| 1 | #!/usr/bin/env python3 |
| 2 | """ |
| 3 | PowerPoint to Markdown Converter |
| 4 | |
| 5 | Extracts slide text, tables, SmartArt node structure, speaker notes, and |
| 6 | embedded pictures from Open XML PowerPoint files into Markdown. |
| 7 | |
| 8 | Primary use case: PPTX source decks -> Markdown for PPT generation input. |
| 9 | |
| 10 | Hyperlinks present in the source deck are preserved: run-level external URLs |
| 11 | and slide-internal jumps are emitted as ``[text](url)`` / ``[text](#slide-N)``, |
| 12 | with a shape-level ``click_action`` fallback. |
| 13 | |
| 14 | Dependency: |
| 15 | pip install python-pptx |
| 16 | |
| 17 | API stability note: |
| 18 | Detecting slide-internal jumps (``ppaction://hlinksldjump``) reads |
| 19 | ``run._r`` (the CT_TextRun lxml element) because python-pptx exposes no |
| 20 | public API to distinguish an internal jump from an external URL. XY chart |
| 21 | extraction likewise reads ``series._element`` for X values and bubble sizes, |
| 22 | which the public chart API does not expose. Keep these private accesses |
| 23 | localized here and covered by conversion smoke tests. |
| 24 | """ |
| 25 | |
| 26 | from __future__ import annotations |
| 27 | |
| 28 | import argparse |
| 29 | import hashlib |
| 30 | import json |
| 31 | import re |
| 32 | import shutil |
| 33 | import sys |
| 34 | import zipfile |
| 35 | from dataclasses import dataclass |
| 36 | from io import BytesIO |
| 37 | from pathlib import Path |
| 38 | from urllib.parse import quote |
| 39 | from xml.etree import ElementTree as ET |
| 40 | |
| 41 | _SCRIPTS_DIR = Path(__file__).resolve().parents[1] |
| 42 | if str(_SCRIPTS_DIR) not in sys.path: |
| 43 | sys.path.insert(0, str(_SCRIPTS_DIR)) |
| 44 | |
| 45 | from console_encoding import configure_utf8_stdio # noqa: E402 |
| 46 | from _batch import run_path_batch # noqa: E402 |
| 47 | from _conversion_profile import write_conversion_profile_best_effort # noqa: E402 |
| 48 | from template_fill_pptx.diagram_read import ( # noqa: E402 |
| 49 | read_smartart_diagrams, |
| 50 | smartart_to_markdown, |
| 51 | ) |
| 52 | |
| 53 | from pptx import Presentation |
| 54 | from pptx.enum.action import PP_ACTION |
| 55 | from pptx.enum.shapes import MSO_SHAPE_TYPE |
| 56 | from pptx.oxml.ns import qn |
| 57 | |
| 58 | configure_utf8_stdio() |
| 59 | |
| 60 | |
| 61 | EMU_PER_INCH = 914400 |
| 62 | DRAWINGML_NS = "http://schemas.openxmlformats.org/drawingml/2006/main" |
| 63 | PRESENTATIONML_NS = "http://schemas.openxmlformats.org/presentationml/2006/main" |
| 64 | RELATIONSHIP_NS = "http://schemas.openxmlformats.org/officeDocument/2006/relationships" |
| 65 | CHARTEX_URI = "http://schemas.microsoft.com/office/drawing/2014/chartex" |
| 66 | OFFICE_VECTOR_EXTENSIONS = {"emf", "wmf"} |
| 67 | IMAGE_EXT_BY_CONTENT_TYPE = { |
| 68 | "image/bmp": "bmp", |
| 69 | "image/gif": "gif", |
| 70 | "image/jpeg": "jpg", |
| 71 | "image/jpg": "jpg", |
| 72 | "image/png": "png", |
| 73 | "image/svg+xml": "svg", |
| 74 | "image/tiff": "tiff", |
| 75 | "image/x-emf": "emf", |
| 76 | "image/x-wmf": "wmf", |
| 77 | } |
| 78 | LEGACY_GENERATED_IMAGE_RE = re.compile(r"^slide_\d{2}_image_\d{2}\.[A-Za-z0-9]+$") |
| 79 | _READBACK_SLIDE_HEADING_RE = re.compile(r"^## Slide\s+\d+\s*$") |
| 80 | _READBACK_NOTES_HEADING_RE = re.compile(r"^### Speaker Notes\s*$") |
| 81 | |
| 82 | # Hyperlink schemes dropped during extraction (a blacklist of known-dangerous |
| 83 | # schemes). PowerPoint also rejects unrecognized schemes at open time, so the |
| 84 | # residual risk from schemes not listed here is low. |
| 85 | UNSUPPORTED_URL_SCHEMES = ("javascript:", "data:", "vbscript:", "file:") |
| 86 | |
| 87 | |
| 88 | SUPPORTED_FORMATS = { |
| 89 | ".pptx": "PowerPoint Presentation", |
| 90 | ".pptm": "Macro-enabled PowerPoint Presentation", |
| 91 | ".ppsx": "PowerPoint Slide Show", |
| 92 | ".ppsm": "Macro-enabled PowerPoint Slide Show", |
| 93 | ".potx": "PowerPoint Template", |
| 94 | ".potm": "Macro-enabled PowerPoint Template", |
| 95 | } |
| 96 | |
| 97 | |
| 98 | @dataclass |
| 99 | class LeafShape: |
| 100 | """Flattened leaf shape with stable position ordering.""" |
| 101 | |
| 102 | shape: object |
| 103 | top: int |
| 104 | left: int |
| 105 | |
| 106 | |
| 107 | @dataclass |
| 108 | class SavedPicture: |
| 109 | """Extracted image asset plus manifest metadata.""" |
| 110 | |
| 111 | filename: str |
| 112 | manifest_entry: dict[str, object] |
| 113 | is_new_asset: bool |
| 114 | |
| 115 | |
| 116 | def normalize_text(value: str) -> str: |
| 117 | """Collapse whitespace while preserving paragraph boundaries elsewhere.""" |
| 118 | value = value.replace("\r\n", "\n").replace("\r", "\n") |
| 119 | lines = [re.sub(r"\s+", " ", line).strip() for line in value.split("\n")] |
| 120 | lines = [line for line in lines if line] |
| 121 | return "\n".join(lines) |
| 122 | |
| 123 | |
| 124 | def _escape_readback_control_lines(value: str) -> str: |
| 125 | """Escape ordinary text lines that collide with converter section markers.""" |
| 126 | lines = value.split("\n") |
| 127 | return "\n".join( |
| 128 | f"\\{line}" |
| 129 | if ( |
| 130 | _READBACK_SLIDE_HEADING_RE.fullmatch(line) |
| 131 | or _READBACK_NOTES_HEADING_RE.fullmatch(line) |
| 132 | ) |
| 133 | else line |
| 134 | for line in lines |
| 135 | ) |
| 136 | |
| 137 | |
| 138 | def normalize_ext(ext: str | None, content_type: str | None = None) -> str: |
| 139 | """Return a lowercase extension without a leading dot.""" |
| 140 | if ext: |
| 141 | ext = ext.lower().lstrip(".") |
| 142 | if ext == "jpeg": |
| 143 | return "jpg" |
| 144 | return ext |
| 145 | if content_type: |
| 146 | return IMAGE_EXT_BY_CONTENT_TYPE.get(content_type.lower(), "bin") |
| 147 | return "bin" |
| 148 | |
| 149 | |
| 150 | def sanitize_filename(value: str) -> str: |
| 151 | """Return a filesystem-safe basename.""" |
| 152 | value = re.sub(r"[^\w.\-]+", "_", value, flags=re.UNICODE) |
| 153 | return value.strip("._") or "asset" |
| 154 | |
| 155 | |
| 156 | def escape_table_cell(value: str) -> str: |
| 157 | """Escape Markdown table syntax inside a cell.""" |
| 158 | normalized = value.replace("\r\n", "\n").replace("\r", "\n") |
| 159 | lines = [re.sub(r"\s+", " ", line).strip() for line in normalized.split("\n")] |
| 160 | with_breaks = "<br>".join(lines) |
| 161 | return with_breaks.replace("|", r"\|") or " " |
| 162 | |
| 163 | |
| 164 | def _safe_position(shape: object, attr: str) -> int: |
| 165 | """Read a shape's ``top`` / ``left`` EMU, tolerating broken inheritance. |
| 166 | |
| 167 | A placeholder with no explicit position resolves it by walking up to its |
| 168 | master. A deck that ships notesSlides without a notesMaster (or any other |
| 169 | partial inheritance chain) makes python-pptx raise on that lookup, so treat |
| 170 | an unresolvable position as 0 rather than aborting the whole conversion. |
| 171 | """ |
| 172 | try: |
| 173 | return int(getattr(shape, attr, 0) or 0) |
| 174 | except Exception: |
| 175 | return 0 |
| 176 | |
| 177 | |
| 178 | def iter_leaf_shapes(shapes: object) -> list[LeafShape]: |
| 179 | """Return a flattened, reading-order list of shapes.""" |
| 180 | items: list[LeafShape] = [] |
| 181 | for shape in shapes: |
| 182 | if shape.shape_type == MSO_SHAPE_TYPE.GROUP: |
| 183 | items.extend(iter_leaf_shapes(shape.shapes)) |
| 184 | continue |
| 185 | items.append( |
| 186 | LeafShape( |
| 187 | shape=shape, |
| 188 | top=_safe_position(shape, "top"), |
| 189 | left=_safe_position(shape, "left"), |
| 190 | ) |
| 191 | ) |
| 192 | items.sort(key=lambda item: (item.top, item.left)) |
| 193 | return items |
| 194 | |
| 195 | |
| 196 | def _is_supported_url(url: str) -> bool: |
| 197 | """Reject empty URLs and the known-dangerous schemes.""" |
| 198 | return bool(url) and not any( |
| 199 | url.lower().startswith(scheme) for scheme in UNSUPPORTED_URL_SCHEMES |
| 200 | ) |
| 201 | |
| 202 | |
| 203 | def _escape_md_link_text(text: str) -> str: |
| 204 | """Backslash-escape characters that would break a Markdown link label. |
| 205 | |
| 206 | A stray ``]`` in anchor text would otherwise close the ``[...]`` early. |
| 207 | """ |
| 208 | return text.replace("\\", "\\\\").replace("[", "\\[").replace("]", "\\]") |
| 209 | |
| 210 | |
| 211 | def _encode_md_url(url: str) -> str: |
| 212 | """Percent-encode a URL so Markdown link syntax stays unambiguous. |
| 213 | |
| 214 | Notably encodes ``(`` / ``)`` to ``%28`` / ``%29`` so a parenthesised URL |
| 215 | does not terminate the ``[text](url)`` form early. |
| 216 | """ |
| 217 | return quote(url, safe="/:?=&%#@!$'*+,;") |
| 218 | |
| 219 | |
| 220 | def _resolve_internal_jump(run: object, shape: object) -> str | None: |
| 221 | """Return ``#slide-N`` for a run carrying a slide-internal jump, else None. |
| 222 | |
| 223 | Reads ``run._r`` (private python-pptx API) because the public |
| 224 | ``run.hyperlink.address`` cannot tell an internal jump apart from an |
| 225 | external URL — see the module docstring's API stability note. |
| 226 | """ |
| 227 | r_id = "" |
| 228 | try: |
| 229 | rpr = run._r.find(qn("a:rPr")) |
| 230 | if rpr is None: |
| 231 | return None |
| 232 | hlink = rpr.find(qn("a:hlinkClick")) |
| 233 | if hlink is None or "hlinksldjump" not in (hlink.get("action", "") or ""): |
| 234 | return None |
| 235 | r_id = hlink.get(qn("r:id"), "") |
| 236 | if not r_id: |
| 237 | return None |
| 238 | target_slide = shape.part.related_part(r_id).slide |
| 239 | prs = shape.part.slide.part.package.presentation_part.presentation |
| 240 | return f"#slide-{list(prs.slides).index(target_slide) + 1}" |
| 241 | except (KeyError, ValueError, AttributeError): |
| 242 | print(f"[WARN] ppt_to_md: could not resolve slide jump rId={r_id}", file=sys.stderr) |
| 243 | return None |
| 244 | |
| 245 | |
| 246 | def _run_url(run: object, shape: object) -> str | None: |
| 247 | """Resolve a run's hyperlink target to a markdown-ready URL, or None.""" |
| 248 | if shape is not None: |
| 249 | internal = _resolve_internal_jump(run, shape) |
| 250 | if internal: |
| 251 | return internal |
| 252 | try: |
| 253 | addr = run.hyperlink.address |
| 254 | except AttributeError: |
| 255 | return None |
| 256 | if _is_supported_url(addr or ""): |
| 257 | return _encode_md_url(addr) |
| 258 | return None |
| 259 | |
| 260 | |
| 261 | def _paragraph_to_markdown(paragraph: object, shape: object) -> str: |
| 262 | """Render one paragraph, merging consecutive runs that share a URL. |
| 263 | |
| 264 | Run text is concatenated verbatim — including the spaces between runs — and |
| 265 | normalized only once over the assembled paragraph, so a link in the middle |
| 266 | of a sentence does not swallow its surrounding spaces. A link group's own |
| 267 | leading / trailing whitespace is kept outside the ``[...]`` so it separates |
| 268 | words rather than padding the anchor text. |
| 269 | """ |
| 270 | parts = [] |
| 271 | current_url = None |
| 272 | current_text = "" |
| 273 | |
| 274 | def flush(): |
| 275 | if not current_text: |
| 276 | return |
| 277 | if current_url is None: |
| 278 | parts.append(current_text) |
| 279 | return |
| 280 | lead = current_text[: len(current_text) - len(current_text.lstrip())] |
| 281 | trail = current_text[len(current_text.rstrip()):] |
| 282 | core = current_text.strip() |
| 283 | display = _escape_md_link_text(core) if core else current_url |
| 284 | parts.append(f"{lead}[{display}]({current_url}){trail}") |
| 285 | |
| 286 | has_run_hyperlink = False |
| 287 | for run in paragraph.runs: |
| 288 | url = _run_url(run, shape) |
| 289 | if url: |
| 290 | has_run_hyperlink = True |
| 291 | if url != current_url: |
| 292 | flush() |
| 293 | current_text = "" |
| 294 | current_url = url |
| 295 | current_text += run.text or "" |
| 296 | flush() |
| 297 | |
| 298 | text = normalize_text("".join(parts)) |
| 299 | |
| 300 | # Shape-level click_action only matters when no run carried its own link. |
| 301 | if not has_run_hyperlink and shape is not None: |
| 302 | text = _apply_shape_click_action(text, shape) |
| 303 | return text |
| 304 | |
| 305 | |
| 306 | def _apply_shape_click_action(text: str, shape: object) -> str: |
| 307 | """Wrap paragraph text in a link from the shape's click_action, if any.""" |
| 308 | try: |
| 309 | action = shape.click_action |
| 310 | if action.action == PP_ACTION.HYPERLINK: |
| 311 | url = action.hyperlink.address or "" |
| 312 | if _is_supported_url(url): |
| 313 | return f"[{_escape_md_link_text(text)}]({_encode_md_url(url)})" |
| 314 | elif action.action == PP_ACTION.NAMED_SLIDE: |
| 315 | target = action.target_slide |
| 316 | if target is not None: |
| 317 | prs = shape.part.slide.part.package.presentation_part.presentation |
| 318 | idx = list(prs.slides).index(target) + 1 |
| 319 | return f"[{_escape_md_link_text(text)}](#slide-{idx})" |
| 320 | except (AttributeError, ValueError): |
| 321 | print("[WARN] ppt_to_md: could not process shape click_action", file=sys.stderr) |
| 322 | return text |
| 323 | |
| 324 | |
| 325 | def _paragraph_has_hyperlink(paragraph: object) -> bool: |
| 326 | """True if any run carries an external URL or an internal slide jump.""" |
| 327 | for run in paragraph.runs: |
| 328 | try: |
| 329 | if run.hyperlink.address: |
| 330 | return True |
| 331 | except AttributeError: |
| 332 | pass |
| 333 | try: |
| 334 | rpr = run._r.find(qn("a:rPr")) |
| 335 | if rpr is not None and rpr.find(qn("a:hlinkClick")) is not None: |
| 336 | return True |
| 337 | except AttributeError: |
| 338 | continue |
| 339 | return False |
| 340 | |
| 341 | |
| 342 | def text_frame_to_markdown(text_frame: object, shape: object = None) -> str: |
| 343 | """Convert a PowerPoint text frame into Markdown, preserving hyperlinks. |
| 344 | |
| 345 | Run-level external URLs and slide-internal jumps are emitted as |
| 346 | ``[text](url)`` / ``[text](#slide-N)``; consecutive runs sharing a URL are |
| 347 | merged. When no run carries a link, the shape's ``click_action`` is used as |
| 348 | a paragraph-level fallback. Pass ``shape`` to enable hyperlink extraction; |
| 349 | without it the frame degrades to plain text. |
| 350 | """ |
| 351 | visible_paragraphs = [ |
| 352 | paragraph for paragraph in text_frame.paragraphs |
| 353 | if normalize_text(paragraph.text) or _paragraph_has_hyperlink(paragraph) |
| 354 | ] |
| 355 | if not visible_paragraphs: |
| 356 | return "" |
| 357 | |
| 358 | list_like = any(paragraph.level > 0 for paragraph in visible_paragraphs) |
| 359 | if not list_like: |
| 360 | list_like = len(visible_paragraphs) > 1 |
| 361 | |
| 362 | paragraphs = [] |
| 363 | for paragraph in visible_paragraphs: |
| 364 | text = _escape_readback_control_lines( |
| 365 | _paragraph_to_markdown(paragraph, shape) |
| 366 | ) |
| 367 | if not text: |
| 368 | continue |
| 369 | if list_like: |
| 370 | indent = " " * max(paragraph.level, 0) |
| 371 | paragraphs.append(f"{indent}- {text}") |
| 372 | else: |
| 373 | paragraphs.append(text) |
| 374 | |
| 375 | if list_like: |
| 376 | return "\n".join(paragraphs) |
| 377 | return "\n\n".join(paragraphs) |
| 378 | |
| 379 | |
| 380 | def table_to_markdown(table: object) -> str: |
| 381 | """Convert a PowerPoint table to a Markdown table.""" |
| 382 | rows = [] |
| 383 | for row in table.rows: |
| 384 | cells = [escape_table_cell(cell.text) for cell in row.cells] |
| 385 | rows.append(cells) |
| 386 | |
| 387 | if not rows: |
| 388 | return "" |
| 389 | |
| 390 | column_count = max(len(row) for row in rows) |
| 391 | normalized_rows = [row + [" "] * (column_count - len(row)) for row in rows] |
| 392 | header = normalized_rows[0] |
| 393 | separator = ["---"] * column_count |
| 394 | body = normalized_rows[1:] |
| 395 | |
| 396 | lines = [ |
| 397 | "| " + " | ".join(header) + " |", |
| 398 | "| " + " | ".join(separator) + " |", |
| 399 | ] |
| 400 | for row in body: |
| 401 | lines.append("| " + " | ".join(row) + " |") |
| 402 | return "\n".join(lines) |
| 403 | |
| 404 | |
| 405 | def _format_chart_value(value: object) -> str: |
| 406 | """Render a chart data point, trimming whole-number floats.""" |
| 407 | if value is None: |
| 408 | return "" |
| 409 | if isinstance(value, float) and value.is_integer(): |
| 410 | return str(int(value)) |
| 411 | return str(value) |
| 412 | |
| 413 | |
| 414 | def _chart_header(chart: object, name: str) -> tuple[str, str]: |
| 415 | """Return the Markdown chart header and its best-effort type label.""" |
| 416 | try: |
| 417 | chart_type = str(chart.chart_type) |
| 418 | except (ValueError, AttributeError, KeyError): |
| 419 | chart_type = "" |
| 420 | raw_name = "" if name is None else str(name) |
| 421 | chart_name = normalize_text(raw_name).replace("\n", " ") or "Chart" |
| 422 | header = f"> [Chart] {chart_name}" + (f" — {chart_type}" if chart_type else "") |
| 423 | return header, chart_type |
| 424 | |
| 425 | |
| 426 | def _chart_warning_lines(warnings: list[str]) -> list[str]: |
| 427 | """Return stable, de-duplicated Markdown warning blocks.""" |
| 428 | lines: list[str] = [] |
| 429 | seen: set[str] = set() |
| 430 | for warning in warnings: |
| 431 | normalized = normalize_text(warning).replace("\n", " ") or "unknown warning" |
| 432 | if normalized in seen: |
| 433 | continue |
| 434 | seen.add(normalized) |
| 435 | lines.append(f"> [Chart data warning: {normalized}]") |
| 436 | return lines |
| 437 | |
| 438 | |
| 439 | def _chart_data_unavailable( |
| 440 | header: str, |
| 441 | reason: str, |
| 442 | *, |
| 443 | warnings: list[str] | None = None, |
| 444 | ) -> str: |
| 445 | """Attach an explicit data-read failure to a chart heading.""" |
| 446 | normalized_reason = normalize_text(reason).replace("\n", " ") or "unknown reason" |
| 447 | lines = [header] |
| 448 | lines.extend(_chart_warning_lines(warnings or [])) |
| 449 | lines.append(f"> [Chart data unavailable: {normalized_reason}]") |
| 450 | return "\n".join(lines) |
| 451 | |
| 452 | |
| 453 | def _chart_value_cell(value: object) -> str: |
| 454 | """Return one chart table cell while keeping missing values visibly empty.""" |
| 455 | rendered = _format_chart_value(value) |
| 456 | return escape_table_cell(rendered) if rendered else "" |
| 457 | |
| 458 | |
| 459 | def _chart_series_element(series: object) -> object | None: |
| 460 | """Return python-pptx's series XML carrier when its public API is insufficient.""" |
| 461 | element = getattr(series, "_element", None) |
| 462 | if element is not None: |
| 463 | return element |
| 464 | return getattr(series, "_ser", None) |
| 465 | |
| 466 | |
| 467 | def _chart_series_name(series: object, index: int) -> str: |
| 468 | """Return a stable, Markdown-safe series name.""" |
| 469 | try: |
| 470 | raw_name = series.name |
| 471 | except (ValueError, AttributeError, KeyError): |
| 472 | raw_name = None |
| 473 | label = str(raw_name) if raw_name not in (None, "") else f"Series {index}" |
| 474 | return escape_table_cell(label) |
| 475 | |
| 476 | |
| 477 | def _chart_numeric_cache_values( |
| 478 | parent: object | None, |
| 479 | ) -> tuple[list[object | None] | None, str | None]: |
| 480 | """Read one XY display cache through python-pptx's OOXML value helpers.""" |
| 481 | if parent is None: |
| 482 | return None, "missing numeric value container" |
| 483 | try: |
| 484 | point_count = int(parent.ptCount_val) |
| 485 | values = [parent.pt_v(index) for index in range(point_count)] |
| 486 | except (AttributeError, IndexError, TypeError, ValueError): |
| 487 | return None, "invalid or unavailable numeric display cache" |
| 488 | if point_count <= 0 or all(value is None for value in values): |
| 489 | return None, "numeric display cache contains no values" |
| 490 | return values, None |
| 491 | |
| 492 | |
| 493 | def _chart_family(chart_type: str, series: list[object]) -> str: |
| 494 | """Classify category, scatter, and bubble charts without misreading XY as category.""" |
| 495 | type_key = chart_type.upper() |
| 496 | if "BUBBLE" in type_key: |
| 497 | return "bubble" |
| 498 | if "SCATTER" in type_key: |
| 499 | return "scatter" |
| 500 | for item in series: |
| 501 | element = _chart_series_element(item) |
| 502 | if element is None: |
| 503 | continue |
| 504 | if element.find(qn("c:bubbleSize")) is not None: |
| 505 | return "bubble" |
| 506 | if element.find(qn("c:xVal")) is not None or element.find(qn("c:yVal")) is not None: |
| 507 | return "scatter" |
| 508 | return "category" |
| 509 | |
| 510 | |
| 511 | def _xy_chart_to_markdown( |
| 512 | series: list[object], |
| 513 | *, |
| 514 | family: str, |
| 515 | header: str, |
| 516 | ) -> str: |
| 517 | """Render scatter/bubble series as typed per-point X/Y[/size] rows.""" |
| 518 | table_header = ["Series", "Point", "X", "Y"] |
| 519 | if family == "bubble": |
| 520 | table_header.append("Size") |
| 521 | rows: list[list[str]] = [] |
| 522 | warnings: list[str] = [] |
| 523 | |
| 524 | for series_index, item in enumerate(series, start=1): |
| 525 | series_name = _chart_series_name(item, series_index) |
| 526 | element = _chart_series_element(item) |
| 527 | if element is None: |
| 528 | x_values = None |
| 529 | warnings.append(f"{series_name}: series XML is unavailable for X data") |
| 530 | else: |
| 531 | x_values, x_error = _chart_numeric_cache_values( |
| 532 | element.find(qn("c:xVal")) |
| 533 | ) |
| 534 | if x_error: |
| 535 | warnings.append(f"{series_name}: X data {x_error}") |
| 536 | try: |
| 537 | y_values = list(item.values) |
| 538 | except (ValueError, TypeError, AttributeError, KeyError): |
| 539 | y_values = [] |
| 540 | warnings.append(f"{series_name}: Y values are unavailable") |
| 541 | |
| 542 | size_values: list[object | None] | None = None |
| 543 | if family == "bubble": |
| 544 | if element is None: |
| 545 | size_error = "missing series XML" |
| 546 | else: |
| 547 | size_values, size_error = _chart_numeric_cache_values( |
| 548 | element.find(qn("c:bubbleSize")) |
| 549 | ) |
| 550 | if size_error: |
| 551 | warnings.append(f"{series_name}: bubble sizes {size_error}") |
| 552 | |
| 553 | x_values = x_values or [] |
| 554 | point_count = max( |
| 555 | len(x_values), |
| 556 | len(y_values), |
| 557 | len(size_values or []), |
| 558 | ) |
| 559 | if point_count == 0: |
| 560 | warnings.append(f"{series_name}: no readable points") |
| 561 | continue |
| 562 | |
| 563 | point_counts = {len(x_values), len(y_values)} |
| 564 | if family == "bubble": |
| 565 | point_counts.add(len(size_values or [])) |
| 566 | if len(point_counts) > 1: |
| 567 | dimensions = "X/Y/size" if family == "bubble" else "X/Y" |
| 568 | warnings.append( |
| 569 | f"{series_name}: {dimensions} point counts differ; " |
| 570 | "missing cells are blank" |
| 571 | ) |
| 572 | |
| 573 | for point_index in range(point_count): |
| 574 | x_value = x_values[point_index] if point_index < len(x_values) else None |
| 575 | y_value = y_values[point_index] if point_index < len(y_values) else None |
| 576 | row = [ |
| 577 | series_name, |
| 578 | str(point_index + 1), |
| 579 | _chart_value_cell(x_value), |
| 580 | _chart_value_cell(y_value), |
| 581 | ] |
| 582 | if family == "bubble": |
| 583 | size_value = ( |
| 584 | size_values[point_index] |
| 585 | if size_values is not None and point_index < len(size_values) |
| 586 | else None |
| 587 | ) |
| 588 | row.append(_chart_value_cell(size_value)) |
| 589 | rows.append(row) |
| 590 | |
| 591 | if not rows: |
| 592 | return _chart_data_unavailable( |
| 593 | header, |
| 594 | "chart has no readable XY points", |
| 595 | warnings=warnings, |
| 596 | ) |
| 597 | lines = [header] |
| 598 | lines.extend(_chart_warning_lines(warnings)) |
| 599 | lines.extend([ |
| 600 | "", |
| 601 | "| " + " | ".join(table_header) + " |", |
| 602 | "| " + " | ".join(["---"] * len(table_header)) + " |", |
| 603 | ]) |
| 604 | lines.extend("| " + " | ".join(row) + " |" for row in rows) |
| 605 | return "\n".join(lines) |
| 606 | |
| 607 | |
| 608 | def _category_chart_to_markdown(chart: object, series: list[object], header: str) -> str: |
| 609 | """Render a conventional category chart through python-pptx's public API.""" |
| 610 | categories: list[str] = [] |
| 611 | warnings: list[str] = [] |
| 612 | has_category_xml = any( |
| 613 | (element := _chart_series_element(item)) is not None |
| 614 | and element.find(qn("c:cat")) is not None |
| 615 | for item in series |
| 616 | ) |
| 617 | try: |
| 618 | plots = list(chart.plots) |
| 619 | if plots: |
| 620 | categories = [ |
| 621 | escape_table_cell(str(category)) if category is not None else "" |
| 622 | for category in plots[0].categories |
| 623 | ] |
| 624 | except (ValueError, TypeError, IndexError, AttributeError, KeyError): |
| 625 | if has_category_xml: |
| 626 | warnings.append("chart categories are unavailable; using point numbers") |
| 627 | if not has_category_xml and not categories: |
| 628 | warnings.append("chart categories are missing; using point numbers") |
| 629 | elif has_category_xml and not categories: |
| 630 | warnings.append("chart categories are empty; using point numbers") |
| 631 | |
| 632 | series_data: list[tuple[str, list[object]]] = [] |
| 633 | for index, item in enumerate(series, start=1): |
| 634 | series_name = _chart_series_name(item, index) |
| 635 | try: |
| 636 | values = list(item.values) |
| 637 | except (ValueError, TypeError, AttributeError, KeyError): |
| 638 | warnings.append(f"{series_name}: series values are unavailable") |
| 639 | continue |
| 640 | series_data.append((series_name, values)) |
| 641 | |
| 642 | row_count = max( |
| 643 | len(categories), |
| 644 | max((len(values) for _, values in series_data), default=0), |
| 645 | ) |
| 646 | if not series_data or row_count == 0: |
| 647 | return _chart_data_unavailable( |
| 648 | header, |
| 649 | "chart has no readable category-series data", |
| 650 | warnings=warnings, |
| 651 | ) |
| 652 | point_counts = {len(values) for _, values in series_data} |
| 653 | if categories: |
| 654 | point_counts.add(len(categories)) |
| 655 | if len(point_counts) > 1: |
| 656 | warnings.append("category/series point counts differ; missing cells are blank") |
| 657 | |
| 658 | table_header = (["Category"] if categories else ["#"]) + [ |
| 659 | series_name for series_name, _ in series_data |
| 660 | ] |
| 661 | lines = [header] |
| 662 | lines.extend(_chart_warning_lines(warnings)) |
| 663 | lines.extend([ |
| 664 | "", |
| 665 | "| " + " | ".join(table_header) + " |", |
| 666 | "| " + " | ".join(["---"] * len(table_header)) + " |", |
| 667 | ]) |
| 668 | for row_index in range(row_count): |
| 669 | if categories: |
| 670 | label = categories[row_index] if row_index < len(categories) else "" |
| 671 | else: |
| 672 | label = str(row_index + 1) |
| 673 | cells = [label] |
| 674 | for _, values in series_data: |
| 675 | value = values[row_index] if row_index < len(values) else None |
| 676 | cells.append(_chart_value_cell(value)) |
| 677 | lines.append("| " + " | ".join(cells) + " |") |
| 678 | return "\n".join(lines) |
| 679 | |
| 680 | |
| 681 | def chart_to_markdown(chart: object, name: str) -> str: |
| 682 | """Render category, scatter, and bubble data without flattening chart semantics. |
| 683 | |
| 684 | A native PowerPoint chart stores its data in embedded XML, not in any text |
| 685 | frame. Public python-pptx APIs cover category values and scatter/bubble Y |
| 686 | values, but not XY X coordinates or bubble sizes. Read only those missing |
| 687 | display caches from the series XML. Preserve every readable value and emit |
| 688 | explicit warnings for missing dimensions or series rather than discarding |
| 689 | the chart's remaining content. |
| 690 | """ |
| 691 | header, chart_type = _chart_header(chart, name) |
| 692 | try: |
| 693 | series = list(chart.series) |
| 694 | except (ValueError, TypeError, AttributeError, KeyError): |
| 695 | return _chart_data_unavailable(header, "chart series are unavailable") |
| 696 | if not series: |
| 697 | return _chart_data_unavailable(header, "chart has no readable series") |
| 698 | |
| 699 | family = _chart_family(chart_type, series) |
| 700 | if family in {"scatter", "bubble"}: |
| 701 | return _xy_chart_to_markdown(series, family=family, header=header) |
| 702 | return _category_chart_to_markdown(chart, series, header) |
| 703 | |
| 704 | |
| 705 | def _chart_reference_id(element: object) -> str | None: |
| 706 | """Return the first chart relationship id carried by an OOXML shape subtree.""" |
| 707 | for descendant in element.iter(): |
| 708 | if descendant.tag.rsplit("}", 1)[-1] != "chart": |
| 709 | continue |
| 710 | relationship_id = descendant.get(f"{{{RELATIONSHIP_NS}}}id") |
| 711 | if relationship_id: |
| 712 | return relationship_id |
| 713 | return None |
| 714 | |
| 715 | |
| 716 | def _unexposed_chartex_markdown( |
| 717 | slide: object, |
| 718 | emitted_relationship_ids: set[str], |
| 719 | ) -> list[str]: |
| 720 | """Report ChartEx objects omitted from ``slide.shapes`` by python-pptx.""" |
| 721 | blocks: list[str] = [] |
| 722 | seen_relationship_ids: set[str] = set() |
| 723 | slide_element = getattr(slide, "element", None) |
| 724 | if slide_element is None: |
| 725 | slide_element = getattr(slide, "_element", None) |
| 726 | if slide_element is None: |
| 727 | return blocks |
| 728 | for graphic_data in slide_element.iter(f"{{{DRAWINGML_NS}}}graphicData"): |
| 729 | if graphic_data.get("uri") != CHARTEX_URI: |
| 730 | continue |
| 731 | relationship_id = _chart_reference_id(graphic_data) |
| 732 | if relationship_id and ( |
| 733 | relationship_id in emitted_relationship_ids |
| 734 | or relationship_id in seen_relationship_ids |
| 735 | ): |
| 736 | continue |
| 737 | if relationship_id: |
| 738 | seen_relationship_ids.add(relationship_id) |
| 739 | |
| 740 | name = "ChartEx chart" |
| 741 | current = graphic_data |
| 742 | while current is not None: |
| 743 | name_element = current.find(f".//{{{PRESENTATIONML_NS}}}cNvPr") |
| 744 | if name_element is not None and name_element.get("name"): |
| 745 | name = name_element.get("name") |
| 746 | break |
| 747 | current = current.getparent() if hasattr(current, "getparent") else None |
| 748 | chart_name = normalize_text(str(name)).replace("\n", " ") or "ChartEx chart" |
| 749 | header = f"> [Chart] {chart_name} — ChartEx" |
| 750 | blocks.append(_chart_data_unavailable(header, "unsupported ChartEx data model")) |
| 751 | return blocks |
| 752 | |
| 753 | |
| 754 | def _image_part_for_shape(shape: object) -> object | None: |
| 755 | """Return the first embedded image part referenced by a shape.""" |
| 756 | element = getattr(shape, "element", None) |
| 757 | part = getattr(shape, "part", None) |
| 758 | if element is None or part is None: |
| 759 | return None |
| 760 | |
| 761 | try: |
| 762 | blips = element.xpath(".//a:blip") |
| 763 | except Exception: |
| 764 | return None |
| 765 | |
| 766 | for blip in blips: |
| 767 | rel_id = blip.get(qn("r:embed")) or blip.get(qn("r:link")) |
| 768 | if not rel_id: |
| 769 | continue |
| 770 | try: |
| 771 | return part.related_part(rel_id) |
| 772 | except Exception: |
| 773 | continue |
| 774 | return None |
| 775 | |
| 776 | |
| 777 | def _image_size_from_bytes(blob: bytes) -> tuple[int | None, int | None]: |
| 778 | """Return bitmap dimensions when Pillow can decode the bytes.""" |
| 779 | try: |
| 780 | from PIL import Image |
| 781 | except ImportError: |
| 782 | return None, None |
| 783 | try: |
| 784 | with Image.open(BytesIO(blob)) as img: |
| 785 | return img.width, img.height |
| 786 | except (OSError, ValueError): |
| 787 | return None, None |
| 788 | |
| 789 | |
| 790 | def _shape_emu(shape: object, attr: str) -> int: |
| 791 | value = getattr(shape, attr, 0) or 0 |
| 792 | return int(value) |
| 793 | |
| 794 | |
| 795 | def _shape_occurrence( |
| 796 | shape: object, |
| 797 | slide_index: int, |
| 798 | ) -> dict[str, object]: |
| 799 | """Return slide-specific image placement metadata.""" |
| 800 | display_width_emu = _shape_emu(shape, "width") |
| 801 | display_height_emu = _shape_emu(shape, "height") |
| 802 | display_ratio = ( |
| 803 | display_width_emu / display_height_emu |
| 804 | if display_width_emu > 0 and display_height_emu > 0 |
| 805 | else None |
| 806 | ) |
| 807 | return { |
| 808 | "slide_index": slide_index, |
| 809 | "shape_name": str(getattr(shape, "name", "")), |
| 810 | "display_left_emu": _shape_emu(shape, "left"), |
| 811 | "display_top_emu": _shape_emu(shape, "top"), |
| 812 | "display_width_emu": display_width_emu, |
| 813 | "display_height_emu": display_height_emu, |
| 814 | "display_width_in": round(display_width_emu / EMU_PER_INCH, 4) if display_width_emu else None, |
| 815 | "display_height_in": round(display_height_emu / EMU_PER_INCH, 4) if display_height_emu else None, |
| 816 | "display_ratio": round(display_ratio, 6) if display_ratio else None, |
| 817 | } |
| 818 | |
| 819 | |
| 820 | def _update_manifest_usage(entry: dict[str, object]) -> None: |
| 821 | """Refresh aggregate fields after adding an occurrence.""" |
| 822 | occurrences = entry.get("occurrences") |
| 823 | if not isinstance(occurrences, list): |
| 824 | occurrences = [] |
| 825 | entry["usage_count"] = len(occurrences) |
| 826 | ratios = sorted({ |
| 827 | occurrence.get("display_ratio") |
| 828 | for occurrence in occurrences |
| 829 | if isinstance(occurrence, dict) |
| 830 | and isinstance(occurrence.get("display_ratio"), (int, float)) |
| 831 | }) |
| 832 | if ratios: |
| 833 | entry["display_ratio_variants"] = ratios |
| 834 | if entry.get("display_ratio") is None: |
| 835 | entry["display_ratio"] = ratios[0] |
| 836 | |
| 837 | |
| 838 | def _manifest_entry( |
| 839 | *, |
| 840 | index: int, |
| 841 | filename: str, |
| 842 | image_part: object, |
| 843 | ext: str, |
| 844 | blob: bytes, |
| 845 | occurrence: dict[str, object], |
| 846 | ) -> dict[str, object]: |
| 847 | """Build image_manifest.json metadata for one unique PowerPoint media part.""" |
| 848 | pixel_width, pixel_height = _image_size_from_bytes(blob) |
| 849 | pixel_ratio = ( |
| 850 | pixel_width / pixel_height |
| 851 | if pixel_width and pixel_height |
| 852 | else None |
| 853 | ) |
| 854 | is_office_vector = ext in OFFICE_VECTOR_EXTENSIONS |
| 855 | partname = str(getattr(image_part, "partname", "")) |
| 856 | content_type = str(getattr(image_part, "content_type", "")) |
| 857 | |
| 858 | entry: dict[str, object] = { |
| 859 | "index": index, |
| 860 | "filename": filename, |
| 861 | "original_filename": filename, |
| 862 | "asset_kind": "office_vector" if is_office_vector else "bitmap", |
| 863 | "svg_renderable": not is_office_vector, |
| 864 | "pptx_native_supported": True, |
| 865 | "source_kind": "pptx_picture", |
| 866 | "source_ext": f".{ext}", |
| 867 | "source_target": partname.lstrip("/"), |
| 868 | "content_type": content_type, |
| 869 | "display_left_emu": occurrence.get("display_left_emu"), |
| 870 | "display_top_emu": occurrence.get("display_top_emu"), |
| 871 | "display_width_emu": occurrence.get("display_width_emu"), |
| 872 | "display_height_emu": occurrence.get("display_height_emu"), |
| 873 | "display_width_in": occurrence.get("display_width_in"), |
| 874 | "display_height_in": occurrence.get("display_height_in"), |
| 875 | "display_ratio": occurrence.get("display_ratio"), |
| 876 | "pixel_width": pixel_width, |
| 877 | "pixel_height": pixel_height, |
| 878 | "pixel_ratio": round(pixel_ratio, 6) if pixel_ratio else None, |
| 879 | "occurrences": [occurrence], |
| 880 | } |
| 881 | if entry["display_ratio"] is None and pixel_ratio: |
| 882 | entry["display_ratio"] = round(pixel_ratio, 6) |
| 883 | _update_manifest_usage(entry) |
| 884 | return entry |
| 885 | |
| 886 | |
| 887 | def _asset_cache_key(image_part: object, blob: bytes) -> str: |
| 888 | """Return a stable key for deduplicating repeated PPTX media references.""" |
| 889 | partname = str(getattr(image_part, "partname", "")) |
| 890 | if partname: |
| 891 | return partname |
| 892 | return hashlib.sha256(blob).hexdigest() |
| 893 | |
| 894 | |
| 895 | def _asset_filename( |
| 896 | image_part: object, |
| 897 | ext: str, |
| 898 | asset_index: int, |
| 899 | used_filenames: set[str], |
| 900 | ) -> str: |
| 901 | """Return a unique asset filename, preferring the PPTX media basename.""" |
| 902 | partname = str(getattr(image_part, "partname", "")) |
| 903 | base = sanitize_filename(Path(partname).name) if partname else f"image_{asset_index:03d}.{ext}" |
| 904 | if "." not in base: |
| 905 | base = f"{base}.{ext}" |
| 906 | if base not in used_filenames: |
| 907 | used_filenames.add(base) |
| 908 | return base |
| 909 | |
| 910 | path = Path(base) |
| 911 | stem = path.stem |
| 912 | suffix = path.suffix or f".{ext}" |
| 913 | counter = 2 |
| 914 | while True: |
| 915 | candidate = f"{stem}_{counter}{suffix}" |
| 916 | if candidate not in used_filenames: |
| 917 | used_filenames.add(candidate) |
| 918 | return candidate |
| 919 | counter += 1 |
| 920 | |
| 921 | |
| 922 | def save_picture( |
| 923 | shape: object, |
| 924 | asset_dir: Path, |
| 925 | slide_index: int, |
| 926 | asset_index: int, |
| 927 | asset_cache: dict[str, SavedPicture], |
| 928 | used_filenames: set[str], |
| 929 | ) -> SavedPicture | None: |
| 930 | """Persist a shape image to the output asset directory.""" |
| 931 | image_part = _image_part_for_shape(shape) |
| 932 | if image_part is None: |
| 933 | return None |
| 934 | |
| 935 | content_type = getattr(image_part, "content_type", None) |
| 936 | part_ext = getattr(getattr(image_part, "partname", None), "ext", None) |
| 937 | ext = normalize_ext(part_ext, content_type) |
| 938 | blob = bytes(getattr(image_part, "blob", b"")) |
| 939 | if not blob: |
| 940 | return None |
| 941 | |
| 942 | occurrence = _shape_occurrence(shape, slide_index) |
| 943 | cache_key = _asset_cache_key(image_part, blob) |
| 944 | cached = asset_cache.get(cache_key) |
| 945 | if cached is not None: |
| 946 | occurrences = cached.manifest_entry.setdefault("occurrences", []) |
| 947 | if isinstance(occurrences, list): |
| 948 | occurrences.append(occurrence) |
| 949 | _update_manifest_usage(cached.manifest_entry) |
| 950 | return SavedPicture( |
| 951 | filename=cached.filename, |
| 952 | manifest_entry=cached.manifest_entry, |
| 953 | is_new_asset=False, |
| 954 | ) |
| 955 | |
| 956 | filename = _asset_filename(image_part, ext, asset_index, used_filenames) |
| 957 | output_path = asset_dir / filename |
| 958 | output_path.write_bytes(blob) |
| 959 | saved = SavedPicture( |
| 960 | filename=filename, |
| 961 | manifest_entry=_manifest_entry( |
| 962 | index=asset_index, |
| 963 | filename=filename, |
| 964 | image_part=image_part, |
| 965 | ext=ext, |
| 966 | blob=blob, |
| 967 | occurrence=occurrence, |
| 968 | ), |
| 969 | is_new_asset=True, |
| 970 | ) |
| 971 | asset_cache[cache_key] = saved |
| 972 | return saved |
| 973 | |
| 974 | |
| 975 | def _reset_generated_asset_dir(asset_dir: Path) -> None: |
| 976 | """Remove a previously generated asset directory.""" |
| 977 | if not asset_dir.exists(): |
| 978 | return |
| 979 | if not (asset_dir / "image_manifest.json").is_file(): |
| 980 | for path in asset_dir.iterdir(): |
| 981 | if path.is_file() and LEGACY_GENERATED_IMAGE_RE.match(path.name): |
| 982 | path.unlink() |
| 983 | return |
| 984 | shutil.rmtree(asset_dir) |
| 985 | |
| 986 | |
| 987 | def extract_notes(slide: object) -> str: |
| 988 | """Extract speaker notes text from a slide, if available.""" |
| 989 | try: |
| 990 | notes_slide = slide.notes_slide |
| 991 | except Exception: |
| 992 | return "" |
| 993 | |
| 994 | blocks = [] |
| 995 | for item in iter_leaf_shapes(notes_slide.shapes): |
| 996 | shape = item.shape |
| 997 | if not getattr(shape, "has_text_frame", False): |
| 998 | continue |
| 999 | text = text_frame_to_markdown(shape.text_frame, shape) |
| 1000 | if text: |
| 1001 | blocks.append(text) |
| 1002 | |
| 1003 | return "\n\n".join(blocks).strip() |
| 1004 | |
| 1005 | |
| 1006 | def convert_presentation_to_markdown( |
| 1007 | input_path: str, |
| 1008 | output_path: str | None = None, |
| 1009 | ) -> str: |
| 1010 | """Convert a supported PowerPoint file to Markdown.""" |
| 1011 | input_file = Path(input_path) |
| 1012 | if not input_file.exists(): |
| 1013 | print(f"[ERROR] File not found: {input_path}") |
| 1014 | return "" |
| 1015 | |
| 1016 | suffix = input_file.suffix.lower() |
| 1017 | if suffix not in SUPPORTED_FORMATS: |
| 1018 | supported = ", ".join(sorted(SUPPORTED_FORMATS.keys())) |
| 1019 | print(f"[ERROR] Unsupported format: {suffix}") |
| 1020 | print(f" Supported: {supported}") |
| 1021 | print(" Legacy .ppt files should be resaved as .pptx or exported to PDF first.") |
| 1022 | return "" |
| 1023 | |
| 1024 | print(f"[INFO] Converting {SUPPORTED_FORMATS[suffix]}: {input_file.name}") |
| 1025 | |
| 1026 | if output_path: |
| 1027 | out_file = Path(output_path) |
| 1028 | else: |
| 1029 | out_file = input_file.with_suffix(".md") |
| 1030 | |
| 1031 | out_file.parent.mkdir(parents=True, exist_ok=True) |
| 1032 | asset_dir = out_file.parent / f"{out_file.stem}_files" |
| 1033 | _reset_generated_asset_dir(asset_dir) |
| 1034 | |
| 1035 | presentation = Presentation(str(input_file)) |
| 1036 | conversion_warnings: list[str] = [] |
| 1037 | diagrams_by_slide: dict[int, list[dict[str, object]]] = {} |
| 1038 | diagram_scan_failures: dict[int, str] = {} |
| 1039 | try: |
| 1040 | with zipfile.ZipFile(input_file) as package: |
| 1041 | for slide_index, slide in enumerate(presentation.slides, 1): |
| 1042 | slide_part = str(slide.part.partname).lstrip("/") |
| 1043 | try: |
| 1044 | diagrams = read_smartart_diagrams(package, slide_part, slide_index) |
| 1045 | except (OSError, RuntimeError, zipfile.BadZipFile, ET.ParseError) as exc: |
| 1046 | diagrams_by_slide[slide_index] = [] |
| 1047 | diagram_scan_failures[slide_index] = str(exc) |
| 1048 | conversion_warnings.append( |
| 1049 | f"Slide {slide_index}: SmartArt scan failed: {exc}" |
| 1050 | ) |
| 1051 | continue |
| 1052 | diagrams_by_slide[slide_index] = diagrams |
| 1053 | for diagram in diagrams: |
| 1054 | issues = [str(item) for item in diagram.get("warnings", []) if item] |
| 1055 | if diagram.get("status") != "ok": |
| 1056 | issues.insert(0, f"status={diagram.get('status')}") |
| 1057 | if not issues: |
| 1058 | continue |
| 1059 | conversion_warnings.append( |
| 1060 | f"Slide {slide_index}, {diagram.get('shape_name') or diagram.get('diagram_id')}: " |
| 1061 | f"SmartArt content {'; '.join(issues)}" |
| 1062 | ) |
| 1063 | except (OSError, RuntimeError, zipfile.BadZipFile, ET.ParseError) as exc: |
| 1064 | conversion_warnings.append(f"SmartArt package scan failed: {exc}") |
| 1065 | for slide_index in range(1, len(presentation.slides) + 1): |
| 1066 | diagrams_by_slide.setdefault(slide_index, []) |
| 1067 | diagram_scan_failures.setdefault(slide_index, str(exc)) |
| 1068 | |
| 1069 | lines = [ |
| 1070 | f"# {input_file.stem}", |
| 1071 | "", |
| 1072 | f"- Source: `{input_file.name}`", |
| 1073 | f"- Total slides: {len(presentation.slides)}", |
| 1074 | "", |
| 1075 | ] |
| 1076 | |
| 1077 | image_count = 0 |
| 1078 | image_ref_count = 0 |
| 1079 | asset_dir_used = False |
| 1080 | image_manifest: list[dict[str, object]] = [] |
| 1081 | asset_cache: dict[str, SavedPicture] = {} |
| 1082 | used_filenames: set[str] = set() |
| 1083 | |
| 1084 | for slide_index, slide in enumerate(presentation.slides, 1): |
| 1085 | lines.append(f"## Slide {slide_index}") |
| 1086 | lines.append("") |
| 1087 | |
| 1088 | blocks = [] |
| 1089 | slide_diagrams = diagrams_by_slide.get(slide_index, []) |
| 1090 | diagrams_by_shape_id = { |
| 1091 | str(diagram.get("shape_id")): diagram |
| 1092 | for diagram in slide_diagrams |
| 1093 | if diagram.get("shape_id") is not None |
| 1094 | } |
| 1095 | emitted_diagram_ids: set[str] = set() |
| 1096 | emitted_chart_relationship_ids: set[str] = set() |
| 1097 | for item in iter_leaf_shapes(slide.shapes): |
| 1098 | shape = item.shape |
| 1099 | |
| 1100 | if getattr(shape, "has_table", False): |
| 1101 | table_md = table_to_markdown(shape.table) |
| 1102 | if table_md: |
| 1103 | blocks.append(table_md) |
| 1104 | continue |
| 1105 | |
| 1106 | shape_id = str(getattr(shape, "shape_id", "")) |
| 1107 | diagram = diagrams_by_shape_id.get(shape_id) |
| 1108 | if diagram is not None: |
| 1109 | blocks.append(smartart_to_markdown(diagram)) |
| 1110 | emitted_diagram_ids.add(str(diagram.get("diagram_id"))) |
| 1111 | continue |
| 1112 | |
| 1113 | is_picture_shape = shape.shape_type in { |
| 1114 | MSO_SHAPE_TYPE.PICTURE, |
| 1115 | MSO_SHAPE_TYPE.LINKED_PICTURE, |
| 1116 | } |
| 1117 | has_shape_image = is_picture_shape or _image_part_for_shape(shape) is not None |
| 1118 | if has_shape_image: |
| 1119 | image_ref_count += 1 |
| 1120 | next_image_index = image_count + 1 |
| 1121 | asset_dir.mkdir(parents=True, exist_ok=True) |
| 1122 | saved_picture = save_picture( |
| 1123 | shape, |
| 1124 | asset_dir, |
| 1125 | slide_index, |
| 1126 | next_image_index, |
| 1127 | asset_cache, |
| 1128 | used_filenames, |
| 1129 | ) |
| 1130 | if saved_picture is None: |
| 1131 | if is_picture_shape: |
| 1132 | blocks.append(f"> [Image] {getattr(shape, 'name', 'Picture')}") |
| 1133 | continue |
| 1134 | else: |
| 1135 | if saved_picture.is_new_asset: |
| 1136 | image_count = next_image_index |
| 1137 | image_manifest.append(saved_picture.manifest_entry) |
| 1138 | asset_dir_used = True |
| 1139 | blocks.append( |
| 1140 | f"![Slide {slide_index} Image {image_ref_count}]" |
| 1141 | f"({asset_dir.name}/{saved_picture.filename})" |
| 1142 | ) |
| 1143 | if is_picture_shape: |
| 1144 | continue |
| 1145 | |
| 1146 | if getattr(shape, "has_text_frame", False): |
| 1147 | text_md = text_frame_to_markdown(shape.text_frame, shape) |
| 1148 | if text_md: |
| 1149 | blocks.append(text_md) |
| 1150 | continue |
| 1151 | |
| 1152 | if getattr(shape, "has_chart", False): |
| 1153 | shape_element = getattr(shape, "element", None) |
| 1154 | if shape_element is None: |
| 1155 | shape_element = getattr(shape, "_element", None) |
| 1156 | relationship_id = ( |
| 1157 | _chart_reference_id(shape_element) |
| 1158 | if shape_element is not None |
| 1159 | else None |
| 1160 | ) |
| 1161 | if relationship_id: |
| 1162 | emitted_chart_relationship_ids.add(relationship_id) |
| 1163 | try: |
| 1164 | blocks.append(chart_to_markdown(shape.chart, getattr(shape, "name", "Chart"))) |
| 1165 | except (ValueError, TypeError, AttributeError, KeyError) as exc: |
| 1166 | raw_name = getattr(shape, "name", "Chart") |
| 1167 | name = normalize_text("" if raw_name is None else str(raw_name)) or "Chart" |
| 1168 | blocks.append( |
| 1169 | _chart_data_unavailable( |
| 1170 | f"> [Chart] {name}", |
| 1171 | f"chart read failed ({type(exc).__name__})", |
| 1172 | ) |
| 1173 | ) |
| 1174 | |
| 1175 | blocks.extend( |
| 1176 | _unexposed_chartex_markdown(slide, emitted_chart_relationship_ids) |
| 1177 | ) |
| 1178 | |
| 1179 | for diagram in slide_diagrams: |
| 1180 | if str(diagram.get("diagram_id")) in emitted_diagram_ids: |
| 1181 | continue |
| 1182 | blocks.append(smartart_to_markdown(diagram)) |
| 1183 | if slide_index in diagram_scan_failures: |
| 1184 | blocks.append( |
| 1185 | f"> [SmartArt scan unavailable: {diagram_scan_failures[slide_index]}]" |
| 1186 | ) |
| 1187 | |
| 1188 | if blocks: |
| 1189 | lines.append("\n\n".join(blocks)) |
| 1190 | lines.append("") |
| 1191 | else: |
| 1192 | lines.append("_No extractable text content._") |
| 1193 | lines.append("") |
| 1194 | |
| 1195 | notes_md = extract_notes(slide) |
| 1196 | if notes_md: |
| 1197 | lines.append("### Speaker Notes") |
| 1198 | lines.append("") |
| 1199 | lines.append(notes_md) |
| 1200 | lines.append("") |
| 1201 | |
| 1202 | markdown_content = "\n".join(lines).strip() + "\n" |
| 1203 | out_file.write_text(markdown_content, encoding="utf-8") |
| 1204 | if image_manifest: |
| 1205 | (asset_dir / "image_manifest.json").write_text( |
| 1206 | json.dumps(image_manifest, ensure_ascii=False, indent=2) + "\n", |
| 1207 | encoding="utf-8", |
| 1208 | ) |
| 1209 | profile_path = write_conversion_profile_best_effort( |
| 1210 | input_path=str(input_file), |
| 1211 | markdown_path=out_file, |
| 1212 | converter="ppt_to_md.py", |
| 1213 | conversion_type=suffix.lstrip("."), |
| 1214 | asset_dir=asset_dir, |
| 1215 | warnings=conversion_warnings, |
| 1216 | ) |
| 1217 | |
| 1218 | print(f"[OK] Saved Markdown to: {out_file}") |
| 1219 | if profile_path: |
| 1220 | print(f" Wrote conversion profile -> {profile_path}") |
| 1221 | if asset_dir_used: |
| 1222 | media_files = [ |
| 1223 | path for path in asset_dir.iterdir() |
| 1224 | if path.is_file() and path.name != "image_manifest.json" |
| 1225 | ] |
| 1226 | print(f" Extracted {len(media_files)} image file(s) -> {asset_dir}") |
| 1227 | if image_ref_count != len(media_files): |
| 1228 | print( |
| 1229 | f" Deduplicated {image_ref_count} image reference(s) " |
| 1230 | f"into {len(media_files)} asset file(s)" |
| 1231 | ) |
| 1232 | print(f" Wrote image manifest -> {asset_dir / 'image_manifest.json'}") |
| 1233 | |
| 1234 | return markdown_content |
| 1235 | |
| 1236 | |
| 1237 | def main() -> int: |
| 1238 | """Run the CLI entry point.""" |
| 1239 | parser = argparse.ArgumentParser( |
| 1240 | description="Convert PowerPoint files to Markdown", |
| 1241 | formatter_class=argparse.RawDescriptionHelpFormatter, |
| 1242 | epilog=""" |
| 1243 | Examples: |
| 1244 | python ppt_to_md.py slides.pptx |
| 1245 | python ppt_to_md.py slides.pptx appendix.pptx |
| 1246 | python ppt_to_md.py ./decks -o ./markdown |
| 1247 | python ppt_to_md.py slides.pptx -o output.md |
| 1248 | python ppt_to_md.py deck.ppsx -o notes/deck.md |
| 1249 | |
| 1250 | Supported formats: |
| 1251 | .pptx .pptm .ppsx .ppsm .potx .potm |
| 1252 | |
| 1253 | Legacy .ppt is not parsed directly. Resave it as .pptx or export it to PDF first. |
| 1254 | """, |
| 1255 | ) |
| 1256 | parser.add_argument("inputs", nargs="+", help="Input PowerPoint file(s) or directories") |
| 1257 | parser.add_argument( |
| 1258 | "-o", |
| 1259 | "--output", |
| 1260 | help="Output Markdown file for one input, or output directory for multiple inputs/directories", |
| 1261 | ) |
| 1262 | |
| 1263 | args = parser.parse_args() |
| 1264 | |
| 1265 | return run_path_batch( |
| 1266 | args.inputs, |
| 1267 | set(SUPPORTED_FORMATS), |
| 1268 | args.output, |
| 1269 | lambda source, output: bool(convert_presentation_to_markdown(str(source), str(output))), |
| 1270 | ) |
| 1271 | |
| 1272 | |
| 1273 | if __name__ == "__main__": |
| 1274 | raise SystemExit(main()) |
| 1275 |