Embed JPEG-encoded images in PDF. Fix #6

If an image is in JPEG format, embed it as-is in the PDF output.
This often results in smaller PDF file size.

(The image is still decoded however,
so there is no rendering speed improvement.)

weasyprint/document.py

@@ -97,6 +97,7 @@ class _TaggedTuple(tuple):
 
     """
 
+
 def _get_metadata(box, bookmarks, links, anchors, matrix):
     bookmark_label = box.bookmark_label
     bookmark_level = box.bookmark_level
@@ -128,6 +129,7 @@ def _get_metadata(box, bookmarks, links, anchors, matrix):
         if has_anchor:
             anchors[anchor_name] = pos_x, pos_y
 
+
 def _prepare(box, bookmarks, links, anchors, matrix):
     transform = _get_matrix(box)
     if transform:

weasyprint/images.py

@@ -27,7 +27,7 @@ from .text import USING_INTROSPECTION
 # Do not try to import PyGObject 3 if we already have PyGTK
 # that tends to segfault.
 if not USING_INTROSPECTION:
-    # Use PyGObject introspection
+    # Use PyGTK
     try:
         from gtk import gdk
         from gtk.gdk import PixbufLoader
@@ -38,7 +38,7 @@ if not USING_INTROSPECTION:
     else:
         def gdkpixbuf_loader(file_obj, string):
             """Load raster images with gdk-pixbuf through PyGTK."""
-            pixbuf = get_pixbuf(file_obj, string)
+            pixbuf, jpeg_data = get_pixbuf(file_obj, string)
             dummy_context = cairo.Context(cairo.ImageSurface(
                 cairo.FORMAT_ARGB32, 1, 1))
             gdk.CairoContext(dummy_context).set_source_pixbuf(pixbuf, 0, 0)
@@ -48,7 +48,14 @@ if not USING_INTROSPECTION:
             # It is therefore not thread-safe and state must be reset
             # before any use.
             get_pattern = dummy_context.get_source
+            if cairo.version_info >= (1, 10, 0):
+                add_jpeg_data(get_pattern().get_surface(), jpeg_data)
             return get_pattern, pixbuf.get_width(), pixbuf.get_height()
+
+        def pixbuf_format(loader):
+            format_ = loader.get_format()
+            if format_:
+                return format_['name']
 else:
     # Use PyGObject introspection
     try:
@@ -58,6 +65,11 @@ else:
         def gdkpixbuf_loader(file_obj, string, pixbuf_error=exception):
             raise pixbuf_error
     else:
+        def pixbuf_format(loader):
+            format_ = loader.get_format()
+            if format_:
+                return format_.get_name()
+
         PIXBUF_VERSION = (GdkPixbuf.PIXBUF_MAJOR,
                           GdkPixbuf.PIXBUF_MINOR,
                           GdkPixbuf.PIXBUF_MICRO)
@@ -75,7 +87,7 @@ else:
                 and Gdk.
 
                 """
-                pixbuf = get_pixbuf(file_obj, string)
+                pixbuf, jpeg_data = get_pixbuf(file_obj, string)
                 dummy_context = cairo.Context(cairo.ImageSurface(
                     cairo.FORMAT_ARGB32, 1, 1))
                 Gdk.cairo_set_source_pixbuf(dummy_context, pixbuf, 0, 0)
@@ -85,6 +97,8 @@ else:
                 # It is therefore not thread-safe and state must be reset
                 # before any use.
                 get_pattern = dummy_context.get_source
+                if cairo.version_info >= (1, 10, 0):
+                    add_jpeg_data(get_pattern().get_surface(), jpeg_data)
                 return get_pattern, pixbuf.get_width(), pixbuf.get_height()
 
         except ImportError:
@@ -94,38 +108,40 @@ else:
                 without Gdk and going through PNG.
 
