Merge pull request #922 from Kozea/perf

Improve performance

weasyprint/__init__.py

@@ -332,7 +332,6 @@ def __init__(self, guess=None, filename=None, url=None, file_obj=None,
         self.base_url = base_url
         self.matcher = matcher or cssselect2.Matcher()
         self.page_rules = [] if page_rules is None else page_rules
-        # TODO: fonts are stored here and should be cleaned after rendering
         self.fonts = []
         preprocess_stylesheet(
             media_type, base_url, stylesheet, url_fetcher, self.matcher,

weasyprint/css/__init__.py

@@ -34,7 +34,8 @@
 from .validation.descriptors import preprocess_descriptors
 
 # Reject anything not in here:
-PSEUDO_ELEMENTS = (None, 'before', 'after', 'first-line', 'first-letter')
+PSEUDO_ELEMENTS = (
+    None, 'before', 'after', 'marker', 'first-line', 'first-letter')
 
 
 PageType = namedtuple('PageType', ['side', 'blank', 'first', 'index', 'name'])
@@ -67,11 +68,14 @@ def __init__(self, html, sheets, presentational_hints, target_collector):
         for specificity, attributes in find_style_attributes(
                 html.etree_element, presentational_hints, html.base_url):
             element, declarations, base_url = attributes
+            style = cascaded_styles.setdefault((element, None), {})
             for name, values, importance in preprocess_declarations(
                     base_url, declarations):
                 precedence = declaration_precedence('author', importance)
                 weight = (precedence, specificity)
-                add_declaration(cascaded_styles, name, values, weight, element)
+                old_weight = style.get(name, (None, None))[1]
+                if old_weight is None or old_weight <= weight:
+                    style[name] = values, weight
 
         # First, add declarations and set computed styles for "real" elements
         # *in tree order*. Tree order is important so that parents have
@@ -84,12 +88,14 @@ def __init__(self, html, sheets, presentational_hints, target_collector):
                 for selector in sheet.matcher.match(element):
                     specificity, order, pseudo_type, declarations = selector
                     specificity = sheet_specificity or specificity
+                    style = cascaded_styles.setdefault(
+                        (element.etree_element, pseudo_type), {})
                     for name, values, importance in declarations:
                         precedence = declaration_precedence(origin, importance)
                         weight = (precedence, specificity)
-                        add_declaration(
-                            cascaded_styles, name, values, weight,
-                            element.etree_element, pseudo_type)
+                        old_weight = style.get(name, (None, None))[1]
+                        if old_weight is None or old_weight <= weight:
+                            style[name] = values, weight
             parent = element.parent.etree_element if element.parent else None
             self.set_computed_styles(
                 element.etree_element, root=html.etree_element, parent=parent,
@@ -165,13 +171,15 @@ def add_page_declarations(self, page_type):
                     specificity, pseudo_type, selector_page_type = selector
                     if self._page_type_match(selector_page_type, page_type):
                         specificity = sheet_specificity or specificity
+                        style = self._cascaded_styles.setdefault(
+                            (page_type, pseudo_type), {})
                         for name, values, importance in declarations:
                             precedence = declaration_precedence(
                                 origin, importance)
                             weight = (precedence, specificity)
-                            add_declaration(
-                                self._cascaded_styles, name, values, weight,
-                                page_type, pseudo_type)
+                            old_weight = style.get(name, (None, None))[1]
+                            if old_weight is None or old_weight <= weight:
+                                style[name] = values, weight
 
     def get_cascaded_styles(self):
         return self._cascaded_styles
@@ -573,19 +581,6 @@ def declaration_precedence(origin, importance):
         return 5
 
