weasyprint 67.0__py3-none-any.whl → 68.1__py3-none-any.whl

weasyprint/layout/block.py

@@ -301,7 +301,7 @@ def _out_of_flow_layout(context, box, index, child, new_children,
     return stop, resume_at, new_child, out_of_flow_resume_at
 
 
-def _break_line(context, box, line, new_children, next_lines, page_is_empty, index,
+def _break_line(context, box, line, new_children, needed, page_is_empty, index,
                 skip_stack, resume_at, absolute_boxes, fixed_boxes):
     """Break line where allowed by orphans and widows.
 
@@ -316,7 +316,6 @@ def _break_line(context, box, line, new_children, next_lines, page_is_empty, ind
         return True, False, resume_at
     # How many lines we need on the next page to satisfy widows
     # -1 for the current line.
-    needed = max(box.style['widows'] - 1 - next_lines, 0)
     if needed > over_orphans and not page_is_empty:
         # Total number of lines < orphans + widows
         remove_placeholders(context, line.children, absolute_boxes, fixed_boxes)
@@ -377,15 +376,21 @@ def _linebox_layout(context, box, index, child, new_children, page_is_empty,
             # If we couldn’t break the line before but can break now, first try to
             # report footnotes and see if we don’t overflow.
             could_break_before = can_break_now = True
-            next_lines = len(tuple(lines_iterator))
+            needed = box.style['widows'] - 1
+            for _ in lines_iterator:
+                needed -= 1
+                # Don’t iterate over all lines as it can be long.
+                if needed == -1:
+                    break
             if len(new_children) + 1 < box.style['orphans']:
                 can_break_now = False
-            elif next_lines < box.style['widows']:
+            elif needed >= 0:
                 can_break_now = False
             if len(new_children) < box.style['orphans']:
                 could_break_before = False
-            elif next_lines + 1 < box.style['widows']:
+            elif needed > 0:
                 could_break_before = False
+            needed = max(0, needed)
             report = not context.in_column and can_break_now and not could_break_before
             reported_footnotes = 0
             while report and context.current_page_footnotes:
@@ -397,9 +402,8 @@ def _linebox_layout(context, box, index, child, new_children, page_is_empty,
                     break
             else:
                 abort, stop, resume_at = _break_line(
-                    context, box, line, new_children, next_lines,
-                    page_is_empty, index, skip_stack, resume_at, absolute_boxes,
-                    fixed_boxes)
+                    context, box, line, new_children, needed, page_is_empty, index,
+                    skip_stack, resume_at, absolute_boxes, fixed_boxes)
 
             # Revert reported footnotes, as they’ve been reported starting from the last
             # one.
@@ -414,8 +418,7 @@ def _linebox_layout(context, box, index, child, new_children, page_is_empty,
         # "When an unforced page break occurs here, both the adjoining
         #  ‘margin-top’ and ‘margin-bottom’ are set to zero."
         # See issue #115.
-        elif page_is_empty and context.overflows_page(
-                bottom_space, new_position_y):
+        elif page_is_empty and context.overflows_page(bottom_space, new_position_y):
             # Remove the top border when a page is empty and the box is
             # too high to be drawn in one page
             new_position_y -= box.margin_top
@@ -433,8 +436,7 @@ def _linebox_layout(context, box, index, child, new_children, page_is_empty,
                 overflow = (
                     overflow or
                     context.reported_footnotes or
-                    context.overflows_page(
-                        bottom_space, new_position_y + offset_y))
+                    context.overflows_page(bottom_space, new_position_y + offset_y))
                 if overflow:
                     context.report_footnote(footnote)
                     # If we've put other content on this page, then we may want
@@ -443,11 +445,15 @@ def _linebox_layout(context, box, index, child, new_children, page_is_empty,
                     # even try.
                     if new_children or not page_is_empty:
                         if footnote.style['footnote_policy'] == 'line':
-                            next_lines = len(tuple(lines_iterator))
+                            if needed := box.style['widows'] - 1:
+                                for _ in lines_iterator:
+                                    needed -= 1
+                                    # Don’t iterate over all lines as it can be long.
+                                    if needed == 0:
+                                        break
                             abort, stop, resume_at = _break_line(
-                                context, box, line, new_children,
-                                next_lines, page_is_empty, index,
-                                skip_stack, resume_at, absolute_boxes,
+                                context, box, line, new_children, needed, page_is_empty,
+                                index, skip_stack, resume_at, absolute_boxes,
                                 fixed_boxes)
                             break_linebox = True
                             break

