Add an option to optimize embedded images size

Author: liZe

docs/install.rst

@@ -14,6 +14,7 @@ WeasyPrint |version| depends on:
 * cssselect2_ ≥ 0.1
 * CairoSVG_ ≥ 2.4.0
 * Pyphen_ ≥ 0.9.1
+* Pillow ≥ 4.0.0
 * GDK-PixBuf_ ≥ 2.25.0 [#]_
 
 .. _CPython: http://www.python.org/

setup.cfg

@@ -57,6 +57,7 @@ install_requires =
   cssselect2>=0.1
   CairoSVG>=2.4.0
   Pyphen>=0.9.1
+  Pillow>=4.0.0
 tests_require =
   pytest-runner
   pytest-cov

weasyprint/__init__.py

@@ -134,8 +134,8 @@ def _get_metadata(self):
         return get_html_metadata(self.wrapper_element, self.base_url)
 
     def render(self, stylesheets=None, enable_hinting=False,
-               presentational_hints=False, font_config=None,
-               counter_style=None):
+               presentational_hints=False, optimize_images=False,
+               font_config=None, counter_style=None):
         """Lay out and paginate the document, but do not (yet) export it
         to PDF or PNG.
 
@@ -158,6 +158,8 @@ def render(self, stylesheets=None, enable_hinting=False,
         :type presentational_hints: bool
         :param presentational_hints: Whether HTML presentational hints are
             followed.
+        :type optimize_images: bool
+        :param optimize_images: Try to optimize the size of embedded images.
         :type font_config: :class:`~fonts.FontConfiguration`
         :param font_config: A font configuration handling ``@font-face`` rules.
         :type counter_style: :class:`~css.counters.CounterStyle`
@@ -167,11 +169,11 @@ def render(self, stylesheets=None, enable_hinting=False,
         """
         return Document._render(
             self, stylesheets, enable_hinting, presentational_hints,
-            font_config, counter_style)
+            optimize_images, font_config, counter_style)
 
     def write_pdf(self, target=None, stylesheets=None, zoom=1,
                   attachments=None, presentational_hints=False,
-                  font_config=None, counter_style=None):
+                  optimize_images=False, font_config=None, counter_style=None):
         """Render the document to a PDF file.
 
         This is a shortcut for calling :meth:`render`, then
@@ -199,6 +201,8 @@ def write_pdf(self, target=None, stylesheets=None, zoom=1,
         :type presentational_hints: bool
         :param presentational_hints: Whether HTML presentational hints are
             followed.
+        :type optimize_images: bool
+        :param optimize_images: Try to optimize the size of embedded images.
         :type font_config: :class:`~fonts.FontConfiguration`
         :param font_config: A font configuration handling ``@font-face`` rules.
         :type counter_style: :class:`~css.counters.CounterStyle`
@@ -212,12 +216,12 @@ def write_pdf(self, target=None, stylesheets=None, zoom=1,
         return self.render(
             stylesheets, enable_hinting=False,
             presentational_hints=presentational_hints,
-            font_config=font_config, counter_style=counter_style).write_pdf(
-                target, zoom, attachments)
+            optimize_images=optimize_images, font_config=font_config,
+            counter_style=counter_style).write_pdf(target, zoom, attachments)
 
     def write_image_surface(self, stylesheets=None, resolution=96,
-                            presentational_hints=False, font_config=None,
-                            counter_style=None):
+                            presentational_hints=False, optimize_images=False,
+                            font_config=None, counter_style=None):
         """Render pages vertically on a cairo image surface.
 
         .. versionadded:: 0.17
@@ -242,6 +246,8 @@ def write_image_surface(self, stylesheets=None, resolution=96,
         :type presentational_hints: bool
         :param presentational_hints: Whether HTML presentational hints are
             followed.
+        :type optimize_images: bool
+        :param optimize_images: Try to optimize the size of embedded images.
         :type font_config: :class:`~fonts.FontConfiguration`
         :param font_config: A font configuration handling ``@font-face`` rules.
         :type counter_style: :class:`~css.counters.CounterStyle`
@@ -250,15 +256,16 @@ def write_image_surface(self, stylesheets=None, resolution=96,
 
         """
         surface, _width, _height = (
-            self.render(stylesheets, enable_hinting=True,
-                        presentational_hints=presentational_hints,
-                        font_config=font_config)
+            self.render(
+                stylesheets, enable_hinting=True,
+                presentational_hints=presentational_hints,
+                font_config=font_config, optimize_images=optimize_images)
             .write_image_surface(resolution))
         return surface
 
     def write_png(self, target=None, stylesheets=None, resolution=96,
-                  presentational_hints=False, font_config=None,
-                  counter_style=None):
+                  presentational_hints=False, optimize_images=False,
+                  font_config=None, counter_style=None):
         """Paint the pages vertically to a single PNG image.
 
