- Full API application, streamlined, de-duplication of document handling code into document_utils.py

- Added meta-data fields to DocumentVersion
- Docker container to support API
This commit is contained in:
Josako
2024-09-09 16:11:42 +02:00
parent 341ba47d1c
commit 76cb825660
39 changed files with 598 additions and 6177 deletions

View File

@@ -1,31 +1,72 @@
from flask_restful import Resource, reqparse
from flask import current_app
import json
from datetime import datetime
import pytz
from flask import current_app, request
from flask_restx import Namespace, Resource, fields, reqparse
from flask_jwt_extended import jwt_required, get_jwt_identity
from werkzeug.datastructures import FileStorage
from werkzeug.utils import secure_filename
from common.utils.document_utils import (
create_document_stack, process_url, start_embedding_task,
validate_file_type, EveAIInvalidLanguageException, EveAIDoubleURLException, EveAIUnsupportedFileType,
process_multiple_urls, prepare_youtube_document
process_multiple_urls, get_documents_list, edit_document, refresh_document, edit_document_version
)
from common.utils.eveai_exceptions import EveAIYoutubeError
class AddDocumentResource(Resource):
def validate_date(date_str):
try:
return datetime.fromisoformat(date_str).replace(tzinfo=pytz.UTC)
except ValueError:
raise ValueError("Invalid date format. Use ISO format (YYYY-MM-DDTHH:MM:SS).")
def validate_json(json_str):
try:
return json.loads(json_str)
except json.JSONDecodeError:
raise ValueError("Invalid JSON format for user_metadata.")
document_ns = Namespace('documents', description='Document related operations')
# Define models for request parsing and response serialization
upload_parser = reqparse.RequestParser()
upload_parser.add_argument('file', location='files', type=FileStorage, required=True, help='The file to upload')
upload_parser.add_argument('name', location='form', type=str, required=False, help='Name of the document')
upload_parser.add_argument('language', location='form', type=str, required=True, help='Language of the document')
upload_parser.add_argument('user_context', location='form', type=str, required=False,
help='User context for the document')
upload_parser.add_argument('valid_from', location='form', type=validate_date, required=False,
help='Valid from date for the document (ISO format)')
upload_parser.add_argument('user_metadata', location='form', type=validate_json, required=False,
help='User metadata for the document (JSON format)')
add_document_response = document_ns.model('AddDocumentResponse', {
'message': fields.String(description='Status message'),
'document_id': fields.Integer(description='ID of the created document'),
'document_version_id': fields.Integer(description='ID of the created document version'),
'task_id': fields.String(description='ID of the embedding task')
})
@document_ns.route('/add_document')
class AddDocument(Resource):
@jwt_required()
@document_ns.expect(upload_parser)
@document_ns.response(201, 'Document added successfully', add_document_response)
@document_ns.response(400, 'Validation Error')
@document_ns.response(500, 'Internal Server Error')
def post(self):
parser = reqparse.RequestParser()
parser.add_argument('file', type=FileStorage, location='files', required=True)
parser.add_argument('name', type=str, required=False)
parser.add_argument('language', type=str, required=True)
parser.add_argument('user_context', type=str, required=False)
parser.add_argument('valid_from', type=str, required=False)
args = parser.parse_args()
"""
Add a new document
"""
tenant_id = get_jwt_identity()
current_app.logger.info(f'Adding document for tenant {tenant_id}')
try:
args = upload_parser.parse_args()
file = args['file']
