From 844dffa2f190a79c0f5986be4446364cf986c4ec Mon Sep 17 00:00:00 2001
From: Guillaume Ayoub <guillaume@courtbouillon.org>
Date: Sat, 3 Aug 2024 11:05:23 +0200
Subject: [PATCH] Add debug PDF format with link annotations on elements with
 id

---
 weasyprint/anchors.py      | 11 ++++++++---
 weasyprint/pdf/__init__.py |  4 ++--
 weasyprint/pdf/debug.py    | 38 ++++++++++++++++++++++++++++++++++++++
 3 files changed, 48 insertions(+), 5 deletions(-)
 create mode 100644 weasyprint/pdf/debug.py

diff --git a/weasyprint/anchors.py b/weasyprint/anchors.py
index c701577a5..7faf8bded 100644
--- a/weasyprint/anchors.py
+++ b/weasyprint/anchors.py
@@ -113,13 +113,18 @@ 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, width, height)
+            pos_x1, pos_y1, pos_x2, pos_y2 = pos_x, pos_y, pos_x + width, pos_y + height
+            print('pos', pos_x1, pos_y1, pos_x2, pos_y2)
+            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)
diff --git a/weasyprint/pdf/__init__.py b/weasyprint/pdf/__init__.py
index 15d12e811..66f38ae41 100644
--- a/weasyprint/pdf/__init__.py
+++ b/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()}
 
 
diff --git a/weasyprint/pdf/debug.py b/weasyprint/pdf/debug.py
new file mode 100644
index 000000000..b0758ae90
--- /dev/null
+++ b/weasyprint/pdf/debug.py
@@ -0,0 +1,38 @@
+"""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
+            })
+            pdf.add_object(annotation)
+            page['Annots'].append(annotation.reference)
+
+
+VARIANTS = {'debug': (debug, {})}