| 1 | """Read SmartArt content and structure from DrawingML diagram parts. |
| 2 | |
| 3 | The reader exposes source facts only. It does not edit DiagramML or promise |
| 4 | native SmartArt regeneration; generated decks continue to redraw the extracted |
| 5 | content through the SVG-to-DrawingML shape pipeline. |
| 6 | """ |
| 7 | |
| 8 | from __future__ import annotations |
| 9 | |
| 10 | import zipfile |
| 11 | from typing import Any |
| 12 | from xml.etree import ElementTree as ET |
| 13 | |
| 14 | from .ooxml import ( |
| 15 | MC_NS, |
| 16 | NS, |
| 17 | _container_geometry, |
| 18 | _normalize_part, |
| 19 | _paragraph_texts, |
| 20 | _qn, |
| 21 | _read_xml, |
| 22 | _rels_name_for_part, |
| 23 | _shape_identity, |
| 24 | _slide_relationships, |
| 25 | ) |
| 26 | |
| 27 | |
| 28 | DIAGRAM_NS = "http://schemas.openxmlformats.org/drawingml/2006/diagram" |
| 29 | DIAGRAM_DRAWING_NS = "http://schemas.microsoft.com/office/drawing/2008/diagram" |
| 30 | DIAGRAM_URI = "http://schemas.openxmlformats.org/drawingml/2006/diagram" |
| 31 | |
| 32 | _DIAGRAM_REL_SUFFIXES = { |
| 33 | "colors": "diagramColors", |
| 34 | "data": "diagramData", |
| 35 | "layout": "diagramLayout", |
| 36 | "quick_style": "diagramQuickStyle", |
| 37 | } |
| 38 | _REL_ATTRS = { |
| 39 | "colors": "cs", |
| 40 | "data": "dm", |
| 41 | "layout": "lo", |
| 42 | "quick_style": "qs", |
| 43 | } |
| 44 | _CONTENT_POINT_TYPES = {"asst", "node"} |
| 45 | |
| 46 | DIAGRAM_NS_MAP = { |
| 47 | **NS, |
| 48 | "dgm": DIAGRAM_NS, |
| 49 | "dsp": DIAGRAM_DRAWING_NS, |
| 50 | "mc": MC_NS, |
| 51 | } |
| 52 | |
| 53 | |
| 54 | def _relationship_matches(rel_type: str, suffix: str) -> bool: |
| 55 | return rel_type.rsplit("/", 1)[-1] == suffix |
| 56 | |
| 57 | |
| 58 | def _read_optional_xml( |
| 59 | zf: zipfile.ZipFile, |
| 60 | part_name: str | None, |
| 61 | ) -> tuple[ET.Element | None, str | None]: |
| 62 | if not part_name: |
| 63 | return None, "missing-part-reference" |
| 64 | try: |
| 65 | return _read_xml(zf, part_name), None |
| 66 | except RuntimeError: |
| 67 | return None, "missing-part" |
| 68 | except ET.ParseError: |
| 69 | return None, "invalid-xml" |
| 70 | |
| 71 | |
| 72 | def _diagram_containers(slide_root: ET.Element) -> list[ET.Element]: |
| 73 | containers: list[ET.Element] = [] |
| 74 | for frame in slide_root.findall(".//p:graphicFrame", DIAGRAM_NS_MAP): |
| 75 | graphic_data = frame.find("a:graphic/a:graphicData", DIAGRAM_NS_MAP) |
| 76 | if graphic_data is not None and graphic_data.attrib.get("uri") == DIAGRAM_URI: |
| 77 | containers.append(frame) |
| 78 | return containers |
| 79 | |
| 80 | |
| 81 | def _fallback_preview_shape_ids(slide_root: ET.Element) -> set[str]: |
| 82 | shape_ids: set[str] = set() |
| 83 | for alternate in slide_root.findall(".//mc:AlternateContent", DIAGRAM_NS_MAP): |
| 84 | frame = alternate.find("mc:Choice//p:graphicFrame", DIAGRAM_NS_MAP) |
| 85 | if frame is None: |
| 86 | continue |
| 87 | graphic_data = frame.find("a:graphic/a:graphicData", DIAGRAM_NS_MAP) |
| 88 | if graphic_data is None or graphic_data.attrib.get("uri") != DIAGRAM_URI: |
| 89 | continue |
| 90 | if alternate.find("mc:Fallback//p:pic", DIAGRAM_NS_MAP) is None: |
| 91 | continue |
| 92 | shape_id, _shape_name = _shape_identity(frame, len(shape_ids) + 1) |
| 93 | shape_ids.add(shape_id) |
| 94 | return shape_ids |
| 95 | |
| 96 | |
| 97 | def _diagram_parts( |
| 98 | rel_ids: ET.Element | None, |
| 99 | relationships: dict[str, dict[str, str]], |
| 100 | slide_part: str, |
| 101 | ) -> tuple[dict[str, str], str | None]: |
| 102 | if rel_ids is None: |
