From 77ee0464635ec2c2cfee980917c8d723efd98cf5 Mon Sep 17 00:00:00 2001 From: shen Date: Thu, 5 Nov 2026 09:49:52 +0800 Subject: [PATCH] fix: preserve list context and nested CSS decoration behavior --- mdhelper/docx_exporter.py | 15 +++- mdhelper/docx_inline.py | 6 +- mdhelper/docx_semantic.py | 62 +++++++------- mdhelper/reader_decoration.py | 82 ++++++++++--------- mdhelper/resources/word_inline_styles.lua | 12 ++- mdhelper/style_dialog.py | 1 + mdhelper/style_sync.py | 37 ++++++++- tests/test_style_revision_053.py | 6 +- tests/test_style_revision_055.py | 99 +++++++++++++++++++++++ tests/word_revision_055_smoke.ps1 | 43 +++++++++- 10 files changed, 284 insertions(+), 79 deletions(-) diff --git a/mdhelper/docx_exporter.py b/mdhelper/docx_exporter.py index 14e3add..b1967e8 100644 --- a/mdhelper/docx_exporter.py +++ b/mdhelper/docx_exporter.py @@ -1,6 +1,7 @@ from __future__ import annotations import html +import re import shutil import subprocess import sys @@ -15,7 +16,7 @@ from markdown_it import MarkdownIt from markdown_it.token import Token from .markdown_compat import cjk_strong_plugin -from .docx_inline import source_style_name, LIST_PARAGRAPH_STYLE, QUOTED_LIST_PARAGRAPH_STYLE, SOURCE_PARAGRAPH_STYLES, CONTINUATION, INNER_QUOTE_STYLE +from .docx_inline import source_style_name, LIST_PARAGRAPH_STYLE, QUOTED_LIST_PARAGRAPH_STYLE, SOURCE_PARAGRAPH_STYLES, CONTINUATION, INNER_QUOTE_STYLE, LIST_TARGET WP_NS = "http://schemas.openxmlformats.org/wordprocessingml/2006/main" @@ -256,6 +257,7 @@ def _add_python_docx_paragraph(document: object, block: ParagraphBlock) -> None: style = name paragraph = document.add_paragraph(style=style) if block.continuation: paragraph._p.set(CONTINUATION, '1') + if block.list_target: paragraph._p.set(LIST_TARGET, block.list_target) if block.list_kind: from docx.oxml import parse_xml paragraph._p.get_or_add_pPr().append(parse_xml(f'')) @@ -336,6 +338,7 @@ class ParagraphBlock: alignment: str = '' continuation: bool = False inner_quote: bool = False + list_target: str = '' @dataclass @@ -354,6 +357,7 @@ class _DocxBuilder: self.list_starts: list[int] = [] self.list_sequence = 1000 self.item_pending: list[bool] = [] + self.item_targets: list[str] = [] self.list_quote_depths: list[int] = [] self.quote_depth = 0 @@ -373,8 +377,10 @@ class _DocxBuilder: self.blocks.append(ParagraphBlock([(line, {"code": True})], style="Code")) elif token.type == 'list_item_open': self.item_pending.append(True) + self.item_targets.append('list_ordered' if self.list_stack[-1] == 'number' else 'list_bullet') elif token.type == 'list_item_close': self.item_pending.pop() + self.item_targets.pop() elif token.type == "blockquote_open": self.quote_depth += 1 elif token.type == "blockquote_close": @@ -401,10 +407,13 @@ class _DocxBuilder: def _add_next_inline(self, style: str, list_kind: str = "") -> None: inline = self.tokens[self.index + 1] if self.index + 1 < len(self.tokens) else None runs = _inline_runs(inline.children if inline is not None else None) + if list_kind and self.item_pending[-1] and inline is not None: + if re.match(r'^\[[ xX]\]\s', inline.content): self.item_targets[-1] = 'list_task' self.blocks.append(ParagraphBlock(runs, style=style, list_kind=list_kind, quote_depth=self.quote_depth, list_level=min(8, max(0, len(self.list_stack) - 1)), list_id=self.list_ids[-1] if self.list_ids else 0, list_start=self.list_starts[-1] if self.list_starts else 1, loose=not self.tokens[self.index].hidden, continuation=bool(list_kind and self.item_pending and not self.item_pending[-1]), + list_target=self.item_targets[-1] if list_kind else '', inner_quote=bool(self.list_quote_depths and self.quote_depth > self.list_quote_depths[-1]))) if self.item_pending: self.item_pending[-1] = False self.index += 3 @@ -512,7 +521,9 @@ def _paragraph_xml(block: ParagraphBlock) -> str: props.append('') if block.alignment: props.append(f'') ppr = f"{''.join(props)}" if props else "" - marker = ' xmlns:mdh="urn:mdhelper:source" mdh:continuation="1"' if block.continuation else '' + marker = ' xmlns:mdh="urn:mdhelper:source"' if block.continuation or block.list_target else '' + if block.continuation: marker += ' mdh:continuation="1"' + if block.list_target: marker += f' mdh:list-target="{block.list_target}"' return f"{ppr}{''.join(_run_xml(text, style) for text, style in block.runs)}" diff --git a/mdhelper/docx_inline.py b/mdhelper/docx_inline.py index 30cc590..445e1da 100644 --- a/mdhelper/docx_inline.py +++ b/mdhelper/docx_inline.py @@ -3,8 +3,12 @@ PREFIX = 'MDHelperSource_' LIST_PARAGRAPH_STYLE = 'MDHelperSourceListParagraph' QUOTED_LIST_PARAGRAPH_STYLE = 'MDHelperSourceQuotedListParagraph' INNER_QUOTE_STYLE = 'MDHelperSourceInnerQuote' -SOURCE_PARAGRAPH_STYLES = (LIST_PARAGRAPH_STYLE, QUOTED_LIST_PARAGRAPH_STYLE, INNER_QUOTE_STYLE) +LIST_TARGETS = ('list_bullet', 'list_ordered', 'list_task') +SOURCE_LIST_STYLES = {base + '_' + target: (base, target) + for base in (LIST_PARAGRAPH_STYLE, QUOTED_LIST_PARAGRAPH_STYLE) for target in LIST_TARGETS} +SOURCE_PARAGRAPH_STYLES = (LIST_PARAGRAPH_STYLE, QUOTED_LIST_PARAGRAPH_STYLE, INNER_QUOTE_STYLE, *SOURCE_LIST_STYLES) CONTINUATION = '{urn:mdhelper:source}continuation' +LIST_TARGET = '{urn:mdhelper:source}list-target' CODES = {'strong': 'B', 'emphasis': 'I', 'strike': 'S', 'link': 'L', 'inline_code': 'C'} diff --git a/mdhelper/docx_semantic.py b/mdhelper/docx_semantic.py index 0068dbe..dceec4b 100644 --- a/mdhelper/docx_semantic.py +++ b/mdhelper/docx_semantic.py @@ -13,7 +13,7 @@ from lxml import etree as ET from .document_style import resolved_rules, normalize_rule, font_points, length_points, number from .style_catalog import CATALOG, CHARACTER_DEFAULTS, explicit_rules, parent_style, style_name, legacy_base from .docx_style import W, NS, child, val, properties, _run_properties, _paragraph_properties, _order_properties -from .docx_inline import source_style_order, ordered_styles, LIST_PARAGRAPH_STYLE, QUOTED_LIST_PARAGRAPH_STYLE, SOURCE_PARAGRAPH_STYLES, CONTINUATION, INNER_QUOTE_STYLE +from .docx_inline import source_style_order, ordered_styles, LIST_PARAGRAPH_STYLE, QUOTED_LIST_PARAGRAPH_STYLE, SOURCE_PARAGRAPH_STYLES, CONTINUATION, INNER_QUOTE_STYLE, LIST_TARGET, LIST_TARGETS, SOURCE_LIST_STYLES from .table_style import HEADER_COLOR, STRIPE_COLOR, BORDER_COLOR, VERTICAL_BORDER_COLOR RUN_FIELDS = 'rFonts sz szCs b bCs i iCs color spacing u strike shd'.split() @@ -327,6 +327,29 @@ class WordStyles: child(style, 'pPr').append(ind) return identifier + def contextual_list_rule(self, target, own, level): + """Compose a real inner paragraph with its list container geometry.""" + base = self.effective[target] + rule = {**base, **own} + for side in ('left', 'right'): + field, unit = f'{side}_indent_value', f'{side}_indent_unit' + if field not in own and base[unit] == 'em' and font_points(base) == font_points(rule): + if side == 'left': rule[field] += level * number(base.get('level_indent', 36)) / font_points(base) + continue + rule[field] = length_points(base[field], base[unit], font_points(base)) + if field in own: + rule[field] += length_points(own[field], own.get(unit, 'pt'), font_points(rule)) + if side == 'left': rule[field] += level * number(base.get('level_indent', 36)) + rule[unit] = 'pt' + # text-indent is inherited as a computed length. An inner p's font + # must not resize an em indent declared on li. Absolute hanging units + # also stop Word's character indent from flattening nested levels. + if rule['indent_mode'] in ('hanging', 'first') and rule.get('indent_unit', 'em') == 'em' and (rule['left_indent_unit'] != 'em' or font_points(rule) != font_points(base)): + origin = rule if 'indent_mode' in own else base + rule['indent_value'] = length_points(rule['indent_value'], 'em', font_points(origin)) + rule['indent_unit'] = 'pt' + return rule + def list_paragraph(self, target, base_id, level=0): """A loose Markdown list contains a real p element; a tight one does not.""" own = normalize_rule(self.rules.get('_list_paragraph')) @@ -335,25 +358,14 @@ class WordStyles: if identifier not in self.created: style = self.new(identifier, 'paragraph', 'MDHelper ' + style_name(target, self.rules) + '(段落)' + base_id.removeprefix(style_id(target)), base_id) base = self.effective[target] - rule = {**base, **own} - # A p's margins are inside its li, so they add to the list's - # indentation rather than replacing it (including nested levels). - for side in ('left', 'right'): - field, unit = f'{side}_indent_value', f'{side}_indent_unit' - if field in own: - rule[field] = length_points(base[field], base[unit], font_points(base)) + length_points(own[field], own.get(unit, 'pt'), font_points(rule)) - if side == 'left': rule[field] += level * number(base.get('level_indent', 36)) - rule[unit] = 'pt' - if 'left_indent_value' in own and rule['indent_mode'] == 'hanging' and rule.get('indent_unit', 'em') == 'em': - # Word otherwise substitutes the character hanging width for - # this absolute left margin, flattening nested list levels. - rule['indent_value'] = length_points(rule['indent_value'], 'em', font_points(rule)) - rule['indent_unit'] = 'pt' + rule = self.contextual_list_rule(target, own, level) + authored = {**own, **{field: rule[field] for field in ('left_indent_unit', 'right_indent_unit', 'indent_unit') + if rule.get(field) != base.get(field)}} for tag, writer in (('pPr', _paragraph_properties), ('rPr', _run_properties)): props, old = child(style, tag), ET.Element(W + tag) writer(props, rule); writer(old, base) if tag == 'rPr': remove_fields(props, ('shd',)); remove_fields(old, ('shd',)) - _delta(props, old, own) + _delta(props, old, authored) return identifier def quoted_list(self, target, base_id, level): @@ -363,22 +375,12 @@ class WordStyles: quote_id = self.paragraph('blockquote') style = self.new(identifier, 'paragraph', 'MDHelper ' + style_name(target, self.rules) + '(引用内段落)' + (str(level + 1) if level else ''), quote_id) own = self.explicit['blockquote'] - rule = {**self.effective[target], **normalize_rule(self.rules.get('_list_paragraph')), **own} # Container typography is inherited by li; a declaration on li # wins over that inherited value. A direct blockquote p rule wins # again. Preserve this distinction from the original CSS cascade. overrides = {field: value for field, value in self.explicit[target].items() if field in self.rules.get('_quote_container', {})} - rule.update(overrides) - # Keep the list's marker layout as context-specific paragraph - # properties; quote typography remains inherited from its style. - if level: - points = font_points(rule) - offset = level * number(rule.get('level_indent', 36)) - if rule['left_indent_unit'] == 'em': rule['left_indent_value'] += offset / points - else: - rule['left_indent_value'] = length_points(rule['left_indent_value'], rule['left_indent_unit'], points) + offset - rule['left_indent_unit'] = 'pt' + rule = self.contextual_list_rule(target, {**normalize_rule(self.rules.get('_list_paragraph')), **own, **overrides}, level) for tag, writer in (('pPr', _paragraph_properties), ('rPr', _run_properties)): props, old = child(style, tag), ET.Element(W + tag) writer(props, rule); writer(old, self.effective['blockquote']) @@ -486,11 +488,17 @@ class WordStyles: def apply(self, document): for paragraph in document.iter(W + 'p'): continuation = paragraph.attrib.pop(CONTINUATION, '') == '1' + list_target = paragraph.attrib.pop(LIST_TARGET, '') ppr = properties(paragraph, 'pPr') original_id = reference(ppr, 'pStyle') or 'Normal' names = {reference(style, 'name') for style in self.chain(original_id)} + for name in tuple(names): + if name in SOURCE_LIST_STYLES: + base, list_target = SOURCE_LIST_STYLES[name] + names.add(base) loose = bool(names & set(SOURCE_PARAGRAPH_STYLES)) target = self.classify(paragraph, original_id) + if target.startswith('list_') and list_target in LIST_TARGETS: target = list_target identifier = self.paragraph(target) if continuation and target == 'blockquote': remove_fields(ppr, ('numPr',)) if target.startswith('list_'): diff --git a/mdhelper/reader_decoration.py b/mdhelper/reader_decoration.py index 53f37d2..fd2b33f 100644 --- a/mdhelper/reader_decoration.py +++ b/mdhelper/reader_decoration.py @@ -47,7 +47,7 @@ def decoration_css(rules, inherited, selectors, source): from .style_catalog import CHARACTER_DEFAULTS, CATALOG if not any(CATALOG.get(key) and CATALOG[key].kind == 'character' and any(rule.get(field) is False for field in ('underline','strike')) - for key,rule in rules.items() if isinstance(rule,dict)): + for key,rule in {key: {**inherited.get(key, {}), **rules.get(key, {})} for key in CATALOG}.items()): return source root='.reader[data-mdh-leaf-decoration]' parts=[MARKER, root+' { --mdh-line-underline: ; --mdh-line-strike: ; }'] @@ -66,52 +66,54 @@ def decoration_css(rules, inherited, selectors, source): # Defaults have zero specificity; authored conditional selectors # can continue to change decorations (including hover effects). scope=selector if field in own else ':where('+selector+')' - parts.append(scope+' { --mdh-decoration-raw: initial; --mdh-line-'+field+': '+(token if rule[field] else '')+'; }') + parts.append(scope+' { --mdh-decoration-mode: initial; --mdh-decoration-raw: initial; --mdh-line-'+field+': '+(token if rule[field] else '')+'; }') parts.append(_decoration_variables(source)) parts.append(', '.join(suppress)+' { text-decoration-line: none !important; }') parts.append(':where('+root+' p, '+root+' strong, '+root+' em, '+root+' a, '+root+' s, '+root+' del, '+root+' code) { text-decoration-color: inherit; text-decoration-style: inherit; text-decoration-thickness: inherit; }') - parts.append(root+' [data-mdh-text] { text-decoration-line: var(--mdh-decoration-raw, var(--mdh-line-underline) var(--mdh-line-strike)); text-decoration-color: inherit; text-decoration-style: inherit; text-decoration-thickness: inherit; }') + # A variable may contain a complete shorthand (including color/style). + # Let the browser resolve it on text leaves; an inner core rule resets + # the inherited mode so it can still cancel decorations independently. + leaf=root+' [data-mdh-text]' + parts.append(leaf+' { text-decoration-line: var(--mdh-decoration-raw, var(--mdh-line-underline) var(--mdh-line-strike)); text-decoration-color: inherit; text-decoration-style: inherit; text-decoration-thickness: inherit; }') + parts.append('@container style(--mdh-decoration-mode: raw) { '+leaf+' { text-decoration: var(--mdh-decoration-shorthand); } }') return '\n'.join(parts) def _decoration_variables(source): """Keep authored selectors, conditions, priority and decoration appearance.""" import tinycss2 as css - from .style_sync import DECORATION_EXPRESSION + from .style_sync import DECORATION_EXPRESSION, _invalid_core_value def visit(nodes): - for node in nodes: - if node.type=='at-rule' and node.content is not None and node.lower_at_keyword in ('media','supports','layer','container'): - children=css.parse_blocks_contents(node.content);visit(children) - node.content=css.parse_component_value_list(css.serialize(children)) - elif node.type=='qualified-rule': - declarations=css.parse_blocks_contents(node.content,skip_whitespace=False,skip_comments=False) - visit(declarations) - output=[] - for declaration in declarations: - # Invalid declarations have no serializable representation. - # Keep valid siblings, including nested qualified rules. - if declaration.type=='error': continue - output.append(declaration) - if declaration.type!='declaration':continue - value=css.serialize(declaration.value).strip() - suffix=' !important' if declaration.important else '' - if declaration.name in ('--mdhelper-underline','--mdhelper-strike') and value in ('true','false'): - field=declaration.name.removeprefix('--mdhelper-') - token=('underline' if field=='underline' else 'line-through') if value=='true' else '' - output.extend(css.parse_declaration_list(f'--mdh-decoration-raw:initial{suffix};--mdh-line-{field}:{token}{suffix};')) - continue - if declaration.lower_name not in ('text-decoration','text-decoration-line'):continue - if value==DECORATION_EXPRESSION:continue - tokens=[t.lower_value for t in declaration.value if t.type=='ident'] - lines=set(tokens)&{'none','underline','line-through','overline'} - if lines and not ('none' in lines and len(lines)>1): - values={'--mdh-decoration-raw':'initial','--mdh-line-underline':'underline' if 'underline' in lines else '', - '--mdh-line-strike':'line-through' if 'line-through' in lines else ''} - if 'overline' in lines:values['--mdh-decoration-raw']=' '.join(x for x in ('underline','line-through','overline') if x in lines) - elif declaration.lower_name=='text-decoration-line' and any(t.type=='function' for t in declaration.value): - values={'--mdh-decoration-raw':value} - else:continue - output.extend(css.parse_declaration_list(''.join(f'{prop}:{val}{suffix};' for prop,val in values.items()))) - node.content=css.parse_component_value_list(css.serialize(output)) - nodes=css.parse_stylesheet(source);visit(nodes) - return css.serialize(nodes) + output=[] + for declaration in nodes: + if declaration.type=='error': continue + if declaration.type=='qualified-rule' or (declaration.type=='at-rule' and declaration.content is not None + and declaration.lower_at_keyword in ('media','supports','layer','container')): + children=css.parse_blocks_contents(declaration.content,skip_whitespace=False,skip_comments=False) + declaration.content=css.parse_component_value_list(css.serialize(visit(children))) + output.append(declaration) + if declaration.type!='declaration': continue + value=css.serialize(declaration.value).strip() + suffix=' !important' if declaration.important else '' + if declaration.name in ('--mdhelper-underline','--mdhelper-strike') and value in ('true','false'): + field=declaration.name.removeprefix('--mdhelper-') + token=('underline' if field=='underline' else 'line-through') if value=='true' else '' + output.extend(css.parse_declaration_list(f'--mdh-decoration-mode:initial{suffix};--mdh-decoration-raw:initial{suffix};--mdh-line-{field}:{token}{suffix};')) + continue + if declaration.lower_name not in ('text-decoration','text-decoration-line'):continue + if value==DECORATION_EXPRESSION:continue + if _invalid_core_value(declaration.lower_name, declaration.value): continue + tokens=[t.lower_value for t in declaration.value if t.type=='ident'] + lines=set(tokens)&{'none','underline','line-through','overline'} + if declaration.lower_name=='text-decoration' and any(t.type=='function' for t in declaration.value): + values={'--mdh-decoration-mode':'raw','--mdh-decoration-shorthand':value} + elif lines and not ('none' in lines and len(lines)>1): + values={'--mdh-decoration-mode':'initial','--mdh-decoration-raw':'initial','--mdh-line-underline':'underline' if 'underline' in lines else '', + '--mdh-line-strike':'line-through' if 'line-through' in lines else ''} + if 'overline' in