weasyprint/layout/grid.py

@@ -1123,6 +1123,7 @@ def grid_layout(context, box, bottom_space, skip_stack, containing_block,
     # Find resume_at row.
     this_page_children = []
     resume_row = None
+    extra_skip_height = 0
     if skip_stack:
         from .block import block_level_layout
         first_skip_row = min(skip_stack)
@@ -1130,7 +1131,6 @@ def grid_layout(context, box, bottom_space, skip_stack, containing_block,
         skip_height = (
             sum(size for size, _ in rows_sizes[:last_skip_row]) +
             (len(rows_sizes[:last_skip_row]) - 1) * row_gap)
-        extra_skip_height = 0
         for child, (x, y, width, height) in children_positions.items():
             if (advancement := box.advancements.get((x, y))) is None:
                 continue
@@ -1271,13 +1271,16 @@ def grid_layout(context, box, bottom_space, skip_stack, containing_block,
         if child.margin_left == 'auto':
             child.margin_left = 0
 
-        child_width = width - (
-            child.margin_left + child.border_left_width + child.padding_left +
-            child.margin_right + child.border_right_width + child.padding_right)
-        child_height = height - (
-            child.margin_bottom + child.border_bottom_width + child.padding_bottom)
+        child_border_width = width - (child.margin_left + child.margin_right)
+        child_content_width = child_border_width - (
+                child.border_left_width + child.padding_left +
+                child.border_right_width + child.padding_right)
+        child_border_height = height - child.margin_bottom
+        child_content_height = child_border_height - (
+            child.border_bottom_width + child.padding_bottom)
         if not child_skip_stack or child.style['box_decoration_break'] == 'clone':
-            child_height -= (
+            child_border_height -= child.margin_top
+            child_content_height -= (
                 child.margin_top + child.border_top_width + child.padding_top)
 
         justify_self = set(child.style['justify_self'])
@@ -1286,6 +1289,9 @@ def grid_layout(context, box, bottom_space, skip_stack, containing_block,
         if justify_self & {'normal', 'stretch'}:
             if child.style['width'] == 'auto':
                 child.style = child.style.copy()
+                child_width = (
+                    child_content_width if child.style['box_sizing'] == 'content-box'
+                    else child_border_width)
                 child.style['width'] = Dimension(child_width, 'px')
         align_self = set(child.style['align_self'])
         if align_self & {'auto'}:
@@ -1293,6 +1299,9 @@ def grid_layout(context, box, bottom_space, skip_stack, containing_block,
         if align_self & {'normal', 'stretch'}:
             if child.style['height'] == 'auto':
                 child.style = child.style.copy()
+                child_height = (
+                    child_content_height if child.style['box_sizing'] == 'content-box'
+                    else child_border_height)
                 child.style['height'] = Dimension(child_height, 'px')
 
         # TODO: Find a better solution for the layout.
@@ -1338,11 +1347,11 @@ def grid_layout(context, box, bottom_space, skip_stack, containing_block,
             continue
 
         if justify_self & {'normal', 'stretch'}:
-            new_child.width = max(child_width, new_child.width)
+            new_child.width = max(child_content_width, new_child.width)
         else:
             if new_child.style['width'] == 'auto':
                 new_child.width = max_content_width(context, new_child, outer=False)
-            diff = child_width - new_child.width
+            diff = child_content_width - new_child.width
             if justify_self & {'center'}:
                 new_child.translate(diff / 2, 0)
             elif justify_self & {'right', 'end', 'flex-end', 'self-end'}:
@@ -1350,9 +1359,9 @@ def grid_layout(context, box, bottom_space, skip_stack, containing_block,
 
