Handle UNC binary directories in AI structure flow

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

@@ -62,12 +62,30 @@ async def html5_ai_structure_run(
     if should_save and save_credentials and username and password:
         save_credentials(project_id, {"username": username, "password": password, "domain": domain})
 
-    binary_source = _detect_binary_input(input_path)
+    work_dir = work_root / f"{effective_project_id}-{uuid4().hex}"
-    if binary_source is not None:
+    try:
+        work_dir.mkdir(parents=True, exist_ok=True)
+        local_input = work_dir / "input" if is_unc_path(input_path) else Path(input_path)
+        local_output = work_dir / "output" if is_unc_path(output_path) else Path(output_path)
+        if is_unc_path(input_path):
+            copy_smb_tree_to_local(
+                source=input_path,
+                target=local_input,
+                username=username,
+                password=password,
+                domain=domain or None,
+            )
+        binary_match = _detect_binary_tree(local_input) or _normalize_binary_match(_detect_binary_input(input_path))
+        if binary_match is not None:
             if start_binary_job is None or save_run_state is None:
                 return render_html5_ai_structure_error("Сервис подготовки CF/CFE через Windows Agent не подключен.")
             try:
-            job = await start_binary_job(project_id=project_id, effective_project_id=effective_project_id, input_path=Path(input_path))
+                job = await start_binary_job(
+                    project_id=project_id,
+                    effective_project_id=effective_project_id,
+                    input_path=input_path,
+                    detected_binary_relative_path=binary_match.get("relative_path"),
+                )
             except HTTPException as error:
                 return render_html5_ai_structure_error(str(error.detail))
             save_run_state(
@@ -92,20 +110,6 @@ async def html5_ai_structure_run(
                 message="Разбор CF/CFE запущен через Windows Agent",
                 logs=getattr(job, "logs", []),
             )
-
-    work_dir = work_root / f"{effective_project_id}-{uuid4().hex}"
-    try:
-        work_dir.mkdir(parents=True, exist_ok=True)
-        local_input = work_dir / "input" if is_unc_path(input_path) else Path(input_path)
-        local_output = work_dir / "output" if is_unc_path(output_path) else Path(output_path)
-        if is_unc_path(input_path):
-            copy_smb_tree_to_local(
-                source=input_path,
-                target=local_input,
-                username=username,
-                password=password,
-                domain=domain or None,
-            )
         result = prepare(
             project_id=effective_project_id,
             input_path=local_input,
@@ -244,5 +248,30 @@ def _detect_binary_input(raw_input_path: str) -> str | None:
     return binary_files[0].suffix.casefold()
 
 
+def _detect_binary_tree(input_path: Path) -> dict[str, str] | None:
+    if not input_path.exists():
+        return None
+    suffixes = {".cf", ".cfe"}
+    if input_path.is_file() and input_path.suffix.casefold() in suffixes:
+        return {"suffix": input_path.suffix.casefold(), "relative_path": input_path.name}
+    if not input_path.is_dir():
+        return None
+    files = sorted(path for path in input_path.rglob("*") if path.is_file())
+    parseable_files = any(path.suffix.casefold() in {".xml", ".mdo", ".bsl"} for path in files)
+    binary_files = [path for path in files if path.suffix.casefold() in suffixes]
+    if parseable_files or not binary_files:
+        return None
+    first = binary_files[0]
+    return {"suffix": first.suffix.casefold(), "relative_path": first.relative_to(input_path).as_posix()}
+
+
+def _normalize_binary_match(value: str | dict[str, str] | None) -> dict[str, str] | None:
+    if value is None:
+        return None
+    if isinstance(value, dict):
+        return value
+    return {"suffix": value, "relative_path": ""}
+
+
 def _enum_text(value: object) -> str:
     return str(getattr(value, "value", value or ""))

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

@@ -4,6 +4,7 @@ import asyncio
 import base64
 import hashlib
 import json
+import ntpath
 import os
 import re
 import shutil
@@ -581,9 +582,15 @@ def _agent_id_for_source(settings: "ProjectSettingsRequest", source: "ImportSour
     return str(agent.get("agent_id") or "").strip()
 
