| 1 | """Core SVG -> DrawingML dispatcher, group handling, and main entry point.""" |
| 2 | |
| 3 | from __future__ import annotations |
| 4 | |
| 5 | import base64 |
| 6 | import binascii |
| 7 | import hashlib |
| 8 | import math |
| 9 | import re |
| 10 | from pathlib import Path |
| 11 | from typing import Any |
| 12 | from xml.etree import ElementTree as ET |
| 13 | |
| 14 | from native_payloads import NativePayloadError, hydrate_native_payload_refs |
| 15 | from hyperlink_contract import ( |
| 16 | SHAPE_HYPERLINK_ATTR, |
| 17 | project_hyperlink_errors, |
| 18 | svg_hyperlink_href, |
| 19 | ) |
| 20 | from pptx_shapes import ( |
| 21 | has_relationship_attributes, |
| 22 | resolve_preset_preview_hash, |
| 23 | svg_preset_preview_fingerprint, |
| 24 | svg_text_fingerprint, |
| 25 | validate_ooxml_xfrm, |
| 26 | ) |
| 27 | from pptx_effects import project_effect_status_errors, txbody_has_run_effects |
| 28 | from pptx_to_svg.preset_authoring import ( |
| 29 | materialize_compact_authored_preset_tree, |
| 30 | validate_authored_preset_tree, |
| 31 | ) |
| 32 | from resource_paths import icon_search_dirs_for_svg |
| 33 | |
| 34 | from .context import ( |
| 35 | TEXT_FLOW_PRESERVE, |
| 36 | TEXT_FLOW_SPLIT, |
| 37 | ConvertContext, |
| 38 | ShapeResult, |
| 39 | resolve_text_flow, |
| 40 | ) |
| 41 | from .hyperlinks import apply_shape_hyperlink |
| 42 | from .paths import ( |
| 43 | project_freeform_geometry_errors, |
| 44 | project_gradient_geometry_errors, |
| 45 | ) |
| 46 | from .theme_colors import ThemeColorSpec |
| 47 | from .theme_fonts import ThemeFontSpec |
| 48 | from .text_properties import ( |
| 49 | materialize_project_text_metrics, |
| 50 | project_text_property_errors, |
| 51 | resolve_project_font_sizes, |
| 52 | resolve_project_letter_spacings, |
| 53 | ) |
| 54 | from .utils import ( |
| 55 | EMU_PER_PX, |
| 56 | SVG_NS, |
| 57 | _extract_inheritable_styles, |
| 58 | _get_attr, |
| 59 | _is_unit_axis_reflection, |
| 60 | is_picture_effect_carrier, |
| 61 | parse_svg_length, |
| 62 | parse_transform_operations, |
| 63 | parse_transform_matrix, |
| 64 | project_definition_errors, |
| 65 | project_definition_index, |
| 66 | project_filter_errors, |
| 67 | project_geometry_length_errors, |
| 68 | project_gradient_errors, |
| 69 | project_image_aspect_ratio_errors, |
| 70 | project_mask_errors, |
| 71 | project_marker_errors, |
| 72 | project_opacity_errors, |
| 73 | project_paint_errors, |
| 74 | project_paint_reference_errors, |
| 75 | project_stroke_style_errors, |
| 76 | project_transform_errors, |
| 77 | resolve_url_id, |
| 78 | supports_full_project_transform, |
| 79 | validate_dml_shape_matrix, |
| 80 | ) |
| 81 | from .styles import ( |
| 82 | build_effect_xml, build_fill_xml, |
| 83 | get_element_opacity, get_fill_opacity, get_stroke_opacity, |
| 84 | ) |
| 85 | from .elements import ( |
| 86 | convert_rect, convert_circle, convert_ellipse, |
| 87 | convert_line, convert_path, |
| 88 | convert_polygon, convert_polyline, |
| 89 | convert_text, convert_image, convert_nested_svg, |
| 90 | project_clip_path_errors, |
| 91 | project_image_errors, |
| 92 | project_nested_svg_crop_errors, |
| 93 | ) |
| 94 | from ..animation_config import is_chrome_id, usable_animation_group_id |
| 95 | from ..canvas_contract import ( |
| 96 | CanvasContractError, |
| 97 | parse_project_svg_root, |
| 98 | parse_project_viewbox, |
| 99 | ) |
| 100 | from ..native_objects import ( |
| 101 | INLINE_FORMULA_ATTR, |
| 102 | NativeMarkerAttributeError, |
| 103 | convert_native_object, |
| 104 | inline_formula_marker_errors, |
| 105 | native_metadata_payload_matches, |
| 106 | native_replacement_kind, |
| 107 | native_marker_transform, |
| 108 | snapshot_native_fallback_freshness, |
| 109 | ) |
| 110 | from ..native_objects.marker_status import native_marker_status_errors |
| 111 | from ..semantic_markers import is_static_page_frame |
| 112 | |
| 113 | |
| 114 | class SvgNativeConversionError(RuntimeError): |
| 115 | """Raised when an SVG cannot be faithfully converted to native DrawingML.""" |
| 116 | |
| 117 | |
| 118 | def _hydrate_native_payloads(root: ET.Element, svg_path: Path) -> int: |
| 119 | """Resolve compressed workspace payload references for native conversion.""" |
| 120 | try: |
| 121 | return hydrate_native_payload_refs(root, svg_path) |
| 122 | except NativePayloadError as exc: |
| 123 | raise SvgNativeConversionError( |
| 124 | f"{svg_path.name}: invalid native payload reference: {exc}" |
| 125 | ) from exc |
| 126 | |
| 127 | |
| 128 | def _require_native_marker_attributes( |
| 129 | root: ET.Element, |
| 130 | svg_path: Path | str, |
| 131 | ) -> None: |
| 132 | """Reject contradictory native marker aliases before conversion.""" |
| 133 | errors: list[str] = [] |
| 134 | for elem in root.iter(): |
| 135 | if elem.tag.rsplit('}', 1)[-1] == 'metadata': |
| 136 | continue |
| 137 | marker_errors = native_marker_status_errors(elem) |
| 138 | if marker_errors: |
| 139 | marker_id = elem.get('id') or elem.get('data-name') or '<unnamed>' |
| 140 | errors.extend(f'{marker_id}: {error}' for error in marker_errors) |
| 141 | continue |
| 142 | marker_id = elem.get('id') or elem.get('data-name') or '<unnamed>' |
| 143 | kind = native_replacement_kind(elem) |
| 144 | if not kind: |
| 145 | continue |
| 146 | for child in elem: |
| 147 | if child.tag.rsplit('}', 1)[-1] != 'metadata': |
| 148 | continue |
| 149 | try: |
| 150 | native_metadata_payload_matches(child, kind) |
| 151 | except NativeMarkerAttributeError as exc: |
| 152 | errors.append(f'{marker_id}: {exc}') |
| 153 | if not errors: |
| 154 | return |
| 155 | preview = '; '.join(errors[:8]) |
| 156 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 157 | raise SvgNativeConversionError( |
| 158 | f'{Path(svg_path).name}: invalid native replacement metadata: ' |
| 159 | f'{preview}{suffix}' |
| 160 | ) |
| 161 | |
| 162 | |
| 163 | def _require_inline_formula_markers( |
| 164 | root: ET.Element, |
| 165 | svg_path: Path | str, |
| 166 | ) -> None: |
| 167 | """Reject malformed inline formula runs before text lowering.""" |
| 168 | errors = inline_formula_marker_errors(root) |
| 169 | if not errors: |
| 170 | return |
| 171 | preview = '; '.join(errors[:8]) |
| 172 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 173 | raise SvgNativeConversionError( |
| 174 | f'{Path(svg_path).name}: invalid {INLINE_FORMULA_ATTR} marker(s): ' |
| 175 | f'{preview}{suffix}' |
| 176 | ) |
| 177 | |
| 178 | |
| 179 | def _require_project_hyperlinks( |
| 180 | root: ET.Element, |
| 181 | svg_path: Path | str, |
| 182 | *, |
| 183 | slide_count: int | None, |
| 184 | ) -> None: |
| 185 | """Reject hyperlinks that cannot be represented faithfully in PPTX.""" |
| 186 | errors = project_hyperlink_errors(root, slide_count=slide_count) |
| 187 | if not errors: |
| 188 | return |
| 189 | preview = '; '.join(errors[:8]) |
| 190 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 191 | raise SvgNativeConversionError( |
| 192 | f'{Path(svg_path).name}: invalid SVG hyperlink(s): {preview}{suffix}' |
| 193 | ) |
| 194 | |
| 195 | |
| 196 | def _native_replacement_enabled(elem: ET.Element, ctx: ConvertContext) -> bool: |
| 197 | """Return whether this marker is active under the current export policy.""" |
| 198 | kind = native_replacement_kind(elem) |
| 199 | if kind == 'formula': |
| 200 | return True |
| 201 | return ctx.native_objects_enabled and kind in {'chart', 'table'} |
| 202 | |
| 203 | |
| 204 | def _contains_enabled_native_replacement( |
| 205 | elem: ET.Element, |
| 206 | ctx: ConvertContext, |
| 207 | ) -> bool: |
| 208 | """Return whether one subtree contains an active native replacement.""" |
| 209 | return any( |
| 210 | descendant.tag.replace(f'{{{SVG_NS}}}', '') != 'metadata' |
| 211 | and _native_replacement_enabled(descendant, ctx) |
| 212 | for descendant in elem.iter() |
| 213 | ) |
| 214 | |
| 215 | |
| 216 | def _require_project_freeform_geometry( |
| 217 | root: ET.Element, |
| 218 | svg_path: Path | str, |
| 219 | ) -> None: |
| 220 | """Reject malformed path and points values with one aggregated error.""" |
| 221 | errors = project_freeform_geometry_errors(root) |
| 222 | if not errors: |
| 223 | return |
| 224 | preview = '; '.join(errors[:8]) |
| 225 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 226 | raise SvgNativeConversionError( |
| 227 | f'{Path(svg_path).name}: invalid project freeform geometry: ' |
| 228 | f'{preview}{suffix}' |
| 229 | ) |
| 230 | |
| 231 | |
| 232 | def _require_project_nested_svg_crops( |
| 233 | root: ET.Element, |
| 234 | svg_path: Path | str, |
| 235 | ) -> None: |
| 236 | """Reject nested SVG outside the imported picture-crop transport.""" |
| 237 | errors = project_nested_svg_crop_errors(root) |
| 238 | if not errors: |
| 239 | return |
| 240 | preview = '; '.join(errors[:8]) |
| 241 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 242 | raise SvgNativeConversionError( |
| 243 | f'{Path(svg_path).name}: invalid nested SVG crop wrapper(s): ' |
| 244 | f'{preview}{suffix}' |
| 245 | ) |
| 246 | |
| 247 | |
| 248 | def _require_project_clip_paths( |
| 249 | root: ET.Element, |
| 250 | svg_path: Path | str, |
| 251 | ) -> None: |
| 252 | """Reject clip references that cannot produce native picture geometry.""" |
| 253 | errors = project_clip_path_errors(root) |
| 254 | if not errors: |
| 255 | return |
| 256 | preview = '; '.join(errors[:8]) |
| 257 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 258 | raise SvgNativeConversionError( |
| 259 | f'{Path(svg_path).name}: invalid project clip-path(s): ' |
| 260 | f'{preview}{suffix}' |
| 261 | ) |
| 262 | |
| 263 | |
| 264 | def _require_project_images( |
| 265 | root: ET.Element, |
| 266 | svg_path: Path | str, |
| 267 | ) -> None: |
| 268 | """Reject invalid picture frames and unresolved or corrupt sources.""" |
| 269 | path = Path(svg_path) |
| 270 | errors = project_image_errors(root, path.parent) |
| 271 | if not errors: |
| 272 | return |
| 273 | preview = '; '.join(errors[:8]) |
| 274 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 275 | raise SvgNativeConversionError( |
| 276 | f'{path.name}: invalid project image(s): {preview}{suffix}' |
| 277 | ) |
| 278 | |
| 279 | |
| 280 | def _require_project_transforms( |
| 281 | root: ET.Element, |
| 282 | svg_path: Path | str, |
| 283 | ) -> None: |
| 284 | """Reject invalid project transform syntax and mappings before conversion.""" |
| 285 | errors = project_transform_errors(root) |
| 286 | if not errors: |
| 287 | return |
| 288 | preview = '; '.join(errors[:8]) |
| 289 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 290 | raise SvgNativeConversionError( |
| 291 | f'{Path(svg_path).name}: invalid project transform(s): ' |
| 292 | f'{preview}{suffix}' |
| 293 | ) |
| 294 | |
| 295 | |
| 296 | def _require_project_stroke_styles( |
| 297 | root: ET.Element, |
| 298 | svg_path: Path | str, |
| 299 | ) -> None: |
| 300 | """Reject invalid project line-style syntax and mappings before conversion.""" |
| 301 | errors = project_stroke_style_errors(root) |
| 302 | if not errors: |
| 303 | return |
| 304 | preview = '; '.join(errors[:8]) |
| 305 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 306 | raise SvgNativeConversionError( |
| 307 | f'{Path(svg_path).name}: invalid project line style(s): ' |
| 308 | f'{preview}{suffix}' |
| 309 | ) |
| 310 | |
| 311 | |
| 312 | def _require_project_image_aspect_ratios( |
| 313 | root: ET.Element, |
| 314 | svg_path: Path | str, |
| 315 | ) -> None: |
| 316 | """Reject ambiguous image fit/crop values before native conversion.""" |
| 317 | errors = project_image_aspect_ratio_errors(root) |
| 318 | if not errors: |
| 319 | return |
| 320 | preview = '; '.join(errors[:8]) |
| 321 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 322 | raise SvgNativeConversionError( |
| 323 | f'{Path(svg_path).name}: invalid project image aspect ratio(s): ' |
| 324 | f'{preview}{suffix}' |
| 325 | ) |
| 326 | |
| 327 | |
| 328 | def _require_project_opacities( |
| 329 | root: ET.Element, |
| 330 | svg_path: Path | str, |
| 331 | ) -> None: |
| 332 | """Reject malformed opacity values before native conversion.""" |
| 333 | errors = project_opacity_errors(root) |
| 334 | if not errors: |
| 335 | return |
| 336 | preview = '; '.join(errors[:8]) |
| 337 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 338 | raise SvgNativeConversionError( |
| 339 | f'{Path(svg_path).name}: invalid project opacity value(s): ' |
| 340 | f'{preview}{suffix}' |
| 341 | ) |
| 342 | |
| 343 | |
| 344 | def _require_project_paints( |
| 345 | root: ET.Element, |
| 346 | svg_path: Path | str, |
| 347 | ) -> None: |
| 348 | """Reject invalid paint values before native conversion.""" |
| 349 | errors = project_paint_errors(root) |
| 350 | if not errors: |
| 351 | return |
| 352 | preview = '; '.join(errors[:8]) |
| 353 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 354 | raise SvgNativeConversionError( |
| 355 | f'{Path(svg_path).name}: invalid project paint value(s): ' |
| 356 | f'{preview}{suffix}' |
| 357 | ) |
| 358 | |
| 359 | |
| 360 | def _require_project_masks( |
| 361 | root: ET.Element, |
| 362 | svg_path: Path | str, |
| 363 | ) -> None: |
| 364 | """Reject SVG masks before native conversion can silently drop them.""" |
| 365 | errors = project_mask_errors(root) |
| 366 | if not errors: |
| 367 | return |
| 368 | preview = '; '.join(errors[:8]) |
| 369 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 370 | raise SvgNativeConversionError( |
| 371 | f'{Path(svg_path).name}: invalid project mask(s): ' |
| 372 | f'{preview}{suffix}' |
| 373 | ) |
| 374 | |
| 375 | |
| 376 | def _require_project_definitions( |
| 377 | root: ET.Element, |
| 378 | svg_path: Path | str, |
| 379 | ) -> None: |
| 380 | """Reject definitions outside the direct, unique local-ref contract.""" |
| 381 | errors = project_definition_errors(root) |
| 382 | if not errors: |
| 383 | return |
| 384 | preview = '; '.join(errors[:8]) |
| 385 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 386 | raise SvgNativeConversionError( |
| 387 | f'{Path(svg_path).name}: invalid project definition(s): ' |
| 388 | f'{preview}{suffix}' |
| 389 | ) |
| 390 | |
| 391 | |
| 392 | def _require_project_paint_references( |
| 393 | root: ET.Element, |
| 394 | svg_path: Path | str, |
| 395 | ) -> None: |
| 396 | """Reject unresolved or context-invalid local paint references.""" |
| 397 | errors = project_paint_reference_errors(root) |
| 398 | if not errors: |
| 399 | return |
| 400 | preview = '; '.join(errors[:8]) |
| 401 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 402 | raise SvgNativeConversionError( |
| 403 | f'{Path(svg_path).name}: invalid project paint reference(s): ' |
| 404 | f'{preview}{suffix}' |
| 405 | ) |
| 406 | |
| 407 | |
| 408 | def _require_project_line_end_markers( |
| 409 | root: ET.Element, |
| 410 | svg_path: Path | str, |
| 411 | ) -> None: |
| 412 | """Reject markers outside the native line-end contract.""" |
| 413 | errors = project_marker_errors(root) |
| 414 | if not errors: |
| 415 | return |
| 416 | preview = '; '.join(errors[:8]) |
| 417 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 418 | raise SvgNativeConversionError( |
| 419 | f'{Path(svg_path).name}: invalid project line-end marker(s): ' |
| 420 | f'{preview}{suffix}' |
| 421 | ) |
| 422 | |
| 423 | |
| 424 | def _require_project_gradients( |
| 425 | root: ET.Element, |
| 426 | svg_path: Path | str, |
| 427 | ) -> None: |
| 428 | """Reject gradients outside the normalized native interface.""" |
| 429 | errors = project_gradient_errors(root) + project_gradient_geometry_errors(root) |
| 430 | if not errors: |
| 431 | return |
| 432 | preview = '; '.join(errors[:8]) |
| 433 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 434 | raise SvgNativeConversionError( |
| 435 | f'{Path(svg_path).name}: invalid project gradient(s): ' |
| 436 | f'{preview}{suffix}' |
| 437 | ) |
| 438 | |
| 439 | |
| 440 | def _require_project_filters( |
| 441 | root: ET.Element, |
| 442 | svg_path: Path | str, |
| 443 | ) -> None: |
| 444 | """Reject filters outside the native shadow/glow interface.""" |
| 445 | errors = project_filter_errors(root) |
| 446 | if not errors: |
| 447 | return |
| 448 | preview = '; '.join(errors[:8]) |
| 449 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 450 | raise SvgNativeConversionError( |
| 451 | f'{Path(svg_path).name}: invalid project filter(s): ' |
| 452 | f'{preview}{suffix}' |
| 453 | ) |
| 454 | |
| 455 | |
| 456 | def _require_project_effect_status( |
| 457 | root: ET.Element, |
| 458 | svg_path: Path | str, |
| 459 | ) -> None: |
| 460 | """Reject source effects that the importer cannot map without distortion.""" |
| 461 | errors = project_effect_status_errors(root) |
| 462 | if not errors: |
| 463 | return |
| 464 | preview = '; '.join(errors[:8]) |
| 465 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 466 | raise SvgNativeConversionError( |
| 467 | f'{Path(svg_path).name}: unsupported imported PPTX effect(s): ' |
| 468 | f'{preview}{suffix}' |
| 469 | ) |
| 470 | |
| 471 | |
| 472 | def _require_project_text_properties( |
| 473 | root: ET.Element, |
| 474 | svg_path: Path | str, |
| 475 | ) -> None: |
| 476 | """Reject text declarations outside the closed DrawingML mapping.""" |
| 477 | errors = project_text_property_errors(root) |
| 478 | if not errors: |
| 479 | return |
| 480 | preview = '; '.join(errors[:8]) |
| 481 | suffix = '' if len(errors) <= 8 else f'; +{len(errors) - 8} more' |
| 482 | raise SvgNativeConversionError( |
| 483 | f'{Path(svg_path).name}: invalid project text property(s): ' |
| 484 | f'{preview}{suffix}' |
| 485 | ) |
| 486 | |
| 487 | |
| 488 | # --------------------------------------------------------------------------- |
| 489 | # Transform & layout helpers |
| 490 | # --------------------------------------------------------------------------- |
| 491 | |
| 492 | def parse_transform(transform_str: str) -> tuple[float, float, float, float, float]: |
| 493 | """Parse an SVG transform list into (dx, dy, sx, sy, angle_deg). |
| 494 | |
| 495 | Composes every translate/scale/rotate/matrix operation rather than picking |
| 496 | the first occurrence — needed for idioms like |
| 497 | ``translate(cx cy) scale(-1 -1) translate(-cx -cy)`` which encode a flip |
| 498 | around a non-origin pivot. |
| 499 | |
| 500 | When the composed matrix has no rotation, the decomposition preserves |
| 501 | signed scale for flips. With rotation, scale uses the column magnitudes and |
| 502 | angle uses the first transformed axis. Zero or non-orthogonal axes fail |
| 503 | before decomposition because this tuple cannot represent them faithfully. |
| 504 | """ |
| 505 | if not transform_str: |
| 506 | return 0.0, 0.0, 1.0, 1.0, 0.0 |
| 507 | |
| 508 | matrix = parse_transform_matrix(transform_str) |
| 509 | validate_dml_shape_matrix(matrix) |
| 510 | a, b, c, d, e, f = matrix |
| 511 | |
| 512 | # No shear / rotation: direct decomposition preserves the original signs of |
| 513 | # sx / sy. ctx_x / ctx_y use the simple ``val * sx + tx`` formula, so this |
| 514 | # is the only form that survives flip-around-pivot composites without |
| 515 | # collapsing them into a rotation that the consumer can't honour. |
| 516 | if abs(b) < 1e-9 and abs(c) < 1e-9: |
| 517 | sx = a if a != 0 else 1.0 |
| 518 | sy = d if d != 0 else 1.0 |
| 519 | return e, f, sx, sy, 0.0 |
| 520 | |
| 521 | sx = math.hypot(a, b) |
| 522 | sy = math.hypot(c, d) |
| 523 | if sx == 0: |
| 524 | sx = 1.0 |
| 525 | if sy == 0: |
| 526 | sy = 1.0 |
| 527 | |
| 528 | angle_deg = math.degrees(math.atan2(b, a)) |
| 529 | return e, f, sx, sy, angle_deg |
| 530 | |
| 531 | |
| 532 | # ``rotate(angle)`` defaults to pivot (0,0); ``rotate(angle, cx, cy)`` rotates |
| 533 | # around (cx, cy). DrawingML grpSp ``rot`` always rotates around the group's |
| 534 | # own bounding-box centre — we need the SVG pivot so ``convert_g`` can |
| 535 | # compensate for the offset between those two centres. |
| 536 | def _root_viewport_size(root: ET.Element) -> tuple[float, float]: |
| 537 | """Return the SVG root viewport size in user units.""" |
| 538 | viewbox = parse_project_viewbox(root.get('viewBox')) |
| 539 | return float(viewbox.width), float(viewbox.height) |
| 540 | |
| 541 | |
| 542 | def _extract_rotate_pivot(transform_str: str) -> tuple[float, float] | None: |
| 543 | """Return the (cx, cy) pivot of a sole ``rotate(...)`` in *transform_str*. |
| 544 | |
| 545 | Returns ``None`` when the transform list contains anything other than one |
| 546 | rotate (other ops compose with rotate in a way the pivot-compensation |
| 547 | fallback can't express). A bare ``rotate(angle)`` returns (0, 0). |
| 548 | """ |
| 549 | if not transform_str: |
| 550 | return None |
| 551 | operations = parse_transform_operations(transform_str) |
| 552 | if len(operations) != 1 or operations[0][0] != 'rotate': |
| 553 | return None |
| 554 | args = operations[0][1] |
| 555 | cx = args[1] if len(args) == 3 else 0.0 |
| 556 | cy = args[2] if len(args) == 3 else 0.0 |
| 557 | return cx, cy |
| 558 | |
| 559 | |
| 560 | def _txbody_metadata(elem: ET.Element) -> ET.Element | None: |
| 561 | for child in elem: |
| 562 | if ( |
| 563 | child.tag.replace(f'{{{SVG_NS}}}', '') == 'metadata' |
| 564 | and child.get('data-pptx-part') == 'txbody' |
| 565 | ): |
| 566 | return child |
| 567 | return None |
| 568 | |
| 569 | |
| 570 | _TXBODY_UNCHANGED_ATTR = 'data-pptx-runtime-txbody-unchanged' |
| 571 | _PREVIEW_UNCHANGED_ATTR = 'data-pptx-runtime-preview-unchanged' |
| 572 | |
| 573 | |
| 574 | def _mark_unchanged_txbody_groups(root: ET.Element) -> None: |
| 575 | """Snapshot author-visible text state before exporter preprocessing.""" |
| 576 | for group in root.iter(): |
| 577 | if group.tag.replace(f'{{{SVG_NS}}}', '') != 'g': |
| 578 | continue |
| 579 | metadata = _txbody_metadata(group) |
| 580 | if metadata is None: |
| 581 | continue |
| 582 | expected = metadata.get('data-pptx-text-sha256') |
| 583 | actual = svg_text_fingerprint(group) |
| 584 | group.set(_TXBODY_UNCHANGED_ATTR, '1' if expected == actual else '0') |
| 585 | |
| 586 | |
| 587 | def _mark_unchanged_preset_previews(root: ET.Element) -> None: |
| 588 | """Snapshot visible preset layers before exporter preprocessing.""" |
| 589 | for group in root.iter(): |
| 590 | if group.tag.replace(f'{{{SVG_NS}}}', '') != 'g': |
| 591 | continue |
| 592 | if ( |
| 593 | group.get('data-pptx-object') not in {'shape', 'connector'} |
| 594 | or group.get('data-pptx-prst') is None |
| 595 | ): |
| 596 | continue |
| 597 | try: |
| 598 | expected = resolve_preset_preview_hash(group) |
| 599 | except ValueError as exc: |
| 600 | raise SvgNativeConversionError( |
| 601 | f'Invalid preset preview fingerprint contract: {exc}' |
| 602 | ) from exc |
| 603 | if expected is None: |
| 604 | continue |
| 605 | actual = svg_preset_preview_fingerprint(group) |
| 606 | group.set(_PREVIEW_UNCHANGED_ATTR, '1' if expected == actual else '0') |
| 607 | |
| 608 | |
| 609 | def _require_unchanged_preset_preview(group: ET.Element) -> None: |
| 610 | try: |
| 611 | expected = resolve_preset_preview_hash(group) |
| 612 | except ValueError as exc: |
| 613 | raise SvgNativeConversionError( |
| 614 | f'Invalid preset preview fingerprint contract: {exc}' |
| 615 | ) from exc |
| 616 | if expected is None: |
| 617 | return |
| 618 | snapshot = group.get(_PREVIEW_UNCHANGED_ATTR) |
| 619 | if snapshot == '1': |
| 620 | return |
| 621 | if snapshot is None and svg_preset_preview_fingerprint(group) == expected: |
| 622 | return |
| 623 | raise SvgNativeConversionError( |
| 624 | 'Visible preset preview was edited without updating its native ' |
| 625 | 'data-pptx-prst/frame/adjustment carrier; export stopped to avoid ' |
| 626 | 'silently discarding the SVG edit' |
| 627 | ) |
| 628 | |
| 629 | |
| 630 | def _decode_unchanged_txbody( |
| 631 | group: ET.Element, |
| 632 | metadata: ET.Element, |
| 633 | *, |
| 634 | trust_runtime_snapshot: bool = True, |
| 635 | ) -> tuple[str, bool] | None: |
| 636 | expected_hash = metadata.get('data-pptx-text-sha256') |
| 637 | if not expected_hash: |
| 638 | raise SvgNativeConversionError('txbody metadata requires a text hash') |
| 639 | snapshot = ( |
| 640 | group.get(_TXBODY_UNCHANGED_ATTR) |
| 641 | if trust_runtime_snapshot else None |
| 642 | ) |
| 643 | unchanged = snapshot == '1' or ( |
| 644 | snapshot != '0' |
| 645 | and svg_text_fingerprint(group) == expected_hash |
| 646 | ) |
| 647 | if metadata.get('data-pptx-encoding') != 'base64': |
| 648 | raise SvgNativeConversionError('txbody metadata requires base64 encoding') |
| 649 | try: |
| 650 | raw = base64.b64decode((metadata.text or '').strip(), validate=True) |
| 651 | txbody = ET.fromstring(raw) |
| 652 | decoded = raw.decode('utf-8') |
| 653 | except (ValueError, binascii.Error, UnicodeDecodeError, ET.ParseError) as exc: |
| 654 | raise SvgNativeConversionError(f'Invalid txbody metadata: {exc}') from exc |
| 655 | if txbody.tag != ( |
| 656 | '{http://schemas.openxmlformats.org/presentationml/2006/main}txBody' |
| 657 | ): |
| 658 | raise SvgNativeConversionError('txbody metadata payload must be p:txBody') |
| 659 | if has_relationship_attributes(txbody): |
| 660 | raise SvgNativeConversionError( |
| 661 | 'txbody metadata must not contain part-local relationship attributes' |
| 662 | ) |
| 663 | if not unchanged: |
| 664 | if txbody_has_run_effects(txbody): |
| 665 | raise SvgNativeConversionError( |
| 666 | 'Visible text or typography was edited while the source ' |
| 667 | 'txBody contains run-level effects; export stopped to avoid ' |
| 668 | 'silently discarding those effects' |
| 669 | ) |
| 670 | return None |
| 671 | return decoded, txbody_has_run_effects(txbody) |
| 672 | |
| 673 | |
| 674 | def _append_shape_text( |
| 675 | shape: ShapeResult, |
| 676 | txbody_xml: str, |
| 677 | ) -> ShapeResult: |
| 678 | if not shape.xml.lstrip().startswith('<p:sp>') or not shape.xml.rstrip().endswith('</p:sp>'): |
| 679 | raise SvgNativeConversionError('Native txBody can only attach to p:sp') |
| 680 | closing = shape.xml.rfind('</p:sp>') |
| 681 | return ShapeResult( |
| 682 | xml=( |
| 683 | shape.xml[:closing] |
| 684 | + txbody_xml |
| 685 | + '\n' |
| 686 | + shape.xml[closing:] |
| 687 | ), |
| 688 | bounds_emu=shape.bounds_emu, |
| 689 | ) |
| 690 | |
| 691 | |
| 692 | def preserved_native_text_body( |
| 693 | group: ET.Element, |
| 694 | *, |
| 695 | trust_runtime_snapshot: bool = True, |
| 696 | ) -> tuple[ET.Element, str] | None: |
| 697 | """Return the geometry carrier and unchanged native text body, if usable.""" |
| 698 | metadata = _txbody_metadata(group) |
| 699 | logical_text_shape = ( |
| 700 | group.get('data-pptx-object') == 'shape' |
| 701 | and ( |
| 702 | group.get('data-pptx-prst') is not None |
| 703 | or group.get('data-pptx-geometry-kind') == 'custom' |
| 704 | ) |
| 705 | and metadata is not None |
| 706 | ) |
| 707 | if not logical_text_shape: |
| 708 | return None |
| 709 | decoded_text = _decode_unchanged_txbody( |
| 710 | group, |
| 711 | metadata, |
| 712 | trust_runtime_snapshot=trust_runtime_snapshot, |
| 713 | ) |
| 714 | carrier_children = [ |
| 715 | child for child in group |
| 716 | if child.get('data-pptx-part') == 'geometry' |
| 717 | ] |
| 718 | allowed_parts = { |
| 719 | 'geometry', |
| 720 | 'geometry-detail', |
| 721 | 'geometry-preview', |
| 722 | 'txbody', |
| 723 | } |
| 724 | has_foreign_visual = any( |
| 725 | child.tag.replace(f'{{{SVG_NS}}}', '') not in {'text', 'metadata'} |
| 726 | and child.get('data-pptx-part') not in allowed_parts |
| 727 | for child in group |
| 728 | ) |
| 729 | if decoded_text is None: |
| 730 | return None |
| 731 | native_text, has_run_effects = decoded_text |
| 732 | if len(carrier_children) != 1 or has_foreign_visual: |
| 733 | if has_run_effects: |
| 734 | raise SvgNativeConversionError( |
| 735 | 'The source txBody contains run-level effects but cannot be ' |
| 736 | 'restored as one native text shape; export stopped to avoid ' |
| 737 | 'silently discarding those effects' |
| 738 | ) |
| 739 | return None |
| 740 | return carrier_children[0], native_text |
| 741 | |
| 742 | |
| 743 | # --------------------------------------------------------------------------- |
| 744 | # Group handling |
| 745 | # --------------------------------------------------------------------------- |
| 746 | |
| 747 | def convert_g(elem: ET.Element, ctx: ConvertContext) -> ShapeResult | None: |
| 748 | """Convert SVG <g> to DrawingML group shape <p:grpSp>. |
| 749 | |
| 750 | Preserves group structure so elements can be selected and moved together |
| 751 | in PowerPoint. Single-child groups are flattened to avoid unnecessary nesting. |
| 752 | |
| 753 | Uses identity coordinate mapping (chOff/chExt == off/ext) so child shapes |
| 754 | keep their absolute slide coordinates unchanged. |
| 755 | """ |
| 756 | transform = elem.get('transform', '') |
| 757 | native_subtree_active = _contains_enabled_native_replacement(elem, ctx) |
| 758 | if native_subtree_active: |
| 759 | dx, dy, sx, sy = native_marker_transform(transform) |
| 760 | angle_deg = 0.0 |
| 761 | else: |
| 762 | dx, dy, sx, sy, angle_deg = parse_transform(transform) |
| 763 | |
| 764 | filter_id = resolve_url_id(elem.get('filter', '')) |
| 765 | style_overrides = _extract_inheritable_styles(elem) |
| 766 | local_opacity = get_element_opacity(elem) |
| 767 | if local_opacity is None: |
| 768 | local_opacity = 1.0 |
| 769 | |
| 770 | elem_id = usable_animation_group_id(elem.get('id')) |
| 771 | semantic_role = elem.get('data-pptx-role') |
| 772 | placeholder = elem.get('data-pptx-placeholder') |
| 773 | has_explicit_semantics = ( |
| 774 | semantic_role is not None or placeholder is not None |
| 775 | ) |
| 776 | is_chrome = ( |
| 777 | is_static_page_frame(semantic_role, placeholder) |
| 778 | if has_explicit_semantics |
| 779 | else is_chrome_id(elem_id) |
| 780 | ) |
| 781 | should_animate_group = ( |
| 782 | ctx.depth == 0 |
| 783 | and elem_id |
| 784 | and ( |
| 785 | not is_chrome |
| 786 | or ( |
| 787 | not has_explicit_semantics |
| 788 | and elem_id in ctx.animation_group_overrides |
| 789 | ) |
| 790 | ) |
| 791 | and elem.get('data-pptx-layer') is None |
| 792 | ) |
| 793 | visual_children = [ |
| 794 | child for child in elem |
| 795 | if child.tag.replace(f'{{{SVG_NS}}}', '') not in _NON_VISUAL_TAGS |
| 796 | ] |
| 797 | unit_axis_reflection = ( |
| 798 | bool(transform) |
| 799 | and _is_unit_axis_reflection(parse_transform_operations(transform)) |
| 800 | ) |
| 801 | matrix_supported = ( |
| 802 | not native_subtree_active |
| 803 | and bool(transform) |
| 804 | and visual_children |
| 805 | and ( |
| 806 | supports_full_project_transform(elem) |
| 807 | or unit_axis_reflection |
| 808 | ) |
| 809 | ) |
| 810 | # A pure ``rotate(angle [cx cy])`` falls through to the fallback path |
| 811 | # below (children are rect/text/path/etc. that don't consume a full |
| 812 | # matrix). Decomposing the matrix produces translation components |
| 813 | # (e, f) that encode the pivot — handing those to children would |
| 814 | # *double-translate* them because grpSp's own ``rot`` already |
| 815 | # rotates around the group's bounding-box centre. Skip the child |
| 816 | # translation here and apply pivot-centre compensation to ``a:off`` |
| 817 | # below instead. |
| 818 | rotate_pivot = _extract_rotate_pivot(transform) if not matrix_supported else None |
| 819 | if rotate_pivot is not None: |
| 820 | angle_deg = parse_transform_operations(transform)[0][1][0] |
| 821 | if matrix_supported: |
| 822 | child_ctx = ctx.child( |
| 823 | 0, 0, 1.0, 1.0, |
| 824 | transform_matrix=parse_transform_matrix(transform), |
| 825 | filter_id=filter_id, |
| 826 | style_overrides=style_overrides, |
| 827 | opacity_multiplier=local_opacity, |
| 828 | ) |
| 829 | elif rotate_pivot is not None: |
| 830 | child_ctx = ctx.child( |
| 831 | 0, 0, 1.0, 1.0, |
| 832 | filter_id=filter_id, |
| 833 | style_overrides=style_overrides, |
| 834 | opacity_multiplier=local_opacity, |
| 835 | ) |
| 836 | else: |
| 837 | child_ctx = ctx.child( |
| 838 | ctx.scale_x * dx if native_subtree_active else dx, |
| 839 | ctx.scale_y * dy if native_subtree_active else dy, |
| 840 | sx, |
| 841 | sy, |
| 842 | filter_id=filter_id, |
| 843 | style_overrides=style_overrides, |
| 844 | opacity_multiplier=local_opacity, |
| 845 | ) |
| 846 | |
| 847 | if native_subtree_active and child_ctx.opacity_multiplier < 1.0: |
| 848 | raise SvgNativeConversionError( |
| 849 | "Group opacity cannot be applied to an active native replacement; " |
| 850 | "remove the group opacity or remove data-pptx-replace-with to keep " |
| 851 | "the subtree as ordinary SVG" |
| 852 | ) |
| 853 | |
| 854 | if _native_replacement_enabled(elem, child_ctx): |
| 855 | native_result = convert_native_object(elem, child_ctx) |
| 856 | if native_result: |
| 857 | ctx.sync_from_child(child_ctx) |
| 858 | if should_animate_group: |
| 859 | shape_match = re.search(r'<p:cNvPr id="(\d+)"', native_result.xml) |
| 860 | if shape_match: |
| 861 | ctx.anim_targets.append((int(shape_match.group(1)), elem_id)) |
| 862 | return native_result |
| 863 | |
| 864 | if ( |
| 865 | elem.get('data-pptx-object') in {'shape', 'connector'} |
| 866 | and elem.get('data-pptx-prst') is not None |
| 867 | ): |
| 868 | _require_unchanged_preset_preview(elem) |
| 869 | |
| 870 | preserved_text = preserved_native_text_body(elem) |
| 871 | if preserved_text is not None: |
| 872 | geometry_carrier, native_text = preserved_text |
| 873 | geometry_ctx = child_ctx |
| 874 | if transform and not native_subtree_active: |
| 875 | geometry_ctx = ctx.child( |
| 876 | 0, 0, 1.0, 1.0, |
| 877 | transform_matrix=parse_transform_matrix(transform), |
| 878 | filter_id=filter_id, |
| 879 | style_overrides=style_overrides, |
| 880 | opacity_multiplier=local_opacity, |
| 881 | ) |
| 882 | geometry_result = convert_element(geometry_carrier, geometry_ctx) |
| 883 | ctx.sync_from_child(geometry_ctx) |
| 884 | if geometry_result is None: |
| 885 | raise SvgNativeConversionError( |
| 886 | 'Logical text shape has no convertible geometry carrier' |
| 887 | ) |
| 888 | restored = _append_shape_text( |
| 889 | geometry_result, |
| 890 | native_text, |
| 891 | ) |
| 892 | if should_animate_group and elem_id: |
| 893 | shape_match = re.search(r'<p:cNvPr id="(\d+)"', restored.xml) |
| 894 | if shape_match: |
| 895 | ctx.anim_targets.append((int(shape_match.group(1)), elem_id)) |
| 896 | return restored |
| 897 | |
| 898 | child_results: list[ShapeResult] = [] |
| 899 | for child in elem: |
| 900 | result = convert_element(child, child_ctx) |
| 901 | if result: |
| 902 | child_results.append(result) |
| 903 | |
| 904 | ctx.sync_from_child(child_ctx) |
| 905 | |
| 906 | if not child_results: |
| 907 | return None |
| 908 | |
| 909 | # A logical imported preset may contain several render-only SVG detail |
| 910 | # paths, but after those are skipped it owns exactly one native object. |
| 911 | # Flatten that carrier even at the top level; otherwise animation grouping |
| 912 | # would turn one source ``p:sp`` into a ``p:grpSp`` wrapper. Retarget an |
| 913 | # optional animation to the restored leaf shape ID. |
| 914 | logical_native_shape_group = ( |
| 915 | elem.get('data-pptx-object') in {'shape', 'connector'} |
| 916 | and ( |
| 917 | elem.get('data-pptx-prst') is not None |
| 918 | or elem.get('data-pptx-geometry-kind') == 'custom' |
| 919 | ) |
| 920 | ) |
| 921 | logical_picture_effect_group = ( |
| 922 | filter_id is not None |
| 923 | and is_picture_effect_carrier(elem) |
| 924 | ) |
| 925 | explicit_native_group = elem.get('data-pptx-object') == 'group' |
| 926 | if ( |
| 927 | len(child_results) == 1 |
| 928 | and not explicit_native_group |
| 929 | and ( |
| 930 | not should_animate_group |
| 931 | or logical_native_shape_group |
| 932 | or logical_picture_effect_group |
| 933 | ) |
| 934 | ): |
| 935 | if should_animate_group and elem_id: |
| 936 | shape_match = re.search(r'<p:cNvPr id="(\d+)"', child_results[0].xml) |
| 937 | if shape_match: |
| 938 | ctx.anim_targets.append((int(shape_match.group(1)), elem_id)) |
| 939 | return child_results[0] |
| 940 | |
| 941 | # Multiple children, or a top-level semantic one-child group: wrap in |
| 942 | # <p:grpSp> so PowerPoint can animate the group as one unit. |
| 943 | min_x = min_y = float('inf') |
| 944 | max_x = max_y = float('-inf') |
| 945 | |
| 946 | for child_result in child_results: |
| 947 | bounds = child_result.bounds_emu |
| 948 | if bounds is None: |
| 949 | continue |
| 950 | min_x = min(min_x, bounds[0]) |
| 951 | min_y = min(min_y, bounds[1]) |
| 952 | max_x = max(max_x, bounds[2]) |
| 953 | max_y = max(max_y, bounds[3]) |
| 954 | |
| 955 | if min_x == float('inf'): |
| 956 | return ShapeResult(xml='\n'.join(result.xml for result in child_results)) |
| 957 | |
| 958 | group_x = int(min_x) |
| 959 | group_y = int(min_y) |
| 960 | group_w = max(int(max_x - min_x), 1) |
| 961 | group_h = max(int(max_y - min_y), 1) |
| 962 | |
| 963 | # ``rotate(angle, cx, cy)`` rotates around the SVG pivot, but DrawingML |
| 964 | # grpSp ``rot`` always rotates around the group's own bbox centre. When |
| 965 | # those centres differ, the visual position drifts by exactly the |
| 966 | # translation a rotate-around-pivot equals. Compensate by offsetting the |
| 967 | # outer <a:off> only; <a:chOff> stays on the unshifted bbox so children |
| 968 | # (still at their original SVG positions because rotate_pivot suppressed |
| 969 | # the dx/dy translation above) remain aligned inside the group. |
| 970 | off_x = group_x |
| 971 | off_y = group_y |
| 972 | if rotate_pivot is not None and angle_deg: |
| 973 | cx_svg, cy_svg = rotate_pivot |
| 974 | pivot_ex = (cx_svg + ctx.translate_x) * EMU_PER_PX |
| 975 | pivot_ey = (cy_svg + ctx.translate_y) * EMU_PER_PX |
| 976 | bbox_cx = group_x + group_w / 2 |
| 977 | bbox_cy = group_y + group_h / 2 |
| 978 | theta = math.radians(angle_deg) |
| 979 | cos_t = math.cos(theta) |
| 980 | sin_t = math.sin(theta) |
| 981 | # Where the bbox centre lands after rotating around the pivot, minus |
| 982 | # where DrawingML's grpSp rot would leave it (i.e. unchanged). |
| 983 | delta_x = (bbox_cx - pivot_ex) * cos_t - (bbox_cy - pivot_ey) * sin_t + pivot_ex - bbox_cx |
| 984 | delta_y = (bbox_cx - pivot_ex) * sin_t + (bbox_cy - pivot_ey) * cos_t + pivot_ey - bbox_cy |
| 985 | off_x = int(round(group_x + delta_x)) |
| 986 | off_y = int(round(group_y + delta_y)) |
| 987 | |
| 988 | shapes_xml = '\n'.join(result.xml for result in child_results) |
| 989 | group_id = ( |
| 990 | ctx.claim_shape_id( |
| 991 | elem.get('data-pptx-shape-id'), |
| 992 | elem.get('data-pptx-shape-scope'), |
| 993 | ) |
| 994 | if elem.get('data-pptx-object') == 'group' |
| 995 | else ctx.next_id() |
| 996 | ) |
| 997 | |
| 998 | # Record top-level semantic groups (e.g. <g id="p02-title">) so the |
| 999 | # PPTX builder can emit per-element object timing. Only the outermost |
| 1000 | # multi-child wrapper qualifies — flattened single-child groups have no |
| 1001 | # <p:grpSp> to anchor a timing target on, and nested groups are |
| 1002 | # ignored to keep the animation budget at ~per-section granularity. |
| 1003 | if should_animate_group: |
| 1004 | ctx.anim_targets.append((group_id, elem_id)) |
| 1005 | |
| 1006 | group_effect = '' |
| 1007 | if filter_id and filter_id in ctx.defs: |
| 1008 | group_effect = build_effect_xml( |
| 1009 | ctx.defs[filter_id], |
| 1010 | child_ctx.opacity_multiplier, |
| 1011 | ) |
| 1012 | |
| 1013 | rot_emu = 0 if matrix_supported else int(angle_deg * 60000) |
| 1014 | rot_attr = f' rot="{rot_emu}"' if rot_emu else '' |
| 1015 | validate_ooxml_xfrm(off_x, off_y, group_w, group_h) |
| 1016 | validate_ooxml_xfrm(group_x, group_y, group_w, group_h) |
