- Addition of general chunking parameters chunking_heading_level and chunking patterns

- Addition of Processor types docx and markdown

eveai_api/api/document_api.py

@@ -11,9 +11,10 @@ from common.utils.document_utils import (
     create_document_stack, process_url, start_embedding_task,
     validate_file_type, EveAIInvalidLanguageException, EveAIDoubleURLException, EveAIUnsupportedFileType,
     get_documents_list, edit_document, refresh_document, edit_document_version,
-    refresh_document_with_info
+    refresh_document_with_info, lookup_document
 )
 from common.utils.eveai_exceptions import EveAIException
+from eveai_api.api.auth import requires_service
 
 
 def validate_date(date_str):
@@ -59,6 +60,7 @@ add_document_response = document_ns.model('AddDocumentResponse', {
 @document_ns.route('/add_document')
 class AddDocument(Resource):
     @jwt_required()
+    @requires_service('DOCAPI')
     @document_ns.expect(upload_parser)
     @document_ns.response(201, 'Document added successfully', add_document_response)
     @document_ns.response(400, 'Validation Error')
@@ -134,6 +136,7 @@ add_url_response = document_ns.model('AddURLResponse', {
 @document_ns.route('/add_url')
 class AddURL(Resource):
     @jwt_required()
+    @requires_service('DOCAPI')
     @document_ns.expect(add_url_model)
     @document_ns.response(201, 'Document added successfully', add_url_response)
     @document_ns.response(400, 'Validation Error')
@@ -190,6 +193,7 @@ document_list_model = document_ns.model('DocumentList', {
 @document_ns.route('/list')
 class DocumentList(Resource):
     @jwt_required()
+    @requires_service('DOCAPI')
     @document_ns.doc('list_documents')
     @document_ns.marshal_list_with(document_list_model, envelope='documents')
     def get(self):
@@ -210,6 +214,7 @@ edit_document_model = document_ns.model('EditDocument', {
 @document_ns.route('/<int:document_id>')
 class DocumentResource(Resource):
     @jwt_required()
+    @requires_service('DOCAPI')
     @document_ns.doc('edit_document')
     @document_ns.expect(edit_document_model)
     @document_ns.response(200, 'Document updated successfully')
@@ -232,6 +237,7 @@ class DocumentResource(Resource):
             return e.to_dict(), e.status_code
 
     @jwt_required()
+    @requires_service('DOCAPI')
     @document_ns.doc('refresh_document')
     @document_ns.response(200, 'Document refreshed successfully')
     def post(self, document_id):
@@ -253,6 +259,7 @@ edit_document_version_model = document_ns.model('EditDocumentVersion', {
 @document_ns.route('/version/<int:version_id>')
 class DocumentVersionResource(Resource):
     @jwt_required()
+    @requires_service('DOCAPI')
     @document_ns.doc('edit_document_version')
     @document_ns.expect(edit_document_version_model)
     @document_ns.response(200, 'Document version updated successfully')
@@ -280,6 +287,7 @@ refresh_document_model = document_ns.model('RefreshDocument', {
 @document_ns.route('/<int:document_id>/refresh')
 class RefreshDocument(Resource):
     @jwt_required()
+    @requires_service('DOCAPI')
     @document_ns.response(200, 'Document refreshed successfully')
     @document_ns.response(404, 'Document not found')
     def post(self, document_id):
@@ -310,6 +318,7 @@ class RefreshDocument(Resource):
 @document_ns.route('/<int:document_id>/refresh_with_info')
 class RefreshDocumentWithInfo(Resource):
     @jwt_required()
+    @requires_service('DOCAPI')
     @document_ns.expect(refresh_document_model)
     @document_ns.response(200, 'Document refreshed successfully')
     @document_ns.response(400, 'Validation Error')
@@ -338,3 +347,112 @@ class RefreshDocumentWithInfo(Resource):
         except Exception as e:
             current_app.logger.error(f'Error refreshing document with info: {str(e)}')
             return {'message': 'Internal server error'}, 500
+
+
+# Define models for lookup requests
+lookup_model = document_ns.model('DocumentLookup', {
+    'lookup_criteria': fields.Raw(required=True,
+                                  description='JSON object containing key-value pairs to match in metadata. '
+                                              'Example: {"external_id": "123", "source": "zapier", "source_type": "google_docs"}'),
+    'metadata_type': fields.String(required=True, enum=['user_metadata', 'system_metadata'],
+                                   description='Which metadata field to search in')
+})
+
+lookup_response = document_ns.model('DocumentLookupResponse', {
+    'document_id': fields.Integer(description='ID of the found document'),
+    'document_version_id': fields.Integer(description='ID of the latest document version'),
+    'name': fields.String(description='Document name'),
+    'metadata': fields.Raw(description='Full metadata of the found document')
+})
+
+
+@document_ns.route('/lookup')
+class DocumentLookup(Resource):
+    @jwt_required()
+    @requires_service('DOCAPI')
+    @document_ns.expect(lookup_model)
+    @document_ns.marshal_with(lookup_response)
+    @document_ns.response(200, 'Document found', lookup_response)
+    @document_ns.response(404, 'No document found matching criteria')
+    def post(self):
+        """
+        Look up a document using metadata criteria
+        """
+        tenant_id = get_jwt_identity()
+        try:
+            data = request.json
+            document, version = lookup_document(
+                tenant_id,
+                data['lookup_criteria'],
+                data['metadata_type']
+            )
+
+            return {
+                'document_id': document.id,
+                'document_version_id': version.id,
+                'name': document.name,
+                'metadata': getattr(version, data['metadata_type'])
+            }
+
+        except EveAIException as e:
+            return e.to_dict(), e.status_code
+
+        except KeyError as e:
+            return {'message': f'Missing required field: {str(e)}'}, 400
+
+
+refresh_content_model = document_ns.model('RefreshDocumentContent', {
+    'file_content': fields.Raw(required=True, description='The new file content'),
+    'language': fields.String(required=False, description='Language of the document'),
+    'user_context': fields.String(required=False, description='User context for the document'),
+    'user_metadata': fields.Raw(required=False, description='Custom metadata fields'),
+    'catalog_properties': fields.Raw(required=False, description='Catalog-specific properties'),
+    'trigger_service': fields.String(required=False, description='Service that triggered the update')
+})
+
+
+@document_ns.route('/<int:document_id>/refresh_content')
+class RefreshDocumentContent(Resource):
+    @jwt_required()
+    @requires_service('DOCAPI')
+    @document_ns.expect(refresh_content_model)
+    @document_ns.response(200, 'Document refreshed successfully')
+    def post(self, document_id):
+        """Refresh a document with new content"""
+        tenant_id = get_jwt_identity()
+        try:
+            data = request.json
+            file_content = data['file_content']
+
+            # Build user_metadata by merging:
+            # 1. Existing metadata (if any)
+            # 2. New metadata from request
+            # 3. Zapier-specific fields
+            user_metadata = data.get('user_metadata', {})
+            user_metadata.update({
+                'source': 'zapier',
+                'trigger_service': data.get('trigger_service')
+            })
+            data['user_metadata'] = user_metadata
+
+            # Keep catalog_properties separate
+            if 'catalog_properties' in data:
+                # We could add validation here against catalog configuration
+                data['catalog_properties'] = data['catalog_properties']
+
+            new_version, task_id = refresh_document_with_content(
+                document_id,
+                tenant_id,
+                file_content,
+                data
+            )
+
+            return {
+                'message': f'Document refreshed successfully. New version: {new_version.id}. Task ID: {task_id}',
+                'document_id': document_id,
+                'document_version_id': new_version.id,
+                'task_id': task_id
+            }, 200
+
+        except EveAIException as e:
+            return e.to_dict(), e.status_code