fix(bioflow): 路径登记目录展开+黑名单+自动挂载

- 登记目录自动展开多文件(扩展名白名单过滤,上限 200)
- 移除白名单根校验(用户数据可在任意位置),改敏感路径黑名单
  (/etc /root /proc /var/lib/docker 等前缀 + .env/*.key/id_ed25519 等文件名)
- 登记文件 symlink 进任务 input + server_registered.json 持久化
- 执行容器自动挂载登记路径(-v src:src:ro),跨重启恢复
- 前端 chat.js 兼容 items[] 响应(单文件亦在 items[0])
- 336 用例全绿

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

bioflow_service/app/config.py

@@ -56,6 +56,21 @@ class Settings(BaseSettings):
     max_upload_gb: float = 100.0             # 上传大小上限（GB；超限 413）
     upload_ext_allow: str = ".fasta,.fa,.faa,.fna,.pep,.cds,.gz,.gff,.gff3,.gtf,.txt,.tsv,.csv,.xlsx,.xls,.bed,.vcf,.vcf.gz,.bam,.cram,.sam,.hmm,.nwk,.newick,.treefile,.json,.yaml,.yml,.pdf,.zip,.py,.r,.sh"
 
+    # ---- 服务器路径登记（FR-BIO-007；用户数据可在任意位置）----
+    max_register_files_per_job: int = 200    # 目录展开单任务文件数上限（防误填大目录）
+    # 敏感路径黑名单：命中前缀即 403（白名单根校验已按用户要求移除）
+    sensitive_path_prefixes: tuple[str, ...] = (
+        "/etc", "/root", "/proc", "/sys", "/dev", "/run", "/boot",
+        "/bin", "/sbin", "/lib", "/lib64", "/usr/bin", "/usr/sbin",
+        "/usr/lib", "/usr/lib64", "/usr/local/bin", "/var/lib", "/var/log",
+        "/opt/neo4j", "/var/lib/docker", "/srv/plantai/install",
+    )
+    # 敏感文件名（正则,re.match 整名）；.env 等含密钥配置、密钥文件不可登记
+    sensitive_filename_patterns: tuple[str, ...] = (
+        r"\.env.*", r".*\.key$", r".*\.pem$", r"id_rsa.*", r"id_ed25519.*",
+        r".*\.p12$", r"\.gitconfig.*",
+    )
+
     # ---- 用户自定义扩展分析（并行链路；存储根 + AI 转换 LLM 键 + 修复上限）----
     extensions_dir: str = "bioflow_data/extensions"  # 扩展存储根（{ext_id}/extension.json...）
     extension_max_repair_rounds: int = 20            # AI 修复循环上限（契约 §6 可配）

bioflow_service/app/context.py

@@ -205,6 +205,17 @@ def _record_from_job(job: dict[str, Any], stages: list[dict[str, Any]]) -> Any:
         auto_continue=bool(job.get("auto_continue")),
         archive_id=job.get("archive_id"),
     )
+    # 登记路径恢复（server_registered.json；供执行容器自动挂载）
+    try:
+        import json as _json
+        reg_file = record.job_dir / "server_registered.json"
+        if reg_file.is_file():
+            record.registered_server_paths = (
+                _json.loads(reg_file.read_text(encoding="utf-8"))
+                .get("registered_server_paths") or []
+            )
+    except Exception:  # noqa: BLE001 - 恢复失败不阻断（挂载缺失→阶段报错可查）
+        pass
     for s in stages:
         record._stage_states[s["stage"]] = {
             "status": s["status"],

bioflow_service/app/routers/biojobs.py

@@ -239,22 +239,87 @@ async def upload_file(
 
 
 async def _register_path(record: Any, server_path: str, settings: Any) -> dict[str, Any]:
-    """服务器路径登记（FR-BIO-007）：限 BIOFLOW_DATA_ROOT/任务目录内。"""
+    """服务器路径登记（FR-BIO-007）：单文件或目录展开，任意数据位置（敏感路径黑名单拦截）。
+
+    目录：递归收集目录内可登记文件（扩展名白名单），登记为多条记录。
+    白名单根校验已按用户要求移除，改为敏感路径黑名单（见 settings）。
+    """
     path = Path(server_path).expanduser().resolve()
     if not path.exists():
         return _err("PATH_NOT_EXIST", f"路径不存在: {server_path}", 400)
-    if not path.is_file():
-        return _err("NOT_A_FILE", f"非文件: {server_path}", 400)
-    allowed_roots = [Path(settings.data_root_abs).resolve(), record.job_dir.resolve()]
-    if not any(path.is_relative_to(root) for root in allowed_roots):
-        return _err("PATH_NOT_ALLOWED", f"路径不在白名单根内: {server_path}", 403)
+    if not path.is_file() and not path.is_dir():
+        return _err("NOT_A_FILE", f"非文件或目录: {server_path}", 400)
+    blocked = _blocked_reason(path, settings)
+    if blocked:
+        return _err("PATH_NOT_ALLOWED", f"路径不允许: {server_path}（{blocked}）", 403)
+
+    files: list[Path] = []
+    if path.is_file():
+        files = [path]
+    else:
+        ext_set = settings.upload_ext_set
+        for p in sorted(path.rglob("*")):
+            if p.is_file():
+                if ext_set and p.suffix.lower() not in ext_set:
+                    continue
+                if _blocked_reason(p, settings):
+                    continue
+                files.append(p)
+                if len(files) >= settings.max_register_files_per_job:
+                    break
+        if not files:
+            return _err("VALIDATION",
+                        f"目录内无可登记文件（允许扩展名 {settings.upload_ext_allow[:80]}…）", 400)
+
+    # 登记文件 symlink 进任务 input（CLI 扫 /job/input 发现数据）+ 持久化挂载清单
+    input_dir = record.input_dir
+    input_dir.mkdir(parents=True, exist_ok=True)
+    for p in files:
+        link = input_dir / p.name
+        try:
+            if not link.exists():
+                link.symlink_to(p)
+        except OSError as exc:  # noqa: BLE001
+            logger.warning("登记 symlink 失败（不阻断）: %s -> %s (%s)", p, link, exc)
+    registered = [str(p) for p in files]
+    try:
+        _persist_registered_paths(record, registered)
+    except Exception as exc:  # noqa: BLE001
+        logger.warning("登记路径持久化失败（不阻断登记）: %s", exc)
+
     # BUG-P7-01：路径登记后自动续跑（await_inputs 任务重推；非等待态 no-op）
     ctx: Context = context_mod.context
     await ctx.runner.retry_after_files(record)
-    return {"ok": True, "data": {
-        "file_id": f"{record.id}:{path.name}", "name": path.name,
-        "size": path.stat().st_size, "path": str(path), "source": "server_path",
-    }}
+    items = [
+        {"file_id": f"{record.id}:{p.name}", "name": p.name,
+         "size": p.stat().st_size, "path": str(p), "source": "server_path"}
+        for p in files
+    ]
+    return {"ok": True, "data": {"items": items, "total": len(items)}}
+
+
+def _persist_registered_paths(record: Any, registered: list[str]) -> None:
+    """登记路径落盘：任务目录 server_registered.json（跨重启可恢复挂载清单）。"""
+    record.registered_server_paths = list(
+        dict.fromkeys((getattr(record, "registered_server_paths", None) or []) + registered)
+    )
+    payload = {"registered_server_paths": record.registered_server_paths}
+    (record.job_dir / "server_registered.json").write_text(
+        json.dumps(payload, ensure_ascii=False), encoding="utf-8"
+    )
+
+
+def _blocked_reason(path: Path, settings: Any) -> str | None:
+    """敏感路径黑名单：命中返回原因字符串，未命中返回 None。"""
+    s = str(path)
+    for prefix in settings.sensitive_path_prefixes:
+        if s == prefix or s.startswith(prefix + "/"):
+            return f"敏感路径前缀 {prefix}"
+    name = path.name
+    for pat in settings.sensitive_filename_patterns:
+        if re.match(pat, name):
+            return f"敏感文件名 {name}"
+    return None
 
 
 @router.get("/jobs/{job_id}/files")

bioflow_service/app/services/bioflow/kernel.py

@@ -71,6 +71,7 @@ class KernelConfig:
     bba_env: dict[str, str] = field(default_factory=dict)
     cpu_allow: str = "4,8,16,32"
     memory_allow: str = "16g,32g,64g"
+    extra_mounts: list[str] = field(default_factory=list)  # 登记路径只读挂载(src==dst)
 
 
 class Kernel:
@@ -118,6 +119,7 @@ class Kernel:
         cpu: str | None = None,
         memory: str | None = None,
         module: str | None = None,
+        extra_mounts: list[str] | None = None,
     ) -> list[str]:
         """拼装 docker run 命令（list，无 shell；TC-BIO-KERN-001 基元）。
 
@@ -140,6 +142,9 @@ class Kernel:
         cmd += ["-v", f"{job_workspace_dir}:{self.cfg.container_workspace_dir}"]
         if self.cfg.data_root:
             cmd += ["-v", f"{self.cfg.data_root}:{self.cfg.container_data_dir}:ro"]
+        # 登记路径只读挂载（src==dst，容器内原路径直接可见；登记时已过黑名单）
+        for m in (extra_mounts or self.cfg.extra_mounts):
+            cmd += ["-v", f"{m}:{m}:ro"]
         eff_cpu = cpu or str(params.get("cpu") or "")
         eff_memory = memory or str(params.get("memory") or "")
         cmd += ["--cpus", self._safe_cpu(eff_cpu or self.cfg.default_cpus)]
@@ -205,9 +210,11 @@ class Kernel:
         log_path: str | Path | None = None,
         cpu: str | None = None,
         memory: str | None = None,
+        extra_mounts: list[str] | None = None,
     ) -> dict[str, Any]:
         """执行一个阶段 docker：返回 {exit_code, timed_out, log_tail, cmd}。
 
+        - extra_mounts：本次执行额外只读挂载的宿主绝对路径（登记路径,src==dst）
         - 超时：terminate → 3s → kill（TC-BIO-KERN-009）
         - 日志全量写 log_path（任务 logs/），返回尾部 log_tail_lines 行
         - 退出码非 0 / 超时 → 失败语义由 runner 判定（本函数只报事实）
@@ -216,7 +223,8 @@ class Kernel:
         cpu = cpu or str(params.get("cpu") or "")
         memory = memory or str(params.get("memory") or "")
         cmd = self.build_command(stage, job_input_dir, job_workspace_dir, params,
-                                 cpu=cpu or None, memory=memory or None)
+                                 cpu=cpu or None, memory=memory or None,
+                                 extra_mounts=extra_mounts)
         log_path = Path(log_path) if log_path else Path(job_workspace_dir) / "logs" / f"{stage}.log"
         log_path.parent.mkdir(parents=True, exist_ok=True)
 

bioflow_service/app/services/bioflow/runner.py

@@ -110,6 +110,7 @@ class JobRecord:
     auto_continue: bool = False
     archive_id: str | None = None
     pause_requested: bool = False
+    registered_server_paths: list[str] = field(default_factory=list)  # 登记路径（自动挂载）
     _stage_states: dict[str, dict[str, Any]] = field(default_factory=dict)
 
     @property
@@ -347,7 +348,8 @@ class Runner:
                 "job_id": record.id, "stage": stage, "attempt": attempt,
             })
             result = await self.executor(stage, record.input_dir, record.workspace_dir,
-                                         params, log_path=record.logs_dir / f"{stage}.log")
+                                         params, log_path=record.logs_dir / f"{stage}.log",
+                                         extra_mounts=self._registered_mounts(record))
             log_tail = str(result.get("log_tail") or "")
             self._set_stage(record, stage, log_tail=log_tail)
             await self._persist_stage(record, stage)
@@ -611,6 +613,17 @@ class Runner:
         await self._persist_job(record)
         self.spawn(self._resume_loop(record))
 
+    def _registered_mounts(self, record: JobRecord) -> list[str]:
+        """登记路径 → 挂载清单：取每个登记文件的父目录（去重）。"""
+        parents: list[str] = []
+        seen: set[str] = set()
+        for p in getattr(record, "registered_server_paths", None) or []:
+            parent = str(Path(p).parent)
+            if parent not in seen:
+                seen.add(parent)
+                parents.append(parent)
+        return parents
+
     async def retry_after_files(self, record: JobRecord) -> None:
         """文件上传/路径登记后的自动续跑（BUG-P7-01：文件引导闭环 FR-BIO-005/006）。
 
@@ -680,6 +693,17 @@ class Runner:
                 auto_continue=bool(row["auto_continue"]),
                 archive_id=row["archive_id"],
             )
+            # 登记路径恢复（server_registered.json）
+            try:
+                import json as _json
+                reg_file = record.job_dir / "server_registered.json"
+                if reg_file.is_file():
+                    record.registered_server_paths = (
+                        _json.loads(reg_file.read_text(encoding="utf-8"))
+                        .get("registered_server_paths") or []
+                    )
+            except Exception:  # noqa: BLE001
+                pass
             for s in stages:
                 record._stage_states[s["stage"]] = {
                     "status": s["status"], "params": params_dict(s["params"]),

bioflow_service/tests/test_api.py

@@ -529,7 +529,7 @@ def test_files_upload_no_job(api):
 
 
 def test_path_register_valid_and_whitelist(api, tmp_path):
-    """TC-BIO-API-042/045：路径登记合法 + 白名单外 403。"""
+    """TC-BIO-API-042/045：路径登记合法 + 敏感黑名单 403。"""
     resp = _create_job(api)
     job_id = _job_id(resp)
     # api_context 的 settings 已指向 tmp_path/data（conftest _isolated_env 派生）
@@ -542,11 +542,13 @@ def test_path_register_valid_and_whitelist(api, tmp_path):
         f.write(">x\nACGT")
     ok = api.post(f"/api/bioflow/jobs/{job_id}/files", json={"server_path": allowed})
     assert ok.status_code == 200
-    assert ok.json()["data"]["source"] == "server_path"
-    # 白名单外
-    outside = str(tmp_path / "outside.fa")
+    body = ok.json()["data"]
+    assert body["items"][0]["source"] == "server_path"
+    assert body["items"][0]["name"] == "SPL.fa"
+    # 敏感黑名单（.env 文件名形态）
+    outside = str(tmp_path / ".env")
     with open(outside, "w") as f:
-        f.write("x")
+        f.write("SECRET=1")
     forbidden = api.post(f"/api/bioflow/jobs/{job_id}/files", json={"server_path": outside})
     assert forbidden.status_code == 403
     assert forbidden.json()["error"]["code"] == "PATH_NOT_ALLOWED"
@@ -557,6 +559,39 @@ def test_path_register_valid_and_whitelist(api, tmp_path):
     assert missing.json()["error"]["code"] == "PATH_NOT_EXIST"
 
 
+def test_path_register_directory_expand(api, tmp_path):
+    """目录登记：递归展开多文件，逐个 items；挂载清单落 server_registered.json。"""
+    resp = _create_job(api)
+    job_id = _job_id(resp)
+    d = tmp_path / "rnaseq_fastq"
+    (d / "s1").mkdir(parents=True)
+    (d / "s2").mkdir(parents=True)
+    (d / "s1" / "CRR1_1.fastq.gz").write_text("g1", encoding="utf-8")
+    (d / "s1" / "CRR1_2.fastq.gz").write_text("g2", encoding="utf-8")
+    (d / "s2" / "CRR2_1.fastq.gz").write_text("g3", encoding="utf-8")
+    (d / "notes.txt").write_text("n", encoding="utf-8")  # 允许扩展名内
+    (d / "skip.exe").write_text("x", encoding="utf-8")   # 扩展名白名单外,跳过
+    ok = api.post(f"/api/bioflow/jobs/{job_id}/files", json={"server_path": str(d)})
+    assert ok.status_code == 200
+    body = ok.json()["data"]
+    names = {it["name"] for it in body["items"]}
+    assert names == {"CRR1_1.fastq.gz", "CRR1_2.fastq.gz", "CRR2_1.fastq.gz", "notes.txt"}
+    assert body["total"] == 4
+    # 挂载清单落盘
+    import json as _json
+    from pathlib import Path as _Path
+    from app.config import get_settings as _gs
+    job_dir = _Path(_gs().jobs_dir_abs)
+    reg = _json.loads((job_dir / job_id / "server_registered.json").read_text(encoding="utf-8"))
+    assert len(reg["registered_server_paths"]) == 4
+    # 空目录报错
+    empty = tmp_path / "empty_dir"
+    empty.mkdir()
+    err = api.post(f"/api/bioflow/jobs/{job_id}/files", json={"server_path": str(empty)})
+    assert err.status_code == 400
+    assert err.json()["error"]["code"] == "VALIDATION"
+
+
 # ---------------------------------------------------------------------------
 # stages 控制 + manifest
 # ---------------------------------------------------------------------------

bioflow_service/tests/test_kernel.py

@@ -66,6 +66,15 @@ def test_build_command_basics(kernel, tmp_path):
     assert "--workspace" in cmd and "/job/workspace" in cmd
 
 
+def test_build_command_extra_mounts(kernel, tmp_path):
+    """登记路径自动挂载：extra_mounts 每个追加 -v src:src:ro。"""
+    extra = [str(tmp_path / "rnaseq"), str(tmp_path / "wgs")]
+    cmd = kernel.build_command("identify", tmp_path / "in", tmp_path / "ws",
+                               extra_mounts=extra)
+    for m in extra:
+        assert f"{m}:{m}:ro" in cmd
+
+
 def test_build_command_params(kernel, tmp_path):
     cmd = kernel.build_command("identify", tmp_path / "in", tmp_path / "ws",
                                {"prefix": "Ft.SPL", "required_pfams": "PF03110",