Possible optimizations - do not merge

CHANGES

@@ -1,6 +1,72 @@
 WeasyPrint changelog
 ====================
 
+
+Versions 0.42.x will only get simple bug fixes backported from the master
+branch. New features, opitimizations and complex bug fixes will only be added
+to the 43+ versions that don't support Python 2 anymore.
+
+Do not rely on future versions, development on the 0.x branch may be stopped at
+any moment.
+
+
+Version 0.42.3
+--------------
+
+Released on 2018-03-27.
+
+Bug fixes:
+
+* `#583 <https://github.com/Kozea/WeasyPrint/issues/583>`_:
+  Fix floating-point number error to fix floating box layout
+* `#586 <https://github.com/Kozea/WeasyPrint/issues/586>`_:
+  Don't optimize resume_at when splitting lines with trailing spaces
+* `#582 <https://github.com/Kozea/WeasyPrint/issues/582>`_:
+  Fix table layout with no overflow
+* `#580 <https://github.com/Kozea/WeasyPrint/issues/580>`_:
+  Fix inline box breaking function
+* `#576 <https://github.com/Kozea/WeasyPrint/issues/576>`_:
+  Split replaced_min_content_width and replaced_max_content_width
+* `#574 <https://github.com/Kozea/WeasyPrint/issues/574>`_:
+  Respect text direction and don't translate rtl columns twice
+* `#569 <https://github.com/Kozea/WeasyPrint/issues/569>`_:
+  Get only first line's width of inline children to get linebox width
+
+
+Version 0.42.2
+--------------
+
+Released on 2018-02-04.
+
+Bug fixes:
+
+* `#560 <https://github.com/Kozea/WeasyPrint/issues/560>`_:
+  Fix a couple of crashes and endless loops when breaking lines.
+
+
+Version 0.42.1
+--------------
+
+Released on 2018-02-01.
+
+Bug fixes:
+
+* `#566 <https://github.com/Kozea/WeasyPrint/issues/566>`_:
+  Don't crash when using @font-config.
+* `#567 <https://github.com/Kozea/WeasyPrint/issues/567>`_:
+  Fix text-indent with text-align: justify.
+* `#465 <https://github.com/Kozea/WeasyPrint/issues/465>`_:
+  Fix string(*, start).
+* `#562 <https://github.com/Kozea/WeasyPrint/issues/562>`_:
+  Handle named pages with pseudo-class.
+* `#507 <https://github.com/Kozea/WeasyPrint/issues/507>`_:
+  Fix running headers.
+* `#557 <https://github.com/Kozea/WeasyPrint/issues/557>`_:
+  Avoid infinite loops in inline_line_width.
+* `#555 <https://github.com/Kozea/WeasyPrint/issues/555>`_:
+  Fix margins, borders and padding in column layouts.
+
+
 Version 0.42
 ------------
 

docs/hacking.rst

@@ -12,13 +12,13 @@ install the `development version`_ of WeasyPrint:
     cd WeasyPrint
     virtualenv --system-site-packages env
     . env/bin/activate
-    pip install pytest Sphinx -e .
+    pip install Sphinx -e .[test]
     weasyprint --help
 
 This will install WeasyPrint in “editable” mode
 (which means that you don’t need to re-install it
 every time you make a change in the source code)
-as well as `py.test <http://pytest.org/>`_
+as well as `pytest <http://pytest.org/>`_
 and `Sphinx <http://sphinx.pocoo.org/>`_.
 
 Lastly, in order to pass unit tests, your system must have two particular fonts:
@@ -43,7 +43,7 @@ The website version is updated automatically when we push to master on GitHub.
 Code changes
 ------------
 
-Use the ``py.test`` command from the ``WeasyPrint`` directory to run the
+Use the ``pytest`` command from the ``WeasyPrint`` directory to run the
 test suite.
 
 Please report any bugs/feature requests and submit patches/pull requests

docs/install.rst

@@ -4,8 +4,8 @@ Installing
 WeasyPrint |version| depends on:
 
 * CPython_ 2.7 or ≥ 3.4
-* cairo_ [#]_
-* Pango_
+* cairo_ ≥ 1.15.0 [#]_
+* Pango_ ≥ 1.29.3
 * CFFI_ ≥ 0.6
 * html5lib_ ≥ 0.999999999
 * cairocffi_ ≥ 0.5
@@ -14,7 +14,7 @@ WeasyPrint |version| depends on:
 * CairoSVG_ ≥ 1.0.20
 * Pyphen_ ≥ 0.8
 * pdfrw_ ≥ 0.4
-* Optional: GDK-PixBuf_ [#]_
+* Optional: GDK-PixBuf_ ≥ 2.25.0 [#]_
 
 .. _CPython: http://www.python.org/
 .. _cairo: http://cairographics.org/
@@ -69,9 +69,11 @@ If everything goes well, you’re ready to :doc:`start using </tutorial>`
 WeasyPrint! Otherwise, please copy the full error message and
 `report the problem <http://weasyprint.org/community/>`_.
 
