Merge pull request codex-team#9 from sebbacon/handle-script-and-styl

Don't output some unsupported tags or attributes

Author: khaydarov

.gitignore

@@ -2,6 +2,7 @@
 build/
 dist/
 venv/
+.idea/
 *.egg-info/
 *.egg
 *.py[cod]

htmlslacker/htmlslacker.py

@@ -22,6 +22,7 @@ def __init__(self, html, *args, **kwargs):
             super().__init__(*args, **kwargs)
         except TypeError:
             HTMLParser.__init__(self, *args, **kwargs)
+        self.skip = False
 
         # slackified string
         self.output = ''
@@ -50,7 +51,10 @@ def handle_starttag(self, tag, attrs):
         if tag == 'a':
             self.output += '<'
             for attr in attrs:
-                self.output += attr[1] + '|'
+                if attr[0] == 'href':
+                    self.output += attr[1] + '|'
+        if tag == 'style' or tag == 'script':
+            self.skip = True
 
     def handle_endtag(self, tag):
         """
@@ -66,14 +70,17 @@ def handle_endtag(self, tag):
             self.output += '>'
         if tag == 'code':
             self.output += '`'
+        if tag == 'style' or tag == 'script':
+            self.skip = False
 
     def handle_data(self, data):
         """
         concatenate TEXT nodes into output
         :param data:
         :return:
         """
-        self.output += data
+        if not self.skip:
+            self.output += data
 
     def handle_comment(self, data):
         pass

setup.py

@@ -8,7 +8,7 @@
 setup(
     name='html-slacker',
     packages=find_packages(),
-    version='0.1.5',
+    version='0.1.6',
     description='Converts HTML to Slack formatted markdown',
     long_description=open(join(dirname(__file__), 'README.rst')).read(),
     keywords='slack bot html convert markdown slackbot pythonbot slack html markdown',

test_general.py

@@ -14,3 +14,24 @@ def test_example_1():
     expected = "*Hello*\n There is _something_ interesting about `this doc` \n And <http://example.com/|here is a link in a paragraph!>"
     output = HTMLSlacker(html).get_output()
     assert(output == expected)
+
+
+def test_style_not_rendered():
+    html = '''<style><!-- /* stuff */ --></style>Dear Prudence'''
+    expected = "Dear Prudence"
+    output = HTMLSlacker(html).get_output()
+    assert(output == expected)
+
+
+def test_script_not_rendered():
+    html = '''<script><!-- /* stuff */ --></script>Dear Prudence'''
+    expected = "Dear Prudence"
+    output = HTMLSlacker(html).get_output()
+    assert(output == expected)
+
+
+def test_link_with_target():
+    html = 'Please click <a href="http://xxx.com/t.html" target="_blank">here</a>'
+    expected = "Please click <http://xxx.com/t.html|here>"
+    output = HTMLSlacker(html).get_output()
+    assert(output == expected)