                 """
-                pixbuf = get_pixbuf(file_obj, string)
+                pixbuf, jpeg_data = get_pixbuf(file_obj, string)
                 _, png = pixbuf.save_to_bufferv('png', ['compression'], ['0'])
-                return cairo_png_loader(None, png)
+                return cairo_png_loader(None, png, jpeg_data)
 
 
 def get_pixbuf(file_obj=None, string=None, chunck_size=16 * 1024):
     """Create a Pixbuf object."""
+    if file_obj:
+        string = file_obj.read()
+    if not string:
+        raise ValueError('Could not load image: empty content')
     loader = PixbufLoader()
     try:
-        if file_obj:
-            while 1:
-                chunck = file_obj.read(chunck_size)
-                if not chunck:
-                    break
-                loader.write(chunck)
-        elif string:
         loader.write(string)
-        else:
-            raise ValueError('Could not load image: empty content')
     finally:
         # Pixbuf is really unhappy if we don’t do this:
         loader.close()
-    return loader.get_pixbuf()
+    jpeg_data = string if pixbuf_format(loader) == 'jpeg' else None
+    return loader.get_pixbuf(), jpeg_data
 
 
-def cairo_png_loader(file_obj, string):
+def cairo_png_loader(file_obj, string, jpeg_data=None):
     """Return a cairo Surface from a PNG byte stream."""
     surface = cairo.ImageSurface.create_from_png(file_obj or BytesIO(string))
+    add_jpeg_data(surface, jpeg_data)
     get_pattern = lambda: cairo.SurfacePattern(surface)
     return get_pattern, surface.get_width(), surface.get_height()
 
 
+def add_jpeg_data(surface, jpeg_data):
+    if jpeg_data and hasattr(surface, 'set_mime_data'):
+        surface.set_mime_data('image/jpeg', jpeg_data)
+
+
 def cairosvg_loader(file_obj, string, uri):
     """Return a cairo Surface from a SVG byte stream.
 

weasyprint/tests/test_draw.py

@@ -151,7 +151,7 @@ def document_to_pixels(document, name, expected_width, expected_height):
 
 
 def png_to_pixels(png_bytes, width, height):
-    pixbuf = get_pixbuf(string=png_bytes)
+    pixbuf, _ = get_pixbuf(string=png_bytes)
     assert (pixbuf.get_width(), pixbuf.get_height()) == (width, height)
     if not pixbuf.get_has_alpha():
         pixbuf = pixbuf.add_alpha(False, 0, 0, 0)  # no substitute color

weasyprint/tests/test_pdf.py

@@ -15,6 +15,7 @@ from __future__ import division, unicode_literals
 import io
 
 import cairo
+import pytest
 
 from .. import CSS
 from .. import pdf
@@ -203,7 +204,6 @@ def test_bookmarks():
              label='a', target=(0, 75, 1425))]
 
 
-
 @assert_no_logs
 def test_links():
     links = get_links('<body>')
@@ -294,3 +294,15 @@ def test_missing_links():
     assert links == [[('internal', (0, 50, 935), (50, 950, 450, 935))]]
     assert len(logs) == 1
     assert 'WARNING: No anchor #missing for internal URI reference' in logs[0]
+
+
+@assert_no_logs
+def test_jpeg():
+    if not hasattr(cairo.ImageSurface, 'set_mime_data'):
+        pytest.xfail()
+    def render(html):
+        return TestHTML(base_url=resource_filename('dummy.html'),
+                        string=html).write_pdf()
+    assert b'/Filter /DCTDecode' not in render('<img src="pattern.gif">')
+    # JPEG-encoded image, embedded in PDF:
+    assert b'/Filter /DCTDecode' in render('<img src="blue.jpg">')

weasyprint/tests/w3_test_suite/run.py

@@ -84,7 +84,7 @@ def make_test_suite():
             HTML(BASE_URL + name).write_png(
                 png_filename, stylesheets=[PAGE_SIZE_STYLESHEET])
             with open(png_filename, 'rb') as image:
-                raw = get_pixbuf(file_obj=image).get_pixels()
+                raw, _ = get_pixbuf(file_obj=image).get_pixels()
             rendered[name] = raw
         return raw