| 103 | return {}, "missing-rel-ids" |
| 104 | |
| 105 | parts: dict[str, str] = {} |
| 106 | data_error: str | None = None |
| 107 | for key, attr_name in _REL_ATTRS.items(): |
| 108 | rel_id = rel_ids.attrib.get(_qn(NS["r"], attr_name), "") |
| 109 | if not rel_id: |
| 110 | if key == "data": |
| 111 | data_error = "missing-data-relationship" |
| 112 | continue |
| 113 | relationship = relationships.get(rel_id) |
| 114 | if relationship is None: |
| 115 | if key == "data": |
| 116 | data_error = "missing-data-relationship" |
| 117 | continue |
| 118 | if not _relationship_matches( |
| 119 | relationship.get("type", ""), |
| 120 | _DIAGRAM_REL_SUFFIXES[key], |
| 121 | ): |
| 122 | if key == "data": |
| 123 | data_error = "invalid-data-relationship" |
| 124 | continue |
| 125 | parts[key] = _normalize_part(relationship["target"], slide_part) |
| 126 | return parts, data_error |
| 127 | |
| 128 | |
| 129 | def _persisted_drawing_part( |
| 130 | zf: zipfile.ZipFile, |
| 131 | data_part: str | None, |
| 132 | data_root: ET.Element | None, |
| 133 | slide_part: str, |
| 134 | slide_relationships: dict[str, dict[str, str]], |
| 135 | ) -> tuple[str | None, list[str]]: |
| 136 | if not data_part or data_root is None: |
| 137 | return None, [] |
| 138 | data_model_ext = data_root.find(".//dsp:dataModelExt", DIAGRAM_NS_MAP) |
| 139 | rel_id = data_model_ext.attrib.get("relId", "") if data_model_ext is not None else "" |
| 140 | if not rel_id: |
| 141 | return None, [] |
| 142 | |
| 143 | warnings: list[str] = [] |
| 144 | relationship = slide_relationships.get(rel_id) |
| 145 | relationship_owner = slide_part |
| 146 | if relationship is not None and not _relationship_matches( |
| 147 | relationship.get("type", ""), |
| 148 | "diagramDrawing", |
| 149 | ): |
| 150 | relationship = None |
| 151 | if relationship is None: |
| 152 | try: |
| 153 | data_relationships = _slide_relationships(zf, _rels_name_for_part(data_part)) |
| 154 | except ET.ParseError: |
| 155 | data_relationships = {} |
| 156 | warnings.append("invalid-persisted-drawing-relationships") |
| 157 | relationship = data_relationships.get(rel_id) |
| 158 | relationship_owner = data_part |
| 159 | if relationship is not None and not _relationship_matches( |
| 160 | relationship.get("type", ""), |
| 161 | "diagramDrawing", |
| 162 | ): |
| 163 | relationship = None |
| 164 | if relationship is None: |
| 165 | warnings.append("unresolved-persisted-drawing-relationship") |
| 166 | return None, warnings |
| 167 | |
| 168 | part_name = _normalize_part(relationship["target"], relationship_owner) |
| 169 | if part_name not in zf.namelist(): |
| 170 | warnings.append("missing-persisted-drawing-part") |
| 171 | return None, warnings |
| 172 | return part_name, warnings |
| 173 | |
| 174 | |
| 175 | def _layout_info( |
| 176 | layout_root: ET.Element | None, |
| 177 | data_root: ET.Element | None, |
| 178 | ) -> dict[str, Any]: |
| 179 | unique_id = layout_root.attrib.get("uniqueId") if layout_root is not None else None |
| 180 | if not unique_id and data_root is not None: |
| 181 | document_properties = data_root.find( |
| 182 | ".//dgm:pt[@type='doc']/dgm:prSet", |
| 183 | DIAGRAM_NS_MAP, |
| 184 | ) |
| 185 | if document_properties is not None: |
| 186 | unique_id = document_properties.attrib.get("loTypeId") |
| 187 | title = layout_root.find("dgm:title", DIAGRAM_NS_MAP) if layout_root is not None else None |
| 188 | name = title.attrib.get("val") if title is not None else None |
| 189 | if not name and unique_id: |
| 190 | name = unique_id.rstrip("/").rsplit("/", 1)[-1] |
