Improvements to Document Interface and correcting embedding workers
This commit is contained in:
@@ -1,3 +1,4 @@
|
||||
import ast
|
||||
import os
|
||||
from datetime import datetime as dt, timezone as tz
|
||||
from flask import request, redirect, flash, render_template, Blueprint, session, current_app
|
||||
@@ -14,10 +15,11 @@ import io
|
||||
|
||||
from common.models.document import Document, DocumentLanguage, DocumentVersion
|
||||
from common.extensions import db
|
||||
from .document_forms import AddDocumentForm, AddURLForm
|
||||
from .document_forms import AddDocumentForm, AddURLForm, EditDocumentForm, EditDocumentLanguageForm
|
||||
from common.utils.middleware import mw_before_request
|
||||
from common.utils.celery_utils import current_celery
|
||||
from common.utils.nginx_utils import prefixed_url_for
|
||||
from common.utils.view_assistants import form_validation_failed, prepare_table_for_macro
|
||||
|
||||
document_bp = Blueprint('document_bp', __name__, url_prefix='/document')
|
||||
|
||||
@@ -29,13 +31,20 @@ def log_before_request():
|
||||
|
||||
@document_bp.after_request
|
||||
def log_after_request(response):
|
||||
current_app.logger.debug(f"After request (document_bp): {request.method} {request.url} - Status: {response.status}")
|
||||
current_app.logger.debug(
|
||||
f"After request (document_bp): {request.method} {request.url} - Status: {response.status}")
|
||||
return response
|
||||
|
||||
|
||||
@document_bp.before_request
|
||||
def before_request():
|
||||
mw_before_request()
|
||||
try:
|
||||
mw_before_request()
|
||||
except Exception as e:
|
||||
current_app.logger.error(f'Error switching schema in Document Blueprint: {e}')
|
||||
for role in current_user.roles:
|
||||
current_app.logger.debug(f'User {current_user.email} has role {role.name}')
|
||||
raise
|
||||
|
||||
|
||||
@document_bp.route('/add_document', methods=['GET', 'POST'])
|
||||
@@ -44,7 +53,7 @@ def add_document():
|
||||
form = AddDocumentForm()
|
||||
|
||||
# If the form is submitted
|
||||
if request.method == 'POST' and form.validate_on_submit():
|
||||
if form.validate_on_submit():
|
||||
current_app.logger.info(f'Adding document for tenant {session["tenant"]["id"]}')
|
||||
file = form.file.data
|
||||
filename = secure_filename(file.filename)
|
||||
@@ -59,8 +68,12 @@ def add_document():
|
||||
current_app.logger.info(f'Embedding creation started for tenant {session["tenant"]["id"]}, '
|
||||
f'Document Version {new_doc_vers.id}. '
|
||||
f'Embedding creation task: {task.id}')
|
||||
flash(f'Processing on document {new_doc.name}, version {new_doc_vers.id} started. Task ID: {task.id}.',
|
||||
'success')
|
||||
|
||||
return redirect(prefixed_url_for('document_bp.documents'))
|
||||
else:
|
||||
form_validation_failed(request, form)
|
||||
|
||||
return render_template('document/add_document.html', form=form)
|
||||
|
||||
@@ -71,7 +84,7 @@ def add_url():
|
||||
form = AddURLForm()
|
||||
|
||||
# If the form is submitted
|
||||
if request.method == 'POST' and form.validate_on_submit():
|
||||
if form.validate_on_submit():
|
||||
current_app.logger.info(f'Adding document for tenant {session["tenant"]["id"]}')
|
||||
url = form.url.data
|
||||
|
||||
@@ -96,8 +109,12 @@ def add_url():
|
||||
current_app.logger.info(f'Embedding creation started for tenant {session["tenant"]["id"]}, '
|
||||
f'Document Version {new_doc_vers.id}. '
|
||||
f'Embedding creation task: {task.id}')
|
||||
flash(f'Processing on document {new_doc.name}, version {new_doc_vers.id} started. Task ID: {task.id}.',
|
||||
'success')
|
||||
|
||||
return redirect(prefixed_url_for('document_bp.documents'))
