返回 ppt-master
context.py
1 """ConvertContext — shared state passed through the SVG → DrawingML pipeline."""
2
3 from __future__ import annotations
4
5 from pathlib import Path
6 from typing import TYPE_CHECKING, Any
7 from xml.etree import ElementTree as ET
8 from dataclasses import dataclass, field
9
10 if TYPE_CHECKING:
11 from .theme_colors import ThemeColorSpec
12 from .theme_fonts import ThemeFontSpec
13
14 AffineMatrix = tuple[float, float, float, float, float, float]
15 IDENTITY_MATRIX: AffineMatrix = (1.0, 0.0, 0.0, 1.0, 0.0, 0.0)
16
17 TEXT_FLOW_PRESERVE = 'preserve'
18 TEXT_FLOW_REFLOW = 'reflow'
19 TEXT_FLOW_SPLIT = 'split'
20 TEXT_FLOW_MODES = frozenset({
21 TEXT_FLOW_PRESERVE,
22 TEXT_FLOW_REFLOW,
23 TEXT_FLOW_SPLIT,
24 })
25
26
27 def resolve_text_flow(
28 text_flow: str | None = None,
29 merge_paragraphs: bool | None = None,
30 ) -> str:
31 """Resolve the public text-layout options to one internal mode."""
32 if text_flow is not None and merge_paragraphs is not None:
33 raise ValueError(
34 'text_flow and legacy merge_paragraphs cannot be used together'
35 )
36 if merge_paragraphs is not None:
37 return TEXT_FLOW_REFLOW if merge_paragraphs else TEXT_FLOW_SPLIT
38 resolved = TEXT_FLOW_PRESERVE if text_flow is None else text_flow
39 if resolved not in TEXT_FLOW_MODES:
40 choices = ', '.join(sorted(TEXT_FLOW_MODES))
41 raise ValueError(
42 f'unsupported text_flow {resolved!r}; expected one of: {choices}'
43 )
44 return resolved
45
46
47 @dataclass
48 class ShapeResult:
49 """Internal conversion result carrying XML plus resolved EMU bounds."""
50
51 xml: str
52 bounds_emu: tuple[int, int, int, int] | None = None
53
54
55 @dataclass
56 class ConvertContext:
57 """Shared context passed through the SVG → DrawingML conversion pipeline.
58
59 Derived via child() during recursive SVG tree traversal to accumulate
60 translate / scale / inherited style information.
61 """
62
63 defs: dict[str, ET.Element] = field(default_factory=dict)
64 id_counter: int = 2 # 1 is reserved for spTree root
65 # Imported PPTX shape ids are reserved before conversion so newly authored
66 # SVG elements cannot steal an id referenced by a native connector.
67 reserved_shape_ids: frozenset[int] = frozenset()
68 source_shape_id_map: dict[tuple[str, str], int] = field(default_factory=dict)
69 claimed_shape_ids: set[int] = field(default_factory=set)
70 referenced_shape_ids: set[int] = field(default_factory=set)
71 slide_num: int = 1
72 translate_x: float = 0.0
73 translate_y: float = 0.0
74 scale_x: float = 1.0
75 scale_y: float = 1.0
76 viewport_width: float = 1280.0
77 viewport_height: float = 720.0
78 transform_matrix: AffineMatrix = IDENTITY_MATRIX
79 use_transform_matrix: bool = False
80 filter_id: str | None = None
81 media_files: dict[str, bytes] = field(default_factory=dict)
82 rel_entries: list[dict[str, str]] = field(default_factory=list)
83 package_files: dict[str, bytes] = field(default_factory=dict)
84 content_type_overrides: dict[str, str] = field(default_factory=dict)
85 rel_id_counter: int = 2 # rId1 reserved for slideLayout
86 svg_dir: Path | None = None
87 inherited_styles: dict[str, str] = field(default_factory=dict)
88 # Effective SVG font sizes keyed by element identity. Shared resolution
89 # keeps relative sizes and em tracking identical across checker/exporter.
90 text_font_sizes: dict[int, float] = field(default_factory=dict)
91 # Effective source-pixel tracking resolved where each declaration occurs.
92 text_letter_spacings: dict[int, float] = field(default_factory=dict)
93 # SVG group opacity is post-compositing, not an inherited presentation
94 # property. DrawingML has no equivalent group alpha, so native export
95 # approximates it by multiplying this value into each descendant object.
