Support combined CF and CFE AI structure imports

services/api-server/src/api_server/html5_ai_structure_controller.py

@@ -2,6 +2,7 @@ from __future__ import annotations
 
 from collections.abc import Callable, Iterable
 from pathlib import Path
+import shutil
 from typing import Any
 from uuid import uuid4
 
@@ -97,6 +98,7 @@ async def html5_ai_structure_run(
             save_run_state(
                 job.job_id,
                 {
+                    **(getattr(job, "state", {}) or {}),
                     "project_id": project_id,
                     "effective_project_id": effective_project_id,
                     "input_path": input_path,
@@ -158,7 +160,7 @@ async def html5_ai_structure_check_path(
     return render_html5_ai_structure_path_check(result)
 
 
-def html5_ai_structure_job(
+async def html5_ai_structure_job(
     *,
     project_id: str,
     job_id: str,
@@ -168,6 +170,7 @@ def html5_ai_structure_job(
     save_run_state: Callable[[str, dict[str, Any]], None],
     load_job: Callable[[str], object | None],
     current_project_source_root: Callable[[str], Path | None],
+    advance_binary_run: Callable[[str, dict[str, Any]], Any] | None = None,
 ) -> str:
     state = load_run_state(job_id)
     if state is None:
@@ -175,6 +178,29 @@ def html5_ai_structure_job(
     if state.get("result") is not None:
         return render_html5_ai_structure_result(dict(state["result"]))
 
+    if state.get("agent_sequence"):
+        if advance_binary_run is None:
+            return render_html5_ai_structure_error("Сервис последовательной выгрузки CF/CFE не подключен.")
+        step_result = await advance_binary_run(job_id, dict(state))
+        if step_result.get("phase") == "error":
+            return render_html5_ai_structure_error(str(step_result.get("error") or "Windows Agent завершил задачу с ошибкой."))
+        if step_result.get("phase") == "running":
+            updated_state = dict(step_result.get("state") or state)
+            save_run_state(job_id, updated_state)
+            return render_html5_ai_structure_job(
+                project_id=project_id,
+                job_id=job_id,
+                status=str(step_result.get("status") or "RUNNING"),
+                source=str(step_result.get("source") or ""),
+                message=str(step_result.get("message") or "Windows Agent выгружает структуру"),
+                logs=list(step_result.get("logs") or []),
+            )
+        if step_result.get("phase") != "completed":
+            return render_html5_ai_structure_error("Не удалось определить состояние последовательной выгрузки CF/CFE.")
+        state = dict(step_result.get("state") or state)
+        save_run_state(job_id, state)
+        source_roots = [Path(path) for path in list(step_result.get("source_roots") or []) if str(path).strip()]
+    else:
         job = load_job(job_id)
         if job is None or str(getattr(job, "project_id", "")) != project_id:
             return render_html5_ai_structure_error(f"Задача агента не найдена: {job_id}")
@@ -204,6 +230,7 @@ def html5_ai_structure_job(
             source_root = Path(source_path) if source_path else None
         if source_root is None or not source_root.exists():
             return render_html5_ai_structure_error("После выгрузки агентом сервер не нашел папку с XML/BSL-структурой для подготовки пакета.")
+        source_roots = [source_root]
 
     output_path = str(state.get("output_path") or "")
     username = str(state.get("username") or "")
@@ -216,6 +243,7 @@ def html5_ai_structure_job(
     try:
         work_dir.mkdir(parents=True, exist_ok=True)
         local_output = work_dir / "output" if is_unc_path(output_path) else Path(output_path)
+        source_root = _compose_ai_structure_source_root(work_dir / "source", source_roots)
         result = prepare(
             project_id=str(state.get("effective_project_id") or project_id),
             input_path=source_root,
@@ -296,5 +324,30 @@ def _normalize_binary_match(value: str | dict[str, str] | None) -> dict[str, str
     return {"suffix": value, "relative_path": "", "binary_relative_paths": []}
 
 
+def _compose_ai_structure_source_root(target_root: Path, source_roots: list[Path]) -> Path:
+    existing_roots = [path for path in source_roots if path.exists()]
+    if not existing_roots:
+        raise FileNotFoundError("После выгрузки агентом сервер не нашел папки с XML/BSL-структурой для подготовки пакета.")
+    if len(existing_roots) == 1:
+        return existing_roots[0]
+    target_root.mkdir(parents=True, exist_ok=True)
+    base_root = next((path for path in existing_roots if path.name.casefold().endswith(".cf") is False and (path / "src").exists()), existing_roots[0])
+    if base_root.exists():
+        shutil.copytree(base_root, target_root, dirs_exist_ok=True)
+    used_names: set[str] = set()
+    for root in existing_roots:
+        if root == base_root:
+            continue
+        name = root.name or f"extension-{len(used_names) + 1}"
+        candidate = name
+        index = 2
+        while candidate in used_names or (target_root / candidate).exists():
+            candidate = f"{name}-{index}"
+            index += 1
+        used_names.add(candidate)
+        shutil.copytree(root, target_root / candidate, dirs_exist_ok=True)
+    return target_root
+
+
 def _enum_text(value: object) -> str:
     return str(getattr(value, "value", value or ""))

services/api-server/src/api_server/main.py

@@ -18,6 +18,7 @@ from collections import Counter
 from difflib import SequenceMatcher
 from pathlib import Path
 from typing import Any, Callable
+from types import SimpleNamespace
 from urllib.parse import quote, urljoin, urlsplit, urlunsplit
 from uuid import uuid4
 
@@ -334,6 +335,9 @@ def _load_ai_structure_agent_run(job_id: str) -> dict[str, Any] | None:
     except (ValueError, UnicodeDecodeError):
         password = ""
     result = payload.get("result")
+    agent_sequence = payload.get("agent_sequence")
+    completed_steps = payload.get("completed_steps")
+    sequence_logs = payload.get("sequence_logs")
     return {
         "project_id": str(payload.get("project_id") or ""),
         "effective_project_id": str(payload.get("effective_project_id") or ""),
@@ -345,6 +349,13 @@ def _load_ai_structure_agent_run(job_id: str) -> dict[str, Any] | None:
         "domain": str(payload.get("domain") or ""),
         "password": password,
         "result": result if isinstance(result, dict) else None,
+        "agent_sequence": agent_sequence if isinstance(agent_sequence, list) else [],
+        "current_step_index": int(payload.get("current_step_index") or 0),
+        "current_agent_job_id": str(payload.get("current_agent_job_id") or ""),
+        "agent_id": str(payload.get("agent_id") or ""),
+        "display_source": str(payload.get("display_source") or ""),
+        "completed_steps": completed_steps if isinstance(completed_steps, list) else [],
+        "sequence_logs": sequence_logs if isinstance(sequence_logs, list) else [],
     }
 
 
@@ -364,6 +375,13 @@ def _save_ai_structure_agent_run(job_id: str, payload: dict[str, Any]) -> None:
             "domain": str(payload.get("domain") or ""),
             "password_b64": base64.b64encode(password.encode("utf-8")).decode("ascii") if password else "",
             "result": payload.get("result") if isinstance(payload.get("result"), dict) else None,
+            "agent_sequence": payload.get("agent_sequence") if isinstance(payload.get("agent_sequence"), list) else [],
+            "current_step_index": int(payload.get("current_step_index") or 0),
+            "current_agent_job_id": str(payload.get("current_agent_job_id") or ""),
+            "agent_id": str(payload.get("agent_id") or ""),
+            "display_source": str(payload.get("display_source") or ""),
+            "completed_steps": payload.get("completed_steps") if isinstance(payload.get("completed_steps"), list) else [],
+            "sequence_logs": payload.get("sequence_logs") if isinstance(payload.get("sequence_logs"), list) else [],
         },
     )
 
@@ -612,6 +630,33 @@ def _ai_structure_agent_root_mismatch_detail(agent_id: str, input_path: str, roo
     )
 
 
+def _ai_structure_step_path(raw_input_path: str, relative_path: str | None) -> Path:
+    if relative_path:
+        return Path(ntpath.join(raw_input_path, relative_path.replace("/", "\\")))
+    return Path(raw_input_path)
+
+
+async def _queue_ai_structure_agent_step(
+    *,
+    project_id: str,
+    source: ImportSourceKind,
+    agent_id: str,
+    local_path: str,
+    metadata: dict[str, Any],
+) -> AgentImportJob:
+    return await create_agent_import_job(
+        project_id,
+        source,
+        AgentImportJobRequest(
+            agent_id=agent_id,
+            source=source,
+            local_path=local_path,
+            mode=ImportMode.FULL_REPLACE,
+            metadata=metadata,
+        ),
+    )
+
+
 async def _start_ai_structure_agent_job(
     *,
     project_id: str,
@@ -631,16 +676,6 @@ async def _start_ai_structure_agent_job(
 
     cf_files = [path for path in binary_files if path.suffix.casefold() == ".cf"]
     cfe_files = [path for path in binary_files if path.suffix.casefold() == ".cfe"]
-    if cf_files and cfe_files:
-        raise HTTPException(
-            status_code=400,
-            detail=(
-                "Во входной папке одновременно лежат .cf и .cfe. "
-                f"Найдены: {_format_binary_file_list(binary_files)}. "
-                "Укажите конкретный файл, который нужно подготовить для ИИ."
-            ),
-        )
-    source = ImportSourceKind.CF_FILE if cf_files else ImportSourceKind.CFE_FILE
     agent_id = _agent_id_for_source(settings, ImportSourceKind.CF_FILE)
     if not agent_id:
         raise HTTPException(
@@ -663,49 +698,85 @@ async def _start_ai_structure_agent_job(
             status_code=400,
             detail=_ai_structure_agent_root_mismatch_detail(agent_id, input_path, network_roots),
         )
-
     agent = settings.agent if isinstance(settings.agent, dict) else {}
-    metadata: dict[str, Any] = {
+    if len(cf_files) > 1:
+        raise HTTPException(
+            status_code=400,
+            detail=(
+                "Для проекта ожидается одна основная конфигурация .cf. "
+                f"Найдены: {_format_binary_file_list(cf_files)}."
+            ),
+        )
+
+    common_metadata: dict[str, Any] = {
         "platform_version": settings.platform_version or None,
         "compatibility_mode": settings.compatibility_mode or None,
     }
-    local_path: str | None = None
-
-    if source == ImportSourceKind.CF_FILE:
-        if len(cf_files) != 1:
-            raise HTTPException(
-                status_code=400,
-                detail=(
-                    "Для прямого разбора .cf укажите один конкретный файл .cf, "
-                    f"а не папку с несколькими конфигурациями. Найдены: {_format_binary_file_list(cf_files)}."
-                ),
+    steps: list[dict[str, Any]] = []
+    if cf_files:
+        cf_file = cf_files[0]
+        steps.append(
+            {
+                "source": ImportSourceKind.CF_FILE,
+                "local_path": str(cf_file),
+                "label": cf_file.name,
+                "metadata": {**common_metadata, "input_mode": "cf_file"},
+            }
         )
-        local_path = str(cf_files[0])
-        metadata["input_mode"] = "cf_file"
-    else:
-        if len(cfe_files) != 1:
-            raise HTTPException(
-                status_code=400,
-                detail=(
-                    "Для прямого разбора расширения укажите один конкретный файл .cfe, "
-                    f"а не папку с несколькими расширениями. Найдены: {_format_binary_file_list(cfe_files)}."
-                ),
+    for cfe_file in sorted(cfe_files):
+        steps.append(
+            {
+                "source": ImportSourceKind.CFE_FILE,
+                "local_path": str(cfe_file),
+                "label": cfe_file.name,
+                "metadata": {
+                    **common_metadata,
+                    "input_mode": "cfe_file",
+                    "one_c_extension": cfe_file.stem,
+                },
+            }
         )
-        cfe_file = cfe_files[0]
-        local_path = str(cfe_file)
-        metadata["one_c_extension"] = cfe_file.stem
-        metadata["input_mode"] = "cfe_file"
+    if not steps:
+        raise HTTPException(status_code=400, detail="Во входном пути не найдены файлы .cf или .cfe.")
 
-    return await create_agent_import_job(
-        effective_project_id,
-        source,
-        AgentImportJobRequest(
+    first = steps[0]
+    first_job = await _queue_ai_structure_agent_step(
+        project_id=effective_project_id,
+        source=first["source"],
         agent_id=agent_id,
-            source=source,
-            local_path=local_path,
-            mode=ImportMode.FULL_REPLACE,
-            metadata=metadata,
-        ),
+        local_path=first["local_path"],
+        metadata=first["metadata"],
+    )
+    binary_kind = "CF_FILE+CFE_FILE" if cf_files and cfe_files else first["source"].value
+    intro_logs = [
+        (
+            f"Найдено файлов для подготовки: конфигураций {len(cf_files)}, "
+            f"расширений {len(cfe_files)}. Сначала выгружается конфигурация, затем все расширения."
+        )
+        if cf_files and cfe_files
+        else f"Найден файл для подготовки: {first['label']}."
+    ]
+    return SimpleNamespace(
+        job_id=first_job.job_id,
+        status=first_job.status,
+        source=binary_kind,
+        logs=[*intro_logs, *list(first_job.logs or [])],
+        state={
+            "agent_sequence": [
+                {
+                    "source": step["source"].value,
+                    "local_path": step["local_path"],
+                    "label": step["label"],
+                    "metadata": step["metadata"],
+                }
+                for step in steps
+            ],
+            "current_step_index": 0,
+            "current_agent_job_id": first_job.job_id,
+            "agent_id": agent_id,
+            "display_source": binary_kind,
+            "completed_steps": [],
+        },
     )
 
 
@@ -802,6 +873,96 @@ async def _check_ai_structure_agent_path(*, project_id: str, input_path: str) ->
     }
 
