Fix images embedded from multiple pages

Fix #1486.

tests/test_pdf.py

@@ -14,6 +14,8 @@ from codecs import BOM_UTF16_BE
 
 import pytest
 from weasyprint import Attachment
+from weasyprint.document import Document, DocumentMetadata
+from weasyprint.text.fonts import FontConfiguration
 from weasyprint.urls import path2url
 
 from .testing_utils import (
@@ -505,6 +507,21 @@ def test_embed_image_once():
         ''').write_pdf().count(b'/Filter /DCTDecode') == 1
 
 
+@assert_no_logs
+def test_embed_images_from_pages():
+    page1, = FakeHTML(
+        base_url=resource_filename('dummy.html'),
+        string='<img src="blue.jpg">').render().pages
+    page2, = FakeHTML(
+        base_url=resource_filename('dummy.html'),
+        string='<img src="not-optimized.jpg">').render().pages
+    document = Document(
+        (page1, page2), metadata=DocumentMetadata(),
+        font_config=FontConfiguration(), url_fetcher=None,
+        optimize_size=()).write_pdf()
+    assert document.count(b'/Filter /DCTDecode') == 2
+
+
 @assert_no_logs
 def test_document_info():
     pdf = FakeHTML(string='''

weasyprint/images.py

@@ -96,10 +96,8 @@ class SVGImage:
 def get_image_from_uri(cache, url_fetcher, optimize_size, url,
                        forced_mime_type=None, context=None):
     """Get an Image instance from an image URI."""
-    missing = object()
-    image = cache.get(url, missing)
-    if image is not missing:
-        return image
+    if url in cache:
+        return cache[url]
 
     try:
         with fetch(url_fetcher, url) as result:
@@ -137,7 +135,7 @@ def get_image_from_uri(cache, url_fetcher, optimize_size, url,
                             raster_exception)
                 else:
                     # Store image id to enable cache in Stream.add_image
-                    image_id = len(cache)
+                    image_id = hash(url)
                     image = RasterImage(pillow_image, image_id, optimize_size)
 
     except (URLFetchingError, ImageLoadingError) as exception: