Rewrite CSS embeded in html.

src/warc2zim/content_rewriting.py

@@ -1,12 +1,12 @@
 
 from html import escape
 from html.parser import HTMLParser
-from tinycss2 import parse_stylesheet, serialize
+from tinycss2 import parse_stylesheet, parse_declaration_list, serialize
 from tinycss2.serializer import serialize_url
 import io
 from .url_rewriting import ArticleUrlRewriter
 
-def process_attr(attr, rewriter):
+def process_attr(attr, url_rewriter, css_rewriter):
     if attr[0] in ('href', 'src', 'data-lazy-src'):
         return (attr[0], url_rewriter(attr[1]))
     if attr[0] in ('srcset', 'data-lazy-srcset'):
@@ -18,6 +18,8 @@ def process_attr(attr, rewriter):
             new_value = " ".join([new_url, *other])
             new_value_list.append(new_value)
         return (attr[0], ", ".join(new_value_list))
+    if attr[0] == "style":
+        return (attr[0], css_rewriter.rewrite_inline(attr[1]))
     return attr
 
 def format_attr(name, value):
@@ -26,8 +28,8 @@ def format_attr(name, value):
     html_escaped_value = escape(value, quote=True)
     return f"{name}=\"{html_escaped_value}\""
 
-def transform_attrs(attrs, url_rewriter):
-    processed_attrs = (process_attr(attr, url_rewriter) for attr in attrs)
+def transform_attrs(attrs, url_rewriter, css_rewriter):
+    processed_attrs = (process_attr(attr, url_rewriter, css_rewriter) for attr in attrs)
     return " ".join(
         format_attr(*attr) for attr in processed_attrs
     )
@@ -36,6 +38,7 @@ class RewriterParser(HTMLParser):
     def __init__(self, host, article_url, head_insert, css_insert, sink):
         super().__init__()
         self.url_rewriter = ArticleUrlRewriter(host, article_url)
+        self.css_rewriter = CSSRewriter(host, article_url)
         self.sink = sink
         self.title = ""
         # This works only for tag without children.
@@ -51,7 +54,7 @@ def handle_starttag(self, tag, attrs, auto_close=False):
         self._active_tag = tag
 
         self.send(f"<{tag} ")
-        self.send(transform_attrs(attrs, self.url_rewriter))
+        self.send(transform_attrs(attrs, self.url_rewriter, self.css_rewriter))
 
         if auto_close:
             self.send(" />")
@@ -73,6 +76,8 @@ def handle_startendtag(self, tag, attrs):
     def handle_data(self, data):
         if self._active_tag == 'title':
             self.title = data.strip()
+        elif self._active_tag == "style":
+            data = self.css_rewriter.rewrite(data)
         self.send(data)
 
     def handle_comment(self, data):
@@ -103,13 +108,20 @@ def __init__(self, host, css_url):
         self.url_rewriter = ArticleUrlRewriter(host, css_url)
 
     def rewrite(self, content):
-        content = content.decode()
+        if isinstance(content, bytes):
+            content = content.decode()
         rules = parse_stylesheet(content)
         self.process_list(rules)
 
         output = serialize(rules)
         return output
 
+    def rewrite_inline(self, content):
+        rules = parse_declaration_list(content)
+        self.process_list(rules)
+        output = serialize(rules)
+        return output
+
     def process_list(self, components):
         if components: # May be null
             for component in components: