Merge pull request #1146 from Tontyna/bookmark-label

Bookmark label

weasyprint/formatting_structure/build.py

@@ -244,6 +244,15 @@ def before_after_to_box(element, pseudo_type, state, style_for,
         target_collector, counter_style))
 
     box.children = children
+
+    # calculate the bookmark-label
+    if style['bookmark_label'] == 'none':
+        box.bookmark_label = ''
+    else:
+        _quote_depth, counter_values, _counter_scopes = state
+        compute_bookmark_label(
+            element, box, style['bookmark_label'], counter_values,
+            target_collector, counter_style)
     return [box]
 
 

weasyprint/layout/__init__.py

@@ -140,23 +140,49 @@ def layout_document(html, root_box, context, max_loops=8):
         if not reloop_content and not reloop_pages:
             break
 
-    # Calculate string-sets and bookmark-label containing page based counters
+    # Calculate string-sets and bookmark-labels containing page based counters
     # when pagination is finished. No need to do that (maybe multiple times) in
     # make_page because they dont create boxes, only appear in MarginBoxes and
     # in the final PDF.
+    # Prevent repetition of bookmarks (see #1145).
+
+    watch_elements = []
+    watch_elements_before = []
+    watch_elements_after = []
     for i, page in enumerate(pages):
         # We need the updated page_counter_values
         resume_at, next_page, right_page, page_state, remake_state = (
             context.page_maker[i + 1])
         page_counter_values = page_state[1]
 
         for child in page.descendants():
+            # Only one bookmark per original box
+            if child.bookmark_label:
+                if child.element_tag.endswith('::before'):
+                    checklist = watch_elements_before
+                elif child.element_tag.endswith('::after'):
+                    checklist = watch_elements_after
+                else:
+                    checklist = watch_elements
+                if child.element in checklist:
+                    child.bookmark_label = ''
+                else:
+                    checklist.append(child.element)
+
             # TODO: remove attribute or set a default value in Box class
             if hasattr(child, 'missing_link'):
                 for (box, css_token), item in (
                         context.target_collector.counter_lookup_items.items()):
                     if child.missing_link == box and css_token != 'content':
+                        if (css_token == 'bookmark-label' and
+                                not child.bookmark_label):
+                            # don't refill it!
+                            continue
                         item.parse_again(page_counter_values)
+                        # string_set is a pointer, but the bookmark_label is
+                        # just a string: copy it
+                        if css_token == 'bookmark-label':
+                            child.bookmark_label = box.bookmark_label
             # Collect the string_sets in the LayoutContext
             string_sets = child.string_set
             if string_sets and string_sets != 'none':

weasyprint/tests/test_pdf.py

@@ -331,6 +331,76 @@ def test_bookmarks_8():
     assert o2.get_value('Title', '(.*)') == b'(g)'
 
 
+@assert_no_logs
+@requires('cairo', (1, 15, 4))
+def test_bookmarks_9():
+    fileobj = io.BytesIO()
+    FakeHTML(string='''
+      <h1 style="bookmark-label: 'h1 on page ' counter(page)">a</h1>
+    ''').write_pdf(target=fileobj)
+    # h1 on page 1
+    pdf_file = pdf.PDFFile(fileobj)
+    outlines = pdf_file.catalog.get_indirect_dict('Outlines', pdf_file)
+    assert outlines.get_type() == 'Outlines'
+    assert outlines.get_value('Count', '(.*)') == b'-1'
+    o1 = outlines.get_indirect_dict('First', pdf_file)
+    assert o1.get_value('Title', '(.*)') == b'(h1 on page 1)'
+
+
+@assert_no_logs
+@requires('cairo', (1, 15, 4))
+def test_bookmarks_10():
+    fileobj = io.BytesIO()
+    FakeHTML(string='''
+      <style>
+      div:before, div:after {
+         content: '';
+         bookmark-level: 1;
+         bookmark-label: 'x';
+      }
+      </style>
+      <div>a</div>
+    ''').write_pdf(target=fileobj)
+    # x
+    # x
+    pdf_file = pdf.PDFFile(fileobj)
+    outlines = pdf_file.catalog.get_indirect_dict('Outlines', pdf_file)
+    assert outlines.get_type() == 'Outlines'
+    assert outlines.get_value('Count', '(.*)') == b'-2'
+    o1 = outlines.get_indirect_dict('First', pdf_file)
+    assert o1.get_value('Title', '(.*)') == b'(x)'
+    o2 = o1.get_indirect_dict('Next', pdf_file)
+    assert o2.get_value('Title', '(.*)') == b'(x)'
+
+
+@assert_no_logs
+@requires('cairo', (1, 15, 4))
+def test_bookmarks_11():
+    fileobj = io.BytesIO()
+    FakeHTML(string='''
+      <div style="display:inline; white-space:pre;
+       bookmark-level:1; bookmark-label:'a'">
+      a
+      a
+      a
+      </div>
+      <div style="bookmark-level:1; bookmark-label:'b'">
+        <div>b</div>
+        <div style="break-before:always">c</div>
+      </div>
+    ''').write_pdf(target=fileobj)
+    # a
+    # b
+    pdf_file = pdf.PDFFile(fileobj)
+    outlines = pdf_file.catalog.get_indirect_dict('Outlines', pdf_file)
+    assert outlines.get_type() == 'Outlines'
+    assert outlines.get_value('Count', '(.*)') == b'-2'
+    o1 = outlines.get_indirect_dict('First', pdf_file)
+    assert o1.get_value('Title', '(.*)') == b'(a)'
+    o2 = o1.get_indirect_dict('Next', pdf_file)
+    assert o2.get_value('Title', '(.*)') == b'(b)'
+
+
 @assert_no_logs
 def test_links_none():
     fileobj = io.BytesIO()