filename = secure_filename(file.filename)
extension = filename.rsplit('.', 1)[1].lower()
@@ -33,10 +74,11 @@ class AddDocumentResource(Resource):
validate_file_type(extension)
api_input = {
'name': args['name'] or filename,
'language': args['language'],
'user_context': args['user_context'],
'valid_from': args['valid_from']
'name': args.get('name') or filename,
'language': args.get('language'),
'user_context': args.get('user_context'),
'valid_from': args.get('valid_from'),
'user_metadata': args.get('user_metadata'),
}
new_doc, new_doc_vers = create_document_stack(api_input, file, filename, extension, tenant_id)
@@ -50,35 +92,57 @@ class AddDocumentResource(Resource):
}, 201
except (EveAIInvalidLanguageException, EveAIUnsupportedFileType) as e:
return {'message': str(e)}, 400
current_app.logger.error(f'Error adding document: {str(e)}')
document_ns.abort(400, str(e))
except Exception as e:
current_app.logger.error(f'Error adding document: {str(e)}')
return {'message': 'Error adding document'}, 500
document_ns.abort(500, 'Error adding document')
class AddURLResource(Resource):
# Models for AddURL
add_url_model = document_ns.model('AddURL', {
'url': fields.String(required=True, description='URL of the document to add'),
'name': fields.String(required=False, description='Name of the document'),
'language': fields.String(required=True, description='Language of the document'),
'user_context': fields.String(required=False, description='User context for the document'),
'valid_from': fields.String(required=False, description='Valid from date for the document'),
'user_metadata': fields.String(required=False, description='User metadata for the document'),
'system_metadata': fields.String(required=False, description='System metadata for the document')
})
add_url_response = document_ns.model('AddURLResponse', {
'message': fields.String(description='Status message'),
'document_id': fields.Integer(description='ID of the created document'),
'document_version_id': fields.Integer(description='ID of the created document version'),
'task_id': fields.String(description='ID of the embedding task')
})
@document_ns.route('/add_url')
class AddURL(Resource):
@jwt_required()
@document_ns.expect(add_url_model)
@document_ns.response(201, 'Document added successfully', add_url_response)
@document_ns.response(400, 'Validation Error')
@document_ns.response(500, 'Internal Server Error')
def post(self):
parser = reqparse.RequestParser()
parser.add_argument('url', type=str, required=True)
parser.add_argument('name', type=str, required=False)
parser.add_argument('language', type=str, required=True)
parser.add_argument('user_context', type=str, required=False)
parser.add_argument('valid_from', type=str, required=False)
args = parser.parse_args()
"""
Add a new document from URL
"""
tenant_id = get_jwt_identity()
current_app.logger.info(f'Adding document from URL for tenant {tenant_id}')
try:
args = document_ns.payload
file_content, filename, extension = process_url(args['url'], tenant_id)
api_input = {
'url': args['url'],
'name': args['name'] or filename,
'name': args.get('name') or filename,
'language': args['language'],
'user_context': args['user_context'],
'valid_from': args['valid_from']
'user_context': args.get('user_context'),
'valid_from': args.get('valid_from'),
'user_metadata': args.get('user_metadata'),
}
new_doc, new_doc_vers = create_document_stack(api_input, file_content, filename, extension, tenant_id)
@@ -92,87 +156,85 @@ class AddURLResource(Resource):
}, 201
except EveAIDoubleURLException:
return {'message': f'A document with URL {args["url"]} already exists.'}, 400
document_ns.abort(400, f'A document with URL {args["url"]} already exists.')