|
||||
else:
|
||||
form_validation_failed(request, form)
|
||||
|
||||
return render_template('document/add_url.html', form=form)
|
||||
|
||||
@@ -108,25 +125,178 @@ def documents():
|
||||
page = request.args.get('page', 1, type=int)
|
||||
per_page = request.args.get('per_page', 10, type=int)
|
||||
|
||||
query = Document.query.order_by(desc(Document.created_at)).options(
|
||||
joinedload(Document.languages).joinedload(DocumentLanguage.versions))
|
||||
query = Document.query.order_by(desc(Document.created_at))
|
||||
|
||||
pagination = query.paginate(page=page, per_page=per_page, error_out=False)
|
||||
docs = pagination.items
|
||||
|
||||
rows = prepare_document_data(docs)
|
||||
rows = prepare_table_for_macro(docs, [('id', ''), ('name', ''), ('valid_from', ''), ('valid_to', '')])
|
||||
|
||||
return render_template('document/documents.html', rows=rows, pagination=pagination)
|
||||
|
||||
|
||||
@document_bp.route('/process_version/<int:version_id>', methods=['POST'])
|
||||
@document_bp.route('/handle_document_selection', methods=['POST'])
|
||||
@roles_accepted('Super User', 'Tenant Admin')
|
||||
def process_version(version_id):
|
||||
version = DocumentVersion.query.get_or_404(version_id)
|
||||
if not version.processing:
|
||||
print(f'Placeholder for processing version: {version_id}')
|
||||
def handle_document_selection():
|
||||
document_identification = request.form['selected_row']
|
||||
doc_id = ast.literal_eval(document_identification).get('value')
|
||||
|
||||
return redirect(prefixed_url_for('documents'))
|
||||
action = request.form['action']
|
||||
|
||||
match action:
|
||||
case 'edit_document':
|
||||
return redirect(prefixed_url_for('document_bp.edit_document', document_id=doc_id))
|
||||
case 'document_languages':
|
||||
return redirect(prefixed_url_for('document_bp.document_languages', document_id=doc_id))
|
||||
|
||||
# Add more conditions for other actions
|
||||
return redirect(prefixed_url_for('document_bp.documents'))
|
||||
|
||||
|
||||
@document_bp.route('/edit_document/<int:document_id>', methods=['GET', 'POST'])
|
||||
@roles_accepted('Super User', 'Tenant Admin')
|
||||
def edit_document(document_id):
|
||||
doc = Document.query.get_or_404(document_id)
|
||||
form = EditDocumentForm(obj=doc)
|
||||
|
||||
if form.validate_on_submit():
|
||||
doc.name = form.name.data
|
||||
doc.valid_from = form.valid_from.data
|
||||
doc.valid_to = form.valid_to.data
|
||||
|
||||
update_logging_information(doc, dt.now(tz.utc))
|
||||
|
||||
try:
|
||||
db.session.add(doc)
|
||||
db.session.commit()
|
||||
flash(f'Document {doc.id} updated successfully', 'success')
|
||||
except SQLAlchemyError as e:
|
||||
db.session.rollback()
|
||||
flash(f'Error updating document: {e}', 'danger')
|
||||
current_app.logger.error(f'Error updating document: {e}')
|
||||
else:
|
||||
form_validation_failed(request, form)
|
||||
|
||||
return render_template('document/edit_document.html', form=form, document_id=document_id)
|
||||
|
||||
|
||||
@document_bp.route('/document_languages/<int:document_id>', methods=['GET', 'POST'])
|
||||
@roles_accepted('Super User', 'Tenant Admin')
|
||||
def document_languages(document_id):
|
||||
doc = Document.query.get_or_404(document_id)
|
||||
doc_desc = f'Document {doc.id}: {doc.name}'
|
||||
page = request.args.get('page', 1, type=int)
|
||||
per_page = request.args.get('per_page', 10, type=int)
|
||||
|
||||
query = DocumentLanguage.query.filter_by(document_id=document_id).order_by(DocumentLanguage.language)
|
||||
|
||||
pagination = query.paginate(page=page, per_page=per_page, error_out=False)
|
||||
doc_langs = pagination.items
|
||||
|
||||
rows = prepare_table_for_macro(doc_langs, [('id', ''), ('language', ''), ('user_context', ''),
|
||||
('system_context', '')])
|
||||
|
||||
return render_template('document/document_languages.html', rows=rows, pagination=pagination, document=doc_desc)
|
||||
|
||||
|
||||
@document_bp.route('/handle_document_language_selection', methods=['POST'])
|
||||
@roles_accepted('Super User', 'Tenant Admin')
|
||||
def handle_document_language_selection():
|
||||
document_language_identification = request.form['selected_row']
|
||||
doc_lang_id = ast.literal_eval(document_language_identification).get('value')
|
||||
|
||||
action = request.form['action']
|
||||
|
||||
match action:
|
||||
case 'edit_document_language':
|
||||
return redirect(prefixed_url_for('document_bp.edit_document_language', document_language_id=doc_lang_id))
|
||||
case 'document_versions':
|
||||
return redirect(prefixed_url_for('document_bp.document_versions', document_language_id=doc_lang_id))
|
||||
|
||||
# Add more conditions for other actions
|
||||
return redirect(prefixed_url_for('document_bp.document_languages'))
|
||||
|
||||
|
||||
@document_bp.route('/edit_document_language/<int:document_language_id>', methods=['GET', 'POST'])
|
||||
@roles_accepted('Super User', 'Tenant Admin')
|
||||
def edit_document_language(document_language_id):
|
||||
doc_lang = DocumentLanguage.query.get_or_404(document_language_id)
|
||||
form = EditDocumentLanguageForm(obj=doc_lang)
|
||||
|
||||
if form.validate_on_submit():
|
||||
doc_lang.user_context = form.user_context.data
|
||||
|
||||
update_logging_information(doc_lang, dt.now(tz.utc))
|
||||
|
||||
try:
|
||||
db.session.add(doc_lang)
|
||||
db.session.commit()
|
||||
flash(f'Document Language {doc_lang.id} updated successfully', 'success')
|
||||
except SQLAlchemyError as e:
|
||||
db.session.rollback()
|
||||
flash(f'Error updating document language: {e}', 'danger')
|
||||
current_app.logger.error(f'Error updating document language {doc_lang.id} '
|
||||
f'for tenant {session['tenant']['id']}: {e}')
|
||||
else:
|
||||
form_validation_failed(request, form)
|
||||
|
||||
return render_template('document/edit_document_language.html', form=form, document_langauge_id=document_language_id,
|
||||
doc_details=f'Document {doc_lang.document.name}')
|
||||
|
||||
|
||||
@document_bp.route('/document_versions/<int:document_language_id>', methods=['GET', 'POST'])
|
||||
@roles_accepted('Super User', 'Tenant Admin')
|
||||
def document_versions(document_language_id):
|
||||
flash(f'Processing documents is a long running process. Please be careful retriggering processing!', 'danger')
|
||||
doc_lang = DocumentLanguage.query.get_or_404(document_language_id)
|
||||
doc_desc = f'Document {doc_lang.document.name}, Language {doc_lang.language}'
|
||||
|
||||
page = request.args.get('page', 1, type=int)
|
||||
per_page = request.args.get('per_page', 10, type=int)
|
||||
|
||||
query = DocumentVersion.query.filter_by(doc_lang_id=document_language_id).order_by(desc(DocumentVersion.id))
|
||||
|
||||
pagination = query.paginate(page=page, per_page=per_page, error_out=False)
|
||||
doc_langs = pagination.items
|
||||
|
||||
rows = prepare_table_for_macro(doc_langs, [('id', ''), ('url', ''), ('file_location', ''),
|
||||
('file_name', ''), ('file_type', ''),
|
||||
('processing', ''), ('processing_started_at', ''),
|
||||
('processing_finished_at', '')])
|
||||
|
||||
return render_template('document/document_versions.html', rows=rows, pagination=pagination, document=doc_desc)
|
||||
|
||||
|
||||