         # TODO: Apply auto margins.
         if align_self & {'normal', 'stretch'}:
-            new_child.height = max(child_height, new_child.height)
+            new_child.height = max(child_content_height, new_child.height)
         else:
-            diff = child_height - new_child.height
+            diff = child_content_height - new_child.height
             if align_self & {'center'}:
                 new_child.translate(0, diff / 2)
             elif align_self & {'end', 'flex-end', 'self-end'}:
@@ -1368,8 +1377,9 @@ def grid_layout(context, box, bottom_space, skip_stack, containing_block,
         context.finish_block_formatting_context(box)
         return None, None, {'break': 'any', 'page': None}, [], False
 
-    old_advancements = box.advancements or {}
-    advancements = box.advancements = {}
+    old_advancements = box.advancements.copy()
+    advancements = box.advancements
+    advancements.clear()
     box = box.copy_with_children(new_children)
     if isinstance(box, boxes.InlineGridBox):
         # TODO: Synthetize a real baseline value.
@@ -1420,7 +1430,8 @@ def grid_layout(context, box, bottom_space, skip_stack, containing_block,
             else:
                 # Child fully drawn, save the extra height added to reach the bottom of
                 # the page to substract it from the advancements.
-                extra_advancement = max(extra_advancement, child.height - child_height)
+                extra_advancement = max(
+                    extra_advancement, child.height - child_content_height)
 
     # Substract the extra height added to reach the bottom of the page from all the
     # advancements.

weasyprint/layout/page.py

@@ -389,7 +389,7 @@ def make_margin_boxes(context, page, state):
         if box.is_generated:
             # @margins mustn't manipulate page-context counters
             margin_state = copy.deepcopy(state)
-            quote_depth, counter_values, counter_scopes = margin_state
+            quote_depth, counter_values, counter_scopes, _page_groups = margin_state
             # TODO: check this, probably useless
             counter_scopes.append(set())
             build.update_counters(margin_state, box.style)
@@ -901,7 +901,7 @@ def _update_page_groups(page_groups, resume_at, next_page, root_box, blank):
             return next_page['page']
 
 
-def remake_page(index, page_groups, context, root_box, html):
+def remake_page(index, context, root_box, html):
     """Return one laid out page without margin boxes.
 
     Start with the initial values from ``context.page_maker[index]``.
@@ -932,6 +932,7 @@ def remake_page(index, page_groups, context, root_box, html):
         (next_page_side == 'right' and not right_page) or
         (context.reported_footnotes and resume_at is None))
     side = 'right' if right_page else 'left'
+    page_groups = page_state[3]
     name = _update_page_groups(page_groups, resume_at, next_page, root_box, blank)
     groups = tuple((name, index) for name, index, _ in page_groups)
     page_type = PageType(side, blank, name, index, groups)
@@ -990,7 +991,6 @@ def make_all_pages(context, root_box, html, pages):
     """
     i = 0
     reported_footnotes = None
-    page_groups = []
     while True:
         remake_state = context.page_maker[i][-1]
         if (len(pages) == 0 or
@@ -1002,7 +1002,7 @@ def make_all_pages(context, root_box, html, pages):
             remake_state['pages_wanted'] = False
             remake_state['anchors'] = []
             remake_state['content_lookups'] = []
-            page, resume_at = remake_page(i, page_groups, context, root_box, html)
+            page, resume_at = remake_page(i, context, root_box, html)
             reported_footnotes = context.reported_footnotes
             yield page
         else:

