feat(bot,wizard): smart message filtering, encrypted standalone wizard, and file persistence

ai procurement/app/api/client.py

@@ -135,3 +135,16 @@ class ErpClient:
         return (await self._request(
             "PATCH", f"/quotes/{quote_id}/reject", json={"reason": reason} if reason else None, actor_id=actor_id
         )).get("data") or {}
+
+    async def upload_quote_file(self, quote_id: str, kind: str, file_name: str, content: bytes,
+                                mime_type: str = "application/pdf", actor_id: str | None = None) -> dict:
+        url = f"{self._base}/quotes/{quote_id}/{kind}"
+        files = {"file": (file_name, content, mime_type)}
+        async with httpx.AsyncClient(timeout=self._timeout) as client:
+            try:
+                resp = await client.post(url, files=files, headers=self._actor_headers(actor_id, None))
+                if resp.status_code in (200, 201):
+                    return resp.json()
+            except Exception as e:
+                logger.warning("Failed to upload quote attachment to ERP: %s", e)
+        return {}

ai procurement/app/config.py

@@ -31,6 +31,13 @@ class Settings(BaseSettings):
     # Frontend URL (for opening Wizard from Bale bot)
     FRONTEND_URL: str = "http://localhost:8081"
 
+    # Security token for standalone wizard review URL
+    DRAFT_TOKEN_SECRET: str = "procurement-draft-secret-change-in-prod-v7"
+    DRAFT_TOKEN_EXPIRE_SECONDS: int = 7200  # 2 hours
+
+    # Persistent storage directory for draft attachments (proformas)
+    ATTACHMENTS_DIR: str = "data/attachments/proformas"
+
     # OpenAI-compatible AI provider (LLM), e.g. OpenAI / GLM / DeepSeek / OpenRouter.
     AI_BASE_URL: str = ""  # e.g. https://api.openai.com/v1
     AI_API_KEY: str = ""

ai procurement/app/handlers/__init__.py

@@ -14,6 +14,7 @@ from app import drafts as draft_store
 from app.config import settings
 from app.nlu import understand
 from app.runtime import BotContext, LinkedUser
+from app.storage import save_proforma_file
 from app.stt import transcribe, transcribe_multi
 from app.texts import CANCELLED, LINK_START, NOTHING_TO_COMMIT, STT_FAILED, UNKNOWN_HINT, VOICE_EMPTY, esc
 
@@ -38,35 +39,15 @@ async def handle_update(ctx: BotContext, update: dict) -> None:
 
 
 async def handle_message(ctx: BotContext, chat_id: int, message: dict) -> None:
-    text = (message.get("text") or "").strip()
+    text = (message.get("text") or message.get("caption") or "").strip()
     source = "text"
     voice_transcripts: list[dict] | None = None
+    message_id = message.get("message_id")
 
     if text.startswith("/"):
         await commands.handle_command(ctx, chat_id, text)
         return
 
-    if not text and (message.get("voice") or message.get("audio") or message.get("document")):
-        source = "voice"
-        await ctx.bale.send_chat_action(chat_id, "typing")
-        file_id = (message.get("voice") or message.get("audio") or message.get("document"))["file_id"]
-        try:
-            audio = await ctx.bale.download_file(file_id)
-            voice_transcripts = await transcribe_multi(
-                settings,
-                audio,
-                mime_type=(message.get("voice") or {}).get("mime_type"),
-                filename=(message.get("voice") or {}).get("file_name"),
-            )
-        except Exception as exc:
-            detail = "هوش مصنوعی تنظیم نشده" if "not configured" in str(exc) else str(exc)[:200]
-            await ctx.send(chat_id, STT_FAILED.format(detail=esc(detail)))
-            return
-        if not voice_transcripts:
-            await ctx.send(chat_id, VOICE_EMPTY)
-            return
-        text = voice_transcripts[0]["text"]
-
     # Pending multi-step interactions take the message first.
     flow = ctx.pending.get(chat_id)
     if flow:
@@ -79,11 +60,43 @@ async def handle_message(ctx: BotContext, chat_id: int, message: dict) -> None:
     if user is None:
         return  # gate either started linking or is mid-linking
 
-    message_id = message.get("message_id")
     reply_to = message.get("reply_to_message")
     reply_to_msg_id = reply_to.get("message_id") if reply_to else None
 
-    target_draft = None
+    def _is_audio(msg: dict) -> bool:
+        if msg.get("voice") or msg.get("audio"):
+            return True
+        doc = msg.get("document")
+        if doc and (doc.get("mime_type") or "").lower().startswith("audio/"):
+            return True
+        return False
+
+    async def _extract_voice() -> list[dict] | None:
+        nonlocal source
+        source = "voice"
+        await ctx.bale.send_chat_action(chat_id, "typing")
+        audio_media = message.get("voice") or message.get("audio") or message.get("document")
+        file_id = audio_media["file_id"]
+        try:
+            audio_bytes = await ctx.bale.download_file(file_id)
+            transcripts = await transcribe_multi(
+                settings,
+                audio_bytes,
+                mime_type=audio_media.get("mime_type"),
+                filename=audio_media.get("file_name"),
+            )
+        except Exception as exc:
+            detail = "هوش مصنوعی تنظیم نشده" if "not configured" in str(exc) else str(exc)[:200]
+            await ctx.send(chat_id, STT_FAILED.format(detail=esc(detail)), reply_to=message_id)
+            return None
+        if not transcripts:
+            await ctx.send(chat_id, VOICE_EMPTY, reply_to=message_id)
+            return None
+        return transcripts
+
+    # =========================================================================
+    # BRANCH 1: EDIT MODE — Strictly requires reply_to_message_id on a draft
+    # =========================================================================
     if reply_to_msg_id is not None:
         target_draft = await draft_store.get_by_message_id(chat_id, reply_to_msg_id)
         if target_draft is None:
@@ -94,6 +107,92 @@ async def handle_message(ctx: BotContext, chat_id: int, message: dict) -> None:
             )
             return
 
+        # Check for proforma attachment (document or photo)
+        proforma_attached = False
+        saved_meta = None
+        doc = message.get("document")
+        photos = message.get("photo")
+
+        if doc and not (doc.get("mime_type") or "").lower().startswith("audio/"):
+            doc_file_id = doc["file_id"]
+            doc_file_name = doc.get("file_name") or f"proforma_{message_id}.pdf"
+            doc_mime = doc.get("mime_type") or "application/pdf"
+        elif photos:
+            largest = photos[-1] if isinstance(photos, list) else photos
+            doc_file_id = largest["file_id"]
+            doc_file_name = f"proforma_{message_id}.jpg"
+            doc_mime = "image/jpeg"
+        else:
+            doc_file_id = None
+            doc_file_name = None
+            doc_mime = None
+
+        if doc_file_id:
+            await ctx.bale.send_chat_action(chat_id, "upload_document")
+            try:
+                content = await ctx.bale.download_file(doc_file_id)
+                saved_meta = await save_proforma_file(
+                    target_draft.id,
+                    doc_file_name,
+                    content,
+                    mime_type=doc_mime,
+                    storage_dir=settings.ATTACHMENTS_DIR,
+                )
+                payload = dict(target_draft.payload or {})
+                payload["proforma_file"] = saved_meta["file_name"]
+                payload["proforma"] = saved_meta
+                target_draft.payload = payload
+                await draft_store.save_payload(target_draft, mark_bale_synced=True)
+                proforma_attached = True
+            except Exception as exc:
+                logger.exception("Failed to persist proforma")
+                await ctx.send(chat_id, f"❌ خطا در ذخیره فایل پیش‌فاکتور: {esc(str(exc)[:100])}", reply_to=message_id)
+                return
+
+        if _is_audio(message):
+            voice_transcripts = await _extract_voice()
+            if not voice_transcripts:
+                return
+            text = voice_transcripts[0]["text"]
+
+        if proforma_attached and not text and not voice_transcripts:
+            await quote.render(ctx, chat_id, target_draft, fresh=False)
+            await ctx.send(
+                chat_id,
+                f"📎 پیش‌فاکتور «{saved_meta['file_name']}» با موفقیت ذخیره و به استعلام ضمیمه شد ✅",
+                reply_to=message_id,
+            )
+            return
+
+        if not proforma_attached and not text and not voice_transcripts:
+            await ctx.send(
+                chat_id,
+                "⚠️ برای ویرایش پیش‌نویس، لطفاً تغییرات را به صورت متن یا پیام صوتی (ویس) ارسال کنید، یا فایل پیش‌فاکتور را بفرستید.",
+                reply_to=message_id,
+            )
+            return
+
+    # =========================================================================
+    # BRANCH 2: NEW QUOTE MODE — Messages without reply
+    # =========================================================================
+    else:
+        target_draft = None
+
+        if _is_audio(message):
+            voice_transcripts = await _extract_voice()
+            if not voice_transcripts:
+                return
+            text = voice_transcripts[0]["text"]
+        elif text:
+            source = "text"
+        else:
+            await ctx.send(
+                chat_id,
+                "⚠️ برای ثبت استعلام خرید جدید، لطفاً اقلام و مشخصات را به صورت پیام متنی یا پیام صوتی (ویس) ارسال فرمایید.",
+                reply_to=message_id,
+            )
+            return
+
     draft_summary = _draft_summary(target_draft) if target_draft else None
     input_payload = voice_transcripts if voice_transcripts else text
     result = await understand(settings, input_payload, draft_summary)

ai procurement/app/handlers/quote.py

@@ -198,9 +198,17 @@ async def render(ctx, chat_id, draft, fresh: bool = True, note: str | None = Non
                  reply_to: int | None = None) -> None:
     from app.config import settings
 
+    from app.security import create_draft_token
+
     payload = draft.payload
     missing = compute_quote_missing(payload)
-    wizard_url = f"{settings.FRONTEND_URL}/wizard?draft={draft.id}"
+    token = create_draft_token(
+        draft.id,
+        chat_id=chat_id,
+        secret=settings.DRAFT_TOKEN_SECRET,
+        expires_in=settings.DRAFT_TOKEN_EXPIRE_SECONDS,
+    )
+    wizard_url = f"{settings.FRONTEND_URL.rstrip('/')}/review?token={token}"
     text = quote_review_text(payload, missing, status_note=note)
     text += f"\n\n🖥 مشاهده و تکمیل در ویزارد:\n{wizard_url}"
     keyboard = _keyboard(payload, str(draft.id), wizard_url=wizard_url)
@@ -302,9 +310,29 @@ async def commit(ctx, chat_id, user, draft, only_resolved: bool = False) -> None
     ctx.pending.pop(chat_id, None)
     ctx.resolver.invalidate()
 
+    quote_id = (created or {}).get("id") or ""
+    # If a proforma file was attached, persist it to the created ERP quote
+    proforma_info = payload.get("proforma")
+    if quote_id and proforma_info and proforma_info.get("saved_path"):
+        try:
+            from pathlib import Path
+            p_path = Path(proforma_info["saved_path"])
+            if p_path.exists():
+                file_bytes = p_path.read_bytes()
+                await ctx.api.upload_quote_file(
+                    quote_id,
+                    "proforma",
+                    proforma_info.get("file_name") or "proforma.pdf",
+                    file_bytes,
+                    mime_type=proforma_info.get("mime_type") or "application/pdf",
+                    actor_id=user.actor_id,
+                )
+                logger.info("Successfully uploaded proforma to quote %s", quote_id)
+        except Exception as e:
+            logger.warning("Could not upload draft proforma to quote %s: %s", quote_id, e)
+
     if draft.review_message_id:
         try:
-            quote_id = (created or {}).get("id") or ""
             msg = f"✅ این پیشنهاد در سیستم با موفقیت ثبت نهایی شد.\nشناسه پیشنهاد: {quote_id}"
             await ctx.edit(chat_id, draft.review_message_id, msg, reply_markup=None)
             await draft_store.set_bale_updated_at(draft.id)