 
-async def _start_ai_structure_agent_job(*, project_id: str, effective_project_id: str, input_path: Path) -> AgentImportJob:
+async def _start_ai_structure_agent_job(
+    *,
+    project_id: str,
+    effective_project_id: str,
+    input_path: str,
+    detected_binary_relative_path: str | None = None,
+) -> AgentImportJob:
     settings = _project_settings_or_404(project_id)
-    binary_files = _ai_structure_binary_files(input_path)
+    binary_files = _ai_structure_binary_files(input_path, detected_binary_relative_path=detected_binary_relative_path)
     if not binary_files:
         raise HTTPException(status_code=400, detail="Во входном пути не найдены файлы .cf или .cfe.")
 
@@ -641,7 +648,14 @@ async def _start_ai_structure_agent_job(*, project_id: str, effective_project_id
     )
 
 
-def _ai_structure_binary_files(input_path: Path) -> list[Path]:
+def _ai_structure_binary_files(raw_input_path: str, detected_binary_relative_path: str | None = None) -> list[Path]:
+    lowered = raw_input_path.strip().casefold()
+    if lowered.endswith(".cf") or lowered.endswith(".cfe"):
+        return [Path(raw_input_path)]
+    if detected_binary_relative_path:
+        windows_path = ntpath.join(raw_input_path, detected_binary_relative_path.replace("/", "\\"))
+        return [Path(windows_path)]
+    input_path = Path(raw_input_path)
     if input_path.is_file() and input_path.suffix.casefold() in {".cf", ".cfe"}:
         return [input_path]
     if not input_path.exists() or not input_path.is_dir():

services/api-server/tests/test_api.py

@@ -1,3 +1,4 @@
+import asyncio
 from pathlib import Path
 import json
 import re
@@ -1961,6 +1962,62 @@ 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_unc_directory_with_cf_through_windows_agent(monkeypatch, tmp_path: Path):
+    from api_server import html5_ai_structure_controller as controller
+
+    copied_root = tmp_path / "copied-unc"
+    copied_root.mkdir()
+    (copied_root / "base.cf").write_bytes(b"binary-cf")
+
+    copied_targets: list[tuple[str, Path]] = []
+
+    def fake_copy_smb_tree_to_local(*, source: str, target: Path, username: str, password: str, domain: str | None = None) -> None:
+        copied_targets.append((source, target))
+        target.mkdir(parents=True, exist_ok=True)
+        for item in copied_root.iterdir():
+            if item.is_file():
+                (target / item.name).write_bytes(item.read_bytes())
+
+    class FakeJob:
+        job_id = "agent-import-test"
+        status = "QUEUED"
+        source = "CF_FILE"
+        logs = ["queued"]
+
+    started: dict[str, object] = {}
+
+    async def fake_start_binary_job(**kwargs):
+        started.update(kwargs)
+        return FakeJob()
+
+    saved_runs: dict[str, dict[str, object]] = {}
+
+    monkeypatch.setattr(controller, "copy_smb_tree_to_local", fake_copy_smb_tree_to_local)
+
+    html = asyncio.run(
+        controller.html5_ai_structure_run(
+            project_id="unc-demo",
+            form={
+                "project_id": ["unc-demo"],
+                "input_path": [r"\\192.168.220.200\mst\1c\MARKA\CODEX\CF"],
+                "output_path": [r"\\192.168.220.200\mst\1c\MARKA\CODEX\CODEX"],
+                "smb_username": ["m"],
+                "smb_password": ["secret"],
+            },
+            prepare=lambda **_: {},
+            work_root=tmp_path / "work",
+            start_binary_job=fake_start_binary_job,
+            save_run_state=lambda job_id, payload: saved_runs.setdefault(job_id, payload),
+        )
+    )
+
+    assert "Windows Agent" in html
+    assert copied_targets
+    assert started["input_path"] == r"\\192.168.220.200\mst\1c\MARKA\CODEX\CF"
+    assert started["detected_binary_relative_path"] == "base.cf"
+    assert "agent-import-test" in saved_runs
+
+
 def test_import_full_replace_replaces_current_normalized_project(tmp_path: Path):
     first = tmp_path / "first"
     second = tmp_path / "second"