| 1017 | |
| 1018 | return ShapeResult(xml=f'''<p:grpSp> |
| 1019 | <p:nvGrpSpPr> |
| 1020 | <p:cNvPr id="{group_id}" name="Group {group_id}"/> |
| 1021 | <p:cNvGrpSpPr/> |
| 1022 | <p:nvPr/> |
| 1023 | </p:nvGrpSpPr> |
| 1024 | <p:grpSpPr> |
| 1025 | <a:xfrm{rot_attr}> |
| 1026 | <a:off x="{off_x}" y="{off_y}"/> |
| 1027 | <a:ext cx="{group_w}" cy="{group_h}"/> |
| 1028 | <a:chOff x="{group_x}" y="{group_y}"/> |
| 1029 | <a:chExt cx="{group_w}" cy="{group_h}"/> |
| 1030 | </a:xfrm> |
| 1031 | {group_effect} |
| 1032 | </p:grpSpPr> |
| 1033 | {shapes_xml} |
| 1034 | </p:grpSp>''', bounds_emu=(group_x, group_y, group_x + group_w, group_y + group_h)) |
| 1035 | |
| 1036 | |
| 1037 | def convert_a(elem: ET.Element, ctx: ConvertContext) -> ShapeResult | None: |
| 1038 | """Convert one standard SVG anchor into a clickable DrawingML object.""" |
| 1039 | result = convert_g(elem, ctx) |
| 1040 | if result is None: |
| 1041 | return None |
| 1042 | return apply_shape_hyperlink(result, ctx, svg_hyperlink_href(elem)) |
| 1043 | |
| 1044 | |
| 1045 | # --------------------------------------------------------------------------- |
| 1046 | # Defs collection & element dispatch |
| 1047 | # --------------------------------------------------------------------------- |
| 1048 | |
| 1049 | _NON_VISUAL_TAGS = frozenset(('defs', 'title', 'desc', 'metadata', 'style')) |
| 1050 | |
| 1051 | _CONVERTERS = { |
| 1052 | 'rect': convert_rect, |
| 1053 | 'circle': convert_circle, |
| 1054 | 'ellipse': convert_ellipse, |
| 1055 | 'line': convert_line, |
| 1056 | 'path': convert_path, |
| 1057 | 'polygon': convert_polygon, |
| 1058 | 'polyline': convert_polyline, |
| 1059 | 'text': convert_text, |
| 1060 | 'image': convert_image, |
| 1061 | 'g': convert_g, |
| 1062 | 'a': convert_a, |
| 1063 | 'svg': convert_nested_svg, |
| 1064 | } |
| 1065 | |
| 1066 | _SUPPORTED_VISUAL_CHILD_TAGS = frozenset(('tspan',)) |
| 1067 | |
| 1068 | |
| 1069 | def _parse_svg_canvas(root: ET.Element) -> tuple[float, float, float, float]: |
| 1070 | """Return the SVG canvas as (x, y, width, height) in SVG units.""" |
| 1071 | viewbox = parse_project_viewbox(root.get('viewBox')) |
| 1072 | return 0.0, 0.0, float(viewbox.width), float(viewbox.height) |
| 1073 | |
| 1074 | |
| 1075 | def _is_full_canvas_rect( |
| 1076 | elem: ET.Element, |
| 1077 | ctx: ConvertContext, |
| 1078 | canvas: tuple[float, float, float, float], |
| 1079 | ) -> bool: |
| 1080 | """Return whether a rect is a safe candidate for native slide background.""" |
| 1081 | if elem.get('transform') or elem.get('filter') or elem.get('clip-path'): |
| 1082 | return False |
| 1083 | if any( |
| 1084 | elem.get(attr) is not None |
| 1085 | for attr in ( |
| 1086 | 'data-pptx-object', |
| 1087 | 'data-pptx-prst', |
| 1088 | 'data-pptx-frame', |
| 1089 | 'data-pptx-geometry-status', |
| 1090 | ) |
| 1091 | ): |
| 1092 | return False |
| 1093 | if ( |
| 1094 | parse_svg_length(elem.get('rx'), 0.0) > 0 |
| 1095 | or parse_svg_length(elem.get('ry'), 0.0) > 0 |
| 1096 | ): |
| 1097 | return False |
| 1098 | |
| 1099 | canvas_x, canvas_y, canvas_w, canvas_h = canvas |
| 1100 | if canvas_w <= 0 or canvas_h <= 0: |
| 1101 | return False |
| 1102 | |
| 1103 | tolerance = 0.5 |
| 1104 | if abs(parse_svg_length(elem.get('x'), 0.0) - canvas_x) > tolerance: |
| 1105 | return False |
| 1106 | if abs(parse_svg_length(elem.get('y'), 0.0) - canvas_y) > tolerance: |
| 1107 | return False |
| 1108 | if abs(parse_svg_length(elem.get('width'), 0.0) - canvas_w) > tolerance: |
| 1109 | return False |
| 1110 | if abs(parse_svg_length(elem.get('height'), 0.0) - canvas_h) > tolerance: |
| 1111 | return False |
| 1112 | |
| 1113 | fill = _get_attr(elem, 'fill', ctx) |
| 1114 | if fill == 'none': |
| 1115 | return False |
| 1116 | |
| 1117 | stroke = _get_attr(elem, 'stroke', ctx) |
| 1118 | stroke_width = parse_svg_length(_get_attr(elem, 'stroke-width', ctx), 1.0) |
| 1119 | stroke_opacity = get_stroke_opacity(elem, ctx) |
| 1120 | if stroke and stroke != 'none' and stroke_width > 0 and stroke_opacity != 0: |
| 1121 | return False |
| 1122 | |
| 1123 | return True |
| 1124 | |
| 1125 | |
| 1126 | def _background_xml_from_rect( |
| 1127 | elem: ET.Element, |
| 1128 | ctx: ConvertContext, |
| 1129 | ) -> str: |
| 1130 | """Build native ``p:bg`` XML from a full-slide SVG background rect.""" |
| 1131 | fill_xml = build_fill_xml( |
| 1132 | elem, |
| 1133 | ctx, |
| 1134 | get_fill_opacity(elem, ctx), |
| 1135 | usage="background", |
| 1136 | ) |
| 1137 | if not fill_xml or '<a:noFill' in fill_xml: |
| 1138 | return '' |
| 1139 | return f'<p:bg><p:bgPr>{fill_xml}<a:effectLst/></p:bgPr></p:bg>' |
| 1140 | |
| 1141 | |
| 1142 | def _extract_background_candidate( |
| 1143 | root: ET.Element, |
| 1144 | ctx: ConvertContext, |
| 1145 | ) -> tuple[str, int | None]: |
| 1146 | """Promote a first-layer SVG background rect to native PowerPoint bgPr. |
| 1147 | |
| 1148 | PowerPoint stores page background fills under ``p:cSld/p:bg/p:bgPr``. |
| 1149 | Keeping the full-canvas SVG rect in ``p:spTree`` makes it an ordinary |
| 1150 | selectable shape, so users hit it during bulk element selection. Only the |
| 1151 | first visual layer is considered, matching pptx_to_svg's round-trip output |
| 1152 | and avoiding accidental promotion of content panels. |
| 1153 | """ |
| 1154 | canvas = _parse_svg_canvas(root) |
| 1155 | for child in root: |
| 1156 | tag = child.tag.replace(f'{{{SVG_NS}}}', '') |
| 1157 | if tag in _NON_VISUAL_TAGS: |
| 1158 | continue |
| 1159 | |
| 1160 | if tag == 'rect' and _is_full_canvas_rect(child, ctx, canvas): |
| 1161 | bg_xml = _background_xml_from_rect(child, ctx) |
| 1162 | if bg_xml: |
| 1163 | return bg_xml, id(child) |
| 1164 | return '', None |
| 1165 | |
| 1166 | if tag != 'g': |
| 1167 | return '', None |
| 1168 | if child.get('transform') or child.get('filter') or child.get('clip-path'): |
| 1169 | return '', None |
| 1170 | style_overrides = _extract_inheritable_styles(child) |
| 1171 | local_opacity = get_element_opacity(child) |
| 1172 | child_ctx = ctx.child( |
| 1173 | style_overrides=style_overrides, |
| 1174 | opacity_multiplier=1.0 if local_opacity is None else local_opacity, |
| 1175 | ) |
| 1176 | visual_children = [ |
| 1177 | grandchild for grandchild in child |
| 1178 | if grandchild.tag.replace(f'{{{SVG_NS}}}', '') not in _NON_VISUAL_TAGS |
| 1179 | ] |
| 1180 | if len(visual_children) != 1: |
| 1181 | return '', None |
| 1182 | only_child = visual_children[0] |
| 1183 | only_tag = only_child.tag.replace(f'{{{SVG_NS}}}', '') |
| 1184 | if only_tag == 'rect' and _is_full_canvas_rect(only_child, child_ctx, canvas): |
| 1185 | bg_xml = _background_xml_from_rect(only_child, child_ctx) |
| 1186 | if bg_xml: |
| 1187 | ctx.sync_from_child(child_ctx) |
| 1188 | return bg_xml, id(child) |
| 1189 | return '', None |
| 1190 | return '', None |
| 1191 | |
| 1192 | return '', None |
| 1193 | |
| 1194 | |
| 1195 | def collect_defs(root: ET.Element) -> dict[str, ET.Element]: |
| 1196 | """Collect all <defs> children into an {id: element} dictionary.""" |
| 1197 | definitions, _duplicates = project_definition_index(root) |
| 1198 | return definitions |
| 1199 | |
| 1200 | |
| 1201 | def _build_source_shape_id_map(root: ET.Element) -> dict[tuple[str, str], int]: |
| 1202 | """Allocate page-unique ids for part-scoped imported shape identities.""" |
| 1203 | source_entries: list[tuple[tuple[str, str], int]] = [] |
| 1204 | seen_keys: set[tuple[str, str]] = set() |
| 1205 | for elem in root.iter(): |
| 1206 | raw_id = elem.get('data-pptx-shape-id') |
| 1207 | if raw_id is None: |
| 1208 | continue |
| 1209 | scope = elem.get('data-pptx-shape-scope') or 'slide' |
| 1210 | if re.fullmatch(r'[A-Za-z0-9_.-]{1,64}', scope) is None: |
| 1211 | raise SvgNativeConversionError( |
| 1212 | f'Invalid data-pptx-shape-scope {scope!r}' |
| 1213 | ) |
| 1214 | try: |
| 1215 | shape_id = int(raw_id) |
| 1216 | except ValueError as exc: |
| 1217 | raise SvgNativeConversionError( |
| 1218 | f'Invalid data-pptx-shape-id {raw_id!r}' |
| 1219 | ) from exc |
| 1220 | if shape_id < 2 or shape_id > 0xFFFFFFFF: |
| 1221 | raise SvgNativeConversionError( |
| 1222 | f'data-pptx-shape-id must be between 2 and 4294967295, got {raw_id!r}' |
| 1223 | ) |
| 1224 | key = (scope, raw_id) |
| 1225 | if key in seen_keys: |
| 1226 | continue |
| 1227 | seen_keys.add(key) |
| 1228 | source_entries.append((key, shape_id)) |
| 1229 | |
| 1230 | preferred_ids = {shape_id for _key, shape_id in source_entries} |
| 1231 | next_fresh = max(preferred_ids, default=1) + 1 |
| 1232 | used: set[int] = set() |
| 1233 | mapping: dict[tuple[str, str], int] = {} |
| 1234 | for key, preferred in source_entries: |
| 1235 | output_id = preferred |
| 1236 | if output_id in used: |
| 1237 | while next_fresh in preferred_ids or next_fresh in used: |
| 1238 | next_fresh += 1 |
| 1239 | if next_fresh > 0xFFFFFFFF: |
| 1240 | raise SvgNativeConversionError('Exhausted PowerPoint shape id range') |
| 1241 | output_id = next_fresh |
| 1242 | next_fresh += 1 |
| 1243 | used.add(output_id) |
| 1244 | mapping[key] = output_id |
| 1245 | return mapping |
| 1246 | |
| 1247 | |
| 1248 | def _geometry_trace_metadata(elem: ET.Element, result: ShapeResult) -> dict[str, Any]: |
| 1249 | """Describe the native geometry decision for conversion diagnostics.""" |
| 1250 | xml = result.xml.lstrip() |
| 1251 | if xml.startswith('<p:grpSp>'): |
| 1252 | return {'output_geometry': 'group', 'fidelity': 'visual-only'} |
| 1253 | if xml.startswith('<p:pic>'): |
| 1254 | return {'output_geometry': 'picture', 'fidelity': 'native-normalized'} |
| 1255 | if xml.startswith('<p:graphicFrame>'): |
| 1256 | return {'output_geometry': 'native-object', 'fidelity': 'native-normalized'} |
| 1257 | if xml.startswith('<mc:AlternateContent'): |
| 1258 | if elem.tag.replace(f'{{{SVG_NS}}}', '') == 'text': |
| 1259 | inline_formula_count = sum( |
| 1260 | 1 for child in elem.iter() |