-.. [#] cairo ≥ 1.12 is best but older versions should work too.
-       The test suite passes on cairo 1.8 and 1.10 with some tests marked as
-       “expected failures” due to behavior changes or bugs in cairo.
+.. [#] cairo ≥ 1.15 is best but older versions may work too. The test suite
+       passes on cairo 1.12 and 1.14, and passes with some tests marked as
+       “expected failures” on 1.8 and 1.10 due to behavior changes or bugs in
+       cairo. With cairo 1.14, WeasyPrint sometimes creates PDF files that may
+       not be correctly interpreted by GhostScript or CUPS.
 
 .. [#] Without it, PNG and SVG are the only supported image formats.
        JPEG, GIF and others are not available.
@@ -100,7 +102,7 @@ Debian 9.0 Stretch or newer, Ubuntu 16.04 Xenial or newer:
 
 .. code-block:: sh
 
-    sudo apt-get install build-essential python3-dev python3-pip python3-cffi libcairo2 libpango-1.0-0 libpangocairo-1.0.0 libgdk-pixbuf2.0-0 libffi-dev shared-mime-info
+    sudo apt-get install build-essential python3-dev python3-pip python3-cffi libcairo2 libpango-1.0-0 libpangocairo-1.0-0 libgdk-pixbuf2.0-0 libffi-dev shared-mime-info
 
 Debian 8.0 Jessie or newer, Ubuntu 14.04 Trusty or newer (with Python 2, but Python 3 may work):
 

docs/tutorial.rst

@@ -243,7 +243,6 @@ by configuring the ``weasyprint`` logger object:
 
     import logging
     logger = logging.getLogger('weasyprint')
-    logger.handlers = []  # Remove the default stderr handler
     logger.addHandler(logging.FileHandler('/path/to/weasyprint.log'))
 
 The ``INFO`` level is used to report the rendering progress. It is useful to

weasyprint/__init__.py

@@ -21,7 +21,7 @@
 import tinycss2
 
 
-VERSION = '0.42'
+VERSION = '0.42.3'
 __version__ = VERSION
 
 # Used for 'User-Agent' in HTTP and 'Creator' in PDF

weasyprint/css/__init__.py

@@ -586,6 +586,82 @@ def computed_from_cascaded(element, cascaded, parent_style, pseudo_type=None,
         base_url))
 
 
+def parse_page_selectors(rule):
+    """Parse a page selector rule.
+
+    Return a list of page data if the rule is correctly parsed. Page data are a
+    dict containing:
+
+    - 'side' ('left', 'right' or None),
+    - 'blank' (True or False),
+    - 'first' (True or False),
+    - 'name' (page name string or None), and
+    - 'spacificity' (list of numbers).
+
+    Return ``None` if something went wrong while parsing the rule.
+
+    """
+    # See https://drafts.csswg.org/css-page-3/#syntax-page-selector
+
+    tokens = list(remove_whitespace(rule.prelude))
+    page_data = []
+
+    # TODO: Specificity is probably wrong, should clean and test that.
+    if not tokens:
+        page_data.append({
+            'side': None, 'blank': False, 'first': False, 'name': None,
+            'specificity': [0, 0, 0]})
+        return page_data
+
+    while tokens:
+        types = {
+            'side': None, 'blank': False, 'first': False, 'name': None,
+            'specificity': [0, 0, 0]}
+
+        if tokens[0].type == 'ident':
+            token = tokens.pop(0)
+            types['name'] = token.value
+            types['specificity'][0] = 1
+
+        if len(tokens) == 1:
+            return None
+        elif not tokens:
+            page_data.append(types)
+            return page_data
+
+        while tokens:
+            literal = tokens.pop(0)
+            if literal.type != 'literal':
+                return None
+
+            if literal.value == ':':
+                if not tokens or tokens[0].type != 'ident':
+                    return None
+                ident = tokens.pop(0)
+                pseudo_class = ident.lower_value
+                if pseudo_class in ('left', 'right'):
+                    if types['side']:
+                        return None
+                    types['side'] = pseudo_class
+                    types['specificity'][2] += 1
+                elif pseudo_class in ('blank', 'first'):
+                    if types[pseudo_class]:
+                        return None
+                    types[pseudo_class] = True
+                    types['specificity'][1] += 1
+                else:
+                    return None
+            elif literal.value == ',':
+                if tokens and any(types['specificity']):
+                    break
+                else:
+                    return None
+
+        page_data.append(types)
+
+    return page_data
+
+
 def preprocess_stylesheet(device_media_type, base_url, stylesheet_rules,
                           url_fetcher, matcher, page_rules, fonts,
                           font_config, ignore_imports=False):
@@ -671,64 +747,44 @@ def preprocess_stylesheet(device_media_type, base_url, stylesheet_rules,
                 matcher, page_rules, fonts, font_config, ignore_imports=True)
 