 
+async def _advance_ai_structure_agent_run(run_id: str, state: dict[str, Any]) -> dict[str, Any]:
+    sequence = list(state.get("agent_sequence") or [])
+    if not sequence:
+        return {"phase": "error", "error": "План выгрузки CF/CFE не найден."}
+    current_job_id = str(state.get("current_agent_job_id") or run_id)
+    current_step_index = int(state.get("current_step_index") or 0)
+    current_job = _agent_import_jobs.get(current_job_id)
+    if current_job is None:
+        return {"phase": "error", "error": f"Задача агента не найдена: {current_job_id}"}
+
+    display_source = str(state.get("display_source") or getattr(current_job.source, "value", current_job.source or ""))
+    logs = list(state.get("sequence_logs") or [])
+    logs.extend(str(item) for item in list(current_job.logs or []))
+    state["sequence_logs"] = logs[-24:]
+
+    if current_job.status in {AgentImportJobStatus.QUEUED, AgentImportJobStatus.RUNNING}:
+        return {
+            "phase": "running",
+            "state": state,
+            "job_id": run_id,
+            "status": current_job.status.value,
+            "source": display_source,
+            "logs": state["sequence_logs"],
+            "message": "Windows Agent выгружает структуру и передает ее на сервер",
+        }
+    if current_job.status != AgentImportJobStatus.SUCCEEDED:
+        error = str(current_job.error or "Windows Agent завершил задачу с ошибкой.")
+        if current_job.logs:
+            error = f"{error} Последние сообщения: {' | '.join(str(item) for item in current_job.logs[-4:])}"
+        return {"phase": "error", "error": error}
+
+    completed_steps = list(state.get("completed_steps") or [])
+    if not any(str(item.get("job_id")) == current_job_id for item in completed_steps):
+        import_summary = current_job.import_summary or {}
+        source_path = str(import_summary.get("source_path") or current_job.server_path or "")
+        if not source_path:
+            return {
+                "phase": "error",
+                "error": "После выгрузки агентом сервер не вернул путь к экспортированной структуре.",
+            }
+        completed_steps.append(
+            {
+                "job_id": current_job_id,
+                "source": sequence[current_step_index]["source"],
+                "label": sequence[current_step_index]["label"],
+                "source_path": source_path,
+            }
+        )
+        state["completed_steps"] = completed_steps
+
+    next_step_index = current_step_index + 1
+    if next_step_index < len(sequence):
+        next_step = sequence[next_step_index]
+        next_job = await _queue_ai_structure_agent_step(
+            project_id=str(state.get("effective_project_id") or state.get("project_id") or ""),
+            source=ImportSourceKind(next_step["source"]),
+            agent_id=str(state.get("agent_id") or ""),
+            local_path=str(next_step["local_path"]),
+            metadata=dict(next_step.get("metadata") or {}),
+        )
+        state["current_step_index"] = next_step_index
+        state["current_agent_job_id"] = next_job.job_id
+        state["sequence_logs"] = [
+            *state["sequence_logs"],
+            f"Шаг {current_step_index + 1} завершен: {sequence[current_step_index]['label']}.",
+            f"Запущен следующий шаг {next_step_index + 1}: {next_step['label']}.",
+            *list(next_job.logs or []),
+        ][-24:]
+        return {
+            "phase": "running",
+            "state": state,
+            "job_id": run_id,
+            "status": next_job.status.value,
+            "source": display_source,
+            "logs": state["sequence_logs"],
+            "message": "Windows Agent продолжает выгрузку конфигурации и расширений",
+        }
+
+    state["sequence_logs"] = [
+        *state["sequence_logs"],
+        f"Шаг {current_step_index + 1} завершен: {sequence[current_step_index]['label']}.",
+        "Все найденные CF/CFE выгружены. Сервер собирает объединенный пакет для ИИ.",
+    ][-24:]
+    return {
+        "phase": "completed",
+        "state": state,
+        "source_roots": [str(item.get("source_path") or "") for item in completed_steps if str(item.get("source_path") or "")],
+    }
+
+
 def _ai_structure_binary_files(
     raw_input_path: str,
     detected_binary_relative_path: str | None = None,
@@ -1978,7 +2139,7 @@ async def html5_project_ai_structure_check_path(project_id: str, request: Reques
 @app.get("/html5/projects/{project_id}/ai-structure/jobs/{job_id}")
 async def html5_project_ai_structure_job(project_id: str, job_id: str) -> Response:
     return _html5_response(
-        _html5_ai_structure_job(
+        await _html5_ai_structure_job(
             project_id=project_id,
             job_id=job_id,
             prepare=_prepare_ai_structure,
@@ -1987,6 +2148,7 @@ async def html5_project_ai_structure_job(project_id: str, job_id: str) -> Respon
             save_run_state=_save_ai_structure_agent_run,
             load_job=lambda current_job_id: _agent_import_jobs.get(current_job_id),
             current_project_source_root=_current_project_source_root,
+            advance_binary_run=_advance_ai_structure_agent_run,
         )
     )
 

services/api-server/tests/test_api.py

@@ -1962,6 +1962,107 @@ def test_html5_ai_structure_routes_binary_cfe_through_windows_agent(tmp_path: Pa
     assert (output / f"codex-1c-context-{project_id}" / "AGENTS.md").exists()
 
 
+def test_html5_ai_structure_routes_cf_and_cfe_as_single_project(tmp_path: Path):
+    base_root = tmp_path / "base-export"
+    base_root.mkdir()
+    (base_root / "metadata.xml").write_text(
+        """
+<Configuration>
+  <Catalog name="Контрагенты" qualifiedName="Справочник.Контрагенты" />
+</Configuration>
+""",
+        encoding="utf-8",
+    )
+    extension_root = tmp_path / "extension-export"
+    extension_root.mkdir()
+    (extension_root / "РасширениеCRM.mdo").write_text(
+        """
+<mdclass:ConfigurationExtension xmlns:mdclass="http://g5.1c.ru/v8/dt/metadata/mdclass/extension">
+  <name>CRM</name>
+  <version>1.0</version>
+  <catalogs>
+    <name>КонтрагентыCRM</name>
+  </catalogs>
+</mdclass:ConfigurationExtension>
+""",
+        encoding="utf-8",
+    )
+    source_dir = tmp_path / "binary"
+    source_dir.mkdir()
+    (source_dir / "1Cv8.cf").write_bytes(b"binary-cf")
+    (source_dir / "Marketplace.cfe").write_bytes(b"binary-cfe")
+    output = tmp_path / "ai-out-mixed"
+    client = TestClient(app)
+    project_id = f"ai-agent-mixed-{uuid4()}"
+    agent_id = f"win-agent-{uuid4()}"
+
+    settings = client.post(
+        f"/projects/{project_id}/settings",
+        json={"name": "AI Agent Mixed Demo", "structure_source": "CF_FILE", "agent": {"cf_agent_id": agent_id}},
+    )
+    assert settings.status_code == 200
+    heartbeat = client.post("/agent/heartbeat", json={"agent_id": agent_id, "host": "test-host"})
+    assert heartbeat.status_code == 200
+
+    queued = client.post(
+        f"/html5/projects/{project_id}/ai-structure/run",
+        data={"project_id": project_id, "input_path": str(source_dir), "output_path": str(output)},
+    )
+    assert queued.status_code == 200
+    assert "конфигураций 1, расширений 1" in queued.text
+    match = re.search(r"/html5/projects/[^/]+/ai-structure/jobs/([A-Za-z0-9-]+)", queued.text)
+    assert match is not None
+    job_id = match.group(1)
+
+    first_claimed = client.get("/agent/jobs/next", params={"agent_id": agent_id})
+    assert first_claimed.status_code == 200
+    assert first_claimed.json()["job_id"] == job_id
+    assert first_claimed.json()["source"] == "CF_FILE"
+
+    completed_cf = client.post(
+        f"/agent/jobs/{job_id}/result",
+        json={"status": "SUCCEEDED", "server_path": str(base_root), "logs": ["Выгрузка конфигурации завершена."]},
+    )
+    assert completed_cf.status_code == 200
+    main._agent_import_jobs[job_id].status = main.AgentImportJobStatus.SUCCEEDED
+    main._agent_import_jobs[job_id].import_summary = {"source_path": str(base_root)}
+
+    deadline = time.monotonic() + 10
+    second_job = None
+    while time.monotonic() < deadline:
+        polled = client.get(f"/html5/projects/{project_id}/ai-structure/jobs/{job_id}")
+        assert polled.status_code == 200
+        second_claimed = client.get("/agent/jobs/next", params={"agent_id": agent_id})
+        assert second_claimed.status_code == 200
+        second_job = second_claimed.json()
+        if second_job:
+            break
+        time.sleep(0.05)
+    assert second_job is not None
+    assert second_job["source"] == "CFE_FILE"
+    second_job_id = second_job["job_id"]
+
+    completed_cfe = client.post(
+        f"/agent/jobs/{second_job_id}/result",
+        json={"status": "SUCCEEDED", "server_path": str(extension_root), "logs": ["Выгрузка расширения завершена."]},
+    )
+    assert completed_cfe.status_code == 200
+    main._agent_import_jobs[second_job_id].status = main.AgentImportJobStatus.SUCCEEDED
+    main._agent_import_jobs[second_job_id].import_summary = {"source_path": str(extension_root)}
+
+    deadline = time.monotonic() + 10
+    fragment = ""
+    while time.monotonic() < deadline:
+        polled = client.get(f"/html5/projects/{project_id}/ai-structure/jobs/{job_id}")
+        assert polled.status_code == 200
+        fragment = polled.text
+        if "готово" in fragment:
+            break
+        time.sleep(0.05)
+    assert "готово" in fragment
+    assert (output / f"codex-1c-context-{project_id}" / "AGENTS.md").exists()
+
+
 def test_html5_ai_structure_routes_unc_directory_with_cf_through_windows_agent(monkeypatch, tmp_path: Path):
     from api_server import html5_ai_structure_controller as controller
 
@@ -2041,7 +2142,7 @@ def test_html5_ai_structure_reports_multiple_binary_files_in_directory(tmp_path:
         data={"project_id": project_id, "input_path": str(tmp_path), "output_path": str(tmp_path / 'out')},
     )
     assert queued.status_code == 200
-    assert "один конкретный файл .cf" in queued.text
+    assert "одна основная конфигурация .cf" in queued.text
     assert "first.cf" in queued.text
     assert "second.cf" in queued.text