except (EveAIInvalidLanguageException, EveAIUnsupportedFileType) as e:
return {'message': str(e)}, 400
document_ns.abort(400, str(e))
except Exception as e:
current_app.logger.error(f'Error adding document from URL: {str(e)}')
return {'message': 'Error adding document from URL'}, 500
document_ns.abort(500, 'Error adding document from URL')
class AddMultipleURLsResource(Resource):
document_list_model = document_ns.model('DocumentList', {
'id': fields.Integer(description='Document ID'),
'name': fields.String(description='Document name'),
'valid_from': fields.DateTime(description='Valid from date'),
'valid_to': fields.DateTime(description='Valid to date'),
})
@document_ns.route('/list')
class DocumentList(Resource):
@jwt_required()
def post(self):
parser = reqparse.RequestParser()
parser.add_argument('urls', type=str, action='append', required=True)
parser.add_argument('name', type=str, required=False)
parser.add_argument('language', type=str, required=True)
parser.add_argument('user_context', type=str, required=False)
parser.add_argument('valid_from', type=str, required=False)
args = parser.parse_args()
tenant_id = get_jwt_identity()
current_app.logger.info(f'Adding multiple documents from URLs for tenant {tenant_id}')
try:
api_input = {
'name': args['name'],
'language': args['language'],
'user_context': args['user_context'],
'valid_from': args['valid_from']
}
results = process_multiple_urls(args['urls'], tenant_id, api_input)
return {
'message': 'Processing of multiple URLs completed',
'results': results
}, 201
except Exception as e:
current_app.logger.error(f'Error adding documents from URLs: {str(e)}')
return {'message': 'Error adding documents from URLs'}, 500
@document_ns.doc('list_documents')
@document_ns.marshal_list_with(document_list_model, envelope='documents')
def get(self):
"""List all documents"""
page = request.args.get('page', 1, type=int)
per_page = request.args.get('per_page', 10, type=int)
pagination = get_documents_list(page, per_page)
return pagination.items, 200
class AddYoutubeResource(Resource):
edit_document_model = document_ns.model('EditDocument', {
'name': fields.String(required=True, description='New name for the document'),
'valid_from': fields.DateTime(required=False, description='New valid from date'),
'valid_to': fields.DateTime(required=False, description='New valid to date'),
})
@document_ns.route('/<int:document_id>')
class DocumentResource(Resource):
@jwt_required()
def post(self):
parser = reqparse.RequestParser()
parser.add_argument('url', type=str, required=True)
parser.add_argument('name', type=str, required=False)
parser.add_argument('language', type=str, required=True)
parser.add_argument('user_context', type=str, required=False)
parser.add_argument('valid_from', type=str, required=False)
args = parser.parse_args()
@document_ns.doc('edit_document')
@document_ns.expect(edit_document_model)
@document_ns.response(200, 'Document updated successfully')
def put(self, document_id):
"""Edit a document"""
data = request.json
updated_doc, error = edit_document(document_id, data['name'], data.get('valid_from'), data.get('valid_to'))
if updated_doc:
return {'message': f'Document {updated_doc.id} updated successfully'}, 200
else:
return {'message': f'Error updating document: {error}'}, 400
tenant_id = get_jwt_identity()
current_app.logger.info(f'Adding YouTube document for tenant {tenant_id}')
try:
api_input = {
'name': args['name'],
'language': args['language'],
'user_context': args['user_context'],
'valid_from': args['valid_from']
}
new_doc, new_doc_vers = prepare_youtube_document(args['url'], tenant_id, api_input)
task_id = start_embedding_task(tenant_id, new_doc_vers.id)
return {
'message': f'Processing on YouTube document {new_doc.name}, version {new_doc_vers.id} started. Task ID: {task_id}.',
'document_id': new_doc.id,
'document_version_id': new_doc_vers.id,
'task_id': task_id
}, 201
except EveAIYoutubeError as e:
return {'message': str(e)}, 400
except (EveAIInvalidLanguageException, EveAIUnsupportedFileType) as e:
return {'message': str(e)}, 400
except Exception as e:
current_app.logger.error(f'Error adding YouTube document: {str(e)}')
return {'message': 'Error adding YouTube document'}, 500
@jwt_required()
@document_ns.doc('refresh_document')
@document_ns.response(200, 'Document refreshed successfully')
def post(self, document_id):
"""Refresh a document"""
new_version, result = refresh_document(document_id)
if new_version:
return {'message': f'Document refreshed. New version: {new_version.id}. Task ID: {result}'}, 200
else:
return {'message': f'Error refreshing document: {result}'}, 400
# You can add more API resources here as needed
edit_document_version_model = document_ns.model('EditDocumentVersion', {
'user_context': fields.String(required=True, description='New user context for the document version'),
})
@document_ns.route('/version/<int:version_id>')
class DocumentVersionResource(Resource):
@jwt_required()
@document_ns.doc('edit_document_version')
@document_ns.expect(edit_document_version_model)
@document_ns.response(200, 'Document version updated successfully')
def put(self, version_id):
"""Edit a document version"""
data = request.json
updated_version, error = edit_document_version(version_id, data['user_context'])
if updated_version:
return {'message': f'Document Version {updated_version.id} updated successfully'}, 200
else:
return {'message': f'Error updating document version: {error}'}, 400