         elif rule.type == 'at-rule' and rule.lower_at_keyword == 'page':
-            tokens = remove_whitespace(rule.prelude)
-            types = {
-                'side': None, 'blank': False, 'first': False, 'name': None}
-            # TODO: Specificity is probably wrong, should clean and test that.
-            if not tokens:
-                specificity = (0, 0, 0)
-            elif (len(tokens) == 2 and
-                    tokens[0].type == 'literal' and
-                    tokens[0].value == ':' and
-                    tokens[1].type == 'ident'):
-                pseudo_class = tokens[1].lower_value
-                if pseudo_class in ('left', 'right'):
-                    types['side'] = pseudo_class
-                    specificity = (0, 0, 1)
-                elif pseudo_class in ('blank', 'first'):
-                    types[pseudo_class] = True
-                    specificity = (0, 1, 0)
-                else:
-                    LOGGER.warning('Unknown @page pseudo-class "%s", '
-                                   'the whole @page rule was ignored '
-                                   'at %s:%s.',
-                                   pseudo_class,
-                                   rule.source_line, rule.source_column)
-                    continue
-            elif len(tokens) == 1 and tokens[0].type == 'ident':
-                types['name'] = tokens[0].value
-                specificity = (1, 0, 0)
-            else:
-                LOGGER.warning('Unsupported @page selector "%s", '
-                               'the whole @page rule was ignored at %s:%s.',
-                               tinycss2.serialize(rule.prelude),
-                               rule.source_line, rule.source_column)
+            data = parse_page_selectors(rule)
+
+            if data is None:
+                LOGGER.warning(
+                    'Unsupported @page selector "%s", '
+                    'the whole @page rule was ignored at %s:%s.',
+                    tinycss2.serialize(rule.prelude),
+                    rule.source_line, rule.source_column)
                 continue
-            ignore_imports = True
-            page_type = PageType(**types)
-            # Use a double lambda to have a closure that holds page_types
-            match = (lambda page_type: lambda page_names: list(
-                matching_page_types(page_type, names=page_names)))(page_type)
-            content = tinycss2.parse_declaration_list(rule.content)
-            declarations = list(preprocess_declarations(base_url, content))
 
-            if declarations:
-                selector_list = [(specificity, None, match)]
-                page_rules.append((rule, selector_list, declarations))
+            ignore_imports = True
+            for page_type in data:
+                specificity = page_type.pop('specificity')
+                page_type = PageType(**page_type)
+                # Use a double lambda to have a closure that holds page_types
+                match = (lambda page_type: lambda page_names: list(
+                    matching_page_types(page_type, names=page_names)))(
+                        page_type)
+                content = tinycss2.parse_declaration_list(rule.content)
+                declarations = list(preprocess_declarations(base_url, content))
 
-            for margin_rule in content:
-                if margin_rule.type != 'at-rule' or (
-                        margin_rule.content is None):
-                    continue
-                declarations = list(preprocess_declarations(
-                    base_url,
-                    tinycss2.parse_declaration_list(margin_rule.content)))
                 if declarations:
-                    selector_list = [(
-                        specificity, '@' + margin_rule.lower_at_keyword,
-                        match)]
-                    page_rules.append(
-                        (margin_rule, selector_list, declarations))
+                    selector_list = [(specificity, None, match)]
+                    page_rules.append((rule, selector_list, declarations))
+
+                for margin_rule in content:
+                    if margin_rule.type != 'at-rule' or (
+                            margin_rule.content is None):
+                        continue
+                    declarations = list(preprocess_declarations(
+                        base_url,
+                        tinycss2.parse_declaration_list(margin_rule.content)))
+                    if declarations:
+                        selector_list = [(
+                            specificity, '@' + margin_rule.lower_at_keyword,
+                            match)]
+                        page_rules.append(
+                            (margin_rule, selector_list, declarations))
 
         elif rule.type == 'at-rule' and rule.lower_at_keyword == 'font-face':
             ignore_imports = True

weasyprint/css/properties.py

@@ -308,6 +308,7 @@
     margin_left
     margin_right
     opacity
+    overflow
     position
     right
     top

weasyprint/css/validation.py

@@ -1585,8 +1585,7 @@ def validate_content_list_token(token):
         elif prototype in (('content', ()), ('content', ('ident',))):
             if not args:
                 return (name, 'text')
-            elif args[0] in ('text', 'after', 'before'):
-                # TODO: first-letter should be allowed here too
+            elif args[0] in ('text', 'after', 'before', 'first-letter'):
                 return (name, args[0])
         elif prototype in (('counter', ('ident',)),
                            ('counters', ('ident', 'string'))):