Compare commits

...

108 Commits

Author SHA1 Message Date
Patrick Jentsch
d4cd313940 Implement /admin using flask-admin. Overall cleanup 2024-12-16 15:37:19 +01:00
Patrick Jentsch
c405061574 Restructure corpora blueprint 2024-12-16 11:39:54 +01:00
Patrick Jentsch
6c1f48eb2f Update corpora package 2024-12-16 10:09:54 +01:00
Patrick Jentsch
cda28910f5 Update settings 2024-12-16 10:07:21 +01:00
Patrick Jentsch
9a805b9d14 Add value to submit button 2024-12-16 09:57:18 +01:00
Patrick Jentsch
16bf891654 fix wtf macros 2024-12-16 09:45:19 +01:00
Patrick Jentsch
cb53b27ebf Rename function 2024-12-12 15:55:39 +01:00
Patrick Jentsch
6684257bc4 added job inputs/results blueprints 2024-12-12 15:32:56 +01:00
Patrick Jentsch
0d1805fb76 Update Job Blueprint package 2024-12-12 15:26:01 +01:00
Patrick Jentsch
bb60a2ba67 Move jobs namespace back to http routes 2024-12-12 10:32:08 +01:00
Patrick Jentsch
328f85ba52 Implement corpora endpoint/socket.io namespace 2024-12-09 16:12:49 +01:00
Patrick Jentsch
93344c9573 Add job namespace and remove old json_routes logic 2024-12-06 11:41:36 +01:00
Patrick Jentsch
1372c86609 Add api and administration links to navigations 2024-12-06 10:14:45 +01:00
Patrick Jentsch
713a7645db Bump nopaque version 2024-12-05 15:34:11 +01:00
Patrick Jentsch
0c64c07925 Update corpus analysis loading modal 2024-12-05 15:33:15 +01:00
Patrick Jentsch
a6ddf4c980 Remove import corpus button 2024-12-05 15:12:53 +01:00
Patrick Jentsch
cab5f7ea05 More js enhancements 2024-12-05 15:07:13 +01:00
Patrick Jentsch
07f09cdbd9 fix cqi_over_socketio 2024-12-05 15:07:03 +01:00
Patrick Jentsch
c97b2a886e Further js refactoring 2024-12-05 14:26:05 +01:00
Patrick Jentsch
df2bffe0fd implement first version of jobs socketio namespace 2024-12-03 16:09:14 +01:00
Patrick Jentsch
aafb3ca3ec Update javascript app structure 2024-12-03 15:59:08 +01:00
Patrick Jentsch
12a3ac1d5d Update JS code structure 2024-12-02 09:34:17 +01:00
Patrick Jentsch
a2904caea2 Update cqpserver image version 2024-11-28 10:02:27 +01:00
Patrick Jentsch
e325552100 Update corpus analysis tabs to look the same as before base template update 2024-11-28 10:02:00 +01:00
Patrick Jentsch
e269156925 fix socketio emits from database event listeners 2024-11-27 15:46:54 +01:00
Patrick Jentsch
9c9de242ca Remove unsed css 2024-11-27 11:35:51 +01:00
Patrick Jentsch
ec54fdc3bb Restore service scheme on pages 2024-11-27 11:34:21 +01:00
Patrick Jentsch
2263a8d27d codestyle enhancements in base template 2024-11-21 11:22:57 +01:00
Patrick Jentsch
143cdd91f9 update workspace settings 2024-11-21 11:22:46 +01:00
Patrick Jentsch
b5f7478e14 Update templates 2024-11-21 11:12:11 +01:00
Patrick Jentsch
a95b8d979d Fix forms 2024-11-20 15:56:48 +01:00
Patrick Jentsch
18d5ab160e Optimize jinja wtf macros 2024-11-20 15:56:29 +01:00
Patrick Jentsch
7439edacef Add background color to job list entries 2024-11-20 15:55:59 +01:00
Patrick Jentsch
99d7a8bdfc Some fixes and improve jinja2 template performance by reducing include statements 2024-11-19 15:28:43 +01:00
Patrick Jentsch
54c4295bf7 Fixes and more descriptions 2024-11-18 13:32:55 +01:00
Patrick Jentsch
1e5c26b8e3 Reorganize Socket.IO code 2024-11-18 12:36:37 +01:00
Patrick Jentsch
9f56647cf7 highlight active items in top navbar 2024-11-18 12:35:53 +01:00
Patrick Jentsch
460257294d Use relative import for sub blueprints 2024-11-18 11:08:28 +01:00
Patrick Jentsch
2c43333c94 Check tos accepted in registration form 2024-11-18 11:03:29 +01:00
Patrick Jentsch
fc8b11fa66 update auth package 2024-11-15 16:07:29 +01:00
Patrick Jentsch
a8ab1bee71 Move some blueprints and rename routes 2024-11-15 15:59:08 +01:00
Patrick Jentsch
ee7f64f5be Design update 2024-11-15 15:21:26 +01:00
Patrick Jentsch
6aacac2419 flatten the contributions blueprint 2024-11-14 14:36:18 +01:00
Patrick Jentsch
ce253f4a65 Make the header span over the complete width 2024-11-13 16:08:18 +01:00
Patrick Jentsch
7b604ce4f2 Remove manual-modal references 2024-11-11 14:51:17 +01:00
Patrick Jentsch
98b20e5cab Remove colors from social area 2024-11-11 13:38:47 +01:00
Patrick Jentsch
a322ffb2f1 Fix README 2024-11-11 12:05:03 +01:00
Patrick Jentsch
29365984a3 fix some namespace responses 2024-11-11 08:45:16 +01:00
Patrick Jentsch
bd0a9c60f8 strictly use socket.io class based namespaces 2024-11-07 12:12:42 +01:00
Patrick Jentsch
d41ebc6efe Fix project vscode settings 2024-11-07 10:51:35 +01:00
Patrick Jentsch
63690222ed Rename cqi extensions file 2024-11-07 10:44:27 +01:00
Patrick Jentsch
b4faa1c695 Code enhancements in vrt file normalizer module 2024-11-07 10:40:25 +01:00
Patrick Jentsch
909b130285 Fix wrong import 2024-11-07 09:48:40 +01:00
Patrick Jentsch
c223f07289 Codestyle enhancements 2024-11-07 08:57:32 +01:00
Patrick Jentsch
fcb49025e9 remove unused socketio event handlers 2024-11-07 08:51:49 +01:00
Patrick Jentsch
191d7813a7 prefix extension name with "nopaque_" 2024-11-07 08:35:02 +01:00
Patrick Jentsch
f255fef631 Remove debug print statement 2024-11-07 08:32:20 +01:00
Patrick Jentsch
76171f306d Remove debug print statements 2024-11-07 08:31:52 +01:00
Patrick Jentsch
5ea6d45f46 Reset all corpora on deploy cli command 2024-11-07 08:31:31 +01:00
Patrick Jentsch
289a551122 Create dedicated '/users' Socket.IO Namespace 2024-11-06 13:04:30 +01:00
Patrick Jentsch
2a28f19660 Move Socket.IO Namespaces to dedicated directory 2024-11-06 12:27:49 +01:00
Patrick Jentsch
fc2ace4b9e Remove unused Socket.IO AdminNamespace 2024-11-05 14:55:48 +01:00
Patrick Jentsch
a174bf968f Remove unused config entry 2024-11-05 14:02:45 +01:00
Patrick Jentsch
551b928dca Add typehints to email code 2024-11-05 09:05:31 +01:00
Patrick Jentsch
eeb5a280b3 move blueprints in dedicated folder 2024-09-30 13:30:13 +02:00
Patrick Jentsch
5fc3015bf1 rename functions to indicate that they should not be imported directly 2024-09-26 15:34:52 +02:00
Patrick Jentsch
5f05cedf5e Make the "daemon" (now tasks) more understandable 2024-09-26 15:33:32 +02:00
Patrick Jentsch
aabea234fe More simplification 2024-09-26 14:45:05 +02:00
Patrick Jentsch
492fdc9d28 modernize type hinting 2024-09-25 17:46:53 +02:00
Patrick Jentsch
02e6c7c16c various updates 2024-09-25 12:08:20 +02:00
Patrick Jentsch
c7ca674b2f Streamline setup process and init code 2024-09-25 10:45:53 +02:00
Patrick Jentsch
81c6f32a35 Simplify logging configuration 2024-08-01 16:29:06 +02:00
Patrick Jentsch
94548ac30c Move sheduler start logic 2024-08-01 12:10:33 +02:00
Patrick Jentsch
158190de1a Codesstyle enhancements 2024-08-01 12:00:52 +02:00
Patrick Jentsch
13e4d461c7 Update .env.tpl 2024-08-01 12:00:34 +02:00
Patrick Jentsch
e51dcafa6f Update vscode settings.json 2024-08-01 11:59:50 +02:00
Patrick Jentsch
f79c6d48b2 Going back to vanilla css 2024-07-01 15:37:34 +02:00
Patrick Jentsch
5ee9edef9f Fix multiple db event listener registrations 2024-06-03 11:08:21 +02:00
Patrick Jentsch
f1ccda6ad7 Fix colors in corpus analysis 2024-06-03 11:03:57 +02:00
Patrick Jentsch
a65b1ff578 remove manual modal 2024-05-27 16:58:51 +02:00
Patrick Jentsch
fe0fcb0e10 fix job status notifications 2024-05-27 09:24:40 +02:00
Patrick Jentsch
32fa632961 move anonymous user to seperate file 2024-05-27 09:23:08 +02:00
Patrick Jentsch
562b8d5ce0 Move clearfix to helpers.scss 2024-05-27 08:58:04 +02:00
Patrick Jentsch
cbd0a41bce style and compatibility update 2024-05-21 10:29:12 +02:00
Patrick Jentsch
c68286e010 more flexible navbar code in base template 2024-05-08 14:01:01 +02:00
Patrick Jentsch
4a29a52f2a Remove implementation of FileSize validator 2024-05-04 17:17:06 +02:00
Patrick Jentsch
991810cff5 Use HTML comments in navbar.html.j2 2024-05-04 17:04:25 +02:00
Patrick Jentsch
6025a4a606 add API back 2024-05-04 15:14:21 +02:00
Patrick Jentsch
e1cfd394fa move socketio decorators 2024-05-04 14:55:05 +02:00
Patrick Jentsch
882987ba68 fixes 2024-05-04 14:41:40 +02:00
Patrick Jentsch
a03b5918d9 More Streamlining 2024-05-03 15:08:57 +02:00
Patrick Jentsch
43b38b2216 better large screen handling 2024-04-30 16:21:23 +02:00
Patrick Jentsch
543276d766 A lot of generalization for better scaling and overview 2024-04-30 16:00:06 +02:00
Patrick Jentsch
485a0155c6 Bump dependencies. Some parts needed to be deactivated for that. They need to be reimplemented.
- breadcrumbs (!flask-breadcrumbs)
- manual modal button
- api blueprint (!flask-marshmallow/!marshmallow-sqlalchemy)
2024-04-30 08:41:29 +02:00
Patrick Jentsch
c29c50feb9 new event system first draft 2024-04-18 15:37:17 +02:00
Patrick Jentsch
c191e7bd4a rename extension extras 2024-04-18 15:35:41 +02:00
Patrick Jentsch
8f960cf359 explicitly set permissions to false for anonymous users 2024-04-11 15:46:58 +02:00
Patrick Jentsch
ccf484c9bc make is_administrator a property, add back db events 2024-04-11 14:33:47 +02:00
Patrick Jentsch
d0d2a8abd6 Move external static content into external directory 2024-04-11 14:13:04 +02:00
Patrick Jentsch
03876f6a39 Self host external css/fonts/js 2024-04-11 11:08:37 +02:00
Patrick Jentsch
cdf6f9fcfd declutter by using default filenames 2024-04-10 13:52:33 +02:00
Patrick Jentsch
268da220d2 Enhance code structure 2024-04-10 13:34:48 +02:00
Patrick Jentsch
84e1755a57 Remove version declaration in docker compose files. 2024-04-10 13:34:14 +02:00
Patrick Jentsch
82d6f6003f Restructure Dockerfile for better caching 2024-03-13 12:58:39 +01:00
Patrick Jentsch
9da74c1c6f Use pathlib where possible 2024-03-07 15:49:04 +01:00
Patrick Jentsch
ec23bd94ee add missing import 2024-03-06 14:46:25 +01:00
Patrick Jentsch
55a62053b0 Make models in package work 2024-03-06 13:20:02 +01:00
Patrick Jentsch
a1e5bd61e0 move models in seperate modules 2024-03-05 16:02:23 +01:00
492 changed files with 1575956 additions and 6992 deletions

View File

@ -5,9 +5,9 @@
!app !app
!migrations !migrations
!tests !tests
!.flaskenv
!boot.sh !boot.sh
!config.py !config.py
!docker-nopaque-entrypoint.sh !docker-nopaque-entrypoint.sh
!nopaque.py
!requirements.txt !requirements.txt
!requirements.freezed.txt
!wsgi.py

View File

@ -1,32 +1,20 @@
############################################################################## ##############################################################################
# Variables for use in Docker Compose YAML files # # Environment variables used by Docker Compose config files. #
############################################################################## ##############################################################################
# HINT: Use this bash command `id -u` # HINT: Use this bash command `id -u`
# NOTE: 0 (= root user) is not allowed # NOTE: 0 (= root user) is not allowed
HOST_UID= HOST_UID=
# HINT: Use this bash command `id -g` # HINT: Use this bash command `id -g`
# NOTE: 0 (= root group) is not allowed
HOST_GID= HOST_GID=
# HINT: Use this bash command `getent group docker | cut -d: -f3` # HINT: Use this bash command `getent group docker | cut -d: -f3`
HOST_DOCKER_GID= HOST_DOCKER_GID=
# DEFAULT: nopaque # DEFAULT: nopaque
# DOCKER_DEFAULT_NETWORK_NAME= NOPAQUE_DOCKER_NETWORK_NAME=nopaque
# DEFAULT: ./volumes/db/data
# NOTE: Use `.` as <project-basedir>
# DOCKER_DB_SERVICE_DATA_VOLUME_SOURCE_PATH=
# DEFAULT: ./volumes/mq/data
# NOTE: Use `.` as <project-basedir>
# DOCKER_MQ_SERVICE_DATA_VOLUME_SOURCE_PATH=
# NOTE: This must be a network share and it must be available on all # NOTE: This must be a network share and it must be available on all
# Docker Swarm nodes, mounted to the same path with the same # Docker Swarm nodes, mounted to the same path.
# user and group ownership. HOST_NOPAQUE_DATA_PATH=/mnt/nopaque
DOCKER_NOPAQUE_SERVICE_DATA_VOLUME_SOURCE_PATH=
# DEFAULT: ./volumes/nopaque/logs
# NOTE: Use `.` as <project-basedir>
# DOCKER_NOPAQUE_SERVICE_LOGS_VOLUME_SOURCE_PATH=.