 
-def add_declaration(cascaded_styles, prop_name, prop_values, weight, element,
-                    pseudo_type=None):
-    """Set the value for a property on a given element.
-
-    The value is only set if there is no value of greater weight defined yet.
-
-    """
-    style = cascaded_styles.setdefault((element, pseudo_type), {})
-    _values, previous_weight = style.get(prop_name, (None, None))
-    if previous_weight is None or previous_weight <= weight:
-        style[prop_name] = prop_values, weight
-
-
 def computed_from_cascaded(element, cascaded, parent_style, pseudo_type=None,
                            root_style=None, base_url=None,
                            target_collector=None):
@@ -602,8 +597,10 @@ def computed_from_cascaded(element, cascaded, parent_style, pseudo_type=None,
         # border-*-style is none, so border-width computes to zero.
         # Other than that, properties that would need computing are
         # border-*-color, but they do not apply.
-        for side in ('top', 'bottom', 'left', 'right'):
-            computed['border_%s_width' % side] = 0
+        computed['border_top_width'] = 0
+        computed['border_bottom_width'] = 0
+        computed['border_left_width'] = 0
+        computed['border_right_width'] = 0
         computed['outline_width'] = 0
         return computed
 
@@ -614,10 +611,10 @@ def computed_from_cascaded(element, cascaded, parent_style, pseudo_type=None,
     if parent_style:
         for name in parent_style:
             if name.startswith('__'):
-                specified[name] = parent_style[name]
+                computed[name] = specified[name] = parent_style[name]
     for name in cascaded:
         if name.startswith('__'):
-            specified[name] = cascaded[name][0]
+            computed[name] = specified[name] = cascaded[name][0]
 
     for name, initial in INITIAL_VALUES.items():
         if name in cascaded:
@@ -761,7 +758,7 @@ def parse_page_selectors(rule):
 
                     types['index'] = (*nth_values, group)
                     # TODO: specificity is not specified yet
-                    # https://github.com/w3c/csswg-drafts/issues/3791
+                    # https://github.com/w3c/csswg-drafts/issues/3524
                     types['specificity'][1] += 1
                     continue
 

weasyprint/css/computed_values.py

@@ -197,29 +197,20 @@ def compute(element, pseudo_type, specified, computed, parent_style,
     """
     from .validation.properties import PROPERTIES
 
-    def computer():
-        """Dummy object that holds attributes."""
-        return 0
-
-    computer.is_root_element = parent_style is None
-    if parent_style is None:
-        parent_style = INITIAL_VALUES
-
-    computer.element = element
-    computer.pseudo_type = pseudo_type
-    computer.specified = specified
-    computer.computed = computed
-    computer.parent_style = parent_style
-    computer.root_style = root_style
-    computer.base_url = base_url
-    computer.target_collector = target_collector
+    computer = {
+        'is_root_element': parent_style is None,
+        'element': element,
+        'pseudo_type': pseudo_type,
+        'specified': specified,
+        'computed': computed,
+        'parent_style': parent_style or INITIAL_VALUES,
+        'root_style': root_style,
+        'base_url': base_url,
+        'target_collector': target_collector,
+    }
 
