def __init__(self):
         super().__init__()
 
+    def reset(self):
+        """Reset the parser instance."""
+        super().reset()
+        self.in_head = False
+        self.finished = False
+        self.stack.clear()
+
     def error(self, message):
+        """Do-nothing override of default error handler.
+
+        This is pointless, but having it appeases pylint.
+        """
         pass
 
     def parse(self, markup):
@@ -29,6 +39,8 @@ def parse(self, markup):
 
         self.feed(markup)
 
+        self.reset()
+
         return self.result
 
     def handle_starttag(self, tag, attrs):
@@ -68,6 +80,3 @@ def handle_data(self, data):
             tag, attrs, text = self.stack.pop()
             text = "{} {}".format(text, data.strip()).strip()
             self.stack.append((tag, attrs, text))
-
-    def handle_entityref(self, name):
-        self.result.append(chr(name2codepoint[name]))
diff --git a/parsers/test_htmlhead.py b/parsers/test_htmlhead.py
@@ -10,20 +10,25 @@ class TestHtmlHeadParser(unittest.TestCase):
     parser = None
 
     @classmethod
-    def setUpClass(cls):
+    def setUp(cls):
         """Create the parser instance."""
         cls.parser = parsers.htmlhead.Parser()
 
     @classmethod
-    def tearDownClass(cls):
+    def tearDown(cls):
         """Destroy the parser."""
         cls.parser = None
 
     def test_simple_parse(self):
-        """A simplistic document is parsed successfully."""
+        """A reasonably-structured document is parsed successfully."""
 
         initial = """
-        <html><head><title id="test">Hello world</title></head></html>
+        <html>
+        <head>
+            <title id="test">Hello world</title>
+            <meta name="keyword" content="whatever" />
+        </head>
+        </html>
         """
         final = self.parser.parse(initial)
 
@@ -32,6 +37,43 @@ def test_simple_parse(self):
         self.assertEqual(final[0][1][0][1], "test")
         self.assertEqual(final[0][2], "Hello world")
         self.assertEqual(len(final[0]), 3)
+        self.assertEqual(len(final), 2)
+
+    def test_no_head(self):
+        """A document with no head is parsed successfully."""
+
+        initial = """
+        <html>this is weird</html>
+        """
+        final = self.parser.parse(initial)
+
+        self.assertEqual(len(final), 0)
+
+    def test_entity(self):
+        """Entities are converted during parsing."""
+
+        initial = """
+        <html><head><title>hello &gt; world</title></head></html>
+        """
+        final = self.parser.parse(initial)
+
+        self.assertEqual(final[0][2], "hello > world")
+
+    def test_only_head(self):
+        """Tags outside the head are ignored."""
+
+        initial = """
+        <html>
+        <head>
+            <title>hello world</title>
+        </head>
+        <body>this is ignored</body>
+        </html>
+        """
+        final = self.parser.parse(initial)
+
+        print(final)
+        self.assertEqual(len(final), 1)
 
 
 if __name__ == "__main__":