Added Mako parsing rules

deezer · Aug 12, 2014 · de963f2 · de963f2
1 parent 8871a8d
commit de963f2
Show file tree

Hide file tree

Showing 3 changed files with 70 additions and 10 deletions.
diff --git a/setup.py b/setup.py
@@ -17,7 +17,7 @@
 
 setup(
     name='template-remover',
-    version='0.1.6',
+    version='0.1.7',
     description='Remove the template markup from html files',
     long_description=open('README.rst').read(),
     author='Sebastian Kreft - Deezer',

diff --git a/template_remover.py b/template_remover.py
@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-"""Replaces the template markup (PHP, Jinja) from an HTML file.
+"""Replaces the template markup (PHP, Jinja, Mako) from an HTML file.
 
 It replaces the markup so the lines and positions of actual HTML content is
 preserved.
@@ -51,6 +51,13 @@
 JINJA_END_TAG = r'%}'
 JINJA_END_ECHO = r'}}'
 
+# Mako patterns
+MAKO_START_ECHO = r'\${'
+MAKO_END_ECHO = r'}'
+MAKO_START_TAG = (
+    r'(</?%|%\s*(end)?(if|else|elif|for|while|try|catch|except|finally)|##)')
+MAKO_END_TAG = r'([%/]?>|$)'
+
 
 def get_pattern(echo_tags, start_tags, end_tags):
     """Constructs a pattern to use with the method clean().
@@ -90,26 +97,32 @@ def get_pattern(echo_tags, start_tags, end_tags):
     [PHP_START_TAG, PHP_START_TAG_SHORT],
     [PHP_END_TAG])
 
+MAKO_PATTERN = get_pattern(
+    [MAKO_START_ECHO],
+    [MAKO_START_TAG],
+    [MAKO_END_TAG, MAKO_END_ECHO])
+
 ALL_PATTERN = get_pattern(
     [PHP_START_ECHO, PHP_START_TAG_WITH_ECHO, PHP_START_TAG_SHORT_WITH_ECHO,
-        JINJA_START_ECHO],
-    [PHP_START_TAG, PHP_START_TAG_SHORT, JINJA_START_TAG],
-    [PHP_END_TAG, JINJA_END_TAG, JINJA_END_ECHO])
+        JINJA_START_ECHO, MAKO_START_ECHO],
+    [PHP_START_TAG, PHP_START_TAG_SHORT, JINJA_START_TAG, MAKO_START_TAG],
+    [PHP_END_TAG, JINJA_END_TAG, JINJA_END_ECHO, MAKO_END_TAG, MAKO_END_ECHO])
 
 
 def _get_tag(match):
     groups = match.groupdict()
-    if groups.get('echo'):
+    if groups.get('echo') is not None:
         return 'ECHO'
-    elif groups.get('end'):
+    elif groups.get('end') is not None:
         return 'END'
-    elif groups.get('start'):
+    elif groups.get('start') is not None:
         return 'START'
-    elif groups.get('newline'):
+    elif groups.get('newline') is not None:
         return 'NEWLINE'
-    elif groups.get('spaces'):
+    elif groups.get('spaces') is not None:
         return 'SPACES'
 
+    print(groups)
     assert False, ('Only the groups "echo", "end", "start", "newline" and ' +
                    '"spaces" are allowed. Please correct your pattern or use ' +
                    'the method get_pattern() to construct it.')
@@ -261,3 +274,11 @@ def clean_jinja(html_content):
     See clean() for more details.
     """
     return clean(html_content, pattern=JINJA_PATTERN)
+
+
+def clean_mako(html_content):
+    """Removes the Mako markup from the supplied string.
+
+    See clean() for more details.
+    """
+    return clean(html_content, pattern=MAKO_PATTERN)
diff --git a/test/test_template_remover.py b/test/test_template_remover.py
@@ -162,3 +162,42 @@ def test_jinja_does_not_complain_with_double_braces(self):
         template = '<script>var dict={foo:{}}</script>'
         expected = '<script>var dict={foo:{}}</script>'
         self.assertEquals(expected, template_remover.clean_jinja(template))
+
+    def test_mako_comprehensive(self):
+        template = '''<html>
+          <body>
+          
+          <a title="${"this is some text" | u}">Foo</a>
+          % if x == 5:
+            <a href="${foo}"><a>
+          % endif
+          <%
+            x = db.get_resource('foo')
+            y = [z.element for z in x if x.frobnizzle==5]
+          %>
+          <%include file="foo.txt"/>
+          ## Comment
+          <%def name="foo" buffered="True">
+            this is a def
+          </%def>
+          
+        '''
+        expected = '''<html>
+          <body>
+          
+          <a title="00000000000000000000000000">Foo</a>
+
+            <a href="000000"><a>
+
+
+
+
+
+
+
+
+            this is a def
+
+          
+        '''
+        self.assertEquals(expected, template_remover.clean_mako(template))