View File

@ -1 +0,0 @@
FLASK_APP=nopaque.py

2
.gitignore vendored
View File

@ -2,8 +2,6 @@
app/static/gen/ app/static/gen/
volumes/ volumes/
docker-compose.override.yml docker-compose.override.yml
logs/
!logs/dummy
*.env *.env
*.pjentsch-testing *.pjentsch-testing

17
.vscode/settings.json vendored
View File

@ -1,9 +1,17 @@
{ {
"editor.rulers": [79], "editor.rulers": [79],
"files.insertFinalNewline": true, "editor.tabSize": 4,
"[css]": { "emmet.includeLanguages": {
"editor.tabSize": 2 "jinja-html": "html"
}, },
"files.associations": {
".flaskenv": "env",
"*.env.tpl": "env",
"*.txt.j2": "jinja"
},
"files.insertFinalNewline": true,
"files.trimFinalNewlines": true,
"files.trimTrailingWhitespace": true,
"[html]": { "[html]": {
"editor.tabSize": 2 "editor.tabSize": 2
}, },
@ -12,8 +20,5 @@
}, },
"[jinja-html]": { "[jinja-html]": {
"editor.tabSize": 2 "editor.tabSize": 2
},
"[scss]": {
"editor.tabSize": 2
} }
} }

View File

@ -4,11 +4,13 @@ FROM python:3.10.13-slim-bookworm
LABEL authors="Patrick Jentsch <p.jentsch@uni-bielefeld.de>" LABEL authors="Patrick Jentsch <p.jentsch@uni-bielefeld.de>"
# Set environment variables
ENV LANG="C.UTF-8" ENV LANG="C.UTF-8"
ENV PYTHONDONTWRITEBYTECODE="1" ENV PYTHONDONTWRITEBYTECODE="1"
ENV PYTHONUNBUFFERED="1" ENV PYTHONUNBUFFERED="1"
# Install system dependencies
RUN apt-get update \ RUN apt-get update \
&& apt-get install --no-install-recommends --yes \ && apt-get install --no-install-recommends --yes \
build-essential \ build-essential \
@ -17,37 +19,39 @@ RUN apt-get update \
&& rm --recursive /var/lib/apt/lists/* && rm --recursive /var/lib/apt/lists/*
# Create a non-root user
RUN useradd --create-home --no-log-init nopaque \ RUN useradd --create-home --no-log-init nopaque \
&& groupadd docker \ && groupadd docker \
&& usermod --append --groups docker nopaque && usermod --append --groups docker nopaque
USER nopaque USER nopaque
WORKDIR /home/nopaque WORKDIR /home/nopaque
# Create a Python virtual environment
ENV NOPAQUE_PYTHON3_VENV_PATH="/home/nopaque/.venv" ENV NOPAQUE_PYTHON3_VENV_PATH="/home/nopaque/.venv"
RUN python3 -m venv "${NOPAQUE_PYTHON3_VENV_PATH}" RUN python3 -m venv "${NOPAQUE_PYTHON3_VENV_PATH}"
ENV PATH="${NOPAQUE_PYTHON3_VENV_PATH}/bin:${PATH}" ENV PATH="${NOPAQUE_PYTHON3_VENV_PATH}/bin:${PATH}"
# Install Python dependencies
COPY --chown=nopaque:nopaque requirements.freezed.txt requirements.freezed.txt
RUN python3 -m pip install --requirement requirements.freezed.txt \
&& rm requirements.freezed.txt
# Install the application
COPY docker-nopaque-entrypoint.sh /usr/local/bin/
COPY --chown=nopaque:nopaque app app COPY --chown=nopaque:nopaque app app
COPY --chown=nopaque:nopaque migrations migrations COPY --chown=nopaque:nopaque migrations migrations
COPY --chown=nopaque:nopaque tests tests COPY --chown=nopaque:nopaque tests tests
COPY --chown=nopaque:nopaque .flaskenv boot.sh config.py nopaque.py requirements.txt ./ COPY --chown=nopaque:nopaque boot.sh config.py wsgi.py ./
RUN python3 -m pip install --requirement requirements.txt \
&& mkdir logs
USER root
COPY docker-nopaque-entrypoint.sh /usr/local/bin/
EXPOSE 5000 EXPOSE 5000
USER root
ENTRYPOINT ["docker-nopaque-entrypoint.sh"] ENTRYPOINT ["docker-nopaque-entrypoint.sh"]

View File

@ -35,7 +35,7 @@ username@hostname:~$ sudo mount --types cifs --options gid=${USER},password=nopa
# Clone the nopaque repository # Clone the nopaque repository
username@hostname:~$ git clone https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque.git username@hostname:~$ git clone https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque.git
# Create data directories # Create data directories
username@hostname:~$ mkdir data/{db,logs,mq} username@hostname:~$ mkdir -p volumes/{db,mq}
username@hostname:~$ cp db.env.tpl db.env username@hostname:~$ cp db.env.tpl db.env
username@hostname:~$ cp .env.tpl .env username@hostname:~$ cp .env.tpl .env
# Fill out the variables within these files. # Fill out the variables within these files.

View File

@ -2,9 +2,10 @@ from apifairy import APIFairy
from config import Config from config import Config
from docker import DockerClient from docker import DockerClient
from flask import Flask from flask import Flask
from flask.logging import default_handler
from flask_admin import Admin
from flask_apscheduler import APScheduler from flask_apscheduler import APScheduler
from flask_assets import Environment from flask_assets import Environment
from flask_breadcrumbs import Breadcrumbs, default_breadcrumb_root
from flask_login import LoginManager from flask_login import LoginManager
from flask_mail import Mail from flask_mail import Mail
from flask_marshmallow import Marshmallow from flask_marshmallow import Marshmallow
@ -13,95 +14,143 @@ from flask_paranoid import Paranoid
from flask_socketio import SocketIO from flask_socketio import SocketIO
from flask_sqlalchemy import SQLAlchemy from flask_sqlalchemy import SQLAlchemy
from flask_hashids import Hashids from flask_hashids import Hashids
from logging import Formatter, StreamHandler
from werkzeug.middleware.proxy_fix import ProxyFix
from .extensions.nopaque_flask_admin_views import AdminIndexView, ModelView
docker_client = DockerClient.from_env()
admin = Admin()
apifairy = APIFairy() apifairy = APIFairy()
assets = Environment() assets = Environment()
breadcrumbs = Breadcrumbs()
db = SQLAlchemy() db = SQLAlchemy()
docker_client = DockerClient()
hashids = Hashids() hashids = Hashids()
login = LoginManager() login = LoginManager()
login.login_view = 'auth.login'
login.login_message = 'Please log in to access this page.'
ma = Marshmallow() ma = Marshmallow()
mail = Mail() mail = Mail()
migrate = Migrate(compare_type=True) migrate = Migrate(compare_type=True)
paranoid = Paranoid() paranoid = Paranoid()
paranoid.redirect_view = '/'
scheduler = APScheduler() scheduler = APScheduler()
socketio = SocketIO() socketio = SocketIO()
def create_app(config: Config = Config) -> Flask: def create_app(config: Config = Config) -> Flask:
''' Creates an initialized Flask (WSGI Application) object. ''' ''' Creates an initialized Flask object. '''
app = Flask(__name__) app = Flask(__name__)
app.config.from_object(config) app.config.from_object(config)
config.init_app(app)
# region Logging
log_formatter = Formatter(
fmt=app.config['NOPAQUE_LOG_FORMAT'],
datefmt=app.config['NOPAQUE_LOG_DATE_FORMAT']
)
log_handler = StreamHandler()
log_handler.setFormatter(log_formatter)
log_handler.setLevel(app.config['NOPAQUE_LOG_LEVEL'])
app.logger.setLevel('DEBUG')
app.logger.removeHandler(default_handler)
app.logger.addHandler(log_handler)
# endregion Logging
# region Middlewares
if app.config['NOPAQUE_PROXY_FIX_ENABLED']:
app.wsgi_app = ProxyFix(
app.wsgi_app,
x_for=app.config['NOPAQUE_PROXY_FIX_X_FOR'],
x_host=app.config['NOPAQUE_PROXY_FIX_X_HOST'],
x_port=app.config['NOPAQUE_PROXY_FIX_X_PORT'],
x_prefix=app.config['NOPAQUE_PROXY_FIX_X_PREFIX'],
x_proto=app.config['NOPAQUE_PROXY_FIX_X_PROTO']
)
# endregion Middlewares
# region Extensions
docker_client.login( docker_client.login(
app.config['NOPAQUE_DOCKER_REGISTRY_USERNAME'], app.config['NOPAQUE_DOCKER_REGISTRY_USERNAME'],
password=app.config['NOPAQUE_DOCKER_REGISTRY_PASSWORD'], password=app.config['NOPAQUE_DOCKER_REGISTRY_PASSWORD'],
registry=app.config['NOPAQUE_DOCKER_REGISTRY'] registry=app.config['NOPAQUE_DOCKER_REGISTRY']
) )
from .models import AnonymousUser, User
admin.init_app(app, index_view=AdminIndexView())
apifairy.init_app(app) apifairy.init_app(app)
assets.init_app(app) assets.init_app(app)
breadcrumbs.init_app(app)
db.init_app(app) db.init_app(app)
hashids.init_app(app) hashids.init_app(app)
login.init_app(app) login.init_app(app)
login.anonymous_user = AnonymousUser
login.login_view = 'auth.login'
login.user_loader(lambda user_id: User.query.get(int(user_id)))
ma.init_app(app) ma.init_app(app)
mail.init_app(app) mail.init_app(app)
migrate.init_app(app, db) migrate.init_app(app, db)
paranoid.init_app(app) paranoid.init_app(app)
paranoid.redirect_view = '/'
scheduler.init_app(app) scheduler.init_app(app)
socketio.init_app(app, message_queue=app.config['NOPAQUE_SOCKETIO_MESSAGE_QUEUE_URI']) # noqa socketio.init_app(app, message_queue=app.config['NOPAQUE_SOCKETIO_MESSAGE_QUEUE_URI'])
# endregion Extensions
from .admin import bp as admin_blueprint # region Blueprints
default_breadcrumb_root(admin_blueprint, '.admin') from .blueprints.api import bp as api_blueprint
app.register_blueprint(admin_blueprint, url_prefix='/admin')
from .api import bp as api_blueprint
app.register_blueprint(api_blueprint, url_prefix='/api') app.register_blueprint(api_blueprint, url_prefix='/api')
from .auth import bp as auth_blueprint from .blueprints.auth import bp as auth_blueprint
default_breadcrumb_root(auth_blueprint, '.')
app.register_blueprint(auth_blueprint) app.register_blueprint(auth_blueprint)
from .contributions import bp as contributions_blueprint from .blueprints.contributions import bp as contributions_blueprint
default_breadcrumb_root(contributions_blueprint, '.contributions')
app.register_blueprint(contributions_blueprint, url_prefix='/contributions') app.register_blueprint(contributions_blueprint, url_prefix='/contributions')
from .corpora import bp as corpora_blueprint from .blueprints.corpora import bp as corpora_blueprint
from .corpora.cqi_over_sio import CQiNamespace
default_breadcrumb_root(corpora_blueprint, '.corpora')
app.register_blueprint(corpora_blueprint, cli_group='corpus', url_prefix='/corpora') app.register_blueprint(corpora_blueprint, cli_group='corpus', url_prefix='/corpora')
socketio.on_namespace(CQiNamespace('/cqi_over_sio'))
from .errors import bp as errors_bp from .blueprints.errors import bp as errors_bp
app.register_blueprint(errors_bp) app.register_blueprint(errors_bp)
from .jobs import bp as jobs_blueprint from .blueprints.jobs import bp as jobs_blueprint
default_breadcrumb_root(jobs_blueprint, '.jobs')
app.register_blueprint(jobs_blueprint, url_prefix='/jobs') app.register_blueprint(jobs_blueprint, url_prefix='/jobs')
from .main import bp as main_blueprint from .blueprints.main import bp as main_blueprint
default_breadcrumb_root(main_blueprint, '.')
app.register_blueprint(main_blueprint, cli_group=None) app.register_blueprint(main_blueprint, cli_group=None)
from .services import bp as services_blueprint from .blueprints.services import bp as services_blueprint
default_breadcrumb_root(services_blueprint, '.services')
app.register_blueprint(services_blueprint, url_prefix='/services') app.register_blueprint(services_blueprint, url_prefix='/services')
from .settings import bp as settings_blueprint from .blueprints.settings import bp as settings_blueprint
default_breadcrumb_root(settings_blueprint, '.settings')
app.register_blueprint(settings_blueprint, url_prefix='/settings') app.register_blueprint(settings_blueprint, url_prefix='/settings')
from .users import bp as users_blueprint from .blueprints.users import bp as users_blueprint
default_breadcrumb_root(users_blueprint, '.users')
app.register_blueprint(users_blueprint, cli_group='user', url_prefix='/users') app.register_blueprint(users_blueprint, cli_group='user', url_prefix='/users')
from .workshops import bp as workshops_blueprint from .blueprints.workshops import bp as workshops_blueprint
app.register_blueprint(workshops_blueprint, url_prefix='/workshops') app.register_blueprint(workshops_blueprint, url_prefix='/workshops')
from .models import _models
for model in _models:
admin.add_view(ModelView(model, db.session, category='Database'))
# endregion Blueprints
# region SocketIO Namespaces
from .namespaces.cqi_over_sio import CQiOverSocketIONamespace
socketio.on_namespace(CQiOverSocketIONamespace('/cqi_over_sio'))
# endregion SocketIO Namespaces
# region Database event Listeners
from .models.event_listeners import register_event_listeners
register_event_listeners()
# endregion Database event Listeners
# region Add scheduler jobs
if app.config['NOPAQUE_IS_PRIMARY_INSTANCE']:
from .jobs import handle_corpora
scheduler.add_job('handle_corpora', handle_corpora, seconds=3, trigger='interval')
from .jobs import handle_jobs
scheduler.add_job('handle_jobs', handle_jobs, seconds=3, trigger='interval')
# endregion Add scheduler jobs
return app return app

View File

@ -1,20 +0,0 @@
from flask import Blueprint
from flask_login import login_required
from app.decorators import admin_required
bp = Blueprint('admin', __name__)
@bp.before_request
@login_required
@admin_required
def before_request():
'''
Ensures that the routes in this package can be visited only by users with
administrator privileges (login_required and admin_required).
'''
pass
from . import json_routes, routes

View File

@ -1,16 +0,0 @@
from flask_wtf import FlaskForm
from wtforms import SelectField, SubmitField
from app.models import Role
class UpdateUserForm(FlaskForm):
role = SelectField('Role')
submit = SubmitField()
def __init__(self, user, *args, **kwargs):
if 'data' not in kwargs:
kwargs['data'] = {'role': user.role.hashid}
if 'prefix' not in kwargs:
kwargs['prefix'] = 'update-user-form'
super().__init__(*args, **kwargs)
self.role.choices = [(x.hashid, x.name) for x in Role.query.all()]

View File

@ -1,23 +0,0 @@
from flask import abort, request
from app import db
from app.decorators import content_negotiation
from app.models import User
from . import bp
@bp.route('/users/<hashid:user_id>/confirmed', methods=['PUT'])
@content_negotiation(consumes='application/json', produces='application/json')
def update_user_role(user_id):
confirmed = request.json
if not isinstance(confirmed, bool):
abort(400)
user = User.query.get_or_404(user_id)
user.confirmed = confirmed
db.session.commit()
response_data = {
'message': (
f'User "{user.username}" is now '
f'{"confirmed" if confirmed else "unconfirmed"}'
)
}
return response_data, 200

View File

@ -1,146 +0,0 @@
from flask import abort, flash, redirect, render_template, url_for
from flask_breadcrumbs import register_breadcrumb
from app import db, hashids
from app.models import Avatar, Corpus, Role, User
from app.users.settings.forms import (
UpdateAvatarForm,
UpdatePasswordForm,
UpdateNotificationsForm,
UpdateAccountInformationForm,
UpdateProfileInformationForm
)
from . import bp
from .forms import UpdateUserForm
from app.users.utils import (
user_endpoint_arguments_constructor as user_eac,
user_dynamic_list_constructor as user_dlc
)
@bp.route('')
@register_breadcrumb(bp, '.', '<i class="material-icons left">admin_panel_settings</i>Administration')
def admin():
return render_template(
'admin/admin.html.j2',
title='Administration'
)
@bp.route('/corpora')
@register_breadcrumb(bp, '.corpora', 'Corpora')
def corpora():
corpora = Corpus.query.all()
return render_template(
'admin/corpora.html.j2',
title='Corpora',
corpora=corpora
)
@bp.route('/users')
@register_breadcrumb(bp, '.users', '<i class="material-icons left">group</i>Users')
def users():
users = User.query.all()
return render_template(
'admin/users.html.j2',
title='Users',
users=users
)
@bp.route('/users/<hashid:user_id>')
@register_breadcrumb(bp, '.users.entity', '', dynamic_list_constructor=user_dlc)
def user(user_id):
user = User.query.get_or_404(user_id)
corpora = Corpus.query.filter(Corpus.user == user).all()
return render_template(
'admin/user.html.j2',
title=user.username,
user=user,
corpora=corpora
)
@bp.route('/users/<hashid:user_id>/settings', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.users.entity.settings', '<i class="material-icons left">settings</i>Settings')
def user_settings(user_id):
user = User.query.get_or_404(user_id)
update_account_information_form = UpdateAccountInformationForm(user)
update_profile_information_form = UpdateProfileInformationForm(user)
update_avatar_form = UpdateAvatarForm()
update_password_form = UpdatePasswordForm(user)
update_notifications_form = UpdateNotificationsForm(user)
update_user_form = UpdateUserForm(user)
# region handle update profile information form
if update_profile_information_form.submit.data and update_profile_information_form.validate():
user.about_me = update_profile_information_form.about_me.data
user.location = update_profile_information_form.location.data
user.organization = update_profile_information_form.organization.data
user.website = update_profile_information_form.website.data
user.full_name = update_profile_information_form.full_name.data
db.session.commit()
flash('Your changes have been saved')
return redirect(url_for('.user_settings', user_id=user.id))
# endregion handle update profile information form
# region handle update avatar form
if update_avatar_form.submit.data and update_avatar_form.validate():
try:
Avatar.create(
update_avatar_form.avatar.data,
user=user
)
except (AttributeError, OSError):
abort(500)
db.session.commit()
flash('Your changes have been saved')
return redirect(url_for('.user_settings', user_id=user.id))
# endregion handle update avatar form
# region handle update account information form
if update_account_information_form.submit.data and update_account_information_form.validate():
user.email = update_account_information_form.email.data
user.username = update_account_information_form.username.data
db.session.commit()
flash('Profile settings updated')
return redirect(url_for('.user_settings', user_id=user.id))
# endregion handle update account information form
# region handle update password form
if update_password_form.submit.data and update_password_form.validate():
user.password = update_password_form.new_password.data
db.session.commit()
flash('Your changes have been saved')
return redirect(url_for('.user_settings', user_id=user.id))
# endregion handle update password form
# region handle update notifications form
if update_notifications_form.submit.data and update_notifications_form.validate():
user.setting_job_status_mail_notification_level = \
update_notifications_form.job_status_mail_notification_level.data
db.session.commit()
flash('Your changes have been saved')
return redirect(url_for('.user_settings', user_id=user.id))
# endregion handle update notifications form
# region handle update user form
if update_user_form.submit.data and update_user_form.validate():
role_id = hashids.decode(update_user_form.role.data)
user.role = Role.query.get(role_id)
db.session.commit()
flash('Your changes have been saved')
return redirect(url_for('.user_settings', user_id=user.id))
# endregion handle update user form
return render_template(
'admin/user_settings.html.j2',
title='Settings',
update_account_information_form=update_account_information_form,
update_avatar_form=update_avatar_form,
update_notifications_form=update_notifications_form,
update_password_form=update_password_form,
update_profile_information_form=update_profile_information_form,
update_user_form=update_user_form,
user=user
)

View File

@ -5,8 +5,8 @@ from flask import abort, Blueprint
from werkzeug.exceptions import InternalServerError from werkzeug.exceptions import InternalServerError
from app import db, hashids from app import db, hashids
from app.models import Job, JobInput, JobStatus, TesseractOCRPipelineModel from app.models import Job, JobInput, JobStatus, TesseractOCRPipelineModel
from .schemas import EmptySchema, JobSchema, SpaCyNLPPipelineJobSchema, TesseractOCRPipelineJobSchema, TesseractOCRPipelineModelSchema
from .auth import auth_error_responses, token_auth from .auth import auth_error_responses, token_auth
from .schemas import EmptySchema, JobSchema, SpaCyNLPPipelineJobSchema, TesseractOCRPipelineJobSchema, TesseractOCRPipelineModelSchema
bp = Blueprint('jobs', __name__) bp = Blueprint('jobs', __name__)
@ -77,7 +77,7 @@ def delete_job(job_id):
job = Job.query.get(job_id) job = Job.query.get(job_id)
if job is None: if job is None:
abort(404) abort(404)
if not (job.user == current_user or current_user.is_administrator()): if not (job.user == current_user or current_user.is_administrator):
abort(403) abort(403)
try: try:
job.delete() job.delete()
@ -97,6 +97,6 @@ def get_job(job_id):
job = Job.query.get(job_id) job = Job.query.get(job_id)
if job is None: if job is None:
abort(404) abort(404)
if not (job.user == current_user or current_user.is_administrator()): if not (job.user == current_user or current_user.is_administrator):
abort(403) abort(403)
return job return job

View File

@ -10,7 +10,7 @@ from app.models import (
User, User,
UserSettingJobStatusMailNotificationLevel UserSettingJobStatusMailNotificationLevel
) )
from app.services import SERVICES from app.blueprints.services import SERVICES

View File

@ -3,11 +3,11 @@ from apifairy import authenticate, body, response
from apifairy.decorators import other_responses from apifairy.decorators import other_responses
from flask import abort, Blueprint from flask import abort, Blueprint
from werkzeug.exceptions import InternalServerError from werkzeug.exceptions import InternalServerError
from app import db
from app.email import create_message, send from app.email import create_message, send
from app import db
from app.models import User from app.models import User
from .schemas import EmptySchema, UserSchema
from .auth import auth_error_responses, token_auth from .auth import auth_error_responses, token_auth
from .schemas import EmptySchema, UserSchema
bp = Blueprint('users', __name__) bp = Blueprint('users', __name__)
@ -60,7 +60,7 @@ def delete_user(user_id):
user = User.query.get(user_id) user = User.query.get(user_id)
if user is None: if user is None:
abort(404) abort(404)
if not (user == current_user or current_user.is_administrator()): if not (user == current_user or current_user.is_administrator):
abort(403) abort(403)
user.delete() user.delete()
db.session.commit() db.session.commit()
@ -78,7 +78,7 @@ def get_user(user_id):
user = User.query.get(user_id) user = User.query.get(user_id)
if user is None: if user is None:
abort(404) abort(404)
if not (user == current_user or current_user.is_administrator()): if not (user == current_user or current_user.is_administrator):
abort(403) abort(403)
return user return user
@ -94,6 +94,6 @@ def get_user_by_username(username):
user = User.query.filter(User.username == username).first() user = User.query.filter(User.username == username).first()
if user is None: if user is None:
abort(404) abort(404)
if not (user == current_user or current_user.is_administrator()): if not (user == current_user or current_user.is_administrator):
abort(403) abort(403)
return user return user

View File

@ -0,0 +1,29 @@
from flask import Blueprint, redirect, request, url_for
from flask_login import current_user
from app import db
bp = Blueprint('auth', __name__)
@bp.before_app_request
def before_request():
if not current_user.is_authenticated:
return
current_user.ping()
db.session.commit()
if (
not current_user.confirmed
and request.endpoint
and request.blueprint != 'auth'
and request.endpoint != 'static'
):
return redirect(url_for('auth.unconfirmed'))
if not current_user.terms_of_use_accepted:
return redirect(url_for('main.terms_of_use'))
from . import routes

View File

@ -60,7 +60,11 @@ class RegistrationForm(FlaskForm):
def validate_username(self, field): def validate_username(self, field):
if User.query.filter_by(username=field.data).first(): if User.query.filter_by(username=field.data).first():
raise ValidationError('Username already in use') raise ValidationError('Username already registered')
def validate_terms_of_use_accepted(self, field):
if not field.data:
raise ValidationError('Terms of Use not accepted')
class LoginForm(FlaskForm): class LoginForm(FlaskForm):

View File

@ -1,5 +1,4 @@
from flask import abort, flash, redirect, render_template, request, url_for from flask import abort, flash, redirect, render_template, request, url_for
from flask_breadcrumbs import register_breadcrumb
from flask_login import current_user, login_user, login_required, logout_user from flask_login import current_user, login_user, login_required, logout_user
from app import db from app import db
from app.email import create_message, send from app.email import create_message, send
@ -13,24 +12,7 @@ from .forms import (
) )
@bp.before_app_request
def before_request():
"""
Checks if a user is unconfirmed when visiting specific sites. Redirects to
unconfirmed view if user is unconfirmed.
"""
if current_user.is_authenticated:
current_user.ping()
db.session.commit()
if (not current_user.confirmed
and request.endpoint
and request.blueprint != 'auth'
and request.endpoint != 'static'):
return redirect(url_for('auth.unconfirmed'))
@bp.route('/register', methods=['GET', 'POST']) @bp.route('/register', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.register', 'Register')
def register(): def register():
if current_user.is_authenticated: if current_user.is_authenticated:
return redirect(url_for('main.dashboard')) return redirect(url_for('main.dashboard'))
@ -67,7 +49,6 @@ def register():
@bp.route('/login', methods=['GET', 'POST']) @bp.route('/login', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.login', 'Login')
def login(): def login():
if current_user.is_authenticated: if current_user.is_authenticated:
return redirect(url_for('main.dashboard')) return redirect(url_for('main.dashboard'))
@ -98,7 +79,6 @@ def logout():
@bp.route('/unconfirmed') @bp.route('/unconfirmed')
@register_breadcrumb(bp, '.unconfirmed', 'Unconfirmed')
@login_required @login_required
def unconfirmed(): def unconfirmed():
if current_user.confirmed: if current_user.confirmed:
@ -141,7 +121,6 @@ def confirm(token):
@bp.route('/reset-password-request', methods=['GET', 'POST']) @bp.route('/reset-password-request', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.reset_password_request', 'Password Reset')
def reset_password_request(): def reset_password_request():
if current_user.is_authenticated: if current_user.is_authenticated:
return redirect(url_for('main.dashboard')) return redirect(url_for('main.dashboard'))
@ -171,7 +150,6 @@ def reset_password_request():
@bp.route('/reset-password/<token>', methods=['GET', 'POST']) @bp.route('/reset-password/<token>', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.reset_password', 'Password Reset')
def reset_password(token): def reset_password(token):
if current_user.is_authenticated: if current_user.is_authenticated:
return redirect(url_for('main.dashboard')) return redirect(url_for('main.dashboard'))

View File

@ -0,0 +1,25 @@
from flask import Blueprint
from flask_login import login_required
bp = Blueprint('contributions', __name__)
@bp.before_request
@login_required
def before_request():
'''
Ensures that the routes in this package can only be visited by users that
are logged in.
'''
pass
from . import routes
from .spacy_nlp_pipeline_models import bp as spacy_nlp_pipeline_models_bp
bp.register_blueprint(spacy_nlp_pipeline_models_bp, url_prefix='/spacy-nlp-pipeline-models')
from .tesseract_ocr_pipeline_models import bp as tesseract_ocr_pipeline_models_bp
bp.register_blueprint(tesseract_ocr_pipeline_models_bp, url_prefix='/tesseract-ocr-pipeline-models')

View File

@ -0,0 +1,7 @@
from flask import render_template
from . import bp
@bp.route('')
def index():
return render_template('contributions/index.html.j2', title='Contributions')

View File

@ -1,8 +1,8 @@
from flask import Blueprint from flask import current_app, Blueprint
from flask_login import login_required from flask_login import login_required
bp = Blueprint('settings', __name__) bp = Blueprint('spacy_nlp_pipeline_models', __name__)
@bp.before_request @bp.before_request
@ -15,4 +15,4 @@ def before_request():
pass pass
from . import routes from . import routes, json_routes

View File

@ -1,7 +1,7 @@
from flask_wtf.file import FileField, FileRequired from flask_wtf.file import FileField, FileRequired
from wtforms import StringField, ValidationError from wtforms import StringField, ValidationError
from wtforms.validators import InputRequired, Length from wtforms.validators import InputRequired, Length
from app.services import SERVICES from app.blueprints.services import SERVICES
from ..forms import ContributionBaseForm, UpdateContributionBaseForm from ..forms import ContributionBaseForm, UpdateContributionBaseForm

View File

@ -1,13 +1,14 @@
from flask import abort, current_app, request from flask import abort, current_app, request
from flask_login import current_user from flask_login import current_user, login_required
from threading import Thread from threading import Thread
from app import db from app import db
from app.decorators import content_negotiation, permission_required from app.decorators import content_negotiation, permission_required
from app.models import SpaCyNLPPipelineModel from app.models import SpaCyNLPPipelineModel
from .. import bp from . import bp
@bp.route('/spacy-nlp-pipeline-models/<hashid:spacy_nlp_pipeline_model_id>', methods=['DELETE']) @bp.route('/<hashid:spacy_nlp_pipeline_model_id>', methods=['DELETE'])
@login_required
@content_negotiation(produces='application/json') @content_negotiation(produces='application/json')
def delete_spacy_model(spacy_nlp_pipeline_model_id): def delete_spacy_model(spacy_nlp_pipeline_model_id):
def _delete_spacy_model(app, spacy_nlp_pipeline_model_id): def _delete_spacy_model(app, spacy_nlp_pipeline_model_id):
@ -15,9 +16,9 @@ def delete_spacy_model(spacy_nlp_pipeline_model_id):
snpm = SpaCyNLPPipelineModel.query.get(spacy_nlp_pipeline_model_id) snpm = SpaCyNLPPipelineModel.query.get(spacy_nlp_pipeline_model_id)
snpm.delete() snpm.delete()
db.session.commit() db.session.commit()
snpm = SpaCyNLPPipelineModel.query.get_or_404(spacy_nlp_pipeline_model_id) snpm = SpaCyNLPPipelineModel.query.get_or_404(spacy_nlp_pipeline_model_id)
if not (snpm.user == current_user or current_user.is_administrator()): if not (snpm.user == current_user or current_user.is_administrator):
abort(403) abort(403)
thread = Thread( thread = Thread(
target=_delete_spacy_model, target=_delete_spacy_model,
@ -31,7 +32,7 @@ def delete_spacy_model(spacy_nlp_pipeline_model_id):
return response_data, 202 return response_data, 202
@bp.route('/spacy-nlp-pipeline-models/<hashid:spacy_nlp_pipeline_model_id>/is_public', methods=['PUT']) @bp.route('/<hashid:spacy_nlp_pipeline_model_id>/is_public', methods=['PUT'])
@permission_required('CONTRIBUTE') @permission_required('CONTRIBUTE')
@content_negotiation(consumes='application/json', produces='application/json') @content_negotiation(consumes='application/json', produces='application/json')
def update_spacy_nlp_pipeline_model_is_public(spacy_nlp_pipeline_model_id): def update_spacy_nlp_pipeline_model_is_public(spacy_nlp_pipeline_model_id):
@ -39,7 +40,7 @@ def update_spacy_nlp_pipeline_model_is_public(spacy_nlp_pipeline_model_id):
if not isinstance(is_public, bool): if not isinstance(is_public, bool):
abort(400) abort(400)
snpm = SpaCyNLPPipelineModel.query.get_or_404(spacy_nlp_pipeline_model_id) snpm = SpaCyNLPPipelineModel.query.get_or_404(spacy_nlp_pipeline_model_id)
if not (snpm.user == current_user or current_user.is_administrator()): if not (snpm.user == current_user or current_user.is_administrator):
abort(403) abort(403)
snpm.is_public = is_public snpm.is_public = is_public
db.session.commit() db.session.commit()

View File

@ -1,6 +1,5 @@
from flask import abort, flash, redirect, render_template, url_for from flask import abort, flash, redirect, render_template, url_for
from flask_breadcrumbs import register_breadcrumb from flask_login import current_user, login_required
from flask_login import current_user
from app import db from app import db
from app.models import SpaCyNLPPipelineModel from app.models import SpaCyNLPPipelineModel
from . import bp from . import bp
@ -8,23 +7,17 @@ from .forms import (
CreateSpaCyNLPPipelineModelForm, CreateSpaCyNLPPipelineModelForm,
UpdateSpaCyNLPPipelineModelForm UpdateSpaCyNLPPipelineModelForm
) )
from .utils import (
spacy_nlp_pipeline_model_dlc as spacy_nlp_pipeline_model_dlc
)
@bp.route('/spacy-nlp-pipeline-models') @bp.route('/')
@register_breadcrumb(bp, '.spacy_nlp_pipeline_models', 'SpaCy NLP Pipeline Models') @login_required
def spacy_nlp_pipeline_models(): def index():
return render_template( return redirect(url_for('contributions.index', _anchor='spacy-nlp-pipeline-models'))
'contributions/spacy_nlp_pipeline_models/spacy_nlp_pipeline_models.html.j2',
title='SpaCy NLP Pipeline Models'
)
@bp.route('/spacy-nlp-pipeline-models/create', methods=['GET', 'POST']) @bp.route('/create', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.spacy_nlp_pipeline_models.create', 'Create') @login_required
def create_spacy_nlp_pipeline_model(): def create():
form = CreateSpaCyNLPPipelineModelForm() form = CreateSpaCyNLPPipelineModelForm()
if form.is_submitted(): if form.is_submitted():
if not form.validate(): if not form.validate():
@ -48,7 +41,7 @@ def create_spacy_nlp_pipeline_model():
abort(500) abort(500)
db.session.commit() db.session.commit()
flash(f'SpaCy NLP Pipeline model "{snpm.title}" created') flash(f'SpaCy NLP Pipeline model "{snpm.title}" created')
return {}, 201, {'Location': url_for('.spacy_nlp_pipeline_models')} return {}, 201, {'Location': url_for('.index')}
return render_template( return render_template(
'contributions/spacy_nlp_pipeline_models/create.html.j2', 'contributions/spacy_nlp_pipeline_models/create.html.j2',
title='Create SpaCy NLP Pipeline Model', title='Create SpaCy NLP Pipeline Model',
@ -56,11 +49,11 @@ def create_spacy_nlp_pipeline_model():
) )
@bp.route('/spacy-nlp-pipeline-models/<hashid:spacy_nlp_pipeline_model_id>', methods=['GET', 'POST']) @bp.route('/<hashid:spacy_nlp_pipeline_model_id>', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.spacy_nlp_pipeline_models.entity', '', dynamic_list_constructor=spacy_nlp_pipeline_model_dlc) @login_required
def spacy_nlp_pipeline_model(spacy_nlp_pipeline_model_id): def entity(spacy_nlp_pipeline_model_id):
snpm = SpaCyNLPPipelineModel.query.get_or_404(spacy_nlp_pipeline_model_id) snpm = SpaCyNLPPipelineModel.query.get_or_404(spacy_nlp_pipeline_model_id)
if not (snpm.user == current_user or current_user.is_administrator()): if not (snpm.user == current_user or current_user.is_administrator):
abort(403) abort(403)
form = UpdateSpaCyNLPPipelineModelForm(data=snpm.to_json_serializeable()) form = UpdateSpaCyNLPPipelineModelForm(data=snpm.to_json_serializeable())
if form.validate_on_submit(): if form.validate_on_submit():
@ -68,9 +61,9 @@ def spacy_nlp_pipeline_model(spacy_nlp_pipeline_model_id):
if db.session.is_modified(snpm): if db.session.is_modified(snpm):
flash(f'SpaCy NLP Pipeline model "{snpm.title}" updated') flash(f'SpaCy NLP Pipeline model "{snpm.title}" updated')
db.session.commit() db.session.commit()
return redirect(url_for('.spacy_nlp_pipeline_models')) return redirect(url_for('.index'))
return render_template( return render_template(
'contributions/spacy_nlp_pipeline_models/spacy_nlp_pipeline_model.html.j2', 'contributions/spacy_nlp_pipeline_models/entity.html.j2',
title=f'{snpm.title} {snpm.version}', title=f'{snpm.title} {snpm.version}',
form=form, form=form,
spacy_nlp_pipeline_model=snpm spacy_nlp_pipeline_model=snpm

View File

@ -2,7 +2,7 @@ from flask import Blueprint
from flask_login import login_required from flask_login import login_required
bp = Blueprint('users', __name__) bp = Blueprint('tesseract_ocr_pipeline_models', __name__)
@bp.before_request @bp.before_request
@ -15,4 +15,4 @@ def before_request():
pass pass
from . import cli, events, json_routes, routes, settings from . import json_routes, routes

View File

@ -1,6 +1,6 @@
from flask_wtf.file import FileField, FileRequired from flask_wtf.file import FileField, FileRequired
from wtforms import ValidationError from wtforms import ValidationError
from app.services import SERVICES from app.blueprints.services import SERVICES
from ..forms import ContributionBaseForm, UpdateContributionBaseForm from ..forms import ContributionBaseForm, UpdateContributionBaseForm
@ -9,7 +9,7 @@ class CreateTesseractOCRPipelineModelForm(ContributionBaseForm):
'File', 'File',
validators=[FileRequired()] validators=[FileRequired()]
) )
def validate_tesseract_model_file(self, field): def validate_tesseract_model_file(self, field):
if not field.data.filename.lower().endswith('.traineddata'): if not field.data.filename.lower().endswith('.traineddata'):
raise ValidationError('traineddata files only!') raise ValidationError('traineddata files only!')

View File

@ -7,7 +7,7 @@ from app.models import TesseractOCRPipelineModel
from . import bp from . import bp
@bp.route('/tesseract-ocr-pipeline-models/<hashid:tesseract_ocr_pipeline_model_id>', methods=['DELETE']) @bp.route('/<hashid:tesseract_ocr_pipeline_model_id>', methods=['DELETE'])
@content_negotiation(produces='application/json') @content_negotiation(produces='application/json')
def delete_tesseract_model(tesseract_ocr_pipeline_model_id): def delete_tesseract_model(tesseract_ocr_pipeline_model_id):
def _delete_tesseract_ocr_pipeline_model(app, tesseract_ocr_pipeline_model_id): def _delete_tesseract_ocr_pipeline_model(app, tesseract_ocr_pipeline_model_id):
@ -17,7 +17,7 @@ def delete_tesseract_model(tesseract_ocr_pipeline_model_id):
db.session.commit() db.session.commit()
topm = TesseractOCRPipelineModel.query.get_or_404(tesseract_ocr_pipeline_model_id) topm = TesseractOCRPipelineModel.query.get_or_404(tesseract_ocr_pipeline_model_id)
if not (topm.user == current_user or current_user.is_administrator()): if not (topm.user == current_user or current_user.is_administrator):
abort(403) abort(403)
thread = Thread( thread = Thread(
target=_delete_tesseract_ocr_pipeline_model, target=_delete_tesseract_ocr_pipeline_model,
@ -31,7 +31,7 @@ def delete_tesseract_model(tesseract_ocr_pipeline_model_id):
return response_data, 202 return response_data, 202
@bp.route('/tesseract-ocr-pipeline-models/<hashid:tesseract_ocr_pipeline_model_id>/is_public', methods=['PUT']) @bp.route('/<hashid:tesseract_ocr_pipeline_model_id>/is_public', methods=['PUT'])
@permission_required('CONTRIBUTE') @permission_required('CONTRIBUTE')
@content_negotiation(consumes='application/json', produces='application/json') @content_negotiation(consumes='application/json', produces='application/json')
def update_tesseract_ocr_pipeline_model_is_public(tesseract_ocr_pipeline_model_id): def update_tesseract_ocr_pipeline_model_is_public(tesseract_ocr_pipeline_model_id):
@ -39,7 +39,7 @@ def update_tesseract_ocr_pipeline_model_is_public(tesseract_ocr_pipeline_model_i
if not isinstance(is_public, bool): if not isinstance(is_public, bool):
abort(400) abort(400)
topm = TesseractOCRPipelineModel.query.get_or_404(tesseract_ocr_pipeline_model_id) topm = TesseractOCRPipelineModel.query.get_or_404(tesseract_ocr_pipeline_model_id)
if not (topm.user == current_user or current_user.is_administrator()): if not (topm.user == current_user or current_user.is_administrator):
abort(403) abort(403)
topm.is_public = is_public topm.is_public = is_public
db.session.commit() db.session.commit()

View File

@ -1,5 +1,4 @@
from flask import abort, flash, redirect, render_template, url_for from flask import abort, flash, redirect, render_template, url_for
from flask_breadcrumbs import register_breadcrumb
from flask_login import current_user from flask_login import current_user
from app import db from app import db
from app.models import TesseractOCRPipelineModel from app.models import TesseractOCRPipelineModel
@ -8,23 +7,15 @@ from .forms import (
CreateTesseractOCRPipelineModelForm, CreateTesseractOCRPipelineModelForm,
UpdateTesseractOCRPipelineModelForm UpdateTesseractOCRPipelineModelForm
) )
from .utils import (
tesseract_ocr_pipeline_model_dlc as tesseract_ocr_pipeline_model_dlc
)
@bp.route('/tesseract-ocr-pipeline-models') @bp.route('/')
@register_breadcrumb(bp, '.tesseract_ocr_pipeline_models', 'Tesseract OCR Pipeline Models') def index():
def tesseract_ocr_pipeline_models(): return redirect(url_for('contributions.index', _anchor='tesseract-ocr-pipeline-models'))
return render_template(
'contributions/tesseract_ocr_pipeline_models/tesseract_ocr_pipeline_models.html.j2',
title='Tesseract OCR Pipeline Models'
)
@bp.route('/tesseract-ocr-pipeline-models/create', methods=['GET', 'POST']) @bp.route('/create', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.tesseract_ocr_pipeline_models.create', 'Create') def create():
def create_tesseract_ocr_pipeline_model():
form = CreateTesseractOCRPipelineModelForm() form = CreateTesseractOCRPipelineModelForm()
if form.is_submitted(): if form.is_submitted():
if not form.validate(): if not form.validate():
@ -47,7 +38,7 @@ def create_tesseract_ocr_pipeline_model():
abort(500) abort(500)
db.session.commit() db.session.commit()
flash(f'Tesseract OCR Pipeline model "{topm.title}" created') flash(f'Tesseract OCR Pipeline model "{topm.title}" created')
return {}, 201, {'Location': url_for('.tesseract_ocr_pipeline_models')} return {}, 201, {'Location': url_for('.index')}
return render_template( return render_template(
'contributions/tesseract_ocr_pipeline_models/create.html.j2', 'contributions/tesseract_ocr_pipeline_models/create.html.j2',
title='Create Tesseract OCR Pipeline Model', title='Create Tesseract OCR Pipeline Model',
@ -55,11 +46,10 @@ def create_tesseract_ocr_pipeline_model():
) )
@bp.route('/tesseract-ocr-pipeline-models/<hashid:tesseract_ocr_pipeline_model_id>', methods=['GET', 'POST']) @bp.route('/<hashid:tesseract_ocr_pipeline_model_id>', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.tesseract_ocr_pipeline_models.entity', '', dynamic_list_constructor=tesseract_ocr_pipeline_model_dlc) def entity(tesseract_ocr_pipeline_model_id):
def tesseract_ocr_pipeline_model(tesseract_ocr_pipeline_model_id):
topm = TesseractOCRPipelineModel.query.get_or_404(tesseract_ocr_pipeline_model_id) topm = TesseractOCRPipelineModel.query.get_or_404(tesseract_ocr_pipeline_model_id)
if not (topm.user == current_user or current_user.is_administrator()): if not (topm.user == current_user or current_user.is_administrator):
abort(403) abort(403)
form = UpdateTesseractOCRPipelineModelForm(data=topm.to_json_serializeable()) form = UpdateTesseractOCRPipelineModelForm(data=topm.to_json_serializeable())
if form.validate_on_submit(): if form.validate_on_submit():
@ -67,9 +57,9 @@ def tesseract_ocr_pipeline_model(tesseract_ocr_pipeline_model_id):
if db.session.is_modified(topm): if db.session.is_modified(topm):
flash(f'Tesseract OCR Pipeline model "{topm.title}" updated') flash(f'Tesseract OCR Pipeline model "{topm.title}" updated')
db.session.commit() db.session.commit()
return redirect(url_for('.tesseract_ocr_pipeline_models')) return redirect(url_for('.index'))
return render_template( return render_template(
'contributions/tesseract_ocr_pipeline_models/tesseract_ocr_pipeline_model.html.j2', 'contributions/tesseract_ocr_pipeline_models/entity.html.j2',
title=f'{topm.title} {topm.version}', title=f'{topm.title} {topm.version}',
form=form, form=form,
tesseract_ocr_pipeline_model=topm tesseract_ocr_pipeline_model=topm

View File

@ -16,4 +16,4 @@ def before_request():
pass pass
from . import cli, files, followers, routes, json_routes from . import cli, files, followers, routes

View File

@ -1,7 +1,7 @@
from app.models import Corpus, CorpusStatus from flask import current_app
import os
import shutil import shutil
from app import db from app import db
from app.models import Corpus, CorpusStatus
from . import bp from . import bp
@ -18,10 +18,17 @@ def reset():
] ]
for corpus in [x for x in Corpus.query.all() if x.status in status]: for corpus in [x for x in Corpus.query.all() if x.status in status]:
print(f'Resetting corpus {corpus}') print(f'Resetting corpus {corpus}')
shutil.rmtree(os.path.join(corpus.path, 'cwb'), ignore_errors=True) corpus_cwb_dir = corpus.path / 'cwb'
os.mkdir(os.path.join(corpus.path, 'cwb')) corpus_cwb_data_dir = corpus_cwb_dir / 'data'
os.mkdir(os.path.join(corpus.path, 'cwb', 'data')) corpus_cwb_registry_dir = corpus_cwb_dir / 'registry'
os.mkdir(os.path.join(corpus.path, 'cwb', 'registry')) try:
shutil.rmtree(corpus.path / 'cwb', ignore_errors=True)
corpus_cwb_dir.mkdir()
corpus_cwb_data_dir.mkdir()
corpus_cwb_registry_dir.mkdir()
except OSError as e:
current_app.logger.error(e)
raise
corpus.status = CorpusStatus.UNPREPARED corpus.status = CorpusStatus.UNPREPARED
corpus.num_analysis_sessions = 0 corpus.num_analysis_sessions = 0
db.session.commit() db.session.commit()

View File

@ -10,7 +10,7 @@ def corpus_follower_permission_required(*permissions):
def decorated_function(*args, **kwargs): def decorated_function(*args, **kwargs):
corpus_id = kwargs.get('corpus_id') corpus_id = kwargs.get('corpus_id')
corpus = Corpus.query.get_or_404(corpus_id) corpus = Corpus.query.get_or_404(corpus_id)
if not (corpus.user == current_user or current_user.is_administrator()): if not (corpus.user == current_user or current_user.is_administrator):
cfa = CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=current_user.id).first() cfa = CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=current_user.id).first()
if cfa is None: if cfa is None:
abort(403) abort(403)
@ -26,7 +26,7 @@ def corpus_owner_or_admin_required(f):
def decorated_function(*args, **kwargs): def decorated_function(*args, **kwargs):
corpus_id = kwargs.get('corpus_id') corpus_id = kwargs.get('corpus_id')
corpus = Corpus.query.get_or_404(corpus_id) corpus = Corpus.query.get_or_404(corpus_id)
if not (corpus.user == current_user or current_user.is_administrator()): if not (corpus.user == current_user or current_user.is_administrator):
abort(403) abort(403)
return f(*args, **kwargs) return f(*args, **kwargs)
return decorated_function return decorated_function

View File

@ -1,7 +1,7 @@
from flask import abort, current_app from flask import current_app
from threading import Thread from threading import Thread
from app import db
from app.decorators import content_negotiation from app.decorators import content_negotiation
from app import db
from app.models import CorpusFile from app.models import CorpusFile
from ..decorators import corpus_follower_permission_required from ..decorators import corpus_follower_permission_required
from . import bp from . import bp

View File

@ -6,25 +6,19 @@ from flask import (
send_from_directory, send_from_directory,
url_for url_for
) )
from flask_breadcrumbs import register_breadcrumb
import os
from app import db from app import db
from app.models import Corpus, CorpusFile, CorpusStatus from app.models import Corpus, CorpusFile, CorpusStatus
from ..decorators import corpus_follower_permission_required from ..decorators import corpus_follower_permission_required
from ..utils import corpus_endpoint_arguments_constructor as corpus_eac
from . import bp from . import bp
from .forms import CreateCorpusFileForm, UpdateCorpusFileForm from .forms import CreateCorpusFileForm, UpdateCorpusFileForm
from .utils import corpus_file_dynamic_list_constructor as corpus_file_dlc
@bp.route('/<hashid:corpus_id>/files') @bp.route('/<hashid:corpus_id>/files')
@register_breadcrumb(bp, '.entity.files', 'Files', endpoint_arguments_constructor=corpus_eac)
def corpus_files(corpus_id): def corpus_files(corpus_id):
return redirect(url_for('.corpus', _anchor='files', corpus_id=corpus_id)) return redirect(url_for('.corpus', _anchor='files', corpus_id=corpus_id))
@bp.route('/<hashid:corpus_id>/files/create', methods=['GET', 'POST']) @bp.route('/<hashid:corpus_id>/files/create', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.entity.files.create', 'Create', endpoint_arguments_constructor=corpus_eac)
@corpus_follower_permission_required('MANAGE_FILES') @corpus_follower_permission_required('MANAGE_FILES')
def create_corpus_file(corpus_id): def create_corpus_file(corpus_id):
corpus = Corpus.query.get_or_404(corpus_id) corpus = Corpus.query.get_or_404(corpus_id)
@ -66,7 +60,6 @@ def create_corpus_file(corpus_id):
@bp.route('/<hashid:corpus_id>/files/<hashid:corpus_file_id>', methods=['GET', 'POST']) @bp.route('/<hashid:corpus_id>/files/<hashid:corpus_file_id>', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.entity.files.entity', '', dynamic_list_constructor=corpus_file_dlc)
@corpus_follower_permission_required('MANAGE_FILES') @corpus_follower_permission_required('MANAGE_FILES')
def corpus_file(corpus_id, corpus_file_id): def corpus_file(corpus_id, corpus_file_id):
corpus_file = CorpusFile.query.filter_by(corpus_id=corpus_id, id=corpus_file_id).first_or_404() corpus_file = CorpusFile.query.filter_by(corpus_id=corpus_id, id=corpus_file_id).first_or_404()
@ -92,9 +85,9 @@ def corpus_file(corpus_id, corpus_file_id):
def download_corpus_file(corpus_id, corpus_file_id): def download_corpus_file(corpus_id, corpus_file_id):
corpus_file = CorpusFile.query.filter_by(corpus_id=corpus_id, id=corpus_file_id).first_or_404() corpus_file = CorpusFile.query.filter_by(corpus_id=corpus_id, id=corpus_file_id).first_or_404()
return send_from_directory( return send_from_directory(
os.path.dirname(corpus_file.path), corpus_file.path.parent,
os.path.basename(corpus_file.path), corpus_file.path.name,
as_attachment=True, as_attachment=True,
attachment_filename=corpus_file.filename, download_name=corpus_file.filename,
mimetype=corpus_file.mimetype mimetype=corpus_file.mimetype
) )

View File

@ -58,7 +58,7 @@ def delete_corpus_follower(corpus_id, follower_id):
current_user.id == follower_id current_user.id == follower_id
or current_user == cfa.corpus.user or current_user == cfa.corpus.user
or CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=current_user.id).first().role.has_permission('MANAGE_FOLLOWERS') or CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=current_user.id).first().role.has_permission('MANAGE_FOLLOWERS')
or current_user.is_administrator()): or current_user.is_administrator):
abort(403) abort(403)
if current_user.id == follower_id: if current_user.id == follower_id:
flash(f'You are no longer following "{cfa.corpus.title}"', 'corpus') flash(f'You are no longer following "{cfa.corpus.title}"', 'corpus')

View File

@ -0,0 +1,299 @@
from datetime import datetime
from flask import (
abort,
current_app,
flash,
Flask,
jsonify,
redirect,
request,
render_template,
url_for
)
from flask_login import current_user
from string import punctuation
from threading import Thread
import nltk
from app import db
from app.models import (
Corpus,
CorpusFollowerAssociation,
CorpusFollowerRole,
User
)
from . import bp
from .decorators import corpus_follower_permission_required
from .forms import CreateCorpusForm
def _delete_corpus(app: Flask, corpus_id: int):
with app.app_context():
corpus: Corpus = Corpus.query.get(corpus_id)
corpus.delete()
db.session.commit()
def _build_corpus(app: Flask, corpus_id: int):
with app.app_context():
corpus = Corpus.query.get(corpus_id)
corpus.build()
db.session.commit()
@bp.route('')
def corpora():
return redirect(url_for('main.dashboard', _anchor='corpora'))
@bp.route('/create', methods=['GET', 'POST'])
def create_corpus():
form = CreateCorpusForm()
if form.validate_on_submit():
try:
corpus = Corpus.create(
title=form.title.data,
description=form.description.data,
user=current_user
)
except OSError:
abort(500)
db.session.commit()
flash(f'Corpus "{corpus.title}" created', 'corpus')
return redirect(corpus.url)
return render_template(
'corpora/create.html.j2',
title='Create corpus',
form=form
)
@bp.route('/<hashid:corpus_id>')
def corpus(corpus_id: int):
corpus = Corpus.query.get_or_404(corpus_id)
cfa = CorpusFollowerAssociation.query.filter_by(
corpus_id=corpus_id,
follower_id=current_user.id
).first()
if cfa is None:
if corpus.user == current_user or current_user.is_administrator:
cfr = CorpusFollowerRole.query.filter_by(name='Administrator').first()
else:
cfr = CorpusFollowerRole.query.filter_by(name='Anonymous').first()
else:
cfr = cfa.role
cfrs = CorpusFollowerRole.query.all()
# TODO: Better solution for filtering admin
users = User.query.filter(
User.is_public == True,
User.id != current_user.id,
User.id != corpus.user.id,
User.role_id < 4
).all()
if (
corpus.user == current_user
or current_user.is_administrator
):
return render_template(
'corpora/corpus.html.j2',
title=corpus.title,
corpus=corpus,
cfr=cfr,
cfrs=cfrs,
users=users
)
if (
current_user.is_following_corpus(corpus)
or corpus.is_public
):
cfas = CorpusFollowerAssociation.query.filter(
Corpus.id == corpus_id,
CorpusFollowerAssociation.follower_id != corpus.user.id
).all()
return render_template(
'corpora/public_corpus.html.j2',
title=corpus.title,
corpus=corpus,
cfrs=cfrs,
cfr=cfr,
cfas=cfas,
users=users
)
abort(403)
@bp.route('/<hashid:corpus_id>', methods=['DELETE'])
def delete_corpus(corpus_id: int):
corpus = Corpus.query.get_or_404(corpus_id)
if not (
corpus.user == current_user
or current_user.is_administrator
):
abort(403)
thread = Thread(
target=_delete_corpus,
args=(current_app._get_current_object(), corpus.id)
)
thread.start()
return jsonify(f'Corpus "{corpus.title}" marked for deletion.'), 202
@bp.route('/<hashid:corpus_id>/build', methods=['POST'])
def build_corpus(corpus_id: int):
corpus = Corpus.query.get_or_404(corpus_id)
cfa = CorpusFollowerAssociation.query.filter_by(
corpus_id=corpus_id,
follower_id=current_user.id
).first()
if not (
cfa is not None and cfa.role.has_permission('MANAGE_FILES')
or corpus.user == current_user
or current_user.is_administrator
):
abort(403)
if len(corpus.files.all()) == 0:
abort(409)
thread = Thread(
target=_build_corpus,
args=(current_app._get_current_object(), corpus.id)
)
thread.start()
return jsonify(f'Corpus "{corpus.title}" marked for building.'), 202
@bp.route('/<hashid:corpus_id>/create-share-link', methods=['POST'])
def create_share_link(corpus_id: int):
data = request.json
expiration_date = data['expiration_date']
if not isinstance(expiration_date, str):
abort(400)
role_name = data['role_name']
if not isinstance(role_name, str):
abort(400)
corpus = Corpus.query.get_or_404(corpus_id)
cfa = CorpusFollowerAssociation.query.filter_by(
corpus_id=corpus_id,
follower_id=current_user.id
).first()
if not (
cfa is not None and cfa.role.has_permission('MANAGE_FOLLOWERS')
or corpus.user == current_user
or current_user.is_administrator
):
abort(403)
_expiration_date = datetime.strptime(expiration_date, '%b %d, %Y')
cfr = CorpusFollowerRole.query.filter_by(name=role_name).first()
if cfr is None:
abort(400)
token = current_user.generate_follow_corpus_token(
corpus.hashid,
role_name,
_expiration_date
)
corpus_share_link = url_for(
'corpora.follow_corpus',
corpus_id=corpus_id,
token=token,
_external=True
)
return jsonify(corpus_share_link)
@bp.route('/<hashid:corpus_id>/analysis')
@corpus_follower_permission_required('VIEW')
def analysis(corpus_id: int):
corpus = Corpus.query.get_or_404(corpus_id)
return render_template(
'corpora/analysis.html.j2',
corpus=corpus,
title=f'Analyse Corpus {corpus.title}'
)
@bp.route('/<hashid:corpus_id>/analysis/stopwords')
def get_stopwords(corpus_id: int):
languages = [
'german',
'english',
'catalan',
'greek',
'spanish',
'french',
'italian',
'russian',
'chinese'
]
nltk.download('stopwords', quiet=True)
stopwords = {
language: nltk.corpus.stopwords.words(language)
for language in languages
}
stopwords['punctuation'] = list(punctuation)
stopwords['punctuation'] += ['', '|', '', '', '', '--']
stopwords['user_stopwords'] = []
return jsonify(stopwords)
@bp.route('/<hashid:corpus_id>/follow/<token>')
def follow_corpus(corpus_id: int, token: str):
corpus = Corpus.query.get_or_404(corpus_id)
if not current_user.follow_corpus_by_token(token):
abort(403)
db.session.commit()
flash(f'You are following "{corpus.title}" now', category='corpus')
return redirect(corpus.url)
@bp.route('/<hashid:corpus_id>/is-public', methods=['PUT'])
def update_is_public(corpus_id):
new_value = request.json
if not isinstance(new_value, bool):
abort(400)
corpus = Corpus.query.get_or_404(corpus_id)
if not (
corpus.user == current_user
or current_user.is_administrator
):
abort(403)
corpus.is_public = new_value
db.session.commit()
return jsonify(f'Corpus "{corpus.title}" is now {"public" if new_value else "private"}'), 200

View File

@ -4,11 +4,17 @@ from . import bp
@bp.app_errorhandler(HTTPException) @bp.app_errorhandler(HTTPException)
def handle_http_exception(error): def handle_http_exception(e: HTTPException):
''' Generic HTTP exception handler ''' ''' Generic HTTP exception handler '''
accept_json = request.accept_mimetypes.accept_json accept_json = request.accept_mimetypes.accept_json
accept_html = request.accept_mimetypes.accept_html accept_html = request.accept_mimetypes.accept_html
if accept_json and not accept_html: if accept_json and not accept_html:
response = jsonify(str(error)) error = {
return response, error.code 'code': e.code,
return render_template('errors/error.html.j2', error=error), error.code 'name': e.name,
'description': e.description
}
return jsonify(error), e.code
return render_template('errors/error.html.j2', error=e), e.code

View File

@ -0,0 +1,13 @@
from flask import Blueprint
bp = Blueprint('jobs', __name__)
from . import routes
from .inputs import bp as inputs_bp
bp.register_blueprint(inputs_bp, url_prefix='/<hashid:job_id>/inputs')
from .results import bp as results_bp
bp.register_blueprint(results_bp, url_prefix='/<hashid:job_id>/results')

View File

@ -1,5 +1,7 @@
from flask import Blueprint from flask import Blueprint
bp = Blueprint('auth', __name__) bp = Blueprint('inputs', __name__)
from . import routes from . import routes

View File

@ -0,0 +1,27 @@
from flask import abort, send_from_directory
from flask_login import current_user, login_required
from app.models import JobInput
from . import bp
@bp.route('/<hashid:job_input_id>/download')
@login_required
def download_job_input(job_id: int, job_input_id: int):
job_input = JobInput.query.filter_by(
job_id=job_id,
id=job_input_id
).first_or_404()
if not (
job_input.job.user == current_user
or current_user.is_administrator
):
abort(403)
return send_from_directory(
job_input.path.parent,
job_input.path.name,
as_attachment=True,
download_name=job_input.filename,
mimetype=job_input.mimetype
)

View File

@ -0,0 +1,7 @@
from flask import Blueprint
bp = Blueprint('results', __name__)
from . import routes

View File

@ -0,0 +1,27 @@
from flask import abort, send_from_directory
from flask_login import current_user, login_required
from app.models import JobResult
from . import bp
@bp.route('/<hashid:job_result_id>/download')
@login_required
def download_job_result(job_id: int, job_result_id: int):
job_result = JobResult.query.filter_by(
job_id=job_id,
id=job_result_id
).first_or_404()
if not (
job_result.job.user == current_user
or current_user.is_administrator
):
abort(403)
return send_from_directory(
job_result.path.parent,
job_result.path.name,
as_attachment=True,
download_name=job_result.filename,
mimetype=job_result.mimetype
)

View File

@ -0,0 +1,111 @@
from flask import (
abort,
current_app,
Flask,
jsonify,
redirect,
render_template,
url_for
)
from flask_login import current_user, login_required
from threading import Thread
from app import db
from app.decorators import admin_required
from app.models import Job, JobStatus
from . import bp
@bp.route('')
@login_required
def index():
return redirect(url_for('main.dashboard', _anchor='jobs'))
@bp.route('/<hashid:job_id>')
@login_required
def job(job_id: int):
job = Job.query.get_or_404(job_id)
if not (
job.user == current_user
or current_user.is_administrator
):
abort(403)
return render_template(
'jobs/job.html.j2',
title='Job',
job=job
)
def _delete_job(app: Flask, job_id: int):
with app.app_context():
job = Job.query.get(job_id)
job.delete()
db.session.commit()
@bp.route('/<hashid:job_id>', methods=['DELETE'])
@login_required
def delete_job(job_id: int):
job = Job.query.get_or_404(job_id)
if not (
job.user == current_user
or current_user.is_administrator
):
abort(403)
thread = Thread(
target=_delete_job,
args=(current_app._get_current_object(), job.id)
)
thread.start()
return jsonify(f'Job "{job.title}" marked for deletion.'), 202
@bp.route('/<hashid:job_id>/log')
@admin_required
def job_log(job_id: int):
job = Job.query.get_or_404(job_id)
if job.status not in [JobStatus.COMPLETED, JobStatus.FAILED]:
abort(409)
log_file_path = job.path / 'pipeline_data' / 'logs' / 'pyflow_log.txt'
with log_file_path.open() as log_file:
log = log_file.read()
return jsonify(log)
def _restart_job(app: Flask, job_id: int):
with app.app_context():
job = Job.query.get(job_id)
job.restart()
db.session.commit()
@bp.route('/<hashid:job_id>/restart', methods=['POST'])
@login_required
def restart_job(job_id: int):
job = Job.query.get_or_404(job_id)
if not (
job.user == current_user
or current_user.is_administrator
):
abort(403)
if job.status != JobStatus.FAILED:
abort(409)
thread = Thread(
target=_restart_job,
args=(current_app._get_current_object(), job.id)
)
thread.start()
return jsonify(f'Job "{job.title}" marked for restarting.'), 202

View File

@ -1,7 +1,9 @@
from flask import current_app from flask import current_app
from flask_migrate import upgrade from flask_migrate import upgrade
import os from pathlib import Path
from app import db
from app.models import ( from app.models import (
Corpus,
CorpusFollowerRole, CorpusFollowerRole,
Role, Role,
SpaCyNLPPipelineModel, SpaCyNLPPipelineModel,
@ -14,25 +16,22 @@ from . import bp
@bp.cli.command('deploy') @bp.cli.command('deploy')
def deploy(): def deploy():
''' Run deployment tasks. ''' ''' Run deployment tasks. '''
# Make default directories
print('Make default directories') print('Make default directories')
base_dir = current_app.config['NOPAQUE_DATA_DIR'] base_dir = current_app.config['NOPAQUE_DATA_DIR']
default_dirs = [ default_dirs: list[Path] = [
os.path.join(base_dir, 'tmp'), base_dir / 'tmp',
os.path.join(base_dir, 'users') base_dir / 'users'
] ]
for dir in default_dirs: for default_dir in default_dirs:
if os.path.exists(dir): if not default_dir.exists():
if not os.path.isdir(dir): default_dir.mkdir()
raise NotADirectoryError(f'{dir} is not a directory') if not default_dir.is_dir():
else: raise NotADirectoryError(f'{default_dir} is not a directory')
os.mkdir(dir)
# migrate database to latest revision
print('Migrate database to latest revision') print('Migrate database to latest revision')
upgrade() upgrade()
# Insert/Update default database values
print('Insert/Update default Roles') print('Insert/Update default Roles')
Role.insert_defaults() Role.insert_defaults()
print('Insert/Update default Users') print('Insert/Update default Users')
@ -44,4 +43,9 @@ def deploy():
print('Insert/Update default TesseractOCRPipelineModels') print('Insert/Update default TesseractOCRPipelineModels')
TesseractOCRPipelineModel.insert_defaults() TesseractOCRPipelineModel.insert_defaults()
print('Stop running analysis sessions')
for corpus in Corpus.query.all():
corpus.num_analysis_sessions = 0
db.session.commit()
# TODO: Implement checks for if the nopaque network exists # TODO: Implement checks for if the nopaque network exists

View File

@ -1,14 +1,11 @@
from flask import flash, redirect, render_template, url_for from flask import flash, redirect, render_template, url_for
from flask_breadcrumbs import register_breadcrumb
from flask_login import current_user, login_required, login_user from flask_login import current_user, login_required, login_user
from app.auth.forms import LoginForm from app.blueprints.auth.forms import LoginForm
from app.models import Corpus, User from app.models import Corpus, User
from sqlalchemy import or_
from . import bp from . import bp
@bp.route('/', methods=['GET', 'POST']) @bp.route('/', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.', '<i class="material-icons">home</i>')
def index(): def index():
form = LoginForm() form = LoginForm()
if form.validate_on_submit(): if form.validate_on_submit():
@ -27,7 +24,6 @@ def index():
@bp.route('/faq') @bp.route('/faq')
@register_breadcrumb(bp, '.faq', 'Frequently Asked Questions')
def faq(): def faq():
return render_template( return render_template(
'main/faq.html.j2', 'main/faq.html.j2',
@ -36,7 +32,6 @@ def faq():
@bp.route('/dashboard') @bp.route('/dashboard')
@register_breadcrumb(bp, '.dashboard', '<i class="material-icons left">dashboard</i>Dashboard')
@login_required @login_required
def dashboard(): def dashboard():
return render_template( return render_template(
@ -45,8 +40,15 @@ def dashboard():
) )
@bp.route('/manual')
def manual():
return render_template(
'main/manual.html.j2',
title='Manual'
)
@bp.route('/news') @bp.route('/news')
@register_breadcrumb(bp, '.news', '<i class="material-icons left">email</i>News')
def news(): def news():
return render_template( return render_template(
'main/news.html.j2', 'main/news.html.j2',
@ -55,7 +57,6 @@ def news():
@bp.route('/privacy_policy') @bp.route('/privacy_policy')
@register_breadcrumb(bp, '.privacy_policy', 'Private statement (GDPR)')
def privacy_policy(): def privacy_policy():
return render_template( return render_template(
'main/privacy_policy.html.j2', 'main/privacy_policy.html.j2',
@ -64,7 +65,6 @@ def privacy_policy():
@bp.route('/terms_of_use') @bp.route('/terms_of_use')
@register_breadcrumb(bp, '.terms_of_use', 'Terms of Use')
def terms_of_use(): def terms_of_use():
return render_template( return render_template(
'main/terms_of_use.html.j2', 'main/terms_of_use.html.j2',
@ -72,17 +72,14 @@ def terms_of_use():
) )
@bp.route('/social-area') @bp.route('/social')
@register_breadcrumb(bp, '.social_area', '<i class="material-icons left">group</i>Social Area')
@login_required @login_required
def social_area(): def social():
print('test')
corpora = Corpus.query.filter(Corpus.is_public == True, Corpus.user != current_user).all() corpora = Corpus.query.filter(Corpus.is_public == True, Corpus.user != current_user).all()
print(corpora)
users = User.query.filter(User.is_public == True, User.id != current_user.id).all() users = User.query.filter(User.is_public == True, User.id != current_user.id).all()
return render_template( return render_template(
'main/social_area.html.j2', 'main/social.html.j2',
title='Social Area', title='Social',
corpora=corpora, corpora=corpora,
users=users users=users
) )

View File

@ -1,12 +1,11 @@
from flask import Blueprint from flask import Blueprint
from flask_login import login_required from flask_login import login_required
import os from pathlib import Path
import yaml import yaml
services_file = \ services_file = Path(__file__).parent / 'services.yml'
os.path.join(os.path.dirname(os.path.abspath(__file__)), 'services.yml') with services_file.open('r') as f:
with open(services_file, 'r') as f:
SERVICES = yaml.safe_load(f) SERVICES = yaml.safe_load(f)
bp = Blueprint('services', __name__) bp = Blueprint('services', __name__)

View File

@ -61,7 +61,7 @@ class CreateTesseractOCRPipelineJobForm(CreateJobBaseForm):
if field.data: if field.data:
if not('methods' in service_info and 'binarization' in service_info['methods']): if not('methods' in service_info and 'binarization' in service_info['methods']):
raise ValidationError('Binarization is not available') raise ValidationError('Binarization is not available')
def validate_pdf(self, field): def validate_pdf(self, field):
if field.data.mimetype != 'application/pdf': if field.data.mimetype != 'application/pdf':
raise ValidationError('PDF files only!') raise ValidationError('PDF files only!')
@ -146,7 +146,7 @@ class CreateSpacyNLPPipelineJobForm(CreateJobBaseForm):
encoding_detection = BooleanField('Encoding detection', render_kw={'disabled': True}) encoding_detection = BooleanField('Encoding detection', render_kw={'disabled': True})
txt = FileField('File', validators=[FileRequired()]) txt = FileField('File', validators=[FileRequired()])
model = SelectField('Model', validators=[InputRequired()]) model = SelectField('Model', validators=[InputRequired()])
def validate_encoding_detection(self, field): def validate_encoding_detection(self, field):
service_info = SERVICES['spacy-nlp-pipeline']['versions'][self.version.data] service_info = SERVICES['spacy-nlp-pipeline']['versions'][self.version.data]
if field.data: if field.data:
@ -167,7 +167,6 @@ class CreateSpacyNLPPipelineJobForm(CreateJobBaseForm):
version = kwargs.pop('version', service_manifest['latest_version']) version = kwargs.pop('version', service_manifest['latest_version'])
super().__init__(*args, **kwargs) super().__init__(*args, **kwargs)
service_info = service_manifest['versions'][version] service_info = service_manifest['versions'][version]
print(service_info)
if self.encoding_detection.render_kw is None: if self.encoding_detection.render_kw is None:
self.encoding_detection.render_kw = {} self.encoding_detection.render_kw = {}
self.encoding_detection.render_kw['disabled'] = True self.encoding_detection.render_kw['disabled'] = True

View File

@ -1,5 +1,4 @@
from flask import abort, current_app, flash, Markup, redirect, render_template, request, url_for from flask import abort, current_app, flash, redirect, render_template, request, url_for
from flask_breadcrumbs import register_breadcrumb
from flask_login import current_user from flask_login import current_user
import requests import requests
from app import db, hashids from app import db, hashids
@ -20,13 +19,11 @@ from .forms import (
@bp.route('/services') @bp.route('/services')
@register_breadcrumb(bp, '.', 'Services')
def services(): def services():
return redirect(url_for('main.dashboard')) return redirect(url_for('main.dashboard'))
@bp.route('/file-setup-pipeline', methods=['GET', 'POST']) @bp.route('/file-setup-pipeline', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.file_setup_pipeline', '<i class="nopaque-icons service-icons left" data-service="file-setup-pipeline"></i>File Setup')
def file_setup_pipeline(): def file_setup_pipeline():
service = 'file-setup-pipeline' service = 'file-setup-pipeline'
service_manifest = SERVICES[service] service_manifest = SERVICES[service]
@ -56,7 +53,7 @@ def file_setup_pipeline():
abort(500) abort(500)
job.status = JobStatus.SUBMITTED job.status = JobStatus.SUBMITTED
db.session.commit() db.session.commit()
message = Markup(f'Job "<a href="{job.url}">{job.title}</a>" created') message = f'Job "<a href="{job.url}">{job.title}</a>" created'
flash(message, 'job') flash(message, 'job')
return {}, 201, {'Location': job.url} return {}, 201, {'Location': job.url}
return render_template( return render_template(
@ -67,7 +64,6 @@ def file_setup_pipeline():
@bp.route('/tesseract-ocr-pipeline', methods=['GET', 'POST']) @bp.route('/tesseract-ocr-pipeline', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.tesseract_ocr_pipeline', '<i class="nopaque-icons service-icons left" data-service="tesseract-ocr-pipeline"></i>Tesseract OCR Pipeline')
def tesseract_ocr_pipeline(): def tesseract_ocr_pipeline():
service_name = 'tesseract-ocr-pipeline' service_name = 'tesseract-ocr-pipeline'
service_manifest = SERVICES[service_name] service_manifest = SERVICES[service_name]
@ -100,7 +96,7 @@ def tesseract_ocr_pipeline():
abort(500) abort(500)
job.status = JobStatus.SUBMITTED job.status = JobStatus.SUBMITTED
db.session.commit() db.session.commit()
message = Markup(f'Job "<a href="{job.url}">{job.title}</a>" created') message = f'Job "<a href="{job.url}">{job.title}</a>" created'
flash(message, 'job') flash(message, 'job')
return {}, 201, {'Location': job.url} return {}, 201, {'Location': job.url}
tesseract_ocr_pipeline_models = [ tesseract_ocr_pipeline_models = [
@ -118,7 +114,6 @@ def tesseract_ocr_pipeline():
@bp.route('/transkribus-htr-pipeline', methods=['GET', 'POST']) @bp.route('/transkribus-htr-pipeline', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.transkribus_htr_pipeline', '<i class="nopaque-icons service-icons left" data-service="transkribus-htr-pipeline"></i>Transkribus HTR Pipeline')
def transkribus_htr_pipeline(): def transkribus_htr_pipeline():
if not current_app.config.get('NOPAQUE_TRANSKRIBUS_ENABLED'): if not current_app.config.get('NOPAQUE_TRANSKRIBUS_ENABLED'):
abort(404) abort(404)
@ -164,7 +159,7 @@ def transkribus_htr_pipeline():
abort(500) abort(500)
job.status = JobStatus.SUBMITTED job.status = JobStatus.SUBMITTED
db.session.commit() db.session.commit()
message = Markup(f'Job "<a href="{job.url}">{job.title}</a>" created') message = f'Job "<a href="{job.url}">{job.title}</a>" created'
flash(message, 'job') flash(message, 'job')
return {}, 201, {'Location': job.url} return {}, 201, {'Location': job.url}
return render_template( return render_template(
@ -176,7 +171,6 @@ def transkribus_htr_pipeline():
@bp.route('/spacy-nlp-pipeline', methods=['GET', 'POST']) @bp.route('/spacy-nlp-pipeline', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.spacy_nlp_pipeline', '<i class="nopaque-icons service-icons left" data-service="spacy-nlp-pipeline"></i>SpaCy NLP Pipeline')
def spacy_nlp_pipeline(): def spacy_nlp_pipeline():
service = 'spacy-nlp-pipeline' service = 'spacy-nlp-pipeline'
service_manifest = SERVICES[service] service_manifest = SERVICES[service]
@ -210,7 +204,7 @@ def spacy_nlp_pipeline():
abort(500) abort(500)
job.status = JobStatus.SUBMITTED job.status = JobStatus.SUBMITTED
db.session.commit() db.session.commit()
message = Markup(f'Job "<a href="{job.url}">{job.title}</a>" created') message = f'Job "<a href="{job.url}">{job.title}</a>" created'
flash(message, 'job') flash(message, 'job')
return {}, 201, {'Location': job.url} return {}, 201, {'Location': job.url}
return render_template( return render_template(
@ -223,7 +217,6 @@ def spacy_nlp_pipeline():
@bp.route('/corpus-analysis') @bp.route('/corpus-analysis')
@register_breadcrumb(bp, '.corpus_analysis', '<i class="nopaque-icons service-icons left" data-service="corpus-analysis"></i>Corpus Analysis')
def corpus_analysis(): def corpus_analysis():
return render_template( return render_template(
'services/corpus_analysis.html.j2', 'services/corpus_analysis.html.j2',

View File

@ -0,0 +1,7 @@
from flask import Blueprint
bp = Blueprint('settings', __name__)
from . import routes

View File

@ -1,6 +1,5 @@
from flask_login import current_user
from flask_wtf import FlaskForm from flask_wtf import FlaskForm
from flask_wtf.file import FileField, FileRequired from flask_wtf.file import FileField, FileRequired, FileSize
from wtforms import ( from wtforms import (
PasswordField, PasswordField,
SelectField, SelectField,
@ -17,7 +16,6 @@ from wtforms.validators import (
Regexp Regexp
) )
from app.models import User, UserSettingJobStatusMailNotificationLevel from app.models import User, UserSettingJobStatusMailNotificationLevel
from app.wtforms.validators import FileSize
class UpdateAccountInformationForm(FlaskForm): class UpdateAccountInformationForm(FlaskForm):
@ -40,8 +38,8 @@ class UpdateAccountInformationForm(FlaskForm):
] ]
) )
submit = SubmitField() submit = SubmitField()
def __init__(self, user, *args, **kwargs): def __init__(self, user: User, *args, **kwargs):
if 'data' not in kwargs: if 'data' not in kwargs:
kwargs['data'] = user.to_json_serializeable() kwargs['data'] = user.to_json_serializeable()
if 'prefix' not in kwargs: if 'prefix' not in kwargs:
@ -66,7 +64,7 @@ class UpdateProfileInformationForm(FlaskForm):
validators=[Length(max=128)] validators=[Length(max=128)]
) )
about_me = TextAreaField( about_me = TextAreaField(
'About me', 'About me',
validators=[ validators=[
Length(max=254) Length(max=254)
] ]
@ -91,7 +89,7 @@ class UpdateProfileInformationForm(FlaskForm):
) )
submit = SubmitField() submit = SubmitField()
def __init__(self, user, *args, **kwargs): def __init__(self, user: User, *args, **kwargs):
if 'data' not in kwargs: if 'data' not in kwargs:
kwargs['data'] = user.to_json_serializeable() kwargs['data'] = user.to_json_serializeable()
if 'prefix' not in kwargs: if 'prefix' not in kwargs:
@ -100,7 +98,7 @@ class UpdateProfileInformationForm(FlaskForm):
class UpdateAvatarForm(FlaskForm): class UpdateAvatarForm(FlaskForm):
avatar = FileField('File', validators=[FileRequired(), FileSize(2)]) avatar = FileField('File', validators=[FileRequired(), FileSize(2_000_000)])
submit = SubmitField() submit = SubmitField()
def validate_avatar(self, field): def validate_avatar(self, field):
@ -132,7 +130,7 @@ class UpdatePasswordForm(FlaskForm):
) )
submit = SubmitField() submit = SubmitField()
def __init__(self, user, *args, **kwargs): def __init__(self, user: User, *args, **kwargs):
if 'prefix' not in kwargs: if 'prefix' not in kwargs:
kwargs['prefix'] = 'update-password-form' kwargs['prefix'] = 'update-password-form'
super().__init__(*args, **kwargs) super().__init__(*args, **kwargs)
@ -154,7 +152,7 @@ class UpdateNotificationsForm(FlaskForm):
) )
submit = SubmitField() submit = SubmitField()
def __init__(self, user, *args, **kwargs): def __init__(self, user: User, *args, **kwargs):
if 'data' not in kwargs: if 'data' not in kwargs:
kwargs['data'] = user.to_json_serializeable() kwargs['data'] = user.to_json_serializeable()
if 'prefix' not in kwargs: if 'prefix' not in kwargs:

View File

@ -0,0 +1,158 @@
from flask import (
abort,
flash,
jsonify,
redirect,
render_template,
request,
url_for
)
from flask_login import current_user, login_required
from app import db
from app.models import Avatar
from . import bp
from .forms import (
UpdateAvatarForm,
UpdatePasswordForm,
UpdateNotificationsForm,
UpdateAccountInformationForm,
UpdateProfileInformationForm
)
@bp.route('', methods=['GET', 'POST'])
@login_required
def index():
update_account_information_form = UpdateAccountInformationForm(current_user)
update_profile_information_form = UpdateProfileInformationForm(current_user)
update_avatar_form = UpdateAvatarForm()
update_password_form = UpdatePasswordForm(current_user)
update_notifications_form = UpdateNotificationsForm(current_user)
# region handle update profile information form
if update_profile_information_form.submit.data and update_profile_information_form.validate():
current_user.about_me = update_profile_information_form.about_me.data
current_user.location = update_profile_information_form.location.data
current_user.organization = update_profile_information_form.organization.data
current_user.website = update_profile_information_form.website.data
current_user.full_name = update_profile_information_form.full_name.data
db.session.commit()
flash('Your changes have been saved')
return redirect(url_for('.index'))
# endregion handle update profile information form
# region handle update avatar form
if update_avatar_form.submit.data and update_avatar_form.validate():
try:
Avatar.create(
update_avatar_form.avatar.data,
user=current_user
)
except (AttributeError, OSError):
abort(500)
db.session.commit()
flash('Your changes have been saved')
return redirect(url_for('.index'))
# endregion handle update avatar form
# region handle update account information form
if update_account_information_form.submit.data and update_account_information_form.validate():
current_user.email = update_account_information_form.email.data
current_user.username = update_account_information_form.username.data
db.session.commit()
flash('Profile settings updated')
return redirect(url_for('.index'))
# endregion handle update account information form
# region handle update password form
if update_password_form.submit.data and update_password_form.validate():
current_user.password = update_password_form.new_password.data
db.session.commit()
flash('Your changes have been saved')
return redirect(url_for('.index'))
# endregion handle update password form
# region handle update notifications form
if update_notifications_form.submit.data and update_notifications_form.validate():
current_user.setting_job_status_mail_notification_level = \
update_notifications_form.job_status_mail_notification_level.data
db.session.commit()
flash('Your changes have been saved')
return redirect(url_for('.index'))
# endregion handle update notifications form
return render_template(
'settings/index.html.j2',
title='Settings',
update_account_information_form=update_account_information_form,
update_avatar_form=update_avatar_form,
update_notifications_form=update_notifications_form,
update_password_form=update_password_form,
update_profile_information_form=update_profile_information_form,
user=current_user
)
@bp.route('/profile-is-public', methods=['PUT'])
@login_required
def update_profile_is_public():
new_value = request.json
if not isinstance(new_value, bool):
abort(400)
current_user.is_public = new_value
db.session.commit()
return jsonify('Your changes have been saved'), 200
@bp.route('/profile-show-email', methods=['PUT'])
@login_required
def update_profile_show_email():
new_value = request.json
if not isinstance(new_value, bool):
abort(400)
if new_value:
current_user.add_profile_privacy_setting('SHOW_EMAIL')
else:
current_user.remove_profile_privacy_setting('SHOW_EMAIL')
db.session.commit()
return jsonify('Your changes have been saved'), 200
@bp.route('/profile-show-last-seen', methods=['PUT'])
@login_required
def update_profile_show_last_seen():
new_value = request.json
if not isinstance(new_value, bool):
abort(400)
if new_value:
current_user.add_profile_privacy_setting('SHOW_LAST_SEEN')
else:
current_user.remove_profile_privacy_setting('SHOW_LAST_SEEN')
db.session.commit()
return jsonify('Your changes have been saved'), 200
@bp.route('/profile-show-member-since', methods=['PUT'])
@login_required
def update_profile_show_member_since():
new_value = request.json
if not isinstance(new_value, bool):
abort(400)
if new_value:
current_user.add_profile_privacy_setting('SHOW_MEMBER_SINCE')
else:
current_user.remove_profile_privacy_setting('SHOW_MEMBER_SINCE')
db.session.commit()
return jsonify('Your changes have been saved'), 200

View File

@ -0,0 +1,7 @@
from flask import Blueprint
bp = Blueprint('users', __name__)
from . import cli, events, routes

View File

@ -1,6 +1,4 @@
from app.models import User from app.models import User
import os
import shutil
from app import db from app import db
from . import bp from . import bp

View File

@ -0,0 +1,58 @@
from flask_login import current_user
from flask_socketio import join_room, leave_room
from app import hashids, socketio
from app.decorators import socketio_login_required
from app.models import User
@socketio.on('SUBSCRIBE User')
@socketio_login_required
def subscribe(user_hashid: str) -> dict:
if not isinstance(user_hashid, str):
return {'status': 400, 'statusText': 'Bad Request'}
user_id = hashids.decode(user_hashid)
if not isinstance(user_id, int):
return {'status': 400, 'statusText': 'Bad Request'}
user = User.query.get(user_id)
if user is None:
return {'status': 404, 'statusText': 'Not Found'}
if not (
user == current_user
or current_user.is_administrator
):
return {'status': 403, 'statusText': 'Forbidden'}
join_room(f'/users/{user.hashid}')
return {'status': 200, 'statusText': 'OK'}
@socketio.on('UNSUBSCRIBE User')
@socketio_login_required
def unsubscribe(user_hashid: str) -> dict:
if not isinstance(user_hashid, str):
return {'status': 400, 'statusText': 'Bad Request'}
user_id = hashids.decode(user_hashid)
if not isinstance(user_id, int):
return {'status': 400, 'statusText': 'Bad Request'}
user = User.query.get(user_id)
if user is None:
return {'status': 404, 'statusText': 'Not Found'}
if not (
user == current_user
or current_user.is_administrator
):
return {'status': 403, 'statusText': 'Forbidden'}
leave_room(f'/users/{user.hashid}')
return {'status': 200, 'statusText': 'OK'}

View File

@ -0,0 +1,150 @@
from flask import (
abort,
current_app,
Flask,
jsonify,
redirect,
render_template,
request,
send_from_directory,
url_for
)
from flask_login import current_user, login_required, logout_user
from threading import Thread
from app import db
from app.models import Avatar, User
from . import bp
@bp.route('')
@login_required
def index():
return redirect(url_for('main.social_area', _anchor='users'))
@bp.route('/<hashid:user_id>')
@login_required
def user(user_id: int):
user = User.query.get_or_404(user_id)
if not (
user.is_public
or user == current_user
or current_user.is_administrator
):
abort(403)
accept_json = request.accept_mimetypes.accept_json
accept_html = request.accept_mimetypes.accept_html
if accept_json and not accept_html:
return user.to_json_serializeable(
backrefs=True,
relationships=True
)
return render_template(
'users/user.html.j2',
title=user.username,
user=user
)
def _delete_user(app: Flask, user_id: int):
with app.app_context():
user = User.query.get(user_id)
user.delete()
db.session.commit()
@bp.route('/<hashid:user_id>', methods=['DELETE'])
@login_required
def delete_user(user_id: int):
user = User.query.get_or_404(user_id)
if not (
user == current_user
or current_user.is_administrator
):
abort(403)
if user == current_user:
logout_user()
thread = Thread(
target=_delete_user,
args=(current_app._get_current_object(), user.id)
)
thread.start()
return jsonify(f'User "{user.username}" marked for deletion'), 202
@bp.route('/<hashid:user_id>/avatar')
@login_required
def user_avatar(user_id: int):
user = User.query.get_or_404(user_id)
if not (
user.is_public
or user == current_user
or current_user.is_administrator
):
abort(403)
if user.avatar is None:
return redirect(url_for('static', filename='images/user_avatar.png'))
return send_from_directory(
user.avatar.path.parent,
user.avatar.path.name,
as_attachment=True,
download_name=user.avatar.filename,
mimetype=user.avatar.mimetype
)
def _delete_avatar(app: Flask, avatar_id: int):
with app.app_context():
avatar = Avatar.query.get(avatar_id)
avatar.delete()
db.session.commit()
@bp.route('/<hashid:user_id>/avatar', methods=['DELETE'])
@login_required
def delete_user_avatar(user_id: int):
user = User.query.get_or_404(user_id)
if user.avatar is None:
abort(409)
if not (
user == current_user
or current_user.is_administrator
):
abort(403)
thread = Thread(
target=_delete_avatar,
args=(current_app._get_current_object(), user.avatar.id)
)
thread.start()
return jsonify('Avatar marked for deletion'), 202
# TODO: Move this to main blueprint(?)
@bp.route('/accept-terms-of-use', methods=['POST'])
@login_required
def accept_terms_of_use():
if not (
current_user.is_authenticated
or current_user.confirmed
):
abort(403)
current_user.terms_of_use_accepted = True
db.session.commit()
return jsonify('You accepted the terms of use'), 202

View File

@ -1,16 +1,13 @@
from flask import redirect, render_template, url_for from flask import redirect, render_template, url_for
from flask_breadcrumbs import register_breadcrumb
from . import bp from . import bp
@bp.route('') @bp.route('')
@register_breadcrumb(bp, '.', '<i class="material-icons left">business_center</i>Workshops')
def workshops(): def workshops():
return redirect(url_for('main.dashboard')) return redirect(url_for('main.dashboard'))
@bp.route('/fgho_sommerschule_2023') @bp.route('/fgho_sommerschule_2023')
@register_breadcrumb(bp, '.fgho_sommerschule_2023', 'FGHO Sommerschule 2023')
def fgho_sommerschule_2023(): def fgho_sommerschule_2023():
return render_template( return render_template(
'workshops/fgho_sommerschule_2023.html.j2', 'workshops/fgho_sommerschule_2023.html.j2',

View File

@ -1,23 +0,0 @@
from flask import Blueprint
from flask_login import login_required
bp = Blueprint('contributions', __name__)
@bp.before_request
@login_required
def before_request():
'''
Ensures that the routes in this package can only be visited by users that
are logged in.
'''
pass
from . import (
routes,
spacy_nlp_pipeline_models,
tesseract_ocr_pipeline_models,
transkribus_htr_pipeline_models
)

View File

@ -1,9 +0,0 @@
from flask import redirect, url_for
from flask_breadcrumbs import register_breadcrumb
from . import bp
@bp.route('')
@register_breadcrumb(bp, '.', '<i class="material-icons left">new_label</i>My Contributions')
def contributions():
return redirect(url_for('main.dashboard', _anchor='contributions'))

View File

@ -1,13 +0,0 @@
from flask import request, url_for
from app.models import SpaCyNLPPipelineModel
def spacy_nlp_pipeline_model_dlc():
snpm_id = request.view_args['spacy_nlp_pipeline_model_id']
snpm = SpaCyNLPPipelineModel.query.get_or_404(snpm_id)
return [
{
'text': f'{snpm.title} {snpm.version}',
'url': url_for('.spacy_nlp_pipeline_model', spacy_nlp_pipeline_model_id=snpm_id)
}
]

View File

@ -1,2 +0,0 @@
from .. import bp
from . import json_routes, routes

View File

@ -1,13 +0,0 @@
from flask import request, url_for
from app.models import TesseractOCRPipelineModel
def tesseract_ocr_pipeline_model_dlc():
topm_id = request.view_args['tesseract_ocr_pipeline_model_id']
topm = TesseractOCRPipelineModel.query.get_or_404(topm_id)
return [
{
'text': f'{topm.title} {topm.version}',
'url': url_for('.tesseract_ocr_pipeline_model', tesseract_ocr_pipeline_model_id=topm_id)
}
]

