use cssutils

qutebrowser/browser/mhtml.py

@@ -35,6 +35,10 @@ from PyQt5.QtCore import QUrl
 from qutebrowser.browser import webelem
 from qutebrowser.utils import log, objreg, message
 
+try:
+    import cssutils
+except ImportError:
+    cssutils = None
 
 _File = collections.namedtuple('_File',
                                ['content', 'content_type', 'content_location',
@@ -50,7 +54,7 @@ _CSS_URL_PATTERNS = [re.compile(x) for x in [
 ]]
 
 
-def _get_css_imports(data):
+def _get_css_imports_regex(data):
     """Return all assets that are referenced in the given CSS document.
 
     The returned URLs are relative to the stylesheet's URL.
@@ -67,6 +71,47 @@ def _get_css_imports(data):
     return urls
 
 
+def _get_css_imports_cssutils(data, inline=False):
+    """Return all assets that are referenced in the given CSS document.
+
+    The returned URLs are relative to the stylesheet's URL.
+
+    Args:
+        data: The content of the stylesheet to scan as string.
+        inline: True if the argument is a inline HTML style attribute.
+    """
+    parser = cssutils.CSSParser(fetcher=lambda url: (None, ""), validate=False)
+    if not inline:
+        sheet = parser.parseString(data)
+        return list(cssutils.getUrls(sheet))
+    else:
+        urls = []
+        declaration = parser.parseStyle(data)
+        # prop = background, color, margin, ...
+        for prop in declaration:
+            # value = red, 10px, url(foobar), ...
+            for value in prop.propertyValue:
+                if isinstance(value, cssutils.css.URIValue):
+                    if value.uri:
+                        urls.append(value.uri)
+        return urls
+
+
+def _get_css_imports(data, inline=False):
+    """Return all assets that are referenced in the given CSS document.
+
+    The returned URLs are relative to the stylesheet's URL.
+
+    Args:
+        data: The content of the stylesheet to scan as string.
+        inline: True if the argument is a inline HTML style attribute.
+    """
+    if cssutils is None:
+        return _get_css_imports_regex(data)
+    else:
+        return _get_css_imports_cssutils(data, inline)
+
+
 MHTMLPolicy = email.policy.default.clone(linesep='\r\n', max_line_length=0)
 
 
@@ -228,7 +273,7 @@ class _Downloader():
         for element in web_frame.findAllElements('[style]'):
             element = webelem.WebElementWrapper(element)
             style = element['style']
-            for element_url in _get_css_imports(style):
+            for element_url in _get_css_imports(style, inline=True):
                 self.fetch_url(web_url.resolved(QUrl(element_url)))
 
         # Shortcut if no assets need to be downloaded, otherwise the file would

requirements.txt

@@ -5,3 +5,4 @@ pyPEG2==2.15.2
 PyYAML==3.11
 colorama==0.3.3
 colorlog==2.6.0
+cssutils==1.0

tests/unit/browser/test_mhtml.py

@@ -250,20 +250,26 @@ def test_removing_file_from_mhtml(checker):
         """)
 
 
-@pytest.mark.parametrize('style, expected_urls', [
-    ("@import 'default.css'", ['default.css']),
-    ('@import "default.css"', ['default.css']),
-    ("@import \t 'tabbed.css'", ['tabbed.css']),
-    ("@import url('default.css')", ['default.css']),
-    ("""body {
+@pytest.mark.parametrize('has_cssutils', [True, False])
+@pytest.mark.parametrize('inline, style, expected_urls', [
+    (False, "@import 'default.css'", ['default.css']),
+    (False, '@import "default.css"', ['default.css']),
+    (False, "@import \t 'tabbed.css'", ['tabbed.css']),
+    (False, "@import url('default.css')", ['default.css']),
+    (False, """body {
     background: url("/bg-img.png")
     }""", ['/bg-img.png']),
-    ('background: url(folder/file.png)', ['folder/file.png']),
-    ('content: url()', []),
+    (True, 'background: url(folder/file.png) no-repeat', ['folder/file.png']),
+    (True, 'content: url()', []),
 ])
-def test_css_url_scanner(style, expected_urls):
+def test_css_url_scanner(monkeypatch, has_cssutils, inline, style,
+                         expected_urls):
+    if has_cssutils:
+        assert mhtml.cssutils is not None
+    else:
+        monkeypatch.setattr('qutebrowser.browser.mhtml.cssutils', None)
     expected_urls.sort()
-    urls = mhtml._get_css_imports(style)
+    urls = mhtml._get_css_imports(style, inline=inline)
     urls.sort()
     assert urls == expected_urls