| 1261 | if child.get(INLINE_FORMULA_ATTR) is not None |
| 1262 | ) |
| 1263 | return { |
| 1264 | 'output_geometry': 'text', |
| 1265 | 'inline_formula_count': inline_formula_count, |
| 1266 | 'fidelity': 'native-normalized', |
| 1267 | } |
| 1268 | return {'output_geometry': 'native-formula', 'fidelity': 'exact'} |
| 1269 | |
| 1270 | preset_match = re.search(r'<a:prstGeom prst="([^"]+)"', xml) |
| 1271 | if preset_match is not None: |
| 1272 | preset = preset_match.group(1) |
| 1273 | source_preset = elem.get('data-pptx-prst') |
| 1274 | is_connector = xml.startswith('<p:cxnSp>') |
| 1275 | fidelity = ( |
| 1276 | 'exact' |
| 1277 | if source_preset == preset |
| 1278 | and elem.get('data-pptx-frame') is not None |
| 1279 | and not is_connector |
| 1280 | else 'native-normalized' |
| 1281 | ) |
| 1282 | return { |
| 1283 | 'output_geometry': 'preset', |
| 1284 | 'preset': preset, |
| 1285 | 'fidelity': fidelity, |
| 1286 | } |
| 1287 | if re.search(r'<a:custGeom(?:\s|>)', xml): |
| 1288 | carrier = next( |
| 1289 | ( |
| 1290 | candidate |
| 1291 | for candidate in elem.iter() |
| 1292 | if candidate.get('data-pptx-part') == 'geometry' |
| 1293 | ), |
| 1294 | elem, |
| 1295 | ) |
| 1296 | source_custom = ( |
| 1297 | carrier.get('data-pptx-geometry-kind') == 'custom' |
| 1298 | and carrier.get('data-pptx-frame') is not None |
| 1299 | ) |
| 1300 | expected_hash = carrier.get('data-pptx-geometry-sha256') |
| 1301 | actual_hash = hashlib.sha256( |
| 1302 | (carrier.get('d') or '').strip().encode('utf-8') |
| 1303 | ).hexdigest() |
| 1304 | unchanged = source_custom and expected_hash == actual_hash |
| 1305 | if unchanged: |
| 1306 | fidelity = 'exact' |
| 1307 | geometry_source = 'preserved-metadata' |
| 1308 | elif source_custom: |
| 1309 | fidelity = 'native-normalized' |
| 1310 | geometry_source = 'svg-recompiled' |
| 1311 | else: |
| 1312 | fidelity = 'visual-only' |
| 1313 | geometry_source = 'svg-authored' |
| 1314 | return { |
| 1315 | 'output_geometry': 'custom', |
| 1316 | 'fidelity': fidelity, |
| 1317 | 'geometry_source': geometry_source, |
| 1318 | } |
| 1319 | return {'output_geometry': 'unknown', 'fidelity': 'visual-only'} |
| 1320 | |
| 1321 | |
| 1322 | def convert_element(elem: ET.Element, ctx: ConvertContext) -> ShapeResult | None: |
| 1323 | """Dispatch an SVG element to the appropriate converter.""" |
| 1324 | tag = elem.tag.replace(f'{{{SVG_NS}}}', '') |
| 1325 | elem_id = elem.get('id') |
| 1326 | |
| 1327 | def trace(decision: str, **metadata: Any) -> None: |
| 1328 | if ctx.trace_events is None: |
| 1329 | return |
| 1330 | event: dict[str, Any] = { |
| 1331 | 'tag': tag, |
| 1332 | 'decision': decision, |
| 1333 | } |
| 1334 | if elem_id: |
| 1335 | event['id'] = elem_id |
| 1336 | for attr in ( |
| 1337 | 'data-pptx-layer', |
| 1338 | 'data-pptx-object', |
| 1339 | 'data-pptx-shape-id', |
| 1340 | 'data-pptx-frame', |
| 1341 | 'data-pptx-prst', |
| 1342 | 'data-pptx-part', |
| 1343 | 'data-pptx-geometry-status', |
| 1344 | 'data-pptx-geometry-reason', |
| 1345 | 'data-pptx-placeholder', |
| 1346 | 'data-pptx-bounds', |
| 1347 | 'data-pptx-carrier', |
| 1348 | 'data-pptx-idx', |
| 1349 | 'data-pptx-role', |
| 1350 | ): |
| 1351 | value = elem.get(attr) |
| 1352 | if value is not None: |
| 1353 | event[attr] = value |
| 1354 | adjustments = { |
| 1355 | attr[len('data-pptx-av-'):]: value |
| 1356 | for attr, value in elem.attrib.items() |
| 1357 | if attr.startswith('data-pptx-av-') |
| 1358 | } |
| 1359 | if adjustments: |
| 1360 | event['adjustments'] = dict(sorted(adjustments.items())) |
| 1361 | event.update(metadata) |
| 1362 | ctx.trace_events.append(event) |
| 1363 | |
| 1364 | if elem.get('data-pptx-part') == 'geometry-detail': |
| 1365 | trace('skip', reason='render-only-preset-geometry-detail') |
| 1366 | return None |
| 1367 | |
| 1368 | converter = _CONVERTERS.get(tag) |
| 1369 | if converter: |
| 1370 | try: |
| 1371 | result = converter(elem, ctx) |
| 1372 | shape_hyperlink = elem.get(SHAPE_HYPERLINK_ATTR) |
| 1373 | if result is not None and shape_hyperlink is not None: |
| 1374 | result = apply_shape_hyperlink(result, ctx, shape_hyperlink) |
| 1375 | except Exception as e: |
| 1376 | trace('error', error=str(e)) |
| 1377 | raise SvgNativeConversionError(f'Failed to convert <{tag}>: {e}') from e |
| 1378 | if result: |
| 1379 | shape_match = re.search(r'<p:cNvPr id="(\d+)"', result.xml) |
| 1380 | metadata: dict[str, Any] = {} |
| 1381 | if shape_match: |
| 1382 | metadata['shape_id'] = int(shape_match.group(1)) |
| 1383 | if result.bounds_emu is not None: |
| 1384 | metadata['bounds_emu'] = list(result.bounds_emu) |
| 1385 | metadata.update(_geometry_trace_metadata(elem, result)) |
| 1386 | trace('native', **metadata) |
| 1387 | else: |
| 1388 | trace('skip', reason='empty-or-non-rendering') |
| 1389 | return result |
| 1390 | |
| 1391 | if tag in _NON_VISUAL_TAGS: |
| 1392 | trace('skip', reason='non-visual') |
| 1393 | return None |
| 1394 | |
| 1395 | trace('unsupported') |
| 1396 | raise SvgNativeConversionError(f'Unsupported visual SVG element <{tag}>') |
| 1397 | |
| 1398 | |
| 1399 | def _local_tag(elem: ET.Element) -> str: |
| 1400 | if not isinstance(elem.tag, str): |
| 1401 | return str(elem.tag) |
| 1402 | prefix = f'{{{SVG_NS}}}' |
| 1403 | return elem.tag[len(prefix):] if elem.tag.startswith(prefix) else elem.tag |
| 1404 | |
| 1405 | |
| 1406 | def collect_unsupported_visuals( |
| 1407 | root: ET.Element, |
| 1408 | *, |
| 1409 | allow_data_icon_use: bool = False, |
| 1410 | ) -> list[str]: |
| 1411 | """Return visual element paths that the native converter cannot dispatch.""" |
| 1412 | issues: list[str] = [] |
| 1413 | |
| 1414 | def walk( |
| 1415 | elem: ET.Element, |
| 1416 | path: str, |
| 1417 | in_defs: bool = False, |
| 1418 | parent_tag: str | None = None, |
| 1419 | ) -> None: |
| 1420 | tag = _local_tag(elem) |
| 1421 | current = f'{path}/{tag}' |
| 1422 | if in_defs: |
| 1423 | return |
| 1424 | if tag in _NON_VISUAL_TAGS: |
| 1425 | return |
| 1426 | is_supported_visual_child = ( |
| 1427 | tag in _SUPPORTED_VISUAL_CHILD_TAGS |
| 1428 | and parent_tag in {'text', 'tspan', 'a'} |
| 1429 | ) |
| 1430 | is_data_icon_placeholder = ( |
| 1431 | allow_data_icon_use |
| 1432 | and tag == 'use' |
| 1433 | and elem.get('data-icon') is not None |
| 1434 | ) |
| 1435 | if (tag not in _CONVERTERS |
| 1436 | and tag not in _NON_VISUAL_TAGS |
| 1437 | and not is_supported_visual_child |
| 1438 | and not is_data_icon_placeholder): |
| 1439 | issues.append(current) |
| 1440 | for idx, child in enumerate(list(elem), start=1): |
| 1441 | walk( |
| 1442 | child, |
| 1443 | f'{current}[{idx}]', |
| 1444 | in_defs=(tag == 'defs'), |
| 1445 | parent_tag=tag, |
| 1446 | ) |
| 1447 | |
| 1448 | for idx, child in enumerate(list(root), start=1): |
| 1449 | walk(child, f'/svg[{idx}]', parent_tag='svg') |
| 1450 | return issues |
| 1451 | |
| 1452 | |
| 1453 | def convert_svg_to_slide_shapes( |
| 1454 | svg_path: str | Path, |
| 1455 | slide_num: int = 1, |
| 1456 | slide_count: int | None = None, |
| 1457 | verbose: bool = False, |
| 1458 | merge_paragraphs: bool | None = None, |
| 1459 | image_optimize: bool = True, |
| 1460 | image_max_dimension: int | None = 2560, |
| 1461 | image_sizing: str = 'cap', |
| 1462 | image_scale: float = 2.0, |
| 1463 | image_quality: int = 85, |
| 1464 | native_objects: bool = False, |
| 1465 | animation_group_overrides: frozenset[str] | None = None, |
| 1466 | theme_font_spec: ThemeFontSpec | None = None, |
| 1467 | theme_color_spec: ThemeColorSpec | None = None, |
| 1468 | primary_language: str | None = None, |
| 1469 | trace_out: list[dict[str, Any]] | None = None, |
| 1470 | promote_background: bool = True, |
| 1471 | text_flow: str | None = None, |
| 1472 | ) -> tuple[ |
| 1473 | str, |
| 1474 | dict[str, bytes], |
| 1475 | list[dict[str, str]], |
| 1476 | list, |
| 1477 | dict[str, bytes], |
| 1478 | dict[str, str], |
| 1479 | ]: |
| 1480 | """Convert an SVG file to a complete DrawingML slide XML. |
| 1481 | |
| 1482 | Args: |
| 1483 | svg_path: Path to the SVG file. |
| 1484 | slide_num: Slide number (for naming). |
| 1485 | slide_count: Public deck size used to validate ``#slide-N`` targets. |
| 1486 | verbose: Print progress info. |
| 1487 | merge_paragraphs: Legacy compatibility option. True selects reflow; |
| 1488 | False selects split. Do not combine with ``text_flow``. |
| 1489 | text_flow: Positional-tspan policy. ``preserve`` keeps authored visual |
| 1490 | line breaks in one frame, ``reflow`` lets PowerPoint wrap the text, |
| 1491 | and ``split`` emits one text frame per visual line. |
| 1492 | image_optimize: Downsample oversized raster images for PPTX export. |
| 1493 | image_max_dimension: Maximum optimized image dimension in pixels. |
| 1494 | image_sizing: ``cap`` to only cap source dimensions, ``display`` to |
| 1495 | size from rendered SVG boxes. |
| 1496 | image_scale: Target image pixels per SVG display pixel. |
| 1497 | image_quality: JPEG quality used for opaque optimized rasters. |
| 1498 | native_objects: Convert explicit ``data-pptx-replace-with`` chart/table |
| 1499 | markers to native PowerPoint Chart/Table objects. Formula markers |
| 1500 | are intrinsically native and do not use this opt-in. Default off. |
| 1501 | animation_group_overrides: Explicit top-level SVG group ids from |
| 1502 | ``animations.json`` that override the legacy chrome-name fallback. |
| 1503 | Explicit structural layer/role/placeholder markers remain excluded. |
| 1504 | theme_font_spec: Optional major/minor theme-font contract. Matching SVG |
| 1505 | families emit DrawingML theme tokens instead of fixed typefaces. |
| 1506 | theme_color_spec: Optional context-aware theme-color contract. Exact |
| 1507 | locked colors emit DrawingML scheme tokens while local colors stay |
| 1508 | fixed. |
| 1509 | primary_language: Canonical BCP-47 project content language. ``None`` |
| 1510 | keeps the legacy per-run script heuristic. |
| 1511 | trace_out: Optional list populated with one per-slide trace dictionary. |
| 1512 | promote_background: Promote the first eligible full-canvas rectangle |
| 1513 | into native ``p:bg``. Structured export disables this generic pass |
| 1514 | and applies its narrower explicit background contract later. |
| 1515 | |
| 1516 | Returns: |
| 1517 | (slide_xml, media_files, rel_entries, anim_targets, |
| 1518 | package_files, content_type_overrides) where: |
| 1519 | - slide_xml: Complete slide XML string. |
| 1520 | - media_files: Dict of {filename: bytes} for media to write. |
| 1521 | - rel_entries: List of relationship entries to add. |
| 1522 | - anim_targets: List of (shape_id, svg_id) tuples for top-level |
| 1523 | semantic groups, in z-order; consumed by the builder's optional |
| 1524 | per-element object-animation timing emitter. |
| 1525 | - package_files: Dict of {pptx internal path: bytes} for non-media |
| 1526 | OOXML parts such as native chart XML and embedded workbooks. |
| 1527 | - content_type_overrides: Dict of {pptx internal path: content type} |
| 1528 | for package_files that require [Content_Types].xml overrides. |
| 1529 | """ |
| 1530 | text_flow = resolve_text_flow(text_flow, merge_paragraphs) |
| 1531 | svg_path = Path(svg_path) |
| 1532 | tree = ET.parse(str(svg_path)) |
| 1533 | root = tree.getroot() |
| 1534 | _hydrate_native_payloads(root, svg_path) |
| 1535 | try: |
| 1536 | parse_project_svg_root( |
| 1537 | root, |
| 1538 | context=svg_path.name, |
| 1539 | ) |
| 1540 | except CanvasContractError as exc: |
| 1541 | raise SvgNativeConversionError(str(exc)) from exc |
| 1542 | _require_native_marker_attributes(root, svg_path) |
| 1543 | _require_inline_formula_markers(root, svg_path) |
| 1544 | _require_project_hyperlinks( |
| 1545 | root, |
| 1546 | svg_path, |
| 1547 | slide_count=slide_count, |
| 1548 | ) |
| 1549 | _require_project_nested_svg_crops(root, svg_path) |
| 1550 | _require_project_clip_paths(root, svg_path) |
| 1551 | authored_errors = validate_authored_preset_tree(root) |
| 1552 | if authored_errors: |
| 1553 | raise SvgNativeConversionError( |
| 1554 | 'Invalid authored preset structure: ' + '; '.join(authored_errors) |
| 1555 | ) |
| 1556 | # Validate the source contract once, then lower compact groups to the |
| 1557 | # established expanded transport IR. Downstream conversion validates the |
| 1558 | # generated preview hash, not the source-format allowlist again. |
| 1559 | try: |
| 1560 | materialize_compact_authored_preset_tree(root) |
| 1561 | except ValueError as exc: |
| 1562 | raise SvgNativeConversionError( |
| 1563 | f'Invalid compact authored preset: {exc}' |
| 1564 | ) from exc |
| 1565 | _mark_unchanged_txbody_groups(root) |
| 1566 | _mark_unchanged_preset_previews(root) |
| 1567 | if native_objects: |
| 1568 | try: |
| 1569 | snapshot_native_fallback_freshness(root) |
| 1570 | except NativeMarkerAttributeError as exc: |
| 1571 | raise SvgNativeConversionError( |
| 1572 | f'{Path(svg_path).name}: conflicting chart/table replacement ' |
| 1573 | f'metadata: {exc}' |
| 1574 | ) from exc |
| 1575 | trace_events: list[dict[str, Any]] | None = [] if trace_out is not None else None |
| 1576 | trace_steps: list[dict[str, Any]] = [] |
| 1577 | |
| 1578 | from ..geometry_properties import ( |
| 1579 | GeometryStyleError, |
| 1580 | materialize_inline_geometry_properties, |
| 1581 | ) |
| 1582 | |
| 1583 | try: |
| 1584 | geometry_count = materialize_inline_geometry_properties(root) |
| 1585 | except GeometryStyleError as exc: |
| 1586 | raise SvgNativeConversionError( |
| 1587 | f'{svg_path.name}: inline geometry materialization failed: {exc}' |
| 1588 | ) from exc |
| 1589 | geometry_trace = None |
| 1590 | if geometry_count: |
| 1591 | geometry_trace = { |
| 1592 | 'action': 'materialize-inline-geometry', |
| 1593 | 'count': geometry_count, |
| 1594 | } |
| 1595 | trace_steps.append(geometry_trace) |
| 1596 | if verbose: |
| 1597 | print(f' Materialized {geometry_count} inline geometry declaration(s)') |
| 1598 | |
| 1599 | geometry_length_errors = project_geometry_length_errors(root) |
| 1600 | if geometry_length_errors: |
| 1601 | preview = '; '.join(geometry_length_errors[:8]) |
| 1602 | suffix = ( |
| 1603 | '' if len(geometry_length_errors) <= 8 |
| 1604 | else f'; +{len(geometry_length_errors) - 8} more' |
| 1605 | ) |
| 1606 | raise SvgNativeConversionError( |
| 1607 | f'{Path(svg_path).name}: invalid project geometry length(s): ' |
| 1608 | f'{preview}{suffix}' |
| 1609 | ) |
| 1610 | |
| 1611 | _require_project_text_properties(root, svg_path) |
| 1612 | _require_project_freeform_geometry(root, svg_path) |
| 1613 | _require_project_stroke_styles(root, svg_path) |
| 1614 | _require_project_opacities(root, svg_path) |
| 1615 | _require_project_paints(root, svg_path) |
| 1616 | _require_project_masks(root, svg_path) |
| 1617 | _require_project_definitions(root, svg_path) |
| 1618 | _require_project_paint_references(root, svg_path) |
| 1619 | _require_project_line_end_markers(root, svg_path) |
| 1620 | _require_project_gradients(root, svg_path) |
| 1621 | _require_project_effect_status(root, svg_path) |
| 1622 | _require_project_filters(root, svg_path) |
| 1623 | _require_project_image_aspect_ratios(root, svg_path) |
| 1624 | _require_project_transforms(root, svg_path) |
| 1625 | |
| 1626 | viewport_width, viewport_height = _root_viewport_size(root) |
| 1627 | |
| 1628 | # Expand project icon placeholders and static same-document <use> |
| 1629 | # references before unsupported-element preflight. |
| 1630 | from ..use_expander import ( |
| 1631 | UseExpansionError, |
| 1632 | expand_local_use_references, |
| 1633 | expand_use_data_icons, |
| 1634 | ) |
| 1635 | |
| 1636 | icons_dir, icons_fallback_dir = icon_search_dirs_for_svg(svg_path) |
| 1637 | if icons_dir.exists(): |
| 1638 | expanded = expand_use_data_icons(root, icons_dir, icons_fallback_dir) |
| 1639 | if expanded: |
| 1640 | trace_steps.append({'action': 'expand-use-data-icons', 'count': expanded}) |
| 1641 | if verbose and expanded: |
| 1642 | print(f' Expanded {expanded} <use data-icon="..."/> placeholder(s)') |
| 1643 | if expanded: |
| 1644 | hydrated = _hydrate_native_payloads(root, svg_path) |
| 1645 | if hydrated: |
| 1646 | trace_steps.append({ |
| 1647 | 'action': 'hydrate-native-payloads-from-icons', |
| 1648 | 'count': hydrated, |
| 1649 | }) |
| 1650 | _mark_unchanged_txbody_groups(root) |
| 1651 | _mark_unchanged_preset_previews(root) |
| 1652 | _require_project_freeform_geometry(root, svg_path) |
| 1653 | |
| 1654 | try: |
| 1655 | injected_geometry_count = materialize_inline_geometry_properties(root) |
| 1656 | except GeometryStyleError as exc: |
| 1657 | raise SvgNativeConversionError( |
| 1658 | f'{svg_path.name}: expanded icon geometry materialization failed: {exc}' |
| 1659 | ) from exc |
| 1660 | if injected_geometry_count: |
| 1661 | geometry_count += injected_geometry_count |
| 1662 | if geometry_trace is None: |
| 1663 | geometry_trace = { |
| 1664 | 'action': 'materialize-inline-geometry', |
| 1665 | 'count': geometry_count, |
| 1666 | } |
| 1667 | trace_steps.append(geometry_trace) |
| 1668 | else: |
| 1669 | geometry_trace['count'] = geometry_count |
| 1670 | if verbose: |
| 1671 | print( |
| 1672 | f' Materialized {injected_geometry_count} inline geometry ' |
| 1673 | 'declaration(s) from expanded icons' |
| 1674 | ) |
| 1675 | |
| 1676 | try: |
| 1677 | expanded_local = expand_local_use_references(root) |
| 1678 | except UseExpansionError as exc: |
| 1679 | raise SvgNativeConversionError( |
| 1680 | f'{svg_path.name}: local <use> expansion failed: {exc}' |
| 1681 | ) from exc |
| 1682 | if expanded_local: |
| 1683 | trace_steps.append({ |
| 1684 | 'action': 'expand-local-use-references', |
| 1685 | 'count': expanded_local, |
| 1686 | }) |
| 1687 | if verbose: |
| 1688 | print(f' Expanded {expanded_local} local <use href="#..."/> instance(s)') |
| 1689 | |
| 1690 | _require_inline_formula_markers(root, svg_path) |
| 1691 | _require_project_hyperlinks( |
| 1692 | root, |
| 1693 | svg_path, |
| 1694 | slide_count=slide_count, |
| 1695 | ) |
| 1696 | |
| 1697 | # Recheck compiler-injected icon/use wrappers and cloned definition trees. |
| 1698 | _require_project_nested_svg_crops(root, svg_path) |
| 1699 | _require_project_images(root, svg_path) |
| 1700 | _require_project_clip_paths(root, svg_path) |
| 1701 | _require_project_text_properties(root, svg_path) |
| 1702 | _require_project_stroke_styles(root, svg_path) |
| 1703 | _require_project_opacities(root, svg_path) |
| 1704 | _require_project_paints(root, svg_path) |
| 1705 | _require_project_masks(root, svg_path) |
| 1706 | _require_project_definitions(root, svg_path) |
| 1707 | _require_project_paint_references(root, svg_path) |
| 1708 | _require_project_gradients(root, svg_path) |
| 1709 | _require_project_effect_status(root, svg_path) |
| 1710 | _require_project_filters(root, svg_path) |
| 1711 | _require_project_image_aspect_ratios(root, svg_path) |
| 1712 | _require_project_transforms(root, svg_path) |
| 1713 | |
| 1714 | try: |
| 1715 | materialize_project_text_metrics(root) |
| 1716 | except ValueError as exc: |
| 1717 | raise SvgNativeConversionError( |
| 1718 | f'{svg_path.name}: text-metric materialization failed: {exc}' |
| 1719 | ) from exc |
| 1720 | |
| 1721 | inline_formula_count_before_text_lowering = sum( |
| 1722 | 1 for elem in root.iter() |
| 1723 | if elem.get(INLINE_FORMULA_ATTR) is not None |
| 1724 | ) |
| 1725 | hyperlink_count_before_text_lowering = sum( |
| 1726 | 1 for elem in root.iter() |
| 1727 | if ( |
| 1728 | elem.tag == f'{{{SVG_NS}}}a' |
| 1729 | or elem.get(SHAPE_HYPERLINK_ATTR) is not None |
| 1730 | ) |
| 1731 | ) |
| 1732 | |
| 1733 | # Flatten positional <tspan> (those with x/y/non-zero dy) into independent |
| 1734 | # <text> elements. DrawingML runs cannot reposition mid-paragraph, so a |