96 opacity_multiplier: float = 1.0
97 # Recursion depth — only the depth==0 (root) context records anim targets.
98 depth: int = 0
99 # Top-level <g id="..."> groups, recorded as (shape_id, svg_id) in z-order.
100 # Used by the PPTX builder to emit per-element object-animation timing.
101 anim_targets: list = field(default_factory=list)
102 # Explicit sidecar group ids may override the legacy chrome-name heuristic.
103 # Explicit structural layer/role/placeholder markers remain non-animatable.
104 animation_group_overrides: frozenset[str] = frozenset()
105 # Text-layout policy for positional tspans: preserve authored line breaks
106 # in one frame, reflow them, or split them into independent frames.
107 text_flow: str = TEXT_FLOW_PRESERVE
108 # Explicit opt-in: replace marked chart/table fallback groups with editable
109 # PowerPoint graphicFrames. Default stays off to preserve SVG output.
110 native_objects_enabled: bool = False
111 # Native PPTX image optimization. Keeps generated decks compact by
112 # downsampling oversized raster assets to their rendered size.
113 image_optimize: bool = True
114 image_max_dimension: int | None = 2560
115 image_sizing: str = 'cap'
116 image_scale: float = 2.0
117 image_quality: int = 85
118 # Optional per-element conversion diagnostics. Shared by child contexts so
119 # callers can inspect native / skipped / unsupported decisions per slide.
120 trace_events: list[dict[str, Any]] | None = None
121 # Optional project theme contract. Matching SVG title/body families emit
122 # DrawingML +mj/+mn tokens instead of fixed typeface names.
123 theme_font_spec: ThemeFontSpec | None = None
124 # Optional project theme-color contract. Exact locked colors are promoted
125 # to context-safe DrawingML scheme slots while local colors stay concrete.
126 theme_color_spec: ThemeColorSpec | None = None
127 # Canonical BCP-47 content language from spec_lock.md. ``None`` preserves
128 # the legacy per-run script heuristic for older projects and lockless quick generation.
129 primary_language: str | None = None
130
131 def next_id(self) -> int:
132 """Allocate the next shape ID."""
133 cid = self.id_counter
134 while cid in self.reserved_shape_ids or cid in self.claimed_shape_ids:
135 cid += 1
136 self.id_counter = cid + 1
137 self.claimed_shape_ids.add(cid)
138 return cid
139
140 def claim_shape_id(
141 self,
142 source_id: str | None,
143 source_scope: str | None = None,
144 ) -> int:
145 """Claim a pre-reserved imported shape id, or allocate a fresh one."""
146 if source_id is None:
147 return self.next_id()
148 scope = source_scope or 'slide'
149 key = (scope, source_id)
150 shape_id = self.source_shape_id_map.get(key)
151 if shape_id is None:
152 raise ValueError(
153 f'Unreserved data-pptx-shape-id {source_id!r} in scope {scope!r}'
154 )
155 if shape_id in self.claimed_shape_ids:
156 raise ValueError(
157 f'Duplicate data-pptx-shape-id {source_id!r} in scope {scope!r}'
158 )
159 self.claimed_shape_ids.add(shape_id)
160 return shape_id
161
162 def reference_shape_id(
163 self,
164 source_id: str,
165 source_scope: str | None = None,
166 ) -> int:
167 """Resolve and record a connector target in the imported id space."""
168 scope = source_scope or 'slide'
169 shape_id = self.source_shape_id_map.get((scope, source_id))
170 if shape_id is None:
171 raise ValueError(
172 f'Unknown connector shape reference {source_id!r} in scope {scope!r}'
173 )
174 self.referenced_shape_ids.add(shape_id)
175 return shape_id
176
177 def next_rel_id(self) -> str:
178 """Allocate the next relationship ID (rIdN)."""
179 rid = f'rId{self.rel_id_counter}'
180 self.rel_id_counter += 1
181 return rid
182
183 def child(
184 self,
185 dx: float = 0,
186 dy: float = 0,
187 sx: float = 1.0,
188 sy: float = 1.0,
189 transform_matrix: AffineMatrix | None = None,
190 filter_id: str | None = None,
191 style_overrides: dict[str, str] | None = None,
192 opacity_multiplier: float = 1.0,
193 ) -> ConvertContext:
194 """Create a child context with accumulated translate / scale / styles.