     getter = COMPUTER_FUNCTIONS.get
 
-    for name in specified:
-        if name.startswith('__'):
-            computed[name] = specified[name]
-
     for name in COMPUTING_ORDER:
         if name in computed:
             # Already computed
@@ -354,24 +345,24 @@ def length(computer, name, value, font_size=None, pixels_only=False):
         result = value.value * LENGTHS_TO_PIXELS[unit]
     elif unit in ('em', 'ex', 'ch', 'rem'):
         if font_size is None:
-            font_size = computer.computed['font_size']
+            font_size = computer['computed']['font_size']
         if unit == 'ex':
             # TODO: cache
-            result = value.value * font_size * ex_ratio(computer.computed)
+            result = value.value * font_size * ex_ratio(computer['computed'])
         elif unit == 'ch':
             # TODO: cache
             # TODO: use context to use @font-face fonts
             layout = text.Layout(
                 context=None, font_size=font_size,
-                style=computer.computed)
+                style=computer['computed'])
             layout.set_text('0')
             line, _ = layout.get_first_line()
-            logical_width, _ = text.get_size(line, computer.computed)
+            logical_width, _ = text.get_size(line, computer['computed'])
             result = value.value * logical_width
         elif unit == 'em':
             result = value.value * font_size
         elif unit == 'rem':
-            result = value.value * computer.root_style['font_size']
+            result = value.value * computer['root_style']['font_size']
     else:
         # A percentage or 'auto': no conversion needed.
         return value
@@ -385,7 +376,7 @@ def length(computer, name, value, font_size=None, pixels_only=False):
 @register_computer('bleed-bottom')
 def bleed(computer, name, value):
     if value == 'auto':
-        if 'crop' in computer.computed['marks']:
+        if 'crop' in computer['computed']['marks']:
             return Dimension(8, 'px')  # 6pt
         else:
             return Dimension(0, 'px')
@@ -418,7 +409,7 @@ def background_size(computer, name, values):
 @register_computer('outline-width')
 def border_width(computer, name, value):
     """Compute the ``border-*-width`` properties."""
-    style = computer.computed[name.replace('width', 'style')]
+    style = computer['computed'][name.replace('width', 'style')]
     if style in ('none', 'hidden'):
         return 0
 
@@ -461,19 +452,19 @@ def compute_attr_function(computer, values):
     func_name, value = values
     assert func_name == 'attr()'
     attr_name, type_or_unit, fallback = value
-    # computer.element sometimes is None
-    # computer.element sometimes is a 'PageType' object without .get()
+    # computer['element'] sometimes is None
+    # computer['element'] sometimes is a 'PageType' object without .get()
     # so wrapt the .get() into try and return None instead of crashing
     try:
-        attr_value = computer.element.get(attr_name, fallback)
+        attr_value = computer['element'].get(attr_name, fallback)
         if type_or_unit == 'string':
             pass  # Keep the string
         elif type_or_unit == 'url':
             if attr_value.startswith('#'):
                 attr_value = ('internal', unquote(attr_value[1:]))
             else:
                 attr_value = (
-                    'external', safe_urljoin(computer.base_url, attr_value))
+                    'external', safe_urljoin(computer['base_url'], attr_value))
         elif type_or_unit == 'color':
             attr_value = parse_color(attr_value.strip())
         elif type_or_unit == 'integer':
@@ -519,12 +510,12 @@ def _content_list(computer, values):
                         (attr,) + value[1][1:]))
             else:
                 computed_value = value
-            if computer.target_collector and computed_value:
-                computer.target_collector.collect_computed_target(
+            if computer['target_collector'] and computed_value:
+                computer['target_collector'].collect_computed_target(
                     computed_value[1][0])
         if computed_value is None:
             LOGGER.warning('Unable to compute %s\'s value for content: %s' % (
-                computer.element, ', '.join(str(item) for item in value)))
+                computer['element'], ', '.join(str(item) for item in value)))
         else:
             computed_values.append(computed_value)
 
@@ -552,7 +543,7 @@ def content(computer, name, values):
     if len(values) == 1:
         value, = values
         if value == 'normal':
-            return 'inhibit' if computer.pseudo_type else 'contents'
+            return 'inhibit' if computer['pseudo_type'] else 'contents'
         elif value == 'none':
             return 'inhibit'
     return _content_list(computer, values)
@@ -565,10 +556,10 @@ def display(computer, name, value):
     See http://www.w3.org/TR/CSS21/visuren.html#dis-pos-flo
 
     """
-    float_ = computer.specified['float']
-    position = computer.specified['position']
+    float_ = computer['specified']['float']
+    position = computer['specified']['position']
     if position in ('absolute', 'fixed') or float_ != 'none' or \
-            computer.is_root_element:
+            computer['is_root_element']:
         if value == 'inline-table':
             return'table'
         elif value in ('inline', 'table-row-group', 'table-column',
@@ -586,7 +577,7 @@ def compute_float(computer, name, value):
     See http://www.w3.org/TR/CSS21/visuren.html#dis-pos-flo
 
