Process whitespaces as soon as box children are set

This early process enables properties with <content-list> values (string-set, bookmark-label, etc.) to be defined using whitespace-processed strings. Fix #607.
Kozea · Oct 13, 2021 · e5beff6 · e5beff6
1 parent 8ff7ed9
commit e5beff6
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 7 deletions.
diff --git a/tests/test_boxes.py b/tests/test_boxes.py
@@ -64,7 +64,7 @@ def test_inline_in_block_1():
                     ('div', 'Text', 'Hello, '),
                     ('em', 'Inline', [
                         ('em', 'Text', 'World')]),
-                    ('div', 'Text', '!\n')])]),
+                    ('div', 'Text', '! ')])]),
             ('p', 'Block', [
                 ('p', 'Line', [
                     ('p', 'Text', 'Lipsum.')])])])]
@@ -86,7 +86,7 @@ def test_inline_in_block_2():
                     ('div', 'Text', 'Hello, '),
                     ('em', 'Inline', [
                         ('em', 'Text', 'World')]),
-                    ('div', 'Text', '!\n')])])])]
+                    ('div', 'Text', '! ')])])])]
     box = parse(source)
     box = build.inline_in_block(box)
     assert_tree(box, expected)
@@ -151,8 +151,7 @@ def test_block_in_inline():
                             ('span', 'Block', [  # This block is "pulled up"
                                 ('span', 'Line', [
                                     ('span', 'Text', 'sit')])]),
-                            # No whitespace processing here.
-                            ('strong', 'Text', '\n      '),
+                            ('strong', 'Text', ' '),
                             ('span', 'Block', [  # This block is "pulled up"
                                 ('span', 'Line', [
                                     ('span', 'Text', 'amet,')])])]),
@@ -180,8 +179,7 @@ def test_block_in_inline():
                     ('p', 'Line', [
                         ('em', 'Inline', [
                             ('strong', 'Inline', [
-                                # Whitespace processing not done yet.
-                                ('strong', 'Text', '\n      ')])])])]),
+                                ('strong', 'Text', ' ')])])])]),
                 ('span', 'Block', [
                     ('span', 'Line', [
                         ('span', 'Text', 'amet,')])]),

diff --git a/tests/test_pdf.py b/tests/test_pdf.py
@@ -279,6 +279,25 @@ def test_bookmarks_13():
     assert re.findall(b'/Title \\((.*)\\)', pdf) == [b'a']
 
 
+@assert_no_logs
+def test_bookmarks_14():
+    pdf = FakeHTML(string='''
+      <h1>a</h1>
+      <h1> b c d </h1>
+      <h1> e
+             f </h1>
+      <h1> g <span> h </span> i </h1>
+    ''').write_pdf()
+    # a
+    # |_ b
+    # |_ c
+    # L_ d
+    # e
+    assert re.findall(b'/Count ([0-9-]*)', pdf)[-1] == b'4'
+    assert re.findall(b'/Title \\((.*)\\)', pdf) == [
+        b'a', b'b c d', b'e f', b'g h i']
+
+
 @assert_no_logs
 def test_links_none():
     pdf = FakeHTML(string='<body>').write_pdf()

diff --git a/weasyprint/formatting_structure/build.py b/weasyprint/formatting_structure/build.py
@@ -72,7 +72,6 @@ def root_style_for(element, pseudo_type=None):
 
     box.is_for_root_element = True
     # If this is changed, maybe update weasy.layout.page.make_margin_boxes()
-    process_whitespace(box)
     box = anonymous_table_boxes(box)
     box = flex_boxes(box)
     box = inline_in_block(box)
@@ -190,6 +189,7 @@ def element_to_box(element, style_for, get_image_from_uri, base_url,
             counter_values.pop(name)
 
     box.children = children
+    process_whitespace(box)
     # calculate string-set and bookmark-label
     set_content_lists(
         element, box, style, counter_values, target_collector, counter_style)