lines:values['--mdh-decoration-raw']=' '.join(x for x in ('underline','line-through','overline') if x in lines) + elif declaration.lower_name=='text-decoration-line' and any(t.type=='function' for t in declaration.value): + values={'--mdh-decoration-mode':'initial','--mdh-decoration-raw':value} + else:continue + output.extend(css.parse_declaration_list(''.join(f'{prop}:{val}{suffix};' for prop,val in values.items()))) + return output + return css.serialize(visit(css.parse_stylesheet(source))) diff --git a/mdhelper/resources/word_inline_styles.lua b/mdhelper/resources/word_inline_styles.lua index 75694bf..6cdc454 100644 --- a/mdhelper/resources/word_inline_styles.lua +++ b/mdhelper/resources/word_inline_styles.lua @@ -48,9 +48,14 @@ local function mark(items, stack) end local function list_paragraphs(list) for _, item in ipairs(list.content) do + local target = list.t == 'OrderedList' and 'list_ordered' or 'list_bullet' + local text = pandoc.utils.stringify(item[1] or {}) + if text:match('^%[[ xX]%]') or text:sub(1, #'☐') == '☐' or text:sub(1, #'☒') == '☒' or text:sub(1, #'☑') == '☑' then + target = 'list_task' + end for index, block in ipairs(item) do if block.t == 'Para' then - item[index] = pandoc.Div({block}, pandoc.Attr('', {}, {['custom-style']='MDHelperSourceListParagraph'})) + item[index] = pandoc.Div({block}, pandoc.Attr('', {}, {['custom-style']='MDHelperSourceListParagraph_' .. target})) end end end @@ -63,8 +68,9 @@ local function quoted_paragraphs(quote) end end return quote:walk({Div=function(block) - if block.attributes['custom-style'] == 'MDHelperSourceListParagraph' then - block.attributes['custom-style'] = 'MDHelperSourceQuotedListParagraph' + local name = block.attributes['custom-style'] or '' + if name:match('^MDHelperSourceListParagraph') then + block.attributes['custom-style'] = name:gsub('^MDHelperSourceListParagraph', 'MDHelperSourceQuotedListParagraph') end return block end}) diff --git a/mdhelper/style_dialog.py b/mdhelper/style_dialog.py index c74419f..51d8d72 100644 --- a/mdhelper/style_dialog.py +++ b/mdhelper/style_dialog.py @@ -425,6 +425,7 @@ class DocumentStyleEditorDialog(QDialog): if hasattr(self.parent(), '_configure_web_settings'): self.parent()._configure_web_settings(self.actual_preview.settings()) sample = '# 标题 1 中的 **加粗** 与 [链接](#预览)\n\n## 标题 2\n\n### 标题 3\n\n#### 标题 4\n\n##### 标题 5\n\n###### 标题 6\n\n正文段落:中文与 English,**加粗**、*强调*、~~删除线~~、`inline code`。\n\n第二段文字,用于观察段间距。\n\n> 引用段落中的 **重点**。\n\n- 无序列表\n - 第二级列表\n\n1. 有序列表\n2. 第二项\n\n- [ ] 任务列表\n\n```text\n代码块 code block\n第二行\n```\n\n| 表头 | Heading |\n|---|---|\n| 单元格 | Cell |' + sample += '\n| 第二行 | Alternating row |\n| 第三行 | Third row |' parent = self.parent() root = getattr(parent, 'project_root', Path.cwd()) content = MarkdownRenderer(root).render_source(sample, root / '__style_preview__.md', diff --git a/mdhelper/style_sync.py b/mdhelper/style_sync.py index 5d083dc..d627551 100644 --- a/mdhelper/style_sync.py +++ b/mdhelper/style_sync.py @@ -11,6 +11,7 @@ import re import tinycss2 as css from tinycss2.color3 import parse_color +from tinycss2.color4 import parse_color as parse_modern_color from .document_style import (STYLE_TARGETS, normalize_rule, css_declarations, build_document_style_css, fmt, font_points, resolved_rules) @@ -253,13 +254,39 @@ def _invalid_core_value(prop, tokens): if prop in ('color','background-color','border-color') and len(tokens) == 1: token=tokens[0] if token.type != 'function': return parse_color(token) is None + if token.lower_name in ('rgb','rgba','hsl','hsla','hwb','lab','lch','oklab','oklch','color'): + # Relative colors and math are valid browser expressions that the + # scalar color parser cannot evaluate. Preserve those as CSS. + if not any(t.type=='function' or t.type=='ident' and t.lower_value=='from' for t in token.arguments): + try: return parse_modern_color(token) is None + except (ValueError, TypeError): return True if prop == 'font-style' and len(tokens) > 1: return not (len(tokens)==2 and tokens[0].type=='ident' and tokens[0].lower_value=='oblique' and tokens[1].type=='dimension' and tokens[1].lower_unit in ('deg','grad','rad','turn')) if prop == 'font': if len(tokens)==1: return not (tokens[0].type=='ident' and tokens[0].lower_value in ('caption','icon','menu','message-box','small-caption','status-bar')) - # Valid but unsupported shorthands still reset the core font fields. + size_index = next((i for i,t in enumerate(tokens) + if t.type in ('dimension','percentage','function') and not (t.type=='dimension' and t.lower_unit in ('deg','grad','rad','turn')) + or t.type=='number' and t.value==0 + or t.type=='ident' and t.lower_value in ('xx-small','x-small','small','medium','large','x-large','xx-large','xxx-large','smaller','larger')), None) + if