修复 Release 资产名被 ASCII 净化导致课程包丢失

GitHub Release 资产名只允许 [A-Za-z0-9._-]，中文课程名上传后被整体
抹成分隔符，纯中文课名全部坍缩成同一个名字并被 --clobber 互相覆盖，
100+ 门课最终只剩 20 个乱名资产。

改为用 DeepSeek 把课程名翻译成英文 slug 作为资产名：
- 新增 Resolve English slugs 步骤，映射持久化为 Release 资产
  course-slugs.json，已有课程永不重译、名字永久稳定，纯 ASCII 课名
  跳过翻译，撞名用课程路径哈希兜底保证唯一
- 生成 manifest.json（英文 slug -> 中文课名/路径）供前端显示中文
- 全量重建时清理历史遗留的乱名资产
- 上传 zip + manifest.json + course-slugs.json

需在仓库 Secrets 配置 DEEPSEEK_API_KEY。

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: Crescentm

.github/workflows/release-updated-courses.yml

@@ -153,71 +153,171 @@ jobs:
           echo "has_changes=true" >> "$GITHUB_OUTPUT"
           fi
 
-      - name: Package changed courses
+          if [ "$EVENT_NAME" = "workflow_dispatch" ] && [ "$FULL_REBUILD" = "true" ]; then
+          echo "full_rebuild=true" >> "$GITHUB_OUTPUT"
+          else
+          echo "full_rebuild=false" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Resolve English slugs
+        id: slugs
         if: steps.detect.outputs.has_changes == 'true'
         shell: bash
+        env:
+          GH_TOKEN: ${{ github.token }}
+          DEEPSEEK_API_KEY: ${{ secrets.DEEPSEEK_API_KEY }}
         run: |
           set -euo pipefail
 
+          # 从 Release 下载已有的 slug 映射（首次不存在则为空）。
+          # 映射存为 Release 资产 course-slugs.json，纯 ASCII，不受资产名净化影响。
+          rm -f course-slugs.json
+          if gh release view "$RELEASE_TAG" >/dev/null 2>&1; then
+            gh release download "$RELEASE_TAG" --pattern course-slugs.json --dir . --clobber 2>/dev/null || true
+          fi
+          [ -f course-slugs.json ] || echo '{}' > course-slugs.json
+
           python3 <<'PY'
-          from pathlib import Path
+          import hashlib
+          import json
+          import os
           import re
-          import zipfile
-
-          dist = Path("dist")
-          dist.mkdir(exist_ok=True)
-
-          exclude_files = {
-              ".DS_Store",
-              "Thumbs.db",
-          }
-
-          exclude_dirs = {
-              ".git",
-              "__MACOSX",
-          }
-
-          def safe_name(name: str) -> str:
-              # 所有空白字符替换成下划线，包括普通空格
-              name = re.sub(r"\s+", "_", name)
+          import sys
+          import urllib.request
+          from pathlib import Path
 
-              # gh release upload 会把 file#label 里的半角 # 当作资产显示名分隔符，
-              # 导致 "数据结构（C#）.zip" 被截断成 "数据结构（C"。
-              # 这里用全角井号 ＃ 替换：gh 不会对它做分隔，且能完整保留语义，
-              # 避免把 # 直接换成 _ 而和 "数据结构（C）" 等课程产生重名。
-              name = name.replace("#", "＃")
+          MAP_PATH = Path("course-slugs.json")          # 持久映射：课程路径 -> 英文 slug
+          RUN_MAP_PATH = Path("slug_map.json")          # 本次需打包课程的子集
 
-              # release asset 文件名中不适合出现的字符也替换掉
-              name = re.sub(r'[\\/:*?"<>|]+', "_", name)
+          api_key = os.environ.get("DEEPSEEK_API_KEY", "").strip()
 
-              # 合并连续下划线
-              name = re.sub(r"_+", "_", name)
+          try:
+              slug_map = json.loads(MAP_PATH.read_text(encoding="utf-8")) or {}
+          except Exception:
+              slug_map = {}
 
-              return name.strip("._") or "unnamed"
+          used = set(slug_map.values())
 
           courses = [
               line.strip()
               for line in Path("changed_courses.txt").read_text(encoding="utf-8").splitlines()
               if line.strip()
           ]
 
-          created = 0
+          def sanitize_slug(text: str) -> str:
+              # 只保留 GitHub 资产名允许的字符 [A-Za-z0-9._-]，其余折叠成下划线
+              text = re.sub(r"[^A-Za-z0-9]+", "_", text.strip())
+              text = re.sub(r"_+", "_", text).strip("._")
+              return text[:80]
+
+          def short_hash(text: str, n: int = 6) -> str:
+              return hashlib.sha1(text.encode("utf-8")).hexdigest()[:n]
+
+          def is_ascii_simple(text: str) -> bool:
+              # 课名本就是 ASCII（如 Matlab）时直接用，省一次 API
+              return bool(re.fullmatch(r"[A-Za-z0-9 ._\-+()]+", text))
+
+          def translate(name: str) -> str:
+              if not api_key:
+                  return ""
+              payload = json.dumps({
+                  "model": "deepseek-chat",
+                  "messages": [
+                      {"role": "system", "content": "You translate Chinese university course names into concise English. Reply with ONLY the English course name in Title Case. No quotes, no explanation, no trailing punctuation."},
+                      {"role": "user", "content": name},
+                  ],
+                  "temperature": 0,
+                  "stream": False,
+              }).encode("utf-8")
+              req = urllib.request.Request(
+                  "https://api.deepseek.com/chat/completions",
+                  data=payload,
+                  headers={
+                      "Authorization": f"Bearer {api_key}",
+                      "Content-Type": "application/json",
+                  },
+                  method="POST",
+              )
+              try:
+                  with urllib.request.urlopen(req, timeout=60) as resp:
+                      data = json.loads(resp.read().decode("utf-8"))
+                  return data["choices"][0]["message"]["content"].strip()
+              except Exception as exc:  # 翻译失败不阻断整个发布，回退到哈希名
+                  print(f"translate failed for {name!r}: {exc}", file=sys.stderr)
+                  return ""
+
+          def assign_slug(course_path: str) -> str:
+              name = course_path.split("/", 1)[1] if "/" in course_path else course_path
+
+              if is_ascii_simple(name):
+                  base = sanitize_slug(name)
+              else:
+                  base = sanitize_slug(translate(name))
+
+              if not base:
+                  base = "course_" + short_hash(course_path, 8)
+
+              candidate = base
+              # 不同课程译名/净化后撞车时，用课程路径哈希保证唯一，绝不再 clobber
+              while candidate in used:
+                  candidate = f"{base}_{short_hash(course_path + candidate)}"
+              return candidate
+
+          run_map = {}
+          changed = False
 
           for course in courses:
+              if course in slug_map:
+                  run_map[course] = slug_map[course]
+                  continue
+              slug = assign_slug(course)
+              slug_map[course] = slug
+              used.add(slug)
+              run_map[course] = slug
+              changed = True
+              print(f"resolved {course} -> {slug}")
+
+          RUN_MAP_PATH.write_text(
+              json.dumps(run_map, ensure_ascii=False, indent=2),
+              encoding="utf-8",
+          )
+          MAP_PATH.write_text(
+              json.dumps(slug_map, ensure_ascii=False, indent=2, sort_keys=True) + "\n",
+              encoding="utf-8",
+          )
+
+          print(f"this run: {len(run_map)} course(s); map total: {len(slug_map)}; map_changed={changed}")
+          PY
+
+      - name: Package changed courses
+        if: steps.detect.outputs.has_changes == 'true'
+        shell: bash
+        run: |
+          set -euo pipefail
+
+          python3 <<'PY'
+          import json
+          import zipfile
+          from pathlib import Path
+
+          dist = Path("dist")
+          dist.mkdir(exist_ok=True)
+
+          exclude_files = {".DS_Store", "Thumbs.db"}
+          exclude_dirs = {".git", "__MACOSX"}
+
+          run_map = json.loads(Path("slug_map.json").read_text(encoding="utf-8"))
+
+          created = 0
+
+          for course, slug in run_map.items():
               course_path = Path(course)
 
               if not course_path.is_dir():
                   print(f"skip deleted course: {course}")
                   continue
 
