Update app/routers/screenshots.py

app/routers/screenshots.py

@@ -1,226 +1,267 @@
-import asyncio
-import uuid
-from datetime import datetime, timedelta, timezone
-
-from fastapi import APIRouter, Depends, File, Form, UploadFile
-
-from app.core.auth import AuthUser, get_current_web_user
-from app.core.config import settings
-from app.models.schemas import ScreenshotIngestResponse, ViolationScreenshot
-from app.services.alert_hub import alert_hub
-from app.services.classifier import (
-    classify_game_with_ocr_llm,
-    classify_sensitive_content,
-)
-from app.services.policy_window import is_restriction_active
-from app.services.repos import (
-    audit_repo,
-    policy_repo,
-    review_repo,
-    url_verdict_repo,
-    violation_repo,
-)
-from app.services.storage import storage_service
-
-router = APIRouter(prefix="/screenshots", tags=["screenshots"])
-
-
-async def _cleanup_expired_violations() -> None:
-    expired = await violation_repo.list_expired()
-    for row in expired:
-        absolute_path = row.get("absolute_path")
-        if isinstance(absolute_path, str):
-            storage_service.delete(absolute_path)
-    await violation_repo.delete_expired()
-
-
-@router.post("/ingest", response_model=ScreenshotIngestResponse)
-async def ingest_screenshot(
-    device_id: str = Form(...),
-    timestamp: str = Form(...),
-    suspected_game: bool = Form(False),
-    file: UploadFile = File(...),
-) -> ScreenshotIngestResponse:
-    screenshot_id = str(uuid.uuid4())
-    _, absolute_path, stored_name = await storage_service.save_upload(file)
-
-    try:
-        moment = datetime.fromisoformat(timestamp.replace("Z", "+00:00"))
-    except ValueError:
-        moment = datetime.now(timezone.utc)
-    policy_doc = await policy_repo.get(device_id)
-    policy_payload = policy_doc.get("policy", {}) if policy_doc else {}
-    windows = policy_payload.get("restriction_windows", [])
-    if not is_restriction_active(windows, moment):
-        storage_service.delete(absolute_path)
-        await audit_repo.log(
-            {
-                "screenshot_id": screenshot_id,
-                "device_id": device_id,
-                "timestamp": timestamp,
-                "filename": file.filename or "unknown.png",
-                "action": "deleted_outside_window",
-                "reason": "outside_restriction_window",
-                "created_at": datetime.now(timezone.utc),
-            }
-        )
-        return ScreenshotIngestResponse(screenshot_id=screenshot_id)
-
-    game_task = asyncio.to_thread(
-        classify_game_with_ocr_llm,
-        absolute_path,
-        file.filename or "capture.png",
-        suspected_game,
-    )
-    sensitive_task = asyncio.to_thread(
-        classify_sensitive_content,
-        absolute_path,
-        file.filename or "capture.png",
-    )
-    game_raw, sensitive_raw = await asyncio.gather(game_task, sensitive_task, return_exceptions=True)
-
-    if isinstance(game_raw, Exception):
-        game_result = {
-            "verdict": "not_game",
-            "confidence": 0.1,
-            "reason": "game-check-error",
-            "urls": [],
-            "source": "error-fallback",
-        }
-    else:
-        game_result = game_raw
-
-    if isinstance(sensitive_raw, Exception):
-        sensitive_detected, sensitive_confidence, sensitive_reason = (False, 0.0, "sensitive-check-error")
-    else:
-        sensitive_detected, sensitive_confidence, sensitive_reason = sensitive_raw
-
-    ocr_urls = [str(x) for x in game_result.get("urls", [])]
-
-    cached_game_url = await url_verdict_repo.get_cached_game_url(ocr_urls)
-    if cached_game_url:
-        game_result = {
-            **game_result,
-            "verdict": "game",
-            "confidence": 0.99,
-            "reason": f"cached-game-url:{cached_game_url}",
-            "source": "url-cache",
-        }
-
-    verdict = str(game_result.get("verdict", "not_game"))
-    confidence = float(game_result.get("confidence", 0.1))
-    reason = str(game_result.get("reason", "no-game-evidence"))
-
-    if sensitive_detected:
-        image_url = f"{settings.media_prefix}/{stored_name}"
-        await violation_repo.insert(
-            {
-                "screenshot_id": screenshot_id,
-                "device_id": device_id,
-                "timestamp": timestamp,
-                "filename": file.filename or "unknown.png",
-                "created_at": datetime.now(timezone.utc).isoformat(),
-                "image_url": image_url,
-                "confidence": sensitive_confidence,
-                "reason": sensitive_reason,
-                "absolute_path": absolute_path,
-                "visibility_scope": "sensitive_admin",
-                "ocr_urls": ocr_urls,
-                "expires_at": datetime.now(timezone.utc) + timedelta(hours=24),
-            }
-        )
-        await alert_hub.broadcast_admin(
-            {
-                "type": "sensitive_detected",
-                "device_id": device_id,
-                "message": "Sensitive content detected. Routed to sensitive admin scope.",
-                "visibility_scope": "sensitive_admin",
-            }
-        )
-        await _cleanup_expired_violations()
-        return ScreenshotIngestResponse(screenshot_id=screenshot_id)
-
-    if verdict == "game":
-        await url_verdict_repo.upsert_game_urls(ocr_urls, source=str(game_result.get("source", "llm")))
-
-    if verdict == "not_game":
-        storage_service.delete(absolute_path)
-        await audit_repo.log(
-            {
-                "screenshot_id": screenshot_id,
-                "device_id": device_id,
-                "timestamp": timestamp,
-                "filename": file.filename or "unknown.png",
-                "action": "deleted_non_violation",
-                "reason": reason,
-                "created_at": datetime.now(timezone.utc),
-            }
-        )
-        return ScreenshotIngestResponse(screenshot_id=screenshot_id)
-
-    image_url = f"{settings.media_prefix}/{stored_name}"
-    await violation_repo.insert(
-        {
-            "screenshot_id": screenshot_id,
-            "device_id": device_id,
-            "timestamp": timestamp,
-            "filename": file.filename or "unknown.png",
-            "created_at": datetime.now(timezone.utc).isoformat(),
-            "image_url": image_url,
-            "confidence": confidence,
-            "reason": reason,
-            "absolute_path": absolute_path,
-            "visibility_scope": "admin",
-            "ocr_urls": ocr_urls,
-            "expires_at": datetime.now(timezone.utc) + timedelta(hours=24),
-        }
-    )
-
-    if verdict == "uncertain":
-        review_id = str(uuid.uuid4())
-        await review_repo.create_pending(
-            {
-                "review_id": review_id,
-                "screenshot_id": screenshot_id,
-                "device_id": device_id,
-                "timestamp": timestamp,
-                "status": "pending",
-                "reason": reason,
-                "ocr_urls": ocr_urls,
-                "created_at": datetime.now(timezone.utc),
-            }
-        )
-        await alert_hub.broadcast_admin(
-            {
-                "type": "game_lock_review_required",
-                "device_id": device_id,
-                "message": "Model uncertain. Admin confirmation is required before lock.",
-                "review_id": review_id,
-                "screenshot_id": screenshot_id,
-                "ocr_urls": ocr_urls,
-            }
-        )
-
-    await _cleanup_expired_violations()
-    return ScreenshotIngestResponse(screenshot_id=screenshot_id)
-
-
-@router.get("/violations", response_model=list[ViolationScreenshot])
-async def list_violations(user: AuthUser = Depends(get_current_web_user)) -> list[ViolationScreenshot]:
-    await _cleanup_expired_violations()
-    rows = await violation_repo.list_active(user.role)
-    return [
-        ViolationScreenshot(
-            screenshot_id=row.get("screenshot_id", ""),
-            device_id=row.get("device_id", ""),
-            timestamp=row.get("timestamp", ""),
-            filename=row.get("filename", ""),
-            created_at=row.get("created_at", ""),
-            image_url=row.get("image_url", ""),
-            confidence=float(row.get("confidence", 0.0)),
-            reason=row.get("reason", ""),
-            visibility_scope=row.get("visibility_scope", "admin"),
-            ocr_urls=[str(x) for x in row.get("ocr_urls", [])],
-        )
-        for row in rows
-    ]
+import asyncio
+import logging
+import uuid
+from datetime import datetime, timedelta, timezone
+
+from fastapi import APIRouter, Depends, File, Form, UploadFile
+
+from app.core.auth import AuthUser, get_current_web_user
+from app.core.config import settings
+from app.models.schemas import ScreenshotIngestResponse, ViolationScreenshot
+from app.services.alert_hub import alert_hub
+from app.services.classifier import (
+    classify_game_with_ocr_llm,
+    classify_sensitive_content,
+)
+from app.services.policy_window import is_restriction_active
+from app.services.repos import (
+    audit_repo,
+    policy_repo,
+    review_repo,
+    url_verdict_repo,
+    violation_repo,
+)
+from app.services.storage import storage_service
+
+router = APIRouter(prefix="/screenshots", tags=["screenshots"])
+logger = logging.getLogger(__name__)
+
+
+async def _cleanup_expired_violations() -> None:
+    expired = await violation_repo.list_expired()
+    for row in expired:
+        absolute_path = row.get("absolute_path")
+        if isinstance(absolute_path, str):
+            storage_service.delete(absolute_path)
+    await violation_repo.delete_expired()
+
+
+@router.post("/ingest", response_model=ScreenshotIngestResponse)
+async def ingest_screenshot(
+    device_id: str = Form(...),
+    timestamp: str = Form(...),
+    suspected_game: bool = Form(False),
+    file: UploadFile = File(...),
+) -> ScreenshotIngestResponse:
+    screenshot_id = str(uuid.uuid4())
+    _, absolute_path, stored_name = await storage_service.save_upload(file)
+    logger.info(
+        "ingest start screenshot_id=%s device_id=%s filename=%s suspected_game=%s",
+        screenshot_id,
+        device_id,
+        file.filename or "unknown.png",
+        suspected_game,
+    )
+
+    try:
+        moment = datetime.fromisoformat(timestamp.replace("Z", "+00:00"))
+    except ValueError:
+        moment = datetime.now(timezone.utc)
+    policy_doc = await policy_repo.get(device_id)
+    policy_payload = policy_doc.get("policy", {}) if policy_doc else {}
+    windows = policy_payload.get("restriction_windows", [])
+    if not is_restriction_active(windows, moment):
+        storage_service.delete(absolute_path)
+        logger.info(
+            "ingest drop screenshot_id=%s reason=outside_restriction_window",
+            screenshot_id,
+        )
+        await audit_repo.log(
+            {
+                "screenshot_id": screenshot_id,
+                "device_id": device_id,
+                "timestamp": timestamp,
+                "filename": file.filename or "unknown.png",
+                "action": "deleted_outside_window",
+                "reason": "outside_restriction_window",
+                "created_at": datetime.now(timezone.utc),
+            }
+        )
+        return ScreenshotIngestResponse(screenshot_id=screenshot_id)
+
+    game_task = asyncio.to_thread(
+        classify_game_with_ocr_llm,
+        absolute_path,
+        file.filename or "capture.png",
+        suspected_game,
+    )
+    sensitive_task = asyncio.to_thread(
+        classify_sensitive_content,
+        absolute_path,
+        file.filename or "capture.png",
+    )
+    game_raw, sensitive_raw = await asyncio.gather(game_task, sensitive_task, return_exceptions=True)
+
+    if isinstance(game_raw, Exception):
+        logger.warning("ingest game-check-error screenshot_id=%s err=%s", screenshot_id, repr(game_raw))
+        game_result = {
+            "verdict": "not_game",
+            "confidence": 0.1,
+            "reason": "game-check-error",
+            "urls": [],
+            "source": "error-fallback",
+        }
+    else:
+        game_result = game_raw
+
+    if isinstance(sensitive_raw, Exception):
+        logger.warning("ingest sensitive-check-error screenshot_id=%s err=%s", screenshot_id, repr(sensitive_raw))
+        sensitive_detected, sensitive_confidence, sensitive_reason = (False, 0.0, "sensitive-check-error")
+    else:
+        sensitive_detected, sensitive_confidence, sensitive_reason = sensitive_raw
+
+    ocr_urls = [str(x) for x in game_result.get("urls", [])]
+
+    cached_game_url = await url_verdict_repo.get_cached_game_url(ocr_urls)
+    if cached_game_url:
+        logger.info(
+            "ingest cache-hit screenshot_id=%s game_url=%s",
+            screenshot_id,
+            cached_game_url,
+        )
+        game_result = {
+            **game_result,
+            "verdict": "game",
+            "confidence": 0.99,
+            "reason": f"cached-game-url:{cached_game_url}",
+            "source": "url-cache",
+        }
+
+    verdict = str(game_result.get("verdict", "not_game"))
+    confidence = float(game_result.get("confidence", 0.1))
+    reason = str(game_result.get("reason", "no-game-evidence"))
+
+    if sensitive_detected:
+        logger.info(
+            "ingest verdict screenshot_id=%s verdict=sensitive confidence=%.2f reason=%s",
+            screenshot_id,
+            sensitive_confidence,
+            sensitive_reason,
+        )
+        image_url = f"{settings.media_prefix}/{stored_name}"
+        await violation_repo.insert(
+            {
+                "screenshot_id": screenshot_id,
+                "device_id": device_id,
+                "timestamp": timestamp,
+                "filename": file.filename or "unknown.png",
