Adding annotations to the PDF to link back its content to its source.

tests/test_api.py

@@ -127,8 +127,10 @@ def _round_meta(pages):
     """Eliminate errors of floating point arithmetic for metadata."""
     for page in pages:
         anchors = page.anchors
-        for anchor_name, (pos_x, pos_y) in anchors.items():
-            anchors[anchor_name] = round(pos_x, 6), round(pos_y, 6)
+        for anchor_name, (x1, y1, x2, y2) in anchors.items():
+            anchors[anchor_name] = (
+                round(x1, 6), round(y1, 6),
+                round(x2, 6), round(y2, 6))
         links = page.links
         for i, link in enumerate(links):
             link_type, target, rectangle, box = link
@@ -884,8 +886,8 @@ def test_links_1():
         ],
         [('internal', 'hello', (0, 0, 200, 30))],
     ], [
-        {'hello': (0, 200)},
-        {'lipsum': (0, 0)}
+        {'hello': (0, 200, 200, 290)},
+        {'lipsum': (0, 0, 200, 90)}
     ], [
         (
             [
@@ -966,7 +968,7 @@ def test_links_6():
         ''', [[
             ('internal', 'lipsum', (5, 10, 195, 10)),
             ('external', 'https://weasyprint.org/', (0, 10, 200, 10))]],
-        [{'lipsum': (5, 10)}],
+        [{'lipsum': (5, 10, 195, 10)}],
         [([('internal', 'lipsum', (5, 10, 195, 10)),
            ('external', 'https://weasyprint.org/', (0, 10, 200, 10))],
           [('lipsum', 5, 10)])],
@@ -982,7 +984,7 @@ def test_links_7():
                         margin: 10px 5px" id="lipsum">
         ''',
         [[('internal', 'lipsum', (5, 10, 195, 10))]],
-        [{'lipsum': (5, 10)}],
+        [{'lipsum': (5, 10, 195, 10)}],
         [([('internal', 'lipsum', (5, 10, 195, 10))], [('lipsum', 5, 10)])],
         base_url=None)
 
@@ -998,7 +1000,7 @@ def test_links_8():
         ''',
         [[('internal', 'lipsum', (0, 0, 200, 15)),
           ('internal', 'missing', (0, 15, 200, 30))]],
-        [{'lipsum': (0, 15)}],
+        [{'lipsum': (0, 15, 200, 30)}],
         [([('internal', 'lipsum', (0, 0, 200, 15))], [('lipsum', 0, 15)])],
         base_url=None,
         warnings=[
@@ -1014,7 +1016,7 @@ def test_links_9():
                 transform: rotate(90deg) scale(2)">
         ''',
         [[('internal', 'lipsum', (30, 10, 70, 210))]],
-        [{'lipsum': (70, 10)}],
+        [{'lipsum': (70, 10, 30, 210)}],
         [([('internal', 'lipsum', (30, 10, 70, 210))], [('lipsum', 70, 10)])],
         round=True)
 

weasyprint/anchors.py

@@ -113,13 +113,17 @@ def gather_anchors(box, anchors, links, bookmarks, forms, parent_matrix=None,
             links.append((link_type, target, rectangle, box))
         if is_input:
             forms[parent_form].append((box.element, box.style, rectangle))
-        if matrix and (has_bookmark or has_anchor):
-            pos_x, pos_y = matrix.transform_point(pos_x, pos_y)
         if has_bookmark:
+            if matrix:
+                pos_x, pos_y = matrix.transform_point(pos_x, pos_y)
             bookmark = (bookmark_level, bookmark_label, (pos_x, pos_y), state)
             bookmarks.append(bookmark)
         if has_anchor:
-            anchors[anchor_name] = pos_x, pos_y
+            pos_x1, pos_y1, pos_x2, pos_y2 = pos_x, pos_y, pos_x + width, pos_y + height
+            if matrix:
+                pos_x1, pos_y1 = matrix.transform_point(pos_x1, pos_y1)
+                pos_x2, pos_y2 = matrix.transform_point(pos_x2, pos_y2)
+            anchors[anchor_name] = (pos_x1, pos_y1, pos_x2, pos_y2)
 
     for child in box.all_children():
         gather_anchors(child, anchors, links, bookmarks, forms, matrix, parent_form)

weasyprint/pdf/__init__.py

@@ -8,7 +8,7 @@
 from ..html import W3C_DATE_RE
 from ..logger import LOGGER, PROGRESS_LOGGER
 from ..matrix import Matrix
-from . import pdfa, pdfua
+from . import debug, pdfa, pdfua
 from .fonts import build_fonts_dictionary
 from .stream import Stream
 
@@ -17,7 +17,7 @@
     write_pdf_attachment)
 
 VARIANTS = {
-    name: data for variants in (pdfa.VARIANTS, pdfua.VARIANTS)
+    name: data for variants in (pdfa.VARIANTS, pdfua.VARIANTS, debug.VARIANTS)
     for (name, data) in variants.items()}
 
 

weasyprint/pdf/anchors.py

@@ -406,7 +406,7 @@ def resolve_links(pages):
     paged_anchors = []
     for i, page in enumerate(pages):
         paged_anchors.append([])
-        for anchor_name, (point_x, point_y) in page.anchors.items():
+        for anchor_name, (point_x, point_y, _, _) in page.anchors.items():
             if anchor_name not in anchors:
                 paged_anchors[-1].append((anchor_name, point_x, point_y))
                 anchors.add(anchor_name)

weasyprint/pdf/debug.py

@@ -0,0 +1,43 @@
+"""PDF generation with debug information."""
+
+import pydyf
+
+from ..matrix import Matrix
+
+
+def debug(pdf, metadata, document, page_streams, attachments, compress):
+    """Set debug PDF metadata."""
+
+    # Add links on ids.
+    pages = zip(pdf.pages['Kids'][::3], document.pages, page_streams)
+    for pdf_page_number, document_page, stream in pages:
+        if not document_page.anchors:
+            continue
+
+        page = pdf.objects[pdf_page_number]
+        if 'Annots' not in page:
+            page['Annots'] = pydyf.Array()
+
+        for id, (x1, y1, x2, y2) in document_page.anchors.items():
+            # TODO: handle zoom correctly.
+            matrix = Matrix(0.75, 0, 0, 0.75) @ stream.ctm
+            x1, y1 = matrix.transform_point(x1, y1)
+            x2, y2 = matrix.transform_point(x2, y2)
+            annotation = pydyf.Dictionary({
+                'Type': '/Annot',
+                'Subtype': '/Link',
+                'Rect': pydyf.Array([x1, y1, x2, y2]),
+                'BS': pydyf.Dictionary({'W': 0}),
+                'P': page.reference,
+                'T': pydyf.String(id),  # id added as metadata
+            })
+
+            # The next line makes all of this relevent to use
+            # with PDFjs
+            annotation['Dest'] = pydyf.String(id)
+
+            pdf.add_object(annotation)
+            page['Annots'].append(annotation.reference)
+
+
+VARIANTS = {'debug': (debug, {})}