@document_bp.route('/handle_document_version_selection', methods=['POST'])
|
||||
@roles_accepted('Super User', 'Tenant Admin')
|
||||
def handle_document_version_selection():
|
||||
document_version_identification = request.form['selected_row']
|
||||
doc_vers_id = ast.literal_eval(document_version_identification).get('value')
|
||||
|
||||
action = request.form['action']
|
||||
|
||||
match action:
|
||||
case 'process_document_version':
|
||||
process_version(doc_vers_id)
|
||||
# Add more conditions for other actions
|
||||
|
||||
doc_vers = DocumentVersion.query.get_or_404(doc_vers_id)
|
||||
return redirect(prefixed_url_for('document_bp.document_versions', document_language_id=doc_vers.doc_lang_id))
|
||||
|
||||
|
||||
def process_version(version_id):
|
||||
task = current_celery.send_task('create_embeddings', queue='embeddings', args=[
|
||||
session['tenant']['id'],
|
||||
version_id,
|
||||
])
|
||||
current_app.logger.info(f'Embedding creation retriggered by user {current_user.id}, {current_user.email} '
|
||||
f'for tenant {session["tenant"]["id"]}, '
|
||||
f'Document Version {version_id}. '
|
||||
f'Embedding creation task: {task.id}')
|
||||
|
||||
flash(f'Processing for document version {version_id} retriggered successfully...', 'success')
|
||||
|
||||
return redirect(prefixed_url_for('document_bp.documents'))
|
||||
|
||||
|
||||
def set_logging_information(obj, timestamp):
|
||||
@@ -136,6 +306,11 @@ def set_logging_information(obj, timestamp):
|
||||
obj.updated_by = current_user.id
|
||||
|
||||
|
||||
def update_logging_information(obj, timestamp):
|
||||
obj.updated_at = timestamp
|
||||
obj.updated_by = current_user.id
|
||||
|
||||
|
||||
def create_document_stack(form, file, filename, extension):
|
||||
# Create the Document
|
||||
new_doc = create_document(form, filename)
|
||||
@@ -155,7 +330,7 @@ def create_document_stack(form, file, filename, extension):
|
||||
db.session.commit()
|
||||
except SQLAlchemyError as e:
|
||||
current_app.logger.error(f'Error adding document for tenant {session["tenant"]["id"]}: {e}')
|
||||
flash('Error adding document.', 'error')
|
||||
flash('Error adding document.', 'alert')
|
||||
db.session.rollback()
|
||||
error = e.args
|
||||
raise
|
||||
@@ -277,25 +452,6 @@ def fetch_html(url):
|
||||
return response.text
|
||||
|
||||
|
||||
# Sample code for adding or updating versions and ensuring latest_version is set in DocumentLanguage
|
||||
# def add_or_update_version(language_id, version_data):
|
||||
# new_version = Version(language_id=language_id, **version_data)
|
||||
# db.session.add(new_version)
|
||||
# db.session.flush() # Ensures new_version gets an ID assigned if it's new
|
||||
#
|
||||
# # Assuming we always call this when we know it's the latest
|
||||
# language = Language.query.get(language_id)
|
||||
# language.latest_version_id = new_version.id
|
||||
# db.session.commit()
|
||||
|
||||
# sample code for using latest_version in the application
|
||||
# @app.route('/language/<int:language_id>')
|
||||
# def show_language(language_id):
|
||||
# language = Language.query.get_or_404(language_id)
|
||||
# latest_version = language.latest_version # This is now a direct, efficient database access
|
||||
# return render_template('language_details.html', language=language, latest_version=latest_version)
|
||||
|
||||
|
||||
def prepare_document_data(docs):
|
||||
rows = []
|
||||
for doc in docs:
|
||||
@@ -340,4 +496,3 @@ def prepare_document_data(docs):
|
||||
'sub_rows': languages_rows})
|
||||
rows.append(doc_row)
|
||||
return rows
|
||||
|
||||
|
||||
Reference in New Issue
Block a user