weasyprint/layout/preferred.py

@@ -101,9 +101,32 @@ def _block_content_width(context, box, function, outer):
             function(context, child, outer=True) for child in box.children
             if not child.is_absolutely_positioned()]
         width = max(children_widths) if children_widths else 0
+    elif box.style['box_sizing'] == 'content-box':
+        width = width.value
     else:
-        assert width.unit.lower() == 'px'
         width = width.value
+        percentages = 0
+
+        for value in ('padding_left', 'padding_right'):
+            style_value = box.style[value]
+            if style_value != 'auto' and not check_math(style_value):
+                if style_value.unit.lower() == 'px':
+                    width -= style_value.value
+                else:
+                    assert style_value.unit == '%'
+                    percentages += style_value.value
+
+        # Same as margin_width().
+        collapse = box.style['border_collapse'] == 'collapse'
+        if collapse and hasattr(box, 'border_left_width'):
+            width -= box.border_left_width
+        else:
+            width -= box.style['border_left_width']
+        if collapse and hasattr(box, 'border_right_width'):
+            width -= box.border_right_width
+        else:
+            width -= box.style['border_right_width']
+        width = (100 - min(100, percentages)) * max(0, width) / 100
 
     return adjust(box, outer, width)
 
@@ -152,7 +175,7 @@ def margin_width(box, width, left=True, right=True):
         (['margin_right', 'padding_right'] if right else [])
     ):
         style_value = box.style[value]
-        if style_value != 'auto':
+        if style_value != 'auto' and not check_math(style_value):
             if style_value.unit.lower() == 'px':
                 width += style_value.value
             else:
@@ -240,7 +263,7 @@ def inline_max_content_width(context, box, outer=True, is_line_start=False):
 def column_group_content_width(context, box):
     """Return the *-content width for a ``TableColumnGroupBox``."""
     width = box.style['width']
-    if width == 'auto' or width.unit == '%':
+    if width == 'auto' or check_math(width) or width.unit == '%':
         width = 0
     else:
         assert width.unit.lower() == 'px'
@@ -362,7 +385,17 @@ def inline_line_widths(context, box, outer, is_line_start, minimum, skip_stack=N
             # "By default, there is a break opportunity
             #  both before and after any inline object."
             if minimum:
-                lines = [None, min_content_width(context, child), None]
+                # "For soft wrap opportunities defined by the boundary between two
+                # characters or atomic inlines, the white-space property on the nearest
+                # common ancestor of the two characters controls breaking; which
+                # elements’ line-break, word-break, and overflow-wrap properties control
+                # the determination of soft wrap opportunities at such boundaries is
+                # undefined in this level." We choose to always follow the parent’s
+                # value here, other parts of the line-breaking algorithm do the same.
+                if box.style['white_space'] in ('normal', 'pre-wrap', 'pre-line'):
+                    lines = [None, min_content_width(context, child), None]
+                else:
+                    lines = [min_content_width(context, child)]
             else:
                 lines = [max_content_width(context, child)]
         # The first text line goes on the current line.
@@ -564,21 +597,22 @@ def table_and_columns_preferred_widths(context, box, outer=True):
     # Define constrainedness
     constrainedness = [False for i in range(grid_width)]
     for i in range(grid_width):
-        if (column_groups[i] and column_groups[i].style['width'] != 'auto' and
-                column_groups[i].style['width'].unit != '%'):
-            constrainedness[i] = True
-            continue
-        if (columns[i] and columns[i].style['width'] != 'auto' and
-                columns[i].style['width'].unit != '%'):
-            constrainedness[i] = True
-            continue
-        for cell in zipped_grid[i]:
-            if (cell and cell.colspan == 1 and
-                    cell.style['width'] != 'auto' and
-                    not check_math(cell.style['width']) and
-                    cell.style['width'].unit != '%'):
+        if column_groups[i]:
+            width = column_groups[i].style['width']
+            if width != 'auto' and not check_math(width) and width.unit != '%':
                 constrainedness[i] = True
-                break
+                continue
+        if columns[i]:
+            width = columns[i].style['width']
+            if width != 'auto' and not check_math(width) and width.unit != '%':
+                constrainedness[i] = True
+                continue
+        for cell in zipped_grid[i]:
+            if cell and cell.colspan == 1:
+                width = cell.style['width']
+                if width != 'auto' and not check_math(width) and width.unit != '%':
+                    constrainedness[i] = True
+                    break
 
     intrinsic_percentages = [
         min(percentage, 100 - sum(intrinsic_percentages[:i]))
@@ -646,7 +680,8 @@ def table_and_columns_preferred_widths(context, box, outer=True):
             sum(max_content_widths), large_percentage_contribution,
             *small_percentage_contributions]))
 
