feat: file upload + expense report creation from Discuss attachments

- Discuss bot now reads ir.attachment from incoming messages; file-only
  messages no longer silently dropped
- ZIP files are described (contents listed) and bot asks clarifying
  question before acting; user's follow-up reply looks back for pending
  attachments so files don't need to be re-uploaded
- receipt_parser: extracts text from ZIP (recursive), JPG/PNG/etc (OCR),
  PDF (pdfplumber), HTML, TXT
- expenses_agent: full rewrite fixing broken method signatures; adds
  create_expense_sheet / create_expense / attach_receipt flow driven by
  LLM receipt parsing (Ollama, HIPAA-locked)
- master_agent: extra_context threads receipts + user_id into directives
- FastAPI /upload multipart endpoint; registered in main.py
- Odoo /ai/upload controller proxies files to agent service
- ab_ai_bot: dispatch_message_with_files() for multipart uploads

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

addons/activeblue_ai/controllers/approval.py

@@ -3,7 +3,7 @@ import logging
 import requests
 
 from odoo import http
-from odoo.http import request
+from odoo.http import request, Response
 
 _logger = logging.getLogger(__name__)
 
@@ -58,3 +58,41 @@ class AiApprovalController(http.Controller):
             session_id=session_id,
         )
         return result
+
+    @http.route('/ai/upload', type='http', auth='user', methods=['POST'], csrf=False)
+    def upload(self, **kwargs):
+        bot = request.env['ab.ai.bot'].sudo().get_active_bot()
+        if not bot:
+            return Response(
+                json.dumps({'error': 'No bot configured'}),
+                content_type='application/json', status=503)
+
+        url = bot._get_service_url() + '/upload'
+        message = request.httprequest.form.get(
+            'message', 'Create an employee expense report from these receipts.')
+        session_id = request.httprequest.form.get('session_id', '')
+
+        files_data = [
+            ('files', (f.filename, f.read(), f.content_type or 'application/octet-stream'))
+            for f in request.httprequest.files.getlist('files')
+        ]
+
+        try:
+            resp = requests.post(
+                url,
+                data={
+                    'user_id': str(request.env.user.id),
+                    'message': message,
+                    'session_id': session_id,
+                },
+                files=files_data or [('files', ('empty', b'', 'application/octet-stream'))],
+                headers=bot._build_headers(),
+                timeout=120,
+            )
+            resp.raise_for_status()
+            return Response(resp.text, content_type='application/json')
+        except Exception as exc:
+            _logger.error('upload proxy failed: %s', exc)
+            return Response(
+                json.dumps({'error': str(exc), 'reply': 'Upload failed. Please try again.'}),
+                content_type='application/json', status=500)

addons/activeblue_ai/models/ab_ai_bot.py

@@ -111,6 +111,44 @@ class AbAiBot(models.Model):
             _logger.error('dispatch_message failed: %s', exc)
             raise UserError(_('Could not reach AI service: %s') % exc)
 
+    def dispatch_message_with_files(self, user_id, message, attachments, context=None, session_id=None):
+        """Send a message with file attachments to the /upload endpoint as multipart."""
+        self.ensure_one()
+        import base64
+        url = self._get_service_url() + '/upload'
+
+        files = []
+        for att in attachments:
+            try:
+                data = base64.b64decode(att.datas) if att.datas else b''
+                files.append(('files', (att.name or 'attachment',
+                                        data,
+                                        att.mimetype or 'application/octet-stream')))
+            except Exception as exc:
+                _logger.warning('Could not encode attachment %s: %s', att.id, exc)
+
+        # Omit Content-Type so requests sets the multipart boundary automatically
+        headers = {}
+        if self.webhook_secret:
+            headers['X-ActiveBlue-Signature'] = self.webhook_secret
+
+        form_data = {
+            'user_id': str(user_id),
+            'message': message or 'Create an employee expense report from these receipts.',
+            'session_id': session_id or '',
+        }
+
+        try:
+            resp = requests.post(url, data=form_data, files=files or [('files', ('empty', b'', 'text/plain'))],
+                                 headers=headers, timeout=120)
+            resp.raise_for_status()
+            return resp.json()
+        except requests.exceptions.Timeout:
+            raise UserError(_('AI service timed out. Please try again.'))
+        except requests.exceptions.RequestException as exc:
+            _logger.error('dispatch_message_with_files failed: %s', exc)
+            raise UserError(_('Could not reach AI service: %s') % exc)
+
     @api.model
     def cron_ping_all(self):
         any_online = False

addons/activeblue_ai/models/ab_ai_mail.py

@@ -1,6 +1,9 @@
 from __future__ import annotations
+import base64
+import io
 import logging
 import re
+import zipfile
 
 from odoo import models, api
 
@@ -8,11 +11,52 @@ _logger = logging.getLogger(__name__)
 
 _HTML_TAG = re.compile(r'<[^>]+>')
 
+# How many recent messages to scan when looking for a pending file upload
+_LOOKBACK_MESSAGES = 10
+
+# File type labels shown in the clarification message
+_EXT_LABELS = {
+    'jpg': 'image', 'jpeg': 'image', 'png': 'image', 'gif': 'image',
+    'bmp': 'image', 'tiff': 'image', 'tif': 'image', 'webp': 'image',
+    'pdf': 'PDF', 'html': 'HTML', 'htm': 'HTML',
+    'txt': 'text', 'csv': 'spreadsheet', 'xlsx': 'spreadsheet',
+    'zip': 'ZIP archive',
+}
+
 
 def _strip_html(html: str) -> str:
     return _HTML_TAG.sub(' ', html or '').strip()
 