-              if len(course_path.parts) < 2:
-                  print(f"skip invalid course path: {course}")
-                  continue
-
-              category = course_path.parts[0]
-              course_name = course_path.parts[1]
-
-              zip_path = dist / f"{safe_name(category)}-{safe_name(course_name)}.zip"
+              zip_path = dist / f"{slug}.zip"
 
               if zip_path.exists():
                   zip_path.unlink()
@@ -232,14 +332,11 @@ jobs:
                   for item in sorted(course_path.rglob("*")):
                       if not item.is_file():
                           continue
-
                       if item.name in exclude_files:
                           continue
-
                       if any(part in exclude_dirs for part in item.parts):
                           continue
-
-                      # zip 内部保留原始路径，只修改外部 zip 文件名
+                      # zip 内部保留原始中文路径，仅外部资产名用英文 slug
                       arcname = item.relative_to(Path("."))
                       zf.write(item, arcname.as_posix())
 
@@ -257,6 +354,38 @@ jobs:
               print(file)
           PY
 
+      - name: Build manifest
+        if: steps.detect.outputs.has_changes == 'true'
+        shell: bash
+        run: |
+          set -euo pipefail
+
+          python3 <<'PY'
+          import json
+          from pathlib import Path
+
+          dist = Path("dist")
+          dist.mkdir(exist_ok=True)
+
+          slug_map = json.loads(Path("course-slugs.json").read_text(encoding="utf-8"))
+
+          courses = []
+          for path, slug in sorted(slug_map.items()):
+              parts = path.split("/")
+              courses.append({
+                  "asset": f"{slug}.zip",
+                  "path": path,
+                  "category": parts[0] if parts else "",
+                  "name": parts[1] if len(parts) > 1 else path,
+              })
+
+          (dist / "manifest.json").write_text(
+              json.dumps({"courses": courses}, ensure_ascii=False, indent=2),
+              encoding="utf-8",
+          )
+          print(f"manifest entries: {len(courses)}")
+          PY
+
       - name: Check archive size
         if: steps.detect.outputs.has_changes == 'true'
         shell: bash
@@ -296,10 +425,29 @@ jobs:
           else
           gh release create "$RELEASE_TAG" \
             --title "课程资料压缩包" \
-            --notes "自动生成课程压缩包。首次可全量生成，后续只更新发生变化的课程。"
+            --notes "自动生成课程压缩包。资产名为英文（见 manifest.json 对应中文课名）。首次可全量生成，后续只更新发生变化的课程。"
           fi
 
-      - name: Upload changed course archives
+      - name: Reset stale assets on full rebuild
+        if: steps.detect.outputs.has_changes == 'true' && steps.detect.outputs.full_rebuild == 'true'
+        shell: bash
+        env:
+          GH_TOKEN: ${{ github.token }}
+        run: |
+          set -euo pipefail
+
+          echo "Full rebuild: clearing existing release assets (keeping map/manifest)."
+          gh release view "$RELEASE_TAG" --json assets --jq '.assets[].name' 2>/dev/null \
+          | while IFS= read -r asset; do
+              [ -z "$asset" ] && continue
+              case "$asset" in
+                course-slugs.json|manifest.json) continue ;;
+              esac
+              echo "deleting stale asset: $asset"
+              gh release delete-asset "$RELEASE_TAG" "$asset" -y || true
+            done
+
+      - name: Upload archives and metadata
         if: steps.detect.outputs.has_changes == 'true'
         shell: bash
         env:
@@ -308,38 +456,29 @@ jobs:
           set -euo pipefail
 
           python3 <<'PY'
-          from pathlib import Path
           import os
           import subprocess
           import time
+          from pathlib import Path
 
           release_tag = os.environ["RELEASE_TAG"]
-          files = sorted(Path("dist").glob("*.zip"))
 
-          if not files:
-              print("No archives to upload.")
-              raise SystemExit(0)
+          files = sorted(Path("dist").glob("*.zip"))
+          files.append(Path("dist/manifest.json"))
+          files.append(Path("course-slugs.json"))
 
           uploaded = 0
-
           for file in files:
-              file_str = str(file)
-              print(f"Uploading: {file_str}")
-
+              if not file.exists():
+                  print(f"skip missing: {file}")
+                  continue
+              print(f"Uploading: {file}")
               subprocess.run(
-                  [
-                      "gh",
-                      "release",
-                      "upload",
-                      release_tag,
-                      file_str,
-                      "--clobber",
-                  ],
+                  ["gh", "release", "upload", release_tag, str(file), "--clobber"],
                   check=True,
               )
-
               uploaded += 1
               time.sleep(1)
 
-          print(f"Uploaded {uploaded} archive(s).")
+          print(f"Uploaded {uploaded} file(s).")
           PY