195
196 Args:
197 dx: X translation delta.
198 dy: Y translation delta.
199 sx: X scale factor.
200 sy: Y scale factor.
201 transform_matrix: Full affine transform to accumulate for
202 converters that can faithfully map it to DrawingML.
203 filter_id: Override filter ID.
204 style_overrides: Style attribute overrides from child element.
205 opacity_multiplier: Local group opacity to multiply into descendants.
206 """
207 local_matrix = transform_matrix or IDENTITY_MATRIX
208 # When first crossing from scalar to matrix mode, fold accumulated
209 # translate_x/y and scale_x/y into the matrix base. Otherwise the
210 # ancestor's scalar transform — which matrix-path readers (e.g.
211 # <image>) never look at — is silently lost, and the descendant
212 # lands at raw SVG coordinates (typically near (0,0)).
213 if transform_matrix is not None and not self.use_transform_matrix:
214 base_matrix: AffineMatrix = (
215 self.scale_x, 0.0,
216 0.0, self.scale_y,
217 self.translate_x, self.translate_y,
218 )
219 else:
220 base_matrix = self.transform_matrix
221 a1, b1, c1, d1, e1, f1 = base_matrix
222 a2, b2, c2, d2, e2, f2 = local_matrix
223 combined_matrix: AffineMatrix = (
224 a1 * a2 + c1 * b2,
225 b1 * a2 + d1 * b2,
226 a1 * c2 + c1 * d2,
227 b1 * c2 + d1 * d2,
228 a1 * e2 + c1 * f2 + e1,
229 b1 * e2 + d1 * f2 + f1,
230 )
231
232 merged = dict(self.inherited_styles)
233 if style_overrides:
234 merged.update(style_overrides)
235
236 local_opacity = max(0.0, min(1.0, opacity_multiplier))
237
238 return ConvertContext(
239 defs=self.defs,
240 id_counter=self.id_counter,
241 reserved_shape_ids=self.reserved_shape_ids,
242 source_shape_id_map=self.source_shape_id_map,
243 claimed_shape_ids=self.claimed_shape_ids,
244 referenced_shape_ids=self.referenced_shape_ids,
245 slide_num=self.slide_num,
246 translate_x=self.translate_x + dx,
247 translate_y=self.translate_y + dy,
248 scale_x=self.scale_x * sx,
249 scale_y=self.scale_y * sy,
250 viewport_width=self.viewport_width,
251 viewport_height=self.viewport_height,
252 transform_matrix=combined_matrix,
253 use_transform_matrix=self.use_transform_matrix or transform_matrix is not None,
254 filter_id=filter_id or self.filter_id,
255 media_files=self.media_files,
256 rel_entries=self.rel_entries,
257 package_files=self.package_files,
258 content_type_overrides=self.content_type_overrides,
259 rel_id_counter=self.rel_id_counter,
260 svg_dir=self.svg_dir,
261 inherited_styles=merged,
262 text_font_sizes=self.text_font_sizes,
263 text_letter_spacings=self.text_letter_spacings,
264 opacity_multiplier=self.opacity_multiplier * local_opacity,
265 depth=self.depth + 1,
266 # anim_targets is intentionally a fresh list on the child;
267 # only the root-level context's list is read by the builder.
268 animation_group_overrides=self.animation_group_overrides,
269 text_flow=self.text_flow,
270 native_objects_enabled=self.native_objects_enabled,
271 image_optimize=self.image_optimize,
272 image_max_dimension=self.image_max_dimension,
273 image_sizing=self.image_sizing,
274 image_scale=self.image_scale,
275 image_quality=self.image_quality,
276 trace_events=self.trace_events,
277 theme_font_spec=self.theme_font_spec,
278 theme_color_spec=self.theme_color_spec,
279 primary_language=self.primary_language,
280 )
281
282 def sync_from_child(self, child_ctx: ConvertContext) -> None:
283 """Sync counters back from a child context."""
284 self.id_counter = child_ctx.id_counter
285 self.rel_id_counter = child_ctx.rel_id_counter
286
286 lines PYTHON