View File

@ -1,2 +0,0 @@
from .. import bp
from . import routes

View File

@ -1,7 +0,0 @@
from flask import abort
from . import bp
@bp.route('/transkribus_htr_pipeline_models')
def transkribus_htr_pipeline_models():
return abort(503)

View File

@ -1,81 +1,69 @@
from datetime import datetime
from flask import current_app from flask import current_app
from pathlib import Path
import json
import shutil
from app import db from app import db
from app.models import User, Corpus, CorpusFile from app.models import User, Corpus, CorpusFile
from datetime import datetime
import json
import os
import shutil
class SandpaperConverter: class SandpaperConverter:
def __init__(self, json_db_file, data_dir): def __init__(self, json_db_file: Path, data_dir: Path):
self.json_db_file = json_db_file self.json_db_file = json_db_file
self.data_dir = data_dir self.data_dir = data_dir
def run(self): def run(self):
with open(self.json_db_file, 'r') as f: with self.json_db_file.open('r') as f:
json_db = json.loads(f.read()) json_db: list[dict] = json.load(f)
for json_user in json_db: for json_user in json_db:
if not json_user['confirmed']: if not json_user['confirmed']:
current_app.logger.info(f'Skip unconfirmed user {json_user["username"]}') current_app.logger.info(f'Skip unconfirmed user {json_user["username"]}')
continue continue
user_dir = os.path.join(self.data_dir, str(json_user['id'])) user_dir = self.data_dir / f'{json_user["id"]}'
self.convert_user(json_user, user_dir) self.convert_user(json_user, user_dir)
db.session.commit() db.session.commit()
def convert_user(self, json_user, user_dir): def convert_user(self, json_user: dict, user_dir: Path):
current_app.logger.info(f'Create User {json_user["username"]}...') current_app.logger.info(f'Create User {json_user["username"]}...')
user = User(
confirmed=json_user['confirmed'],
email=json_user['email'],
last_seen=datetime.fromtimestamp(json_user['last_seen']),
member_since=datetime.fromtimestamp(json_user['member_since']),
password_hash=json_user['password_hash'], # TODO: Needs to be added manually
username=json_user['username']
)
db.session.add(user)
db.session.flush(objects=[user])
db.session.refresh(user)
try: try:
user.makedirs() user = User.create(
except OSError as e: confirmed=json_user['confirmed'],
current_app.logger.error(e) email=json_user['email'],
db.session.rollback() last_seen=datetime.fromtimestamp(json_user['last_seen']),
member_since=datetime.fromtimestamp(json_user['member_since']),
password_hash=json_user['password_hash'], # TODO: Needs to be added manually
username=json_user['username']
)
except OSError:
raise Exception('Internal Server Error') raise Exception('Internal Server Error')
for json_corpus in json_user['corpora'].values(): for json_corpus in json_user['corpora'].values():
if not json_corpus['files'].values(): if not json_corpus['files'].values():
current_app.logger.info(f'Skip empty corpus {json_corpus["title"]}') current_app.logger.info(f'Skip empty corpus {json_corpus["title"]}')
continue continue
corpus_dir = os.path.join(user_dir, 'corpora', str(json_corpus['id'])) corpus_dir = user_dir / 'corpora' / f'{json_corpus["id"]}'
self.convert_corpus(json_corpus, user, corpus_dir) self.convert_corpus(json_corpus, user, corpus_dir)
current_app.logger.info('Done') current_app.logger.info('Done')
def convert_corpus(self, json_corpus, user, corpus_dir): def convert_corpus(self, json_corpus: dict, user: User, corpus_dir: Path):
current_app.logger.info(f'Create Corpus {json_corpus["title"]}...') current_app.logger.info(f'Create Corpus {json_corpus["title"]}...')
corpus = Corpus(
user=user,
creation_date=datetime.fromtimestamp(json_corpus['creation_date']),
description=json_corpus['description'],
title=json_corpus['title']
)
db.session.add(corpus)
db.session.flush(objects=[corpus])
db.session.refresh(corpus)
try: try:
corpus.makedirs() corpus = Corpus.create(
except OSError as e: user=user,
current_app.logger.error(e) creation_date=datetime.fromtimestamp(json_corpus['creation_date']),
db.session.rollback() description=json_corpus['description'],
title=json_corpus['title']
)
except OSError:
raise Exception('Internal Server Error') raise Exception('Internal Server Error')
for json_corpus_file in json_corpus['files'].values(): for json_corpus_file in json_corpus['files'].values():
self.convert_corpus_file(json_corpus_file, corpus, corpus_dir) self.convert_corpus_file(json_corpus_file, corpus, corpus_dir)
current_app.logger.info('Done') current_app.logger.info('Done')
def convert_corpus_file(self, json_corpus_file, corpus, corpus_dir): def convert_corpus_file(self, json_corpus_file: dict, corpus: Corpus, corpus_dir: Path):
current_app.logger.info(f'Create CorpusFile {json_corpus_file["title"]}...') current_app.logger.info(f'Create CorpusFile {json_corpus_file["title"]}...')
corpus_file = CorpusFile( corpus_file = CorpusFile(
corpus=corpus, corpus=corpus,
@ -99,13 +87,13 @@ class SandpaperConverter:
db.session.refresh(corpus_file) db.session.refresh(corpus_file)
try: try:
shutil.copy2( shutil.copy2(
os.path.join(corpus_dir, json_corpus_file['filename']), corpus_dir / json_corpus_file['filename'],
corpus_file.path corpus_file.path
) )
except: except:
current_app.logger.warning( current_app.logger.warning(
'Can not convert corpus file: ' 'Can not convert corpus file: '
f'{os.path.join(corpus_dir, json_corpus_file["filename"])}' f'{corpus_dir / json_corpus_file["filename"]}'
' -> ' ' -> '
f'{corpus_file.path}' f'{corpus_file.path}'
) )

View File

@ -1,69 +1,25 @@
from flask import current_app from flask import current_app
from pathlib import Path
def normalize_vrt_file(input_file, output_file): def normalize_vrt_file(input_file: Path, output_file: Path):
def check_pos_attribute_order(vrt_lines):
# The following orders are possible:
# since 26.02.2019: 'word,lemma,simple_pos,pos,ner'
# since 26.03.2021: 'word,pos,lemma,simple_pos,ner'
# since 27.01.2022: 'word,pos,lemma,simple_pos'
# This Function tries to find out which order we have by looking at the
# number of attributes and the position of the simple_pos attribute
SIMPLE_POS_LABELS = [
'ADJ', 'ADP', 'ADV', 'AUX', 'CONJ',
'DET', 'INTJ', 'NOUN', 'NUM', 'PART',
'PRON', 'PROPN', 'PUNCT', 'SCONJ', 'SYM',
'VERB', 'X'
]
for line in vrt_lines:
if line.startswith('<'):
continue
pos_attrs = line.rstrip('\n').split('\t')
num_pos_attrs = len(pos_attrs)
if num_pos_attrs == 4:
if pos_attrs[3] in SIMPLE_POS_LABELS:
return ['word', 'pos', 'lemma', 'simple_pos']
continue
elif num_pos_attrs == 5:
if pos_attrs[2] in SIMPLE_POS_LABELS:
return ['word', 'lemma', 'simple_pos', 'pos', 'ner']
elif pos_attrs[3] in SIMPLE_POS_LABELS:
return ['word', 'pos', 'lemma', 'simple_pos', 'ner']
continue
return None
def check_has_ent_as_s_attr(vrt_lines):
for line in vrt_lines:
if line.startswith('<ent'):
return True
return False
def pos_attrs_to_string_1(pos_attrs):
return f'{pos_attrs[0]}\t{pos_attrs[3]}\t{pos_attrs[1]}\t{pos_attrs[2]}\n'
def pos_attrs_to_string_2(pos_attrs):
return f'{pos_attrs[0]}\t{pos_attrs[1]}\t{pos_attrs[2]}\t{pos_attrs[3]}\n'
current_app.logger.info(f'Converting {input_file}...') current_app.logger.info(f'Converting {input_file}...')
with open(input_file) as f: with input_file.open() as f:
input_vrt_lines = f.readlines() input_vrt_lines = f.readlines()
pos_attr_order = check_pos_attribute_order(input_vrt_lines) pos_attr_order = _check_pos_attribute_order(input_vrt_lines)
has_ent_as_s_attr = check_has_ent_as_s_attr(input_vrt_lines) has_ent_as_s_attr = _check_has_ent_as_s_attr(input_vrt_lines)
current_app.logger.info(f'Detected pos_attr_order: [{",".join(pos_attr_order)}]') current_app.logger.info(f'Detected pos_attr_order: [{",".join(pos_attr_order)}]')
current_app.logger.info(f'Detected has_ent_as_s_attr: {has_ent_as_s_attr}') current_app.logger.info(f'Detected has_ent_as_s_attr: {has_ent_as_s_attr}')
if pos_attr_order == ['word', 'lemma', 'simple_pos', 'pos', 'ner']: if pos_attr_order == ['word', 'lemma', 'simple_pos', 'pos', 'ner']:
pos_attrs_to_string_function = pos_attrs_to_string_1 pos_attrs_to_string_function = _pos_attrs_to_string_1
elif pos_attr_order == ['word', 'pos', 'lemma', 'simple_pos', 'ner']: elif pos_attr_order == ['word', 'pos', 'lemma', 'simple_pos', 'ner']:
pos_attrs_to_string_function = pos_attrs_to_string_2 pos_attrs_to_string_function = _pos_attrs_to_string_2
elif pos_attr_order == ['word', 'pos', 'lemma', 'simple_pos']: elif pos_attr_order == ['word', 'pos', 'lemma', 'simple_pos']:
pos_attrs_to_string_function = pos_attrs_to_string_2 pos_attrs_to_string_function = _pos_attrs_to_string_2
else: else:
raise Exception('Can not handle format') raise Exception('Can not handle format')
@ -113,5 +69,49 @@ def normalize_vrt_file(input_file, output_file):
current_ent = pos_attrs[4] current_ent = pos_attrs[4]
output_vrt += pos_attrs_to_string_function(pos_attrs) output_vrt += pos_attrs_to_string_function(pos_attrs)
with open(output_file, 'w') as f: with output_file.open(mode='w') as f:
f.write(output_vrt) f.write(output_vrt)
def _check_pos_attribute_order(vrt_lines: list[str]) -> list[str]:
# The following orders are possible:
# since 26.02.2019: 'word,lemma,simple_pos,pos,ner'
# since 26.03.2021: 'word,pos,lemma,simple_pos,ner'
# since 27.01.2022: 'word,pos,lemma,simple_pos'
# This Function tries to find out which order we have by looking at the
# number of attributes and the position of the simple_pos attribute
SIMPLE_POS_LABELS = [
'ADJ', 'ADP', 'ADV', 'AUX', 'CONJ', 'DET', 'INTJ', 'NOUN', 'NUM',
'PART', 'PRON', 'PROPN', 'PUNCT', 'SCONJ', 'SYM', 'VERB', 'X'
]
for line in vrt_lines:
if line.startswith('<'):
continue
pos_attrs = line.rstrip('\n').split('\t')
num_pos_attrs = len(pos_attrs)
if num_pos_attrs == 4:
if pos_attrs[3] in SIMPLE_POS_LABELS:
return ['word', 'pos', 'lemma', 'simple_pos']
continue
elif num_pos_attrs == 5:
if pos_attrs[2] in SIMPLE_POS_LABELS:
return ['word', 'lemma', 'simple_pos', 'pos', 'ner']
elif pos_attrs[3] in SIMPLE_POS_LABELS:
return ['word', 'pos', 'lemma', 'simple_pos', 'ner']
continue
# TODO: raise exception "can't determine attribute order"
def _check_has_ent_as_s_attr(vrt_lines: list[str]) -> bool:
for line in vrt_lines:
if line.startswith('<ent'):
return True
return False
def _pos_attrs_to_string_1(pos_attrs: list[str]) -> str:
return f'{pos_attrs[0]}\t{pos_attrs[3]}\t{pos_attrs[1]}\t{pos_attrs[2]}\n'
def _pos_attrs_to_string_2(pos_attrs: list[str]) -> str:
return f'{pos_attrs[0]}\t{pos_attrs[1]}\t{pos_attrs[2]}\t{pos_attrs[3]}\n'

View File

@ -1,131 +0,0 @@
from cqi.models.corpora import Corpus as CQiCorpus
from cqi.models.subcorpora import Subcorpus as CQiSubcorpus
from typing import Dict, List
def lookups_by_cpos(corpus: CQiCorpus, cpos_list: List[int]) -> Dict:
lookups = {}
lookups['cpos_lookup'] = {cpos: {} for cpos in cpos_list}
for attr in corpus.positional_attributes.list():
cpos_attr_values: List[str] = attr.values_by_cpos(cpos_list)
for i, cpos in enumerate(cpos_list):
lookups['cpos_lookup'][cpos][attr.name] = cpos_attr_values[i]
for attr in corpus.structural_attributes.list():
# We only want to iterate over non subattributes, identifiable by
# attr.has_values == False
if attr.has_values:
continue
cpos_attr_ids: List[int] = attr.ids_by_cpos(cpos_list)
for i, cpos in enumerate(cpos_list):
if cpos_attr_ids[i] == -1:
continue
lookups['cpos_lookup'][cpos][attr.name] = cpos_attr_ids[i]
occured_attr_ids = [x for x in set(cpos_attr_ids) if x != -1]
if len(occured_attr_ids) == 0:
continue
subattrs = corpus.structural_attributes.list(filters={'part_of': attr})
if len(subattrs) == 0:
continue
lookup_name: str = f'{attr.name}_lookup'
lookups[lookup_name] = {}
for attr_id in occured_attr_ids:
lookups[lookup_name][attr_id] = {}
for subattr in subattrs:
subattr_name = subattr.name[(len(attr.name) + 1):] # noqa
for i, subattr_value in enumerate(subattr.values_by_ids(occured_attr_ids)): # noqa
lookups[lookup_name][occured_attr_ids[i]][subattr_name] = subattr_value # noqa
return lookups
def partial_export_subcorpus(
subcorpus: CQiSubcorpus,
match_id_list: List[int],
context: int = 25
) -> Dict:
if subcorpus.size == 0:
return {"matches": []}
match_boundaries = []
for match_id in match_id_list:
if match_id < 0 or match_id >= subcorpus.size:
continue
match_boundaries.append(
(
match_id,
subcorpus.dump(subcorpus.fields['match'], match_id, match_id)[0],
subcorpus.dump(subcorpus.fields['matchend'], match_id, match_id)[0]
)
)
cpos_set = set()
matches = []
for match_boundary in match_boundaries:
match_num, match_start, match_end = match_boundary
c = (match_start, match_end)
if match_start == 0 or context == 0:
lc = None
cpos_list_lbound = match_start
else:
lc_lbound = max(0, (match_start - context))
lc_rbound = match_start - 1
lc = (lc_lbound, lc_rbound)
cpos_list_lbound = lc_lbound
if match_end == (subcorpus.collection.corpus.size - 1) or context == 0:
rc = None
cpos_list_rbound = match_end
else:
rc_lbound = match_end + 1
rc_rbound = min(
(match_end + context),
(subcorpus.collection.corpus.size - 1)
)
rc = (rc_lbound, rc_rbound)
cpos_list_rbound = rc_rbound
match = {'num': match_num, 'lc': lc, 'c': c, 'rc': rc}
matches.append(match)
cpