Preprocess: don't parse and reparse HTML unnecessarily

mokibit · mokibit · commit 16c9e13deb76 · 2018-06-05T20:21:46.000+03:00
diff --git a/commands/preprocess_cssless.py b/commands/preprocess_cssless.py
@@ -39,28 +39,26 @@ def preprocess_html_merge_css(src_path, dst_path):
 
     with open(src_path, 'r') as a_file:
         content = a_file.read()
+        parser = etree.HTMLParser()
+        stripped = content.strip()
+        root = etree.fromstring(stripped, parser)
 
     with warnings.catch_warnings():
         warnings.simplefilter("ignore")
-        premailer = Premailer(content, base_url=src_path, preserve_all_links=True,
+        premailer = Premailer(root, base_url=src_path, preserve_all_links=True,
                               remove_classes=True)
-        content = premailer.transform()
+
+        root = premailer.transform().getroot()
 
     head = os.path.dirname(dst_path)
     os.makedirs(head, exist_ok=True)
 
-    with open(dst_path,"w") as a_file:
-        a_file.write(content)
-
-    with open(dst_path, encoding='utf-8') as a_file:
-        # completely remove content of style tags and tags
-        parser = etree.HTMLParser()
-        tree = etree.parse(StringIO(content), parser)
-        nondata_tags = ['style']
-        strip_elements(tree, *nondata_tags)
+    # completely remove content of style tags and tags
+    nondata_tags = ['style']
+    strip_elements(root, *nondata_tags)
 
     with open(dst_path, 'wb') as a_file:
-        tree.write(dst_path, pretty_print=True, method="html",
-                   encoding='utf-8')
+        root.getroottree().write(a_file, pretty_print=True, method="html",
+                                 encoding='utf-8')
 
     return output.getvalue()
