Skip to content

AI turns documents or topics into real, native PowerPoint decks—with native shapes, transitions and animations, data-backed charts and tables on demand, audio narration from speaker notes, and support for your own .pptx templates. · by Hugo He

docs:讲稿格式要求恢复严格,拆分工具保留容错说明

hugohe354d0f059bc05
diff --git a/roles/Strategist.md b/roles/Strategist.md
index 12e895d7..e50b137f 100644
--- a/roles/Strategist.md
+++ b/roles/Strategist.md
@@ -190,7 +190,7 @@
1. 演讲总时长:[分钟]
2. 备注风格:正式/口语化/互动式
3. 讲演目的:告知/说服/激励/指导/汇报
-4. 特殊要求:不允许任何#标签行出现在备注文件中。
+4. 特殊要求:拆分后的备注文件中不允许出现 `#` 标题行(`notes/total.md` 作为汇总讲稿必须使用 `#` 标题行)。
---
diff --git a/tools/README.md b/tools/README.md
index 94b46f2f..3f3b4497 100644
--- a/tools/README.md
+++ b/tools/README.md
@@ -690,10 +690,14 @@ python3 tools/total_md_split.py projects/<svg 标题>_ppt169_YYYYMMDD -q
讲稿内容...
```
-- 每个章节以 `# ` 开头的一级标题开始
-- 章节之间用 `---` 分隔
+- 每个章节以 `# ` 开头的一级标题开始(**严格要求**)
+- 标题文本需与对应 SVG 文件名一致(**严格要求**)
+- 章节之间必须用 `---` 分隔(**严格要求**)
- 讲稿内容在标题之后,直到下一个标题或文件结束
+**容错说明(工具侧)**:
+即便未完全符合上述格式,`total_md_split.py` 仍会尝试根据标题/页码/名称相似度进行拆分,但结果不保证正确。请以严格格式为准。
+
**错误处理**:
如果存在 SVG 文件没有对应的讲稿,工具会输出错误信息:
diff --git a/tools/total_md_split.py b/tools/total_md_split.py
index b321e308..6aa8dc05 100644
--- a/tools/total_md_split.py
+++ b/tools/total_md_split.py
@@ -29,6 +29,69 @@ import re
from pathlib import Path
from typing import List, Dict, Tuple, Optional
+HEADING_RE = re.compile(r'^(#{1,6})\s*(.+?)\s*$')
+HR_RE = re.compile(r'^\s*[-*]{3,}\s*$')
+
+
+def normalize_title(title: str) -> str:
+ """Normalize titles for fuzzy matching with SVG stems."""
+ if not title:
+ return ''
+ text = title.strip()
+ # Replace any non-alnum / non-CJK run with underscore
+ text = re.sub(r'[^0-9A-Za-z\u4e00-\u9fff]+', '_', text)
+ text = re.sub(r'_+', '_', text).strip('_')
+ return text.lower()
+
+
+def extract_leading_number(text: str) -> Optional[int]:
+ """Extract leading slide number if present."""
+ if not text:
+ return None
+ m = re.match(r'^(\d{1,3})', text.strip())
+ if not m:
+ return None
+ try:
+ return int(m.group(1))
+ except ValueError:
+ return None
+
+
+def build_match_maps(svg_stems: List[str]) -> Tuple[set, Dict[str, List[str]], Dict[int, List[str]]]:
+ exact = set(svg_stems)
+ norm_map: Dict[str, List[str]] = {}
+ num_map: Dict[int, List[str]] = {}
+ for stem in svg_stems:
+ norm = normalize_title(stem)
+ if norm:
+ norm_map.setdefault(norm, []).append(stem)
+ num = extract_leading_number(stem)
+ if num is not None:
+ num_map.setdefault(num, []).append(stem)
+ return exact, norm_map, num_map
+
+
+def match_title(
+ raw_title: str,
+ exact: set,
+ norm_map: Dict[str, List[str]],
+ num_map: Dict[int, List[str]],
+ svg_stems: Optional[List[str]] = None,
+) -> Optional[str]:
+ if raw_title in exact:
+ return raw_title
+ norm = normalize_title(raw_title)
+ if norm in norm_map and len(norm_map[norm]) == 1:
+ return norm_map[norm][0]
+ num = extract_leading_number(raw_title)
+ if num is not None and num in num_map and len(num_map[num]) == 1:
+ return num_map[num][0]
+ if norm and svg_stems:
+ candidates = [s for s in svg_stems if norm in normalize_title(s)]
+ if len(candidates) == 1:
+ return candidates[0]
+ return None
+
def find_svg_files(project_path: Path) -> List[Path]:
"""
@@ -49,7 +112,7 @@ def find_svg_files(project_path: Path) -> List[Path]:
return sorted(svg_dir.glob('*.svg'))
-def parse_total_md(md_path: Path) -> Dict[str, str]:
+def parse_total_md(md_path: Path, svg_stems: Optional[List[str]] = None, verbose: bool = True) -> Dict[str, str]:
"""
解析 total.md 文件,提取每个一级标题对应的讲稿内容
@@ -70,35 +133,51 @@ def parse_total_md(md_path: Path) -> Dict[str, str]:
print(f"错误: 无法读取文件 {md_path}: {e}")
return {}
- # 按分隔符拆分内容
- sections = re.split(r'\n---\n', content)
-
- notes = {}
-
- for section in sections:
- section = section.strip()
- if not section:
+ svg_stems = svg_stems or []
+ exact, norm_map, num_map = build_match_maps(svg_stems)
+
+ # 按标题解析(支持 # / ## / ###)
+ notes: Dict[str, str] = {}
+ current_key: Optional[str] = None
+ current_lines: List[str] = []
+ unmatched_headings: List[str] = []
+
+ lines = content.splitlines()
+ for line in lines:
+ m = HEADING_RE.match(line)
+ if m:
+ raw_title = m.group(2).strip()
+ matched = match_title(raw_title, exact, norm_map, num_map, svg_stems)
+ if matched:
+ if current_key is not None:
+ text = '\n'.join(current_lines).strip()
+ if current_key in notes and text:
+ notes[current_key] = (notes[current_key].rstrip() + "\n\n" + text).strip()
+ elif current_key not in notes:
+ notes[current_key] = text
+ current_key = matched
+ current_lines = []
+ continue
+ unmatched_headings.append(raw_title)
+
+ if HR_RE.match(line):
continue
-
- # 查找一级标题
- lines = section.split('\n')
- title_line = None
- content_lines = []
-
- for line in lines:
- if line.startswith('# '):
- title_line = line
- elif title_line is not None:
- content_lines.append(line)
-
- if title_line:
- # 提取标题文本(去掉 # 和空格)
- title = re.sub(r'^#+\s*', '', title_line).strip()
- # 提取讲稿内容(去掉空行)
- content = '\n'.join(content_lines).strip()
-
- if title and content:
- notes[title] = content
+ if current_key is not None:
+ current_lines.append(line)
+
+ if current_key is not None:
+ text = '\n'.join(current_lines).strip()
+ if current_key in notes and text:
+ notes[current_key] = (notes[current_key].rstrip() + "\n\n" + text).strip()
+ elif current_key not in notes:
+ notes[current_key] = text
+
+ if verbose and unmatched_headings:
+ print("\n[提示] 发现未匹配的标题(已忽略):")
+ for t in unmatched_headings[:10]:
+ print(f" - {t}")
+ if len(unmatched_headings) > 10:
+ print(f" ... 以及另外 {len(unmatched_headings) - 10} 个")
return notes
@@ -226,7 +305,8 @@ def main():
# 解析 total.md
total_md_path = project_path / 'notes' / 'total.md'
- notes = parse_total_md(total_md_path)
+ svg_stems = [p.stem for p in svg_files]
+ notes = parse_total_md(total_md_path, svg_stems, verbose)
if not notes:
print("错误: 未找到讲稿内容")