size_index is None or _invalid_core_value('font-size', [tokens[size_index]]): return True + prefix_names = {'normal','italic','oblique','small-caps','bold','bolder','lighter', + 'ultra-condensed','extra-condensed','condensed','semi-condensed', + 'semi-expanded','expanded','extra-expanded','ultra-expanded'} + for token in tokens[:size_index]: + if token.type=='ident' and token.lower_value in prefix_names: continue + if token.type=='number' and 1 <= token.value <= 1000: continue + if token.type=='dimension' and token.lower_unit in ('deg','grad','rad','turn'): continue + return True + end = size_index + 1 + if end < len(tokens) and tokens[end].type=='literal' and tokens[end].value=='/': + end += 2 + if end > len(tokens) or _invalid_core_value('line-height', [tokens[end-1]]): return True + if end >= len(tokens): return True + try: _font_names(_text(tokens[end:])) + except ValueError: return True + # Retain valid, unsupported prefixes and relative sizes as CSS. return False scalar = {'font-size','line-height','letter-spacing','text-indent','font-weight','color','background-color','text-align', 'margin-top','margin-bottom','margin-left','margin-right'} @@ -270,6 +297,14 @@ def _invalid_core_value(prop, tokens): selected = [name for name in names if name in lines] if 'none' in selected and len(selected)>1 or len(selected)!=len(set(selected)): return True if prop == 'text-decoration-line': return any(t.type!='ident' or t.lower_value not in lines for t in tokens) + for token in tokens: + if token.type=='ident' and token.lower_value in lines | {'solid','double','dotted','dashed','wavy','auto','from-font'}: continue + if token.type=='function': + if _invalid_core_value('color', [token]): return True + continue # Includes variable colors/thickness. + if parse_color(token) is not None: continue + if token.type in ('dimension','percentage','number') and not _invalid_core_value('font-size',[token]): continue + return True if len(tokens) != 1: return False token = tokens[0] if token.type == 'string' and prop != 'font-family' and not prop.startswith('--'): return True diff --git a/tests/test_style_revision_053.py b/tests/test_style_revision_053.py index addfe6f..2174768 100644 --- a/tests/test_style_revision_053.py +++ b/tests/test_style_revision_053.py @@ -152,13 +152,15 @@ def test_nested_quoted_lists_keep_depth_numbering_and_emphasis(tmp_path,engine): values={} for p in effective_document(doc,styles).iter(W+'p'): text=''.join(p.itertext()); props=p.find(W+'pPr') - values[text]=props.find(W+'ind').get(W+'leftChars') + # Absolute indents are needed when the quote and list use different + # font sizes. Compare the resulting distance, not its storage unit. + values[text]=int(props.find(W+'ind').get(W+'left')) assert props.find(W+'numPr') is not None if text in ('OuterQuote','NestedQuote'): run=next(r for r in p.iter(W+'r') if ''.join(r.itertext())==text) assert run.find('w:rPr/w:b',NS).get(W+'val')=='1' assert run.find('w:rPr/w:color',NS).get(W+'val')=='ff0000' - assert int(values['NestedQuote'])>int(values['OuterQuote']) + assert values['NestedQuote']-values['OuterQuote']==36*20 def test_graphical_quote_edit_overrides_important_broad_p(qt_application,tmp_path): diff --git a/tests/test_style_revision_055.py b/tests/test_style_revision_055.py index b709548..e1dd323 100644 --- a/tests/test_style_revision_055.py +++ b/tests/test_style_revision_055.py @@ -12,6 +12,11 @@ ENGINES = ('builtin', 'python-docx', 'pandoc') CODE_SOURCE = '# Parent\n\n`Inline`\n\n```text\nBlock\n```' CODE_RULES = {'h1': {'font_color': '#ff0000', 'font_size': 20, 'font_size_unit': 'pt'}, '_metadata': {'code': {'based_on': 'h1'}}} +ORDERED = '1. First\n\n Continuation\n\n2. Second' +TASK = '- [ ] First\n\n Continuation\n\n- [x] Second' +QUOTED = '> - Outer\n>\n> - Inner\n>\n> - NestedSecond\n>\n> - Second' +MARGINS = 'p {margin-left:12pt;} li {margin-left:24pt;}' +MIXED = 'li {font-size:12pt;margin-left:24pt;text-indent:-1em;} p {font-size:20pt;margin-left:12pt;}' @pytest.mark.parametrize('nested', ['& strong {color:red}', '@media screen { & strong {color:red} }']) @@ -60,3 +65,97 @@ def test_word_code_keeps_editable_parent_chain(tmp_path, engine): run = next(r for r in effective_document(doc, styles).iter(W+'r') if ''.join(r.itertext())=='Inline') assert run.find('w:rPr/w:color', NS).get(W+'val') == '0000ff' assert run.find('w:rPr/w:sz', NS).get(W+'val') == '56' + + +@pytest.mark.parametrize('engine', ENGINES) +@pytest.mark.parametrize('source,target', [(ORDERED,'list_ordered'), (TASK,'list_task')]) +def test_word_continuation_preserves_semantic_list_type(tmp_path, engine, source, target): + rules = {target: {'font_color': '#ff0000'}, 'list_bullet': {'font_color': '#0000ff'}} + doc, styles = parts(export_markdown_to_docx(source,tmp_path/'continuation.docx',engine=engine,style_rules=rules)) + assert not any(a.startswith('{urn:mdhelper:source}') for p in doc.iter(W+'p') for a in p.attrib) + from mdhelper.docx_semantic import style_id + base = styles.find(f"w:style[@w:styleId='{style_id(target)}']",NS) + base.find('w:rPr/w:color',NS).set(W+'val','008000') + for p in effective_document(doc,styles).iter(W+'p'): + for run in p.iter(W+'r'): + if ''.join(run.itertext()).strip(): assert run.find('w:rPr/w:color',NS).get(W+'val')=='008000' + + +@pytest.mark.parametrize('engine', ENGINES) +def test_quoted_list_keeps_additive_nested_indents(tmp_path, engine): + doc, styles = parts(export_markdown_to_docx(QUOTED,tmp_path/'quoted.docx',engine=engine,style_rules=normalize_payload({'css':MARGINS})['rules'])) + for p in effective_document(doc,styles).iter(W+'p'): + text=''.join(p.itertext()) + if text in ('Outer','Inner'): + ind=p.find('w:pPr/w:ind',NS) + assert int(ind.get(W+'left'))/20 == (36 if text=='Outer' else 72) + assert ind.get(W+'hangingChars','0')=='0' + + +@pytest.mark.parametrize('engine', ENGINES) +def test_inherited_list_hanging_uses_container_font(tmp_path, engine): + doc,styles=parts(export_markdown_to_docx('- Outer\n\n- Second',tmp_path/'hanging.docx',engine=engine,style_rules=normalize_payload({'css':MIXED})['rules'])) + for p in effective_document(doc,styles).iter(W+'p'): + assert p.find('w:pPr/w:ind',NS).get(W+'hanging') == '240' + + +def test_legacy_code_can_cancel_outer_decoration(qt_application,tmp_path): + payload=normalize_payload({'rules':{'strong':{'underline':True},'code':{'underline':False}}}) + result=browser(qt_application,tmp_path,payload,'**Outer `Inner` outer**', + "[document.querySelector('strong'),document.querySelector('strong > [data-mdh-text]'),document.querySelector('code [data-mdh-text]')].map(e=>getComputedStyle(e).textDecorationLine)") + assert result==['none','underline','none'] + + +@pytest.mark.parametrize('bad',['color:rgb(nope)','color:rgba(1,2)','color:hsl(nope)','font:bold garbage','font:italic 12pt','font:bold nonsense 20pt Arial']) +def test_invalid_function_and_font_keep_valid_core_values(qt_application,tmp_path,bad): + payload=normalize_payload({'css':'p{color:red;font-size:20pt;'+bad+'}'}) + assert payload['rules']['paragraph']['font_color']=='#ff0000' + assert payload['rules']['paragraph']['font_size']==20 + assert normalize_payload(payload)==payload + assert browser(qt_application,tmp_path,payload,'Fallback',"getComputedStyle(document.querySelector('p')).color")=='rgb(255, 0, 0)' + + +@pytest.mark.parametrize('invalid', ['underline nonsense','underline rgb(nope)']) +def test_invalid_decoration_does_not_draw_an_extra_line(qt_application,tmp_path,invalid): + payload=normalize_payload({'css':'em{text-decoration:none} strong{text-decoration:line-through;text-decoration:'+invalid+'}'}) + assert browser(qt_application,tmp_path,payload,'**Text**',"getComputedStyle(document.querySelector('strong [data-mdh-text]')).textDecorationLine")=='line-through' + + +@pytest.mark.parametrize('value,line,style,color', [('line-through','line-through','solid','rgb(23, 105, 170)'), + ('underline line-through wavy red','underline line-through','wavy','rgb(255, 0, 0)')]) +def test_variable_decoration_shorthand_keeps_browser_effect(qt_application,tmp_path,value,line,style,color): + payload=normalize_payload({'css':'em{text-decoration:none} a{text-decoration:var(--custom-line);--custom-line:'+value+'}'}) + assert browser(qt_application,tmp_path,payload,'[Link](#link)',"(()=>{const s=getComputedStyle(document.querySelector('a [data-mdh-text]'));return [s.textDecorationLine,s.textDecorationStyle,s.textDecorationColor]})()") == [line,style,color] + + +@pytest.mark.parametrize('scope', ['a { %s }','a { @media screen { %s } }','@supports (display:grid) { a { %s } }']) +def test_variable_decorations_survive_nested_conditions(qt_application,tmp_path,scope): + payload=normalize_payload({'css':'em{text-decoration:none} '+scope % 'text-decoration:var(--line);--line:line-through wavy red;'}) + actual=browser(qt_application,tmp_path,payload,'[Outer *Inner*](#link)', + "[...document.querySelectorAll('a [data-mdh-text]')].map(e=>getComputedStyle(e).textDecorationLine)") + assert actual==['line-through','none'] + + +@pytest.mark.parametrize('css', ['color:rgb(255 0 0 / .5)', 'color:rgb(from red r g b)', 'color:rgb(calc(255 / 2) 0 0)', 'font:italic 2em Arial']) +def test_valid_advanced_values_remain_css_only(css): + from mdhelper.style_sync import analyze_css + result=analyze_css('p{color:blue;font-size:20pt;'+css+'}') + assert not result.errors and not any('已忽略无效' in note for note in result.notes) + + +def test_actual_table_preview_shows_alternating_rows(qt_application): + import json,time + dialog=DocumentStyleEditorDialog({'rules':{'table':{'table_banded_rows':True,'table_stripe_color':'#ff0000'}}}) + try: + dialog.tabs.setCurrentIndex(2) + values=[] + end=time.monotonic()+20 + while time.monotonic()getComputedStyle(tr.cells[0]).backgroundColor))",values.append) + while not values and time.monotonic()=3: break + qt_application.processEvents();time.sleep(.01) + colors=json.loads(values[0]) + assert len(colors)>=3 and colors[1]=='rgb(255, 0, 0)' and colors[0]!