 
+def _ext(filename: str) -> str:
+    return filename.rsplit('.', 1)[-1].lower() if '.' in filename else ''
+
+
+def _describe_zip(datas_b64: str, zip_name: str) -> str:
+    """Return a short HTML summary of a ZIP's contents without extracting data."""
+    try:
+        raw = base64.b64decode(datas_b64)
+        with zipfile.ZipFile(io.BytesIO(raw)) as zf:
+            members = [m for m in zf.namelist() if not m.endswith('/')]
+        if not members:
+            return f'<b>{zip_name}</b> (empty archive)'
+        # Count by type
+        counts: dict[str, int] = {}
+        for m in members:
+            label = _EXT_LABELS.get(_ext(m), 'file')
+            counts[label] = counts.get(label, 0) + 1
+        type_summary = ', '.join(f'{n} {t}(s)' for t, n in counts.items())
+        lines = [f'<b>{zip_name}</b> — {len(members)} item(s): {type_summary}']
+        for m in members[:8]:
+            lines.append(f'&nbsp;&nbsp;• {m}')
+        if len(members) > 8:
+            lines.append(f'&nbsp;&nbsp;… and {len(members) - 8} more')
+        return '<br/>'.join(lines)
+    except Exception as exc:
+        _logger.warning('_describe_zip failed for %s: %s', zip_name, exc)
+        return f'<b>{zip_name}</b> (could not inspect contents)'
+
+
 class DiscussChannel(models.Model):
     _inherit = 'discuss.channel'
 
@@ -40,10 +84,28 @@ class DiscussChannel(models.Model):
             return result
 
         text = _strip_html(body)
-        if not text:
+        attachments = result.attachment_ids
+
+        # Nothing to work with
+        if not text and not attachments:
             return result
 
-        # Identify the human sender
+        # ── Case 1: file(s) with no instruction ──────────────────────────────
+        # Show the user what we received and ask what to do with it.
+        if attachments and not text:
+            self._post_file_clarification(attachments, bot_partner)
+            return result
+
+        # ── Case 2: text only — look back for a pending file upload ──────────
+        # If the user just replied to our clarification question, find the
+        # attachment(s) they uploaded earlier in this conversation.
+        pending = self.env['ir.attachment'].browse()
+        if text and not attachments:
+            pending = self._find_pending_attachments(bot_partner)
+
+        effective_attachments = attachments or pending
+
+        # ── Case 3: text (+ possibly pending files) → dispatch ───────────────
         human_partner = member_partners.filtered(lambda p: p != bot_partner)[:1]
         user = self.env['res.users'].search([('partner_id', '=', human_partner.id)], limit=1)
         uid = user.id if user else self.env.uid
@@ -52,11 +114,21 @@ class DiscussChannel(models.Model):
             bot = self.env['ab.ai.bot'].sudo().search([('active', '=', True)], limit=1)
             if not bot:
                 return result
-            response = bot.dispatch_message(
-                user_id=uid,
-                message=text,
-                context={'channel_id': self.id, 'source': 'discuss'},
-            )
+
+            if effective_attachments:
+                response = bot.dispatch_message_with_files(
+                    user_id=uid,
+                    message=text,
+                    attachments=effective_attachments,
+                    context={'channel_id': self.id, 'source': 'discuss'},
+                )
+            else:
+                response = bot.dispatch_message(
+                    user_id=uid,
+                    message=text,
+                    context={'channel_id': self.id, 'source': 'discuss'},
+                )
+
             reply = (response or {}).get('reply') or (response or {}).get('message') or \
                     'I could not process your request right now.'
             self.sudo().message_post(
@@ -69,3 +141,52 @@ class DiscussChannel(models.Model):
             _logger.error('AI bot Discuss reply failed: %s', exc)
 
         return result
+
+    def _post_file_clarification(self, attachments, bot_partner):
+        """Describe the uploaded file(s) and ask the user what to do with them."""
+        lines = []
+        for att in attachments:
+            name = att.name or 'file'
+            ext = _ext(name)
+            if ext == 'zip' and att.datas:
+                lines.append(_describe_zip(att.datas, name))
+            else:
+                label = _EXT_LABELS.get(ext, 'file')
+                lines.append(f'<b>{name}</b> ({label})')
+
+        file_summary = '<br/>'.join(lines)
+        question = (
+            f'I received the following file(s):<br/>{file_summary}<br/><br/>'
+            'What would you like me to do with them? Some options:<br/>'
+            '&bull; <b>Create an expense report</b> from these receipts<br/>'
+            '&bull; <b>Import products</b> from this data<br/>'
+            '&bull; <b>Something else</b> — just tell me what you need'
+        )
+        self.sudo().message_post(
+            body=question,
+            author_id=bot_partner.id,
+            message_type='comment',
+            subtype_xmlid='mail.mt_comment',
+        )
+
+    def _find_pending_attachments(self, bot_partner):
+        """
+        Scan the last _LOOKBACK_MESSAGES messages in this channel for the most
+        recent human-sent message that has attachments.  Only returns them if
+        the immediately following bot message looks like a clarification question
+        (i.e. the bot hasn't already acted on those files).
+        """
+        messages = self.message_ids.sorted('date', reverse=True)[:_LOOKBACK_MESSAGES]
+        prev_was_bot_question = False
+        for msg in messages:
+            is_bot = msg.author_id == bot_partner
+            if is_bot:
+                # Check whether this bot message was a clarification question
+                if 'what would you like me to do' in (msg.body or '').lower():
+                    prev_was_bot_question = True
+                continue
+            # Human message
+            if msg.attachment_ids and prev_was_bot_question:
+                return msg.attachment_ids
+            prev_was_bot_question = False
+        return self.env['ir.attachment'].browse()