Cache images to avoid downloading them over and over.

(In particular, page background images were downloaded for each page.)

weasy/document.py

@@ -34,6 +34,7 @@ from .css.computed_values import LENGTHS_TO_PIXELS
 from .formatting_structure.build import build_formatting_structure
 from .layout import layout
 from . import draw
+from .draw import helpers as draw_helpers
 from . import utils
 
 
@@ -67,6 +68,7 @@ class Document(object):
         self._computed_styles = None
         self._formatting_structure = None
         self._pages = None
+        self._image_cache = {}
 
     @property
     def base_url(self):
@@ -142,6 +144,18 @@ class Document(object):
             self._pages = layout(self)
         return self._pages
 
+    def get_image_surface_from_uri(self, uri):
+        if uri in self._image_cache:
+            return self._image_cache[uri]
+        try:
+            surface = draw_helpers.get_image_surface_from_uri(uri)
+        # TODO: have a more specific list of exception for network errors
+        # and PNG parsing errors.
+        except Exception:
+            surface = None
+        self._image_cache[uri] = surface
+        return surface
+
 
 class PNGDocument(Document):
     """PNG output document."""

weasy/draw/helpers.py

@@ -161,11 +161,8 @@ def draw_background(context, box, clip=True):
         if bg_image.type != 'URI':
             return
 
-        try:
-            surface = get_image_surface_from_uri(bg_image.absoluteUri)
-        # TODO: have a more specific list of exception for network errors
-        # and PNG parsing errors.
-        except Exception:
+        surface = box.document.get_image_surface_from_uri(bg_image.absoluteUri)
+        if surface is None:
             return
 
         image_width = surface.get_width()

weasy/formatting_structure/build.py

@@ -118,14 +118,20 @@ def add_list_marker(box):
 
     """
     image = box.style.list_style_image
-    if get_single_keyword(image) == 'none':
+    if get_single_keyword(image) != 'none':
+        # surface may be None here too, in case the image is not available.
+        surface = box.document.get_image_surface_from_uri(image[0].absoluteUri)
+    else:
+        surface = None
+
+    if surface is None:
         type_ = get_single_keyword(box.style.list_style_type)
         if type_ == 'none':
             return
         marker = GLYPH_LIST_MARKERS[type_]
         marker_box = boxes.TextBox(box.document, box.element, marker)
     else:
-        replacement = html.ImageReplacement(image[0].absoluteUri)
+        replacement = html.ImageReplacement(surface)
         marker_box = boxes.ImageMarkerBox(
             box.document, box.element, replacement)
 

weasy/html/__init__.py

@@ -117,11 +117,11 @@ def handle_img(document, element):
     src = get_url_attribute(element, 'src')
     alt = element.get('alt')
     if src:
-        try:
-            replacement = ImageReplacement(src)
-        # TODO: have a more specific list of exception for network errors
-        # and image parsing errors.
-        except Exception:
+        surface = document.get_image_surface_from_uri(src)
+        if surface is not None:
+            replacement = ImageReplacement(surface)
+            return make_replaced_box(document, element, replacement)
+        else:
             # Invalid image, use the alt-text.
             if alt:
                 return make_text_box(document, element, alt)
@@ -133,8 +133,6 @@ def handle_img(document, element):
                 # TODO: find some indicator that an image is missing.
                 # For now, just remove the image.
                 return None
-        else:
-            return make_replaced_box(document, element, replacement)
     else:
         if alt:
             return make_text_box(document, element, alt)
@@ -172,11 +170,11 @@ class Replacement(object):
 class ImageReplacement(Replacement):
     """Replaced ``<img>`` element.
 
-    :param image_uri: uri where to get the image.
+    :param surface: a cairo :class:`ImageSurface` object.
 
     """
-    def __init__(self, image_uri):
-        self.surface = get_image_surface_from_uri(image_uri)
+    def __init__(self, surface):
+        self.surface = surface
 
     def intrinsic_width(self):
         if self.surface: