notebook

EmailsImport/DockerCustomApp/app.py

@@ -1,8 +1,9 @@
-# app.py | v1.7 | 2026-06-05
+# app.py | v1.9 | 2026-06-08
 # FastAPI server pro příjem .msg a .db souborů, upload do Dropboxu a import do Graph API.
 # Endpointy: /upload (.msg → /msgs + Graph import), /upload-db (.db → /msgs/db),
 #            /upload-dropbox (→ Dropbox /!!!Days/Downloads Z230),
 #            /message-delete, /message-update (sync: smazání, přečtení, přesun složky),
+#            /mirror-plan (diff manifestu z JNJ vůči schránce → smaže přebytky, vrátí to_add),
 #            /pending-files (seznam souborů k odeslání na JNJ), /download-file/{filename}.
 
 from fastapi import FastAPI, UploadFile, File, Form, Header, HTTPException, Response
@@ -151,6 +152,55 @@ def _map_jnj_folder(folder: str) -> list[str]:
     return prefix + rest if rest else prefix
 
 
+def _norm_mid(mid: str) -> str:
+    """Normalizuj Internet Message-ID pro porovnání (osekej <> a whitespace)."""
+    return (mid or "").strip().strip("<>").strip()
+
+
+def _enumerate_jnj_mailbox(cutoff_iso: str) -> dict[str, str]:
+    """Vrať {normalizované internetMessageId: graph_id} pro všechny zprávy ve
+    složkách JNJ/* schránky, které mají receivedDateTime >= cutoff_iso.
+
+    Slouží jako 'co už ve schránce je' pro mirror diff. Starší zprávy než cutoff
+    (např. únorový archiv) se nenačtou — mirror se jich tedy nikdy nedotkne.
+    """
+    jnj_id = _ensure_folder([GRAPH_ROOT_FOLDER])
+
+    # BFS přes JNJ root + všechny podsložky
+    all_folders = [jnj_id]
+    i = 0
+    while i < len(all_folders):
+        fid = all_folders[i]
+        i += 1
+        url = f"{GRAPH_URL}/users/{GRAPH_MAILBOX}/mailFolders/{fid}/childFolders?$top=100"
+        while url:
+            r = _retry_graph(http_requests.get, url, _graph_headers, timeout=20)
+            data = r.json()
+            for f in data.get("value", []):
+                all_folders.append(f["id"])
+            url = data.get("@odata.nextLink")
+
+    # Posbírej message-id z každé složky (filtrováno na okno)
+    result: dict[str, str] = {}
+    cutoff_enc = cutoff_iso.replace(":", "%3A")
+    for fid in all_folders:
+        url = (
+            f"{GRAPH_URL}/users/{GRAPH_MAILBOX}/mailFolders/{fid}/messages"
+            f"?$filter=receivedDateTime ge {cutoff_enc}"
+            f"&$select=id,internetMessageId&$top=200"
+        )
+        while url:
+            r = _retry_graph(http_requests.get, url, _graph_headers, timeout=30)
+            data = r.json()
+            for m in data.get("value", []):
+                mid = _norm_mid(m.get("internetMessageId", ""))
+                if mid:
+                    result[mid] = m["id"]
+            url = data.get("@odata.nextLink")
+
+    return result
+
+
 def _make_recipient(addr: str) -> dict:
     if "<" in addr and ">" in addr:
         name = addr[: addr.index("<")].strip().strip('"')
@@ -221,6 +271,20 @@ def _import_msg_to_graph(msg_path: Path, folder: str) -> Optional[str]:
         folder_parts = _map_jnj_folder(folder)
         folder_id = _ensure_folder(folder_parts)
 
+        ext_props = [{"id": "Integer 0x0E07", "value": "1"}]
+
+        if date_raw:
+            try:
+                dt = dtparser.parse(str(date_raw))
+                dt_str = dt.astimezone(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ")
+                # PR_MESSAGE_DELIVERY_TIME (0x0E06) — jediný způsob jak nastavit
+                # receivedDateTime přes Graph API (přímé pole je read-only)
+                ext_props.append({"id": "SystemTime 0x0E06", "value": dt_str})
+            except Exception:
+                dt_str = None
+        else:
+            dt_str = None
+
         payload = {
             "subject": subject,
             "body": {
@@ -231,19 +295,11 @@ def _import_msg_to_graph(msg_path: Path, folder: str) -> Optional[str]:
             "toRecipients": [_make_recipient(a) for a in to_list],
             "ccRecipients": [_make_recipient(a) for a in cc_list],
             "isRead": True,
-            "singleValueExtendedProperties": [
-                {"id": "Integer 0x0E07", "value": "1"}
-            ],
+            "singleValueExtendedProperties": ext_props,
         }
 
-        if date_raw:
-            try:
-                dt = dtparser.parse(str(date_raw))
-                payload["receivedDateTime"] = dt.astimezone(timezone.utc).strftime(
-                    "%Y-%m-%dT%H:%M:%SZ"
-                )
-            except Exception:
-                pass
+        if dt_str:
+            payload["sentDateTime"] = dt_str
 
         if att_list:
             payload["attachments"] = att_list
@@ -393,6 +449,56 @@ async def message_update(req: MessageUpdateRequest, authorization: str = Header(
     return result
 
 
+class MirrorPlanRequest(BaseModel):
+    manifest: list[dict]   # [{"message_id": ..., "folder": ..., "is_read": ...}]
+    cutoff: str            # ISO8601 UTC, např. "2026-05-09T00:00:00Z"
+
+
+@app.post("/mirror-plan")
+async def mirror_plan(req: MirrorPlanRequest, authorization: str = Header(None)):
+    """Porovná manifest zpráv z JNJ (posledních 30 dní) se stavem schránky.
+
+    - smaže ze schránky zprávy které v manifestu nejsou (smazané v JNJ / vypadlé z okna)
+    - vrátí to_add = message_id které ve schránce chybí (klient je pak nahraje na /upload)
+
+    Maže POUZE v rámci okna (cutoff) — starší archiv zůstává nedotčen.
+    """
+    if authorization != f"Bearer {TOKEN}":
+        raise HTTPException(status_code=401, detail="Unauthorized")
+
+    # manifest: normalizované id → původní message_id (pro echo zpět klientovi)
+    manifest_map: dict[str, str] = {}
+    for e in req.manifest:
+        mid = _norm_mid(e.get("message_id", ""))
+        if mid:
+            manifest_map[mid] = e["message_id"]
+
+    mailbox = _enumerate_jnj_mailbox(req.cutoff)  # {norm_mid: graph_id}
+
+    to_add = [orig for nmid, orig in manifest_map.items() if nmid not in mailbox]
+    to_delete = [(nmid, gid) for nmid, gid in mailbox.items() if nmid not in manifest_map]
+
+    deleted = 0
+    for nmid, gid in to_delete:
+        url = f"{GRAPH_URL}/users/{GRAPH_MAILBOX}/messages/{gid}"
+        r = _retry_graph(http_requests.delete, url, _graph_headers, timeout=15)
+        if r.status_code in (200, 204):
+            deleted += 1
+        else:
+            log.error("mirror delete FAIL [%d]: %s", r.status_code, r.text[:150])
+
+    log.info(
+        "mirror-plan: manifest=%d mailbox=%d → add=%d delete=%d",
+        len(manifest_map), len(mailbox), len(to_add), deleted,
+    )
+    return {
+        "to_add": to_add,
+        "deleted": deleted,
+        "manifest_count": len(manifest_map),
+        "mailbox_count": len(mailbox),
+    }
+
+
 @app.post("/upload-file")
 async def upload_file(
     file: UploadFile = File(...),