     """
-    if computer.specified['position'] in ('absolute', 'fixed'):
+    if computer['specified']['position'] in ('absolute', 'fixed'):
         return 'none'
     else:
         return value
@@ -599,7 +590,7 @@ def font_size(computer, name, value):
         return FONT_SIZE_KEYWORDS[value]
     # TODO: support 'larger' and 'smaller'
 
-    parent_font_size = computer.parent_style['font_size']
+    parent_font_size = computer['parent_style']['font_size']
     if value.unit == '%':
         return value.value * parent_font_size / 100.
     else:
@@ -615,7 +606,7 @@ def font_weight(computer, name, value):
     elif value == 'bold':
         return 700
     elif value in ('bolder', 'lighter'):
-        parent_value = computer.parent_style['font_weight']
+        parent_value = computer['parent_style']['font_weight']
         return FONT_WEIGHT_RELATIVE[value][parent_value]
     else:
         return value
@@ -630,7 +621,7 @@ def line_height(computer, name, value):
         return ('NUMBER', value.value)
     elif value.unit == '%':
         factor = value.value / 100.
-        font_size_value = computer.computed['font_size']
+        font_size_value = computer['computed']['font_size']
         pixels = factor * font_size_value
     else:
         pixels = length(computer, name, value, pixels_only=True)
@@ -642,8 +633,8 @@ def anchor(computer, name, values):
     """Compute the ``anchor`` property."""
     if values != 'none':
         _, key = values
-        anchor_name = computer.element.get(key) or None
-        computer.target_collector.collect_anchor(anchor_name)
+        anchor_name = computer['element'].get(key) or None
+        computer['target_collector'].collect_anchor(anchor_name)
         return anchor_name
 
 
@@ -656,7 +647,7 @@ def link(computer, name, values):
         type_, value = values
         if type_ == 'attr()':
             return get_link_attribute(
-                computer.element, value, computer.base_url)
+                computer['element'], value, computer['base_url'])
         else:
             return values
 
@@ -669,7 +660,7 @@ def lang(computer, name, values):
     else:
         type_, key = values
         if type_ == 'attr()':
-            return computer.element.get(key) or None
+            return computer['element'].get(key) or None
         elif type_ == 'string':
             return key
 
@@ -703,11 +694,11 @@ def vertical_align(computer, name, value):
                  'top', 'bottom'):
         return value
     elif value == 'super':
-        return computer.computed['font_size'] * 0.5
+        return computer['computed']['font_size'] * 0.5
     elif value == 'sub':
-        return computer.computed['font_size'] * -0.5
+        return computer['computed']['font_size'] * -0.5
     elif value.unit == '%':
-        height, _ = strut_layout(computer.computed)
+        height, _ = strut_layout(computer['computed'])
         return height * value.value / 100.
     else:
         return length(computer, name, value, pixels_only=True)

weasyprint/css/html5_ua.css

@@ -473,6 +473,7 @@ track { display: none; }
 tt { font-family: monospace; }
 u { text-decoration: underline; }
 
+::marker { unicode-bidi: isolate; font-variant-numeric: tabular-nums; }
 ul { display: block; list-style-type: disc; margin-bottom: 1em; margin-top: 1em; padding-left: 40px; unicode-bidi: isolate; }
 
 *[dir=ltr] ul { padding-left: 40px; padding-right: 0; }

weasyprint/css/tests_ua.css

@@ -32,3 +32,5 @@ h3 { bookmark-level: 3; bookmark-label: content(text); }
 h4 { bookmark-level: 4; bookmark-label: content(text); }
 h5 { bookmark-level: 5; bookmark-label: content(text); }
 h6 { bookmark-level: 6; bookmark-label: content(text); }
+
+::marker { unicode-bidi: isolate; font-variant-numeric: tabular-nums; }