From 9b0a672acd67c559b8280bf80f7fc97b1d13ce19 Mon Sep 17 00:00:00 2001
From: aschmitz <29508+aschmitz@users.noreply.github.com>
Date: Tue, 1 Feb 2022 18:13:39 -0600
Subject: [PATCH] fix: handle float rounding when extending boxes over page
 breaks

When a box would break over the edge of a page, its height is extended
to the bottom of that page (per
https://www.w3.org/TR/css-break-3/#box-splitting , primarily to allow
backgrounds and borders to continue to the end of the page).

When this happened, sometimes the values that would be calculated for
the height of the extended element would be rounded *over* the
calculated height that remained on the page, forcing the entire
containing box to wrap to the next page.

Rather than trying to carefully manage the order of operations to try to
be safe in IEEE floats for directions, we apply a small "fudge factor":
if an element fits very nearly (within a thousandth of a pixel) into the
remaining space, it is still accepted.
---
 tests/layout/test_page.py  | 34 ++++++++++++++++++++++++++++++++++
 weasyprint/layout/block.py |  8 +++++++-
 2 files changed, 41 insertions(+), 1 deletion(-)

diff --git a/tests/layout/test_page.py b/tests/layout/test_page.py
index 36e9a026f..878112869 100644
--- a/tests/layout/test_page.py
+++ b/tests/layout/test_page.py
@@ -141,6 +141,40 @@ def test_page_breaks(html):
         for page_child in page_children] == [[10, 40], [10, 40], [10]]
 
 
+@assert_no_logs
+def test_page_breaks_box_split():
+    # If floats round the wrong way, a block that gets filled to the end of a
+    # page due to breaking over the page may be forced onto the next page
+    # because it is slightly taller than can fit on the previous page, even if
+    # it wouldn't have been without being filled. These numbers aren't ideal,
+    # but they do seem to trigger the issue.
+    page_1, page_2 = render_pages('''
+      <style>
+        @page { size: 982.4146981627297px; margin: 0 }
+        div { font-size: 5px; height: 200.0123456789px; margin: 0; padding: 0 }
+        figure { margin: 0; padding: 0 }
+      </style>
+      <div>text</div>
+      <div>text</div><!-- no page break here -->
+      <section>
+        <div>line1</div>
+        <div>line2</div><!-- page break here -->
+        <div>line3</div>
+        <div>line4</div>
+      </section>
+    ''')
+    html, = page_1.children
+    body, = html.children
+    assert len(body.children) == 3
+    div1, div2, section = body.children
+    assert len(section.children) == 2
+
+    html, = page_2.children
+    body, = html.children
+    section, = body.children
+    assert len(section.children) == 2
+
+
 @assert_no_logs
 def test_page_breaks_complex_1():
     page_1, page_2, page_3, page_4 = render_pages('''
diff --git a/weasyprint/layout/block.py b/weasyprint/layout/block.py
index 75ad3bb17..64089185b 100644
--- a/weasyprint/layout/block.py
+++ b/weasyprint/layout/block.py
@@ -466,7 +466,13 @@ def _in_flow_layout(context, box, index, child, new_children, page_is_empty,
             new_position_y = (
                 new_child.border_box_y() + new_child.border_height())
 
-            if (new_content_position_y > context.page_bottom - bottom_space and
+            # Use a small fudge factor on this due to box splitting setting the
+            # height of some elements to the remaining height of the page:
+            # https://www.w3.org/TR/css-break-3/#box-splitting
+            # (Occasionally the order of this calculation would otherwise come
+            # out with unequal float values, forcing the box to the next page.)
+            if (new_content_position_y >
+                    context.page_bottom - bottom_space + 0.001 and
                     not page_is_empty_with_no_children):
                 # The child content overflows the page area, display it on the
                 # next page.