数字我 Digital Me：全量入库（大文件走 Git LFS）

tools/qa_woyanmo_lyrics_20shots.py

@@ -0,0 +1,86 @@
+from __future__ import annotations
+
+import json
+import zipfile
+from pathlib import Path
+
+from docx import Document
+from docx.oxml.ns import qn
+
+
+path = Path(r"D:\workspace\digital_me\artifacts\《我为你研墨》完整歌词脚本与20镜分镜提示词（每镜10秒）.docx")
+issues = []
+with zipfile.ZipFile(path) as zf:
+    bad = zf.testzip()
+    media = [n for n in zf.namelist() if n.startswith("word/media/")]
+    if bad:
+        issues.append(f"Corrupt member: {bad}")
+    if media:
+        issues.append(f"Unexpected embedded images: {media}")
+
+doc = Document(path)
+text = "\n".join(p.text for p in doc.paragraphs)
+for table in doc.tables:
+    for row in table.rows:
+        text += "\n" + "\t".join(cell.text for cell in row.cells)
+
+scene_headings = [p.text for p in doc.paragraphs if p.style.name == "Heading 1" and p.text.startswith("分镜 ")]
+if len(scene_headings) != 20:
+    issues.append(f"Expected 20 shot headings, found {len(scene_headings)}")
+if text.count("10秒分镜提示词（可直接复制）") != 20:
+    issues.append("Expected 20 prompt sections")
+if text.count("生成时长严格10秒") != 20:
+    issues.append("Every prompt is not explicitly fixed to 10 seconds")
+
+lyrics = [
+    "你杀了我祖父的那一年", "我还不会写字", "如今我替你 把诏书写得漂亮",
+    "一笔一划 都是你要的锋芒", "掖庭里长大 罪臣的孙女", "我把恨 磨成了墨",
+    "一滴 不许它响", "该恨你 还是该谢你", "你毁了我的家 又给了我笔",
+    "我为你研墨 也为自己活", "仇人的江山 我替她说", "你看不透我 我也看不透我",
+    "这盘棋里 谁是谁的 因果", "掌上起风 婉儿落笔", "你不问我恨不恨 你只看我写得好不好",
+    "我也学会了 把心 藏进批注的小字号", "你老了 手也抖了 我扶你上朝",
+    "这一路 我们谁也没提过 那道诏", "祖父啊 我没替你报仇",
+    "我活成了 比报仇 更长的一句咒", "或者 只是一个 想活下去的人",
+    "墨干了 诏成了", "风 还在掌上",
+]
+missing = [line for line in lyrics if line not in text]
+if missing:
+    issues.append(f"Missing lyric lines: {missing}")
+
+for marker in ("TODO", "TBD", "【统一风格前缀】", "placeholder"):
+    if marker in text:
+        issues.append(f"Unexpected placeholder: {marker}")
+
+tables = []
+for index, table in enumerate(doc.tables, start=1):
+    tblpr = table._tbl.tblPr
+    tblw = tblpr.find(qn("w:tblW"))
+    grid = [int(x.get(qn("w:w"))) for x in table._tbl.tblGrid.findall(qn("w:gridCol"))]
+    expected = sum(grid)
+    if tblw is None or int(tblw.get(qn("w:w"))) != expected or expected != 9360:
+        issues.append(f"Table {index}: invalid total geometry")
+    for row in table.rows:
+        widths = []
+        for cell in row.cells:
+            tcw = cell._tc.get_or_add_tcPr().find(qn("w:tcW"))
+            widths.append(int(tcw.get(qn("w:w"))) if tcw is not None else None)
+        if widths != grid:
+            issues.append(f"Table {index}: cell/grid mismatch")
+            break
+    tables.append({"table": index, "rows": len(table.rows), "cols": len(table.columns), "width_dxa": expected})
+
+result = {
+    "file": str(path),
+    "bytes": path.stat().st_size,
+    "paragraphs": len(doc.paragraphs),
+    "tables": tables,
+    "shot_headings": len(scene_headings),
+    "prompt_sections": text.count("10秒分镜提示词（可直接复制）"),
+    "fixed_10s_prompts": text.count("生成时长严格10秒"),
+    "embedded_images": 0,
+    "missing_lyrics": missing,
+    "issues": issues,
+    "status": "PASS" if not issues else "FAIL",
+}
+print(json.dumps(result, ensure_ascii=False, indent=2))
+raise SystemExit(0 if not issues else 1)