+                "created_at": datetime.now(timezone.utc).isoformat(),
+                "image_url": image_url,
+                "confidence": sensitive_confidence,
+                "reason": sensitive_reason,
+                "absolute_path": absolute_path,
+                "visibility_scope": "sensitive_admin",
+                "ocr_urls": ocr_urls,
+                "expires_at": datetime.now(timezone.utc) + timedelta(hours=24),
+            }
+        )
+        await alert_hub.broadcast_admin(
+            {
+                "type": "sensitive_detected",
+                "device_id": device_id,
+                "message": "Sensitive content detected. Routed to sensitive admin scope.",
+                "visibility_scope": "sensitive_admin",
+            }
+        )
+        await _cleanup_expired_violations()
+        return ScreenshotIngestResponse(screenshot_id=screenshot_id)
+
+    if verdict == "game":
+        await url_verdict_repo.upsert_game_urls(ocr_urls, source=str(game_result.get("source", "llm")))
+
+    if verdict == "not_game":
+        storage_service.delete(absolute_path)
+        logger.info(
+            "ingest verdict screenshot_id=%s verdict=not_game confidence=%.2f reason=%s urls=%d",
+            screenshot_id,
+            confidence,
+            reason,
+            len(ocr_urls),
+        )
+        await audit_repo.log(
+            {
+                "screenshot_id": screenshot_id,
+                "device_id": device_id,
+                "timestamp": timestamp,
+                "filename": file.filename or "unknown.png",
+                "action": "deleted_non_violation",
+                "reason": reason,
+                "created_at": datetime.now(timezone.utc),
+            }
+        )
+        return ScreenshotIngestResponse(screenshot_id=screenshot_id)
+
+    image_url = f"{settings.media_prefix}/{stored_name}"
+    logger.info(
+        "ingest verdict screenshot_id=%s verdict=%s confidence=%.2f reason=%s urls=%d",
+        screenshot_id,
+        verdict,
+        confidence,
+        reason,
+        len(ocr_urls),
+    )
+    await violation_repo.insert(
+        {
+            "screenshot_id": screenshot_id,
+            "device_id": device_id,
+            "timestamp": timestamp,
+            "filename": file.filename or "unknown.png",
+            "created_at": datetime.now(timezone.utc).isoformat(),
+            "image_url": image_url,
+            "confidence": confidence,
+            "reason": reason,
+            "absolute_path": absolute_path,
+            "visibility_scope": "admin",
+            "ocr_urls": ocr_urls,
+            "expires_at": datetime.now(timezone.utc) + timedelta(hours=24),
+        }
+    )
+
+    if verdict == "uncertain":
+        review_id = str(uuid.uuid4())
+        await review_repo.create_pending(
+            {
+                "review_id": review_id,
+                "screenshot_id": screenshot_id,
+                "device_id": device_id,
+                "timestamp": timestamp,
+                "status": "pending",
+                "reason": reason,
+                "ocr_urls": ocr_urls,
+                "created_at": datetime.now(timezone.utc),
+            }
+        )
+        await alert_hub.broadcast_admin(
+            {
+                "type": "game_lock_review_required",
+                "device_id": device_id,
+                "message": "Model uncertain. Admin confirmation is required before lock.",
+                "review_id": review_id,
+                "screenshot_id": screenshot_id,
+                "ocr_urls": ocr_urls,
+            }
+        )
+
+    await _cleanup_expired_violations()
+    return ScreenshotIngestResponse(screenshot_id=screenshot_id)
+
+
+@router.get("/violations", response_model=list[ViolationScreenshot])
+async def list_violations(user: AuthUser = Depends(get_current_web_user)) -> list[ViolationScreenshot]:
+    await _cleanup_expired_violations()
+    rows = await violation_repo.list_active(user.role)
+    return [
+        ViolationScreenshot(
+            screenshot_id=row.get("screenshot_id", ""),
+            device_id=row.get("device_id", ""),
+            timestamp=row.get("timestamp", ""),
+            filename=row.get("filename", ""),
+            created_at=row.get("created_at", ""),
+            image_url=row.get("image_url", ""),
+            confidence=float(row.get("confidence", 0.0)),
+            reason=row.get("reason", ""),
+            visibility_scope=row.get("visibility_scope", "admin"),
+            ocr_urls=[str(x) for x in row.get("ocr_urls", [])],
+        )
+        for row in rows
+    ]