-    if table.style['width'] != 'auto' and table.style['width'].unit.lower() == 'px':
+    width = table.style['width']
+    if width != 'auto' and not check_math(width) and width.unit.lower() == 'px':
         # "percentages on the following properties are treated instead as
         # though they were the following: width: auto"
         # https://dbaron.org/css/intrinsic/#outer-intrinsic
@@ -681,12 +716,16 @@ def replaced_min_content_width(box, outer=True):
     width = box.style['width']
     if width == 'auto':
         height = box.style['height']
-        if height == 'auto' or height.unit == '%':
+        if height == 'auto' or check_math(height) or height.unit == '%':
             height = 'auto'
         else:
             assert height.unit.lower() == 'px'
             height = height.value
-        if box.style['max_width'] != 'auto' and box.style['max_width'].unit == '%':
+        unknown_max_width = (
+            box.style['max_width'] != 'auto' and
+            not check_math(box.style['max_width']) and
+            box.style['max_width'].unit == '%')
+        if unknown_max_width:
             # See https://drafts.csswg.org/css-sizing/#intrinsic-contribution
             width = 0
         else:
@@ -697,7 +736,7 @@ def replaced_min_content_width(box, outer=True):
             width, _ = default_image_sizing(
                 intrinsic_width, intrinsic_height, intrinsic_ratio, 'auto',
                 height, default_width=0, default_height=0)
-    elif box.style['width'].unit == '%':
+    elif check_math(box.style['width']) or box.style['width'].unit == '%':
         # See https://drafts.csswg.org/css-sizing/#intrinsic-contribution
         width = 0
     else:
@@ -711,7 +750,7 @@ def replaced_max_content_width(box, outer=True):
     width = box.style['width']
     if width == 'auto':
         height = box.style['height']
-        if height == 'auto' or height.unit == '%':
+        if height == 'auto' or check_math(height) or height.unit == '%':
             height = 'auto'
         else:
             assert height.unit.lower() == 'px'
@@ -723,7 +762,7 @@ def replaced_max_content_width(box, outer=True):
         width, _ = default_image_sizing(
             intrinsic_width, intrinsic_height, intrinsic_ratio, 'auto', height,
             default_width=300, default_height=150)
-    elif box.style['width'].unit == '%':
+    elif check_math(box.style['width']) or box.style['width'].unit == '%':
         # See https://drafts.csswg.org/css-sizing/#intrinsic-contribution
         width = 0
     else:

weasyprint/pdf/__init__.py

@@ -9,6 +9,7 @@ from .. import VERSION, Attachment
 from ..html import W3C_DATE_RE
 from ..logger import LOGGER, PROGRESS_LOGGER
 from ..matrix import Matrix
+from ..urls import select_source
 from . import debug, pdfa, pdfua, pdfx
 from .fonts import build_fonts_dictionary
 from .stream import Stream
@@ -273,14 +274,24 @@ def generate_pdf(document, target, zoom, **options):
             key = key.encode('ascii', errors='ignore').decode()
             if key:
                 pdf.info[key] = pydyf.String(value)
+    if options['xmp_metadata']:
+        for url in options['xmp_metadata']:
+            result = select_source(url)
+            with result as (file_obj, base_url, charset, _):
+                xmp_metadata = file_obj.read()
+                if charset:
+                    xmp_metadata = xmp_metadata.decode(charset).encode()
+                metadata.xmp_metadata.append(xmp_metadata)
 
     # Embedded files
     attachments = metadata.attachments.copy()
     if options['attachments']:
+        relationships = iter(options['attachment_relationships'] or [])
         for attachment in options['attachments']:
             if not isinstance(attachment, Attachment):
                 attachment = Attachment(
-                    attachment, url_fetcher=document.url_fetcher)
+                    attachment, url_fetcher=document.url_fetcher,
+                    relationship=next(relationships, 'Unspecified'))
             attachments.append(attachment)
     pdf_attachments = []
     for attachment in attachments:

weasyprint/pdf/anchors.py

@@ -1,7 +1,6 @@
 """Insert anchors, links, bookmarks and inputs in PDFs."""
 