=colors[1] and colors[2]!=colors[1] + finally: dialog.close() diff --git a/tests/word_revision_055_smoke.ps1 b/tests/word_revision_055_smoke.ps1 index d056310..09e693d 100644 --- a/tests/word_revision_055_smoke.ps1 +++ b/tests/word_revision_055_smoke.ps1 @@ -10,23 +10,60 @@ sys.path.insert(0,str(Path('tests').resolve())) from test_style_revision_055 import * folder=Path(tempfile.mkdtemp(prefix='MDHelper-055-Word-')) for engine in ENGINES: - export_markdown_to_docx(CODE_SOURCE,folder/f'{engine}-code.docx',engine=engine,style_rules=CODE_RULES) + for name,source,rules in [('code',CODE_SOURCE,CODE_RULES), + ('ordered',ORDERED,{'list_ordered':{'font_color':'#ff0000'},'list_bullet':{'font_color':'#0000ff'}}), + ('task',TASK,{'list_task':{'font_color':'#ff0000'},'list_bullet':{'font_color':'#0000ff'}}), + ('quoted',QUOTED,normalize_payload({'css':MARGINS})['rules']), + ('hanging','- Outer\n\n- Second',normalize_payload({'css':MIXED})['rules']), + ('invalid','Fallback',normalize_payload({'css':'p{color:red;color:rgb(nope);font-size:20pt;font:bold garbage;}'})['rules']), + ('legacy','**Outer `Inner` outer**',{'strong':{'underline':True},'code':{'underline':False}})]: + export_markdown_to_docx(source,folder/f'{engine}-{name}.docx',engine=engine,style_rules=rules) print(folder) '@ | & $Python -B -).Trim() if ($LASTEXITCODE -ne 0) {throw 'Fixture generation failed'} $word=New-Object -ComObject Word.Application $word.Visible=$false;$word.DisplayAlerts=0 foreach($engine in @('builtin','python-docx','pandoc')) { - $doc=$word.Documents.Open((Join-Path $folder "$engine-code.docx"),$false,$true) + foreach($case in @('code','ordered','task','quoted','hanging','invalid','legacy')) { + $doc=$word.Documents.Open((Join-Path $folder "$engine-$case.docx"),$false,$true) try { + if($case -eq 'code') { $range=$doc.Content.Duplicate;[void]$range.Find.Execute('Parent') $range.Style.Font.Color=16711680;$range.Style.Font.Size=28 foreach($text in @('Inline','Block')) { $range=$doc.Content.Duplicate;[void]$range.Find.Execute($text) if($range.Font.Color -ne 16711680 -or $range.Font.Size -ne 28){throw "$engine $text does not follow ancestor: $($range.Font.Color), $($range.Font.Size)"} } - Write-Output "$engine code ancestor edits passed" + } elseif($case -in @('ordered','task')) { + $range=$doc.Content.Duplicate;[void]$range.Find.Execute('First') + $name=if($case -eq 'ordered'){'MDHelper 有序列表'}else{'MDHelper 任务列表'} + $base=$doc.Styles.Item($name) + $base.Font.Color=65280 + foreach($text in @('First','Continuation','Second')) { + $range=$doc.Content.Duplicate;[void]$range.Find.Execute($text) + if($range.Font.Color -ne 65280){throw "$engine $case $text lost editable list base: $($range.Font.Color)"} + if($text -eq 'Continuation' -and $range.ListFormat.ListString.Trim() -ne ''){throw "$engine continuation consumed a marker"} + } + } elseif($case -eq 'quoted') { + foreach($pair in @(@('Outer',36),@('Inner',72))) { + $range=$doc.Content.Duplicate;[void]$range.Find.Execute($pair[0]) + if([Math]::Abs($range.ParagraphFormat.LeftIndent-$pair[1]) -gt 0.1){throw "$engine quoted $($pair[0]) indent: $($range.ParagraphFormat.LeftIndent)"} + } + } elseif($case -eq 'hanging') { + $range=$doc.Content.Duplicate;[void]$range.Find.Execute('Outer') + if($range.ParagraphFormat.FirstLineIndent -ne -12 -or $range.Font.Size -ne 20){throw "$engine inherited hanging indent resized: $($range.ParagraphFormat.FirstLineIndent), font $($range.Font.Size)"} + } elseif($case -eq 'invalid') { + $range=$doc.Content.Duplicate;[void]$range.Find.Execute('Fallback') + if($range.Font.Color -ne 255 -or $range.Font.Size -ne 20){throw "$engine invalid CSS discarded valid fallback"} + } elseif($case -eq 'legacy') { + foreach($pair in @(@('Outer',1),@('Inner',0))) { + $range=$doc.Content.Duplicate;[void]$range.Find.Execute($pair[0]) + if($range.Font.Underline -ne $pair[1] -or $range.Font.Bold -ne -1){throw "$engine legacy decoration or emphasis changed"} + } + } + Write-Output "$engine $case passed" } finally {$doc.Close(0);[void][Runtime.InteropServices.Marshal]::FinalReleaseComObject($doc)} + } } } finally { if($word){$word.Quit();[void][Runtime.InteropServices.Marshal]::FinalReleaseComObject($word)}