ai procurement/app/security.py

@@ -0,0 +1,72 @@
+"""Security and token generation for standalone draft review."""
+import base64
+import hashlib
+import hmac
+import json
+import time
+from typing import Any
+from uuid import UUID
+
+
+def _b64url_encode(data: bytes) -> str:
+    return base64.urlsafe_b64encode(data).rstrip(b"=").decode("ascii")
+
+
+def _b64url_decode(s: str) -> bytes:
+    padding = 4 - (len(s) % 4)
+    if padding != 4:
+        s += "=" * padding
+    return base64.urlsafe_b64decode(s.encode("ascii"))
+
+
+def create_draft_token(
+    draft_id: str | UUID,
+    chat_id: int | None = None,
+    user_id: str | None = None,
+    secret: str = "procurement-draft-secret-change-in-prod-v7",
+    expires_in: int = 7200,
+) -> str:
+    """Generate an HMAC-SHA256 signed token for secure draft access."""
+    now = int(time.time())
+    payload = {
+        "draft_id": str(draft_id),
+        "chat_id": chat_id,
+        "user_id": str(user_id) if user_id else None,
+        "iat": now,
+        "exp": now + expires_in,
+    }
+    payload_bytes = json.dumps(payload, separators=(",", ":"), ensure_ascii=False).encode("utf-8")
+    payload_b64 = _b64url_encode(payload_bytes)
+
+    sig = hmac.new(secret.encode("utf-8"), payload_b64.encode("ascii"), hashlib.sha256).digest()
+    sig_b64 = _b64url_encode(sig)
+
+    return f"{payload_b64}.{sig_b64}"
+
+
+def verify_draft_token(
+    token: str,
+    secret: str = "procurement-draft-secret-change-in-prod-v7",
+) -> dict[str, Any] | None:
+    """Verify signature and expiration of a draft token. Return payload or None."""
+    if not token or "." not in token:
+        return None
+    try:
+        payload_b64, sig_b64 = token.split(".", 1)
+        expected_sig = hmac.new(secret.encode("utf-8"), payload_b64.encode("ascii"), hashlib.sha256).digest()
+        actual_sig = _b64url_decode(sig_b64)
+
+        if not hmac.compare_digest(expected_sig, actual_sig):
+            return None
+
+        payload_bytes = _b64url_decode(payload_b64)
+        payload = json.loads(payload_bytes.decode("utf-8"))
+
+        # Check expiration
+        exp = payload.get("exp")
+        if exp is not None and int(exp) < int(time.time()):
+            return None
+
+        return payload
+    except Exception:
+        return None

ai procurement/app/storage.py

@@ -0,0 +1,53 @@
+"""Persistent storage for draft attachments (proforma files)."""
+from datetime import datetime, timezone
+import logging
+import os
+from pathlib import Path
+import re
+from uuid import UUID, uuid4
+
+logger = logging.getLogger("bot.storage")
+
+
+def sanitize_filename(filename: str) -> str:
+    """Sanitize filename preserving Persian/Arabic/English letters, numbers, and extensions."""
+    raw = (filename or "proforma.pdf").strip()
+    name = Path(raw).name
+    # Remove control characters and path traversal
+    clean = re.sub(r'[\/\\:\*\?"<>\|\r\n\t]', "_", name)
+    clean = clean.strip(". ")
+    return clean or f"proforma_{uuid4().hex[:8]}.pdf"
+
+
+async def save_proforma_file(
+    draft_id: str | UUID,
+    file_name: str,
+    content: bytes,
+    mime_type: str | None = None,
+    storage_dir: str = "data/attachments/proformas",
+) -> dict:
+    """Save proforma bytes permanently to disk.
+
+    Returns dict metadata for persisting in draft payload.
+    """
+    clean_name = sanitize_filename(file_name)
+    target_dir = Path(storage_dir) / str(draft_id)
+    target_dir.mkdir(parents=True, exist_ok=True)
+
+    unique_prefix = uuid4().hex[:8]
+    saved_filename = f"{unique_prefix}_{clean_name}"
+    target_path = target_dir / saved_filename
+
+    # Write bytes to disk
+    target_path.write_bytes(content)
+    logger.info("Persisted proforma (%d bytes) to %s", len(content), target_path)
+
+    return {
+        "file_name": clean_name,
+        "saved_filename": saved_filename,
+        "saved_path": str(target_path.resolve()),
+        "relative_path": str(target_path.as_posix()),
+        "size_bytes": len(content),
+        "mime_type": mime_type or "application/octet-stream",
+        "saved_at": datetime.now(timezone.utc).isoformat(),
+    }