| 191 | categories = ( |
| 192 | [ |
| 193 | category.attrib["type"] |
| 194 | for category in layout_root.findall("dgm:catLst/dgm:cat", DIAGRAM_NS_MAP) |
| 195 | if category.attrib.get("type") |
| 196 | ] |
| 197 | if layout_root is not None |
| 198 | else [] |
| 199 | ) |
| 200 | return { |
| 201 | "name": name, |
| 202 | "unique_id": unique_id, |
| 203 | "categories": categories, |
| 204 | } |
| 205 | |
| 206 | |
| 207 | def _integer_or_none(value: str | None) -> int | None: |
| 208 | if value is None: |
| 209 | return None |
| 210 | try: |
| 211 | return int(value) |
| 212 | except ValueError: |
| 213 | return None |
| 214 | |
| 215 | |
| 216 | def _point_text(point: ET.Element) -> str: |
| 217 | text_body = point.find("dgm:t", DIAGRAM_NS_MAP) |
| 218 | if text_body is None: |
| 219 | return "" |
| 220 | return "\n".join(_paragraph_texts(text_body)).strip() |
| 221 | |
| 222 | |
| 223 | def _connections(data_root: ET.Element) -> list[dict[str, Any]]: |
| 224 | connections: list[dict[str, Any]] = [] |
| 225 | for connection in data_root.findall( |
| 226 | ".//dgm:cxnLst/dgm:cxn", |
| 227 | DIAGRAM_NS_MAP, |
| 228 | ): |
| 229 | source_id = connection.attrib.get("srcId", "") |
| 230 | destination_id = connection.attrib.get("destId", "") |
| 231 | if not source_id or not destination_id: |
| 232 | continue |
| 233 | connections.append( |
| 234 | { |
| 235 | "type": connection.attrib.get("type") or "parOf", |
| 236 | "source_id": source_id, |
| 237 | "destination_id": destination_id, |
| 238 | "source_order": _integer_or_none(connection.attrib.get("srcOrd")), |
| 239 | "destination_order": _integer_or_none(connection.attrib.get("destOrd")), |
| 240 | } |
| 241 | ) |
| 242 | return connections |
| 243 | |
| 244 | |
| 245 | def _nearest_content_parent( |
| 246 | node_id: str, |
| 247 | parent_by_id: dict[str, str], |
| 248 | content_ids: set[str], |
| 249 | ) -> str | None: |
| 250 | current = parent_by_id.get(node_id) |
| 251 | visited = {node_id} |
| 252 | while current: |
| 253 | if current in content_ids: |
| 254 | return current |
| 255 | if current in visited: |
| 256 | return None |
| 257 | visited.add(current) |
| 258 | current = parent_by_id.get(current) |
| 259 | return None |
| 260 | |
| 261 | |
| 262 | def _break_parent_cycles( |
| 263 | parent_by_node: dict[str, str | None], |
| 264 | source_order: dict[str, int], |
| 265 | ) -> list[str]: |
| 266 | """Break cycles in the Markdown tree projection while retaining raw connections.""" |
| 267 | warnings: list[str] = [] |
| 268 | for start_id in sorted(parent_by_node, key=lambda node_id: source_order[node_id]): |
| 269 | path: list[str] = [] |
| 270 | path_index: dict[str, int] = {} |
| 271 | current_id: str | None = start_id |
| 272 | while current_id is not None and current_id in parent_by_node: |
| 273 | if current_id in path_index: |
| 274 | cycle = path[path_index[current_id] :] |
| 275 | root_id = min(cycle, key=lambda node_id: source_order[node_id]) |
| 276 | parent_by_node[root_id] = None |
| 277 | warnings.append(f"parent-cycle-broken-at:{root_id}") |
| 278 | break |
| 279 | path_index[current_id] = len(path) |
| 280 | path.append(current_id) |
| 281 | current_id = parent_by_node.get(current_id) |
| 282 | return warnings |
| 283 | |
| 284 | |
| 285 | def _ordered_nodes( |
| 286 | data_root: ET.Element, |
| 287 | ) -> tuple[list[dict[str, Any]], list[dict[str, Any]], list[str]]: |
| 288 | raw_points: list[dict[str, Any]] = [] |
| 289 | for source_index, point in enumerate( |
| 290 | data_root.findall(".//dgm:ptLst/dgm:pt", DIAGRAM_NS_MAP), |
| 291 | ): |
| 292 | model_id = point.attrib.get("modelId", "") |
| 293 | point_type = point.attrib.get("type") or "node" |
| 294 | text = _point_text(point) |
| 295 | if not model_id or point_type not in _CONTENT_POINT_TYPES: |
| 296 | continue |
| 297 | raw_points.append( |
| 298 | { |
| 299 | "id": model_id, |
| 300 | "type": point_type, |
| 301 | "text": text, |
| 302 | "source_index": source_index, |
| 303 | } |
| 304 | ) |
| 305 | |
| 306 | all_connections = _connections(data_root) |
| 307 | parent_by_id: dict[str, str] = {} |
| 308 | order_by_id: dict[str, int | None] = {} |
| 309 | for connection in all_connections: |
| 310 | if connection["type"] != "parOf": |
| 311 | continue |
| 312 | destination_id = connection["destination_id"] |
| 313 | parent_by_id.setdefault(destination_id, connection["source_id"]) |
| 314 | order_by_id.setdefault(destination_id, connection["source_order"]) |
| 315 | |
| 316 | content_ids = {point["id"] for point in raw_points} |
| 317 | source_order = {point["id"]: int(point["source_index"]) for point in raw_points} |
| 318 | parent_by_node = { |
| 319 | point["id"]: _nearest_content_parent( |
| 320 | point["id"], |
| 321 | parent_by_id, |
| 322 | content_ids, |
| 323 | ) |
| 324 | for point in raw_points |
| 325 | } |
| 326 | structure_warnings = _break_parent_cycles(parent_by_node, source_order) |
| 327 | nodes_by_id: dict[str, dict[str, Any]] = {} |
| 328 | for point in raw_points: |
| 329 | node = { |
| 330 | "id": point["id"], |
| 331 | "type": point["type"], |
| 332 | "text": point["text"], |
| 333 | "parent_id": parent_by_node[point["id"]], |
| 334 | "order": order_by_id.get(point["id"]), |
| 335 | "depth": 0, |
| 336 | "_source_index": point["source_index"], |
| 337 | } |
| 338 | nodes_by_id[point["id"]] = node |
| 339 | |
| 340 | children: dict[str | None, list[dict[str, Any]]] = {} |
| 341 | for node in nodes_by_id.values(): |
| 342 | children.setdefault(node["parent_id"], []).append(node) |
| 343 | |
| 344 | def sort_key(node: dict[str, Any]) -> tuple[int, int]: |
| 345 | order = node["order"] |
| 346 | return ( |
| 347 | order if isinstance(order, int) else 1_000_000, |
| 348 | int(node["_source_index"]), |
| 349 | ) |
| 350 | |
| 351 | for siblings in children.values(): |
| 352 | siblings.sort(key=sort_key) |
| 353 | |
| 354 | ordered: list[dict[str, Any]] = [] |
| 355 | visited: set[str] = set() |
| 356 | |
| 357 | def visit(node: dict[str, Any], depth: int) -> None: |
| 358 | node_id = str(node["id"]) |
| 359 | if node_id in visited: |
| 360 | return |
| 361 | visited.add(node_id) |
| 362 | node["depth"] = depth |
| 363 | ordered.append(node) |
| 364 | for child in children.get(node_id, []): |
| 365 | visit(child, depth + 1) |
| 366 | |
| 367 | for root in children.get(None, []): |
| 368 | visit(root, 0) |
| 369 | for node in sorted(nodes_by_id.values(), key=sort_key): |
| 370 | if node["id"] not in visited: |
| 371 | visit(node, 0) |
| 372 | |
| 373 | for node in ordered: |
| 374 | node.pop("_source_index", None) |
| 375 | |
| 376 | visible_connections = [ |
| 377 | connection |
| 378 | for connection in all_connections |
| 379 | if connection["type"] == "parOf" |
| 380 | and connection["source_id"] in content_ids |
| 381 | and connection["destination_id"] in content_ids |
| 382 | ] |
| 383 | return ordered, visible_connections, structure_warnings |
| 384 | |
| 385 | |
| 386 | def _read_diagram_container( |
| 387 | zf: zipfile.ZipFile, |
| 388 | container: ET.Element, |
| 389 | *, |
| 390 | slide_part: str, |
| 391 | slide_index: int, |
| 392 | order: int, |
| 393 | relationships: dict[str, dict[str, str]], |
| 394 | relationship_error: str | None, |
| 395 | fallback_shape_ids: set[str], |
| 396 | ) -> dict[str, Any]: |
| 397 | shape_id, shape_name = _shape_identity(container, order) |
| 398 | graphic_data = container.find("a:graphic/a:graphicData", DIAGRAM_NS_MAP) |
| 399 | rel_ids = ( |
| 400 | graphic_data.find("dgm:relIds", DIAGRAM_NS_MAP) |
| 401 | if graphic_data is not None |
| 402 | else None |
| 403 | ) |
| 404 | parts, diagram_relation_error = _diagram_parts( |
| 405 | rel_ids, |
| 406 | relationships, |
| 407 | slide_part, |
| 408 | ) |
| 409 | data_root, data_error = _read_optional_xml(zf, parts.get("data")) |
| 410 | layout_root, layout_error = _read_optional_xml(zf, parts.get("layout")) |
| 411 | metadata_warnings = [f"layout:{layout_error}"] if layout_error else [] |
| 412 | nodes: list[dict[str, Any]] = [] |
| 413 | connections: list[dict[str, Any]] = [] |
| 414 | structure_warnings: list[str] = [] |
| 415 | if data_root is not None: |
| 416 | nodes, connections, structure_warnings = _ordered_nodes(data_root) |
| 417 | |
| 418 | status = relationship_error or diagram_relation_error or data_error |
| 419 | if status is None and structure_warnings: |
| 420 | status = "structure-cycle" |
| 421 | persisted_drawing, drawing_warnings = _persisted_drawing_part( |
| 422 | zf, |
| 423 | parts.get("data"), |
| 424 | data_root, |
| 425 | slide_part, |
| 426 | relationships, |
| 427 | ) |
| 428 | text_items = [node["text"] for node in nodes if node["text"]] |
| 429 | return { |
| 430 | "diagram_id": f"s{slide_index:02d}_dgm{shape_id}", |
| 431 | "kind": "smartart", |
| 432 | "shape_id": shape_id, |
| 433 | "shape_name": shape_name, |
| 434 | "geometry": _container_geometry(container), |
| 435 | "layout": _layout_info(layout_root, data_root), |
| 436 | "root_ids": [node["id"] for node in nodes if node["parent_id"] is None], |
| 437 | "nodes": nodes, |
| 438 | "connections": connections, |
| 439 | "text_items": text_items, |
| 440 | "node_count": len(nodes), |
| 441 | "text_count": len(text_items), |
| 442 | "connection_count": len(connections), |
| 443 | "max_depth": max((int(node["depth"]) for node in nodes), default=0), |
| 444 | "text_extracted": data_root is not None, |
| 445 | "has_persisted_drawing": persisted_drawing is not None, |
| 446 | "has_fallback_preview": shape_id in fallback_shape_ids, |
| 447 | "status": status or "ok", |
| 448 | "warnings": metadata_warnings + structure_warnings + drawing_warnings, |
| 449 | } |
| 450 | |
| 451 | |
| 452 | def _failed_diagram( |
| 453 | container: ET.Element, |
| 454 | *, |
| 455 | slide_index: int, |
| 456 | order: int, |
| 457 | error: Exception, |
| 458 | fallback_shape_ids: set[str], |
| 459 | ) -> dict[str, Any]: |
| 460 | shape_id, shape_name = _shape_identity(container, order) |
| 461 | return { |
| 462 | "diagram_id": f"s{slide_index:02d}_dgm{shape_id}", |
| 463 | "kind": "smartart", |
| 464 | "shape_id": shape_id, |
| 465 | "shape_name": shape_name, |
| 466 | "geometry": _container_geometry(container), |
| 467 | "layout": {"name": None, "unique_id": None, "categories": []}, |
| 468 | "root_ids": [], |
| 469 | "nodes": [], |
| 470 | "connections": [], |
| 471 | "text_items": [], |
| 472 | "node_count": 0, |
| 473 | "text_count": 0, |
| 474 | "connection_count": 0, |
| 475 | "max_depth": 0, |
| 476 | "text_extracted": False, |
| 477 | "has_persisted_drawing": False, |
| 478 | "has_fallback_preview": shape_id in fallback_shape_ids, |
| 479 | "status": "diagram-read-error", |
| 480 | "warnings": [f"{type(error).__name__}:{error}"], |
| 481 | } |
| 482 | |
| 483 | |
| 484 | def read_smartart_diagrams( |
| 485 | zf: zipfile.ZipFile, |
| 486 | slide_part: str, |
| 487 | slide_index: int, |
| 488 | ) -> list[dict[str, Any]]: |
| 489 | """Return SmartArt source facts for one slide part.""" |
| 490 | slide_root = _read_xml(zf, slide_part) |
| 491 | relationship_error: str | None = None |
| 492 | try: |
| 493 | relationships = _slide_relationships(zf, _rels_name_for_part(slide_part)) |
| 494 | except ET.ParseError: |
| 495 | relationships = {} |
| 496 | relationship_error = "invalid-slide-relationships" |
| 497 | fallback_shape_ids = _fallback_preview_shape_ids(slide_root) |
| 498 | diagrams: list[dict[str, Any]] = [] |
| 499 | |
| 500 | for order, container in enumerate(_diagram_containers(slide_root), start=1): |
| 501 | try: |
| 502 | diagram = _read_diagram_container( |
| 503 | zf, |
| 504 | container, |
| 505 | slide_part=slide_part, |
| 506 | slide_index=slide_index, |
| 507 | order=order, |
| 508 | relationships=relationships, |
| 509 | relationship_error=relationship_error, |
| 510 | fallback_shape_ids=fallback_shape_ids, |
| 511 | ) |
| 512 | except (OSError, RuntimeError, zipfile.BadZipFile, ET.ParseError, KeyError, ValueError) as exc: |
| 513 | diagram = _failed_diagram( |
| 514 | container, |
| 515 | slide_index=slide_index, |
| 516 | order=order, |
| 517 | error=exc, |
| 518 | fallback_shape_ids=fallback_shape_ids, |
| 519 | ) |
| 520 | diagrams.append(diagram) |
| 521 | return diagrams |
| 522 | |
| 523 | |
| 524 | def smartart_to_markdown(diagram: dict[str, Any]) -> str: |
| 525 | """Render one extracted SmartArt diagram as hierarchical Markdown.""" |
| 526 | name = str(diagram.get("shape_name") or diagram.get("diagram_id") or "SmartArt") |
| 527 | layout = diagram.get("layout") or {} |
| 528 | layout_name = str(layout.get("name") or "") if isinstance(layout, dict) else "" |
| 529 | heading = f"### SmartArt: {name}" |
| 530 | if layout_name and layout_name.lower() not in name.lower(): |
| 531 | heading += f" — {layout_name}" |
| 532 | lines = [heading, ""] |
| 533 | nodes = diagram.get("nodes") or [] |
| 534 | text_nodes = [node for node in nodes if str(node.get("text") or "").strip()] |
| 535 | if text_nodes: |
| 536 | nodes_by_id = { |
| 537 | str(node.get("id")): node |
| 538 | for node in nodes |
| 539 | if node.get("id") is not None |
| 540 | } |
| 541 | rendered_depths: dict[str, int] = {} |
| 542 | |
| 543 | def rendered_depth(node: dict[str, Any]) -> int: |
| 544 | node_id = str(node.get("id") or "") |
| 545 | if node_id in rendered_depths: |
| 546 | return rendered_depths[node_id] |
| 547 | parent = nodes_by_id.get(str(node.get("parent_id") or "")) |
| 548 | if parent is None: |
| 549 | depth = 0 |
| 550 | else: |
| 551 | depth = rendered_depth(parent) |
| 552 | if str(parent.get("text") or "").strip(): |
| 553 | depth += 1 |
| 554 | rendered_depths[node_id] = depth |
| 555 | return depth |
| 556 | |
| 557 | for node in text_nodes: |
| 558 | text = " / ".join(str(node.get("text") or "").splitlines()).strip() |
| 559 | indent = " " * rendered_depth(node) |
| 560 | lines.append(f"{indent}- {text}") |
| 561 | elif nodes: |
| 562 | lines.append(f"> [SmartArt structure has {len(nodes)} node(s), but no text]") |
| 563 | else: |
| 564 | status = str(diagram.get("status") or "content-unavailable") |
| 565 | if diagram.get("text_extracted") and status == "ok": |
| 566 | lines.append("> [SmartArt data is readable but has no semantic nodes]") |
| 567 | else: |
| 568 | lines.append(f"> [SmartArt content unavailable: {status}]") |
| 569 | return "\n".join(lines).rstrip() |
| 570 |