 import collections
-import io
 import mimetypes
 from hashlib import md5
 from os.path import basename
@@ -330,18 +329,12 @@ def write_pdf_attachment(pdf, attachment, compress):
     """Write an attachment to the PDF stream."""
     # Attachments from document links like <link> or <a> can only be URLs.
     # They're passed in as tuples
-    url = None
-    uncompressed_length = 0
-    stream = b''
+    url = mime_type = None
     try:
-        with attachment.source as (_, source, url, _):
-            if isinstance(source, str):
-                source = source.encode()
-            if isinstance(source, bytes):
-                source = io.BytesIO(source)
-            for data in iter(lambda: source.read(4096), b''):
-                uncompressed_length += len(data)
-                stream += data
+        with attachment.source as (file_obj, url, _, mime_type):
+            stream = file_obj.read()
+            if isinstance(stream, str):
+                stream = stream.encode()
     except URLFetchingError as exception:
         LOGGER.error('Failed to load attachment: %s', exception)
         LOGGER.debug('Error while loading attachment:', exc_info=exception)
@@ -356,9 +349,10 @@ def write_pdf_attachment(pdf, attachment, compress):
         filename = basename(unquote(urlsplit(url).path))
     else:
         filename = 'attachment.bin'
-    mime_type = mimetypes.guess_type(filename, strict=False)[0]
-    if not mime_type:
-        mime_type = 'application/octet-stream'
+    mime_type = (
+        mime_type or
+        mimetypes.guess_type(filename, strict=False)[0] or
+        'application/octet-stream')
 
     creation = pydyf.String(attachment.created.strftime('D:%Y%m%d%H%M%SZ'))
     mod = pydyf.String(attachment.modified.strftime('D:%Y%m%d%H%M%SZ'))
@@ -367,7 +361,7 @@ def write_pdf_attachment(pdf, attachment, compress):
         'Subtype': f'/{mime_type.replace("/", "#2f")}',
         'Params': pydyf.Dictionary({
             'CheckSum': f'<{attachment.md5}>',
-            'Size': uncompressed_length,
+            'Size': len(stream),
             'CreationDate': creation,
             'ModDate': mod,
         })

weasyprint/pdf/fonts.py

@@ -618,14 +618,23 @@ def _build_vector_font_dictionary(font_dictionary, pdf, font, widths, compress,
     if font.missing:
         # Add CMap that doesn’t include missing glyphs, so that they can be replaced by
         # .notdef.
+        cmap_extra = pydyf.Dictionary({
+            'Type': '/CMap',
+            'CMapName': '/WP-Encod-0',
+            'CIDSystemInfo': pydyf.Dictionary({
+                'Registry': pydyf.String('Adobe'),
+                'Ordering': pydyf.String('Identity'),
+                'Supplement': 0,
+            }),
+        })
         encoding = pydyf.Stream([
             b'/CIDInit /ProcSet findresource begin',
             b'12 dict begin',
             b'begincmap',
             b'/CIDSystemInfo',
             b'3 dict dup begin',
-            b'/Registry (WP) def',
-            b'/Ordering (Encod) def',
+            b'/Registry (Adobe) def',
+            b'/Ordering (Identity) def',
             b'/Supplement 0 def',
             b'end def',
             b'/CMapName /WP-Encod-0 def',
@@ -633,7 +642,7 @@ def _build_vector_font_dictionary(font_dictionary, pdf, font, widths, compress,
             b'1 begincodespacerange',
             b'<0000> <ffff>',
             b'endcodespacerange',
-        ], compress=compress)
+        ], cmap_extra, compress=compress)
         available = tuple(font.to_unicode)
         available_length = len(available)
         for i in range(ceil(available_length / 100)):