         There is no decoration around pages other than those specified in CSS
@@ -284,6 +291,8 @@ def write_png(self, target=None, stylesheets=None, resolution=96,
         :type presentational_hints: bool
         :param presentational_hints: Whether HTML presentational hints are
             followed.
+        :type optimize_images: bool
+        :param optimize_images: Try to optimize the size of embedded images.
         :type font_config: :class:`~fonts.FontConfiguration`
         :param font_config: A font configuration handling ``@font-face`` rules.
         :type counter_style: :class:`~css.counters.CounterStyle`
@@ -298,7 +307,8 @@ def write_png(self, target=None, stylesheets=None, resolution=96,
             self.render(
                 stylesheets, enable_hinting=True,
                 presentational_hints=presentational_hints,
-                font_config=font_config, counter_style=counter_style)
+                optimize_images=optimize_images, font_config=font_config,
+                counter_style=counter_style)
             .write_png(target, resolution))
         return png_bytes
 

weasyprint/__main__.py

@@ -87,6 +87,10 @@ def main(argv=None, stdout=None, stdin=None):
         <https://www.w3.org/TR/html/rendering.html\
         #the-css-user-agent-style-sheet-and-presentational-hints>`_.
 
+    .. option:: -o, --optimize-images
+
+        Try to optimize the size of embedded images.
+
     .. option:: -v, --verbose
 
         Show warnings and information messages.
@@ -133,6 +137,8 @@ def main(argv=None, stdout=None, stdin=None):
                              'to attach to the PDF document')
     parser.add_argument('-p', '--presentational-hints', action='store_true',
                         help='Follow HTML presentational hints.')
+    parser.add_argument('-o', '--optimize-images', action='store_true',
+                        help='Try to optimize the size of embedded images.')
     parser.add_argument('-v', '--verbose', action='store_true',
                         help='Show warnings and information messages.')
     parser.add_argument('-d', '--debug', action='store_true',
@@ -175,7 +181,8 @@ def main(argv=None, stdout=None, stdin=None):
 
     kwargs = {
         'stylesheets': args.stylesheet,
-        'presentational_hints': args.presentational_hints}
+        'presentational_hints': args.presentational_hints,
+        'optimize_images': args.optimize_images}
     if args.resolution:
         if format_ == 'png':
             kwargs['resolution'] = args.resolution

weasyprint/document.py

@@ -349,7 +349,8 @@ class Document:
 
     @classmethod
     def _build_layout_context(cls, html, stylesheets, enable_hinting,
-                              presentational_hints=False, font_config=None,
+                              presentational_hints=False,
+                              optimize_images=False, font_config=None,
                               counter_style=None):
         if font_config is None:
             font_config = FontConfiguration()
@@ -368,7 +369,7 @@ def _build_layout_context(cls, html, stylesheets, enable_hinting,
             html, user_stylesheets, presentational_hints, font_config,
             counter_style, page_rules, target_collector)
         get_image_from_uri = functools.partial(
-            original_get_image_from_uri, {}, html.url_fetcher)
+            original_get_image_from_uri, {}, html.url_fetcher, optimize_images)
         PROGRESS_LOGGER.info('Step 4 - Creating formatting structure')
         context = LayoutContext(
             enable_hinting, style_for, get_image_from_uri, font_config,
@@ -377,8 +378,8 @@ def _build_layout_context(cls, html, stylesheets, enable_hinting,
 
     @classmethod
     def _render(cls, html, stylesheets, enable_hinting,
-                presentational_hints=False, font_config=None,
-                counter_style=None):
+                presentational_hints=False, optimize_images=False,
+                font_config=None, counter_style=None):
         if font_config is None:
             font_config = FontConfiguration()
 
@@ -387,7 +388,7 @@ def _render(cls, html, stylesheets, enable_hinting,
 
         context = cls._build_layout_context(
             html, stylesheets, enable_hinting, presentational_hints,
-            font_config, counter_style)
+            optimize_images, font_config, counter_style)
 
         root_box = build_formatting_structure(
             html.etree_element, context.style_for, context.get_image_from_uri,

weasyprint/images.py

@@ -13,6 +13,7 @@
 import cairocffi
 import cairosvg.parser
 import cairosvg.surface
+from PIL import Image
 
 from .layout.percentages import percentage
 from .logger import LOGGER
@@ -173,7 +174,8 @@ def draw(self, context, concrete_width, concrete_height, _image_rendering):
                 'Failed to draw an SVG image at %s : %s', self._base_url, e)
 
 
-def get_image_from_uri(cache, url_fetcher, url, forced_mime_type=None):
+def get_image_from_uri(cache, url_fetcher, optimize_images, url,
+                       forced_mime_type=None):
     """Get a cairo Pattern from an image URI."""
     missing = object()
     image = cache.get(url, missing)
@@ -195,6 +197,7 @@ def get_image_from_uri(cache, url_fetcher, url, forced_mime_type=None):
                 # Try to rely on given mimetype
                 try:
                     if mime_type == 'image/png':
+                        # Cairo already optimizes PNG images, no PIL needed
                         try:
                             surface = cairocffi.ImageSurface.create_from_png(
                                 BytesIO(string))
@@ -216,6 +219,17 @@ def get_image_from_uri(cache, url_fetcher, url, forced_mime_type=None):
                     try:
                         image = SVGImage(string, url, url_fetcher)
                     except BaseException:
+                        if optimize_images:
+                            try:
+                                image = Image.open(BytesIO(string))
+                                output = BytesIO()
+                                image.save(
+                                    output, format=image.format, optimize=True)
+                            except BaseException:
+                                # Optimization did not work, keep the original
+                                pass
+                            else:
+                                string = output.getvalue()
                         try:
                             surface, format_name = (
                                 pixbuf.decode_to_image_surface(string))