| 1735 | # dy-stacked block of tspans would otherwise collapse onto one baseline, |
| 1736 | # and an x-anchored tspan would render in the wrong column. finalize_svg |
| 1737 | # does the same flattening on disk; doing it here keeps native pptx output |
| 1738 | # correct when reading raw svg_output/. |
| 1739 | # Preserve/reflow modes additionally fold conservative paragraph blocks |
| 1740 | # into one annotated <text>. Preserve keeps visual lines as DrawingML hard |
| 1741 | # breaks; reflow joins wrapping lines; split keeps one frame per line. |
| 1742 | from ..tspan_flattener import flatten_positional_tspans |
| 1743 | flattened = flatten_positional_tspans( |
| 1744 | tree, |
| 1745 | merge_paragraphs=text_flow != TEXT_FLOW_SPLIT, |
| 1746 | preserve_line_breaks=text_flow == TEXT_FLOW_PRESERVE, |
| 1747 | ) |
| 1748 | if flattened: |
| 1749 | trace_steps.append({ |
| 1750 | 'action': 'flatten-positional-tspans', |
| 1751 | 'text_flow': text_flow, |
| 1752 | # Compatibility field for older trace readers. |
| 1753 | 'merge_paragraphs': text_flow != TEXT_FLOW_SPLIT, |
| 1754 | }) |
| 1755 | if verbose: |
| 1756 | print(f' Lowered positional <tspan> using {text_flow} text flow') |
| 1757 | |
| 1758 | inline_formula_count_after_text_lowering = sum( |
| 1759 | 1 for elem in root.iter() |
| 1760 | if elem.get(INLINE_FORMULA_ATTR) is not None |
| 1761 | ) |
| 1762 | hyperlink_count_after_text_lowering = sum( |
| 1763 | 1 for elem in root.iter() |
| 1764 | if ( |
| 1765 | elem.tag == f'{{{SVG_NS}}}a' |
| 1766 | or elem.get(SHAPE_HYPERLINK_ATTR) is not None |
| 1767 | ) |
| 1768 | ) |
| 1769 | if ( |
| 1770 | inline_formula_count_after_text_lowering |
| 1771 | != inline_formula_count_before_text_lowering |
| 1772 | ): |
| 1773 | raise SvgNativeConversionError( |
| 1774 | f'{svg_path.name}: positional text lowering changed inline formula ' |
| 1775 | f'marker count from {inline_formula_count_before_text_lowering} ' |
| 1776 | f'to {inline_formula_count_after_text_lowering}' |
| 1777 | ) |
| 1778 | if hyperlink_count_after_text_lowering != hyperlink_count_before_text_lowering: |
| 1779 | raise SvgNativeConversionError( |
| 1780 | f'{svg_path.name}: positional text lowering changed hyperlink ' |
| 1781 | f'count from {hyperlink_count_before_text_lowering} ' |
| 1782 | f'to {hyperlink_count_after_text_lowering}' |
| 1783 | ) |
| 1784 | _require_inline_formula_markers(root, svg_path) |
| 1785 | _require_project_hyperlinks( |
| 1786 | root, |
| 1787 | svg_path, |
| 1788 | slide_count=slide_count, |
| 1789 | ) |
| 1790 | |
| 1791 | _require_project_text_properties(root, svg_path) |
| 1792 | try: |
| 1793 | text_font_sizes = resolve_project_font_sizes(root) |
| 1794 | text_letter_spacings = resolve_project_letter_spacings( |
| 1795 | root, |
| 1796 | text_font_sizes, |
| 1797 | ) |
| 1798 | except ValueError as exc: |
| 1799 | raise SvgNativeConversionError( |
| 1800 | f'{svg_path.name}: invalid project text-metric inheritance: {exc}' |
| 1801 | ) from exc |
| 1802 | |
| 1803 | unsupported = collect_unsupported_visuals(root) |
| 1804 | if unsupported: |
| 1805 | preview = '; '.join(unsupported[:8]) |
| 1806 | suffix = '' if len(unsupported) <= 8 else f'; +{len(unsupported) - 8} more' |
| 1807 | raise SvgNativeConversionError( |
| 1808 | f'{svg_path.name}: unsupported visual SVG element(s): {preview}{suffix}' |
| 1809 | ) |
| 1810 | |
| 1811 | defs = collect_defs(root) |
| 1812 | source_shape_id_map = _build_source_shape_id_map(root) |
| 1813 | ctx = ConvertContext( |
| 1814 | defs=defs, |
| 1815 | reserved_shape_ids=frozenset(source_shape_id_map.values()), |
| 1816 | source_shape_id_map=source_shape_id_map, |
| 1817 | slide_num=slide_num, |
| 1818 | slide_count=slide_count, |
| 1819 | viewport_width=viewport_width, |
| 1820 | viewport_height=viewport_height, |
| 1821 | svg_dir=Path(svg_path).parent, |
| 1822 | text_flow=text_flow, |
| 1823 | image_optimize=image_optimize, |
| 1824 | image_max_dimension=image_max_dimension, |
| 1825 | image_sizing=image_sizing, |
| 1826 | image_scale=image_scale, |
| 1827 | image_quality=image_quality, |
| 1828 | native_objects_enabled=native_objects, |
| 1829 | animation_group_overrides=animation_group_overrides or frozenset(), |
| 1830 | trace_events=trace_events, |
| 1831 | theme_font_spec=theme_font_spec, |
| 1832 | theme_color_spec=theme_color_spec, |
| 1833 | primary_language=primary_language, |
| 1834 | inherited_styles=_extract_inheritable_styles(root), |
| 1835 | text_font_sizes=text_font_sizes, |
| 1836 | text_letter_spacings=text_letter_spacings, |
| 1837 | ) |
| 1838 | |
| 1839 | shapes: list[str] = [] |
| 1840 | converted = 0 |
| 1841 | skipped = 0 |
| 1842 | has_top_level_group = any( |
| 1843 | child.tag.replace(f'{{{SVG_NS}}}', '') == 'g' |
| 1844 | for child in root |
| 1845 | ) |
| 1846 | background_xml, background_skip_id = ( |
| 1847 | _extract_background_candidate(root, ctx) |
| 1848 | if promote_background |
| 1849 | else ('', None) |
| 1850 | ) |
| 1851 | promoted_backgrounds = 1 if background_xml else 0 |
| 1852 | if background_xml and trace_events is not None: |
| 1853 | trace_events.append({ |
| 1854 | 'tag': 'rect', |
| 1855 | 'decision': 'native-background', |
| 1856 | 'reason': 'promoted-full-canvas-rect-to-bgPr', |
| 1857 | }) |
| 1858 | # Per-element shape ids of every top-level child, used as an animation |
| 1859 | # fallback when no <g id="..."> groups are present at the root. |
| 1860 | fallback_targets: list = [] |
| 1861 | |
| 1862 | for child in root: |
| 1863 | tag = child.tag.replace(f'{{{SVG_NS}}}', '') |
| 1864 | if tag == 'defs': |
| 1865 | continue |
| 1866 | if id(child) == background_skip_id: |
| 1867 | continue |
| 1868 | result = convert_element(child, ctx) |
| 1869 | if result: |
| 1870 | shapes.append(result.xml) |
| 1871 | converted += 1 |
| 1872 | m = re.search(r'<p:cNvPr id="(\d+)"', result.xml) |
| 1873 | elem_id = child.get('id') |
| 1874 | role = child.get('data-pptx-role') |
| 1875 | placeholder = child.get('data-pptx-placeholder') |
| 1876 | has_explicit_semantics = role is not None or placeholder is not None |
| 1877 | structurally_static = ( |
| 1878 | child.get('data-pptx-layer') is not None |
| 1879 | or ( |
| 1880 | has_explicit_semantics |
| 1881 | and is_static_page_frame(role, placeholder) |
| 1882 | ) |
| 1883 | ) |
| 1884 | legacy_chrome = ( |
| 1885 | not has_explicit_semantics |
| 1886 | and is_chrome_id(elem_id) |
| 1887 | ) |
| 1888 | explicit_legacy_override = ( |
| 1889 | elem_id is not None |
| 1890 | and elem_id in ctx.animation_group_overrides |
| 1891 | ) |
| 1892 | if ( |
| 1893 | m |
| 1894 | and not structurally_static |
| 1895 | and (not legacy_chrome or explicit_legacy_override) |
| 1896 | ): |
| 1897 | fallback_targets.append((int(m.group(1)), elem_id or tag)) |
| 1898 | else: |
| 1899 | if tag not in _NON_VISUAL_TAGS: |
| 1900 | skipped += 1 |
| 1901 | |
| 1902 | unresolved_connector_targets = sorted( |
| 1903 | ctx.referenced_shape_ids - ctx.claimed_shape_ids |
| 1904 | ) |
| 1905 | if unresolved_connector_targets: |
| 1906 | raise SvgNativeConversionError( |
| 1907 | 'Connector target shape ids were reserved but not restored: ' |
| 1908 | + ', '.join(str(shape_id) for shape_id in unresolved_connector_targets) |
| 1909 | ) |
| 1910 | |
| 1911 | # Animation target fallback. Semantic <g id="..."> groups are the |
| 1912 | # preferred anchors (set inside convert_g). When the SVG has none |
| 1913 | # at the root we fall back to top-level primitives, but only when |
| 1914 | # the count is reasonable. Presenter-click animation should reveal |
| 1915 | # semantic blocks, not atomized drawing primitives, so fallback is |
| 1916 | # intentionally capped at a low count. |
| 1917 | _ANIM_FALLBACK_CAP = 8 |
| 1918 | if ( |
| 1919 | not has_top_level_group |
| 1920 | and not ctx.anim_targets |
| 1921 | and 0 < len(fallback_targets) <= _ANIM_FALLBACK_CAP |
| 1922 | ): |
| 1923 | ctx.anim_targets = fallback_targets |
| 1924 | |
| 1925 | if verbose: |
| 1926 | promoted = ( |
| 1927 | f', promoted {promoted_backgrounds} background' |
| 1928 | if promoted_backgrounds else '' |
| 1929 | ) |
| 1930 | print(f' Converted {converted} elements, skipped {skipped}{promoted}') |
| 1931 | |
| 1932 | if trace_out is not None: |
| 1933 | trace_out.append({ |
| 1934 | 'slide_num': slide_num, |
| 1935 | 'svg': str(svg_path), |
| 1936 | 'page_role': root.get('data-pptx-page-role'), |
| 1937 | 'summary': { |
| 1938 | 'converted': converted, |
| 1939 | 'skipped': skipped, |
| 1940 | 'promoted_backgrounds': promoted_backgrounds, |
| 1941 | 'media_files': len(ctx.media_files), |
| 1942 | 'package_files': len(ctx.package_files), |
| 1943 | 'relationships': len(ctx.rel_entries), |
| 1944 | 'animation_targets': len(ctx.anim_targets), |
| 1945 | }, |
| 1946 | 'preprocess': trace_steps, |
| 1947 | 'events': trace_events or [], |
| 1948 | }) |
| 1949 | |
| 1950 | shapes_xml = '\n'.join(shapes) |
| 1951 | |
| 1952 | slide_xml = f'''<?xml version="1.0" encoding="UTF-8" standalone="yes"?> |
| 1953 | <p:sld xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main" |
| 1954 | xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships" |
| 1955 | xmlns:p="http://schemas.openxmlformats.org/presentationml/2006/main"> |
| 1956 | <p:cSld> |
| 1957 | {background_xml} |
| 1958 | <p:spTree> |
| 1959 | <p:nvGrpSpPr> |
| 1960 | <p:cNvPr id="1" name=""/> |
| 1961 | <p:cNvGrpSpPr/><p:nvPr/> |
| 1962 | </p:nvGrpSpPr> |
| 1963 | <p:grpSpPr> |
| 1964 | <a:xfrm><a:off x="0" y="0"/><a:ext cx="0" cy="0"/> |
| 1965 | <a:chOff x="0" y="0"/><a:chExt cx="0" cy="0"/></a:xfrm> |
| 1966 | </p:grpSpPr> |
| 1967 | {shapes_xml} |
| 1968 | </p:spTree> |
| 1969 | </p:cSld> |
| 1970 | <p:clrMapOvr><a:masterClrMapping/></p:clrMapOvr> |
| 1971 | </p:sld>''' |
| 1972 | |
| 1973 | return ( |
| 1974 | slide_xml, |
| 1975 | ctx.media_files, |
| 1976 | ctx.rel_entries, |
| 1977 | ctx.anim_targets, |
| 1978 | ctx.package_files, |
| 1979 | ctx.content_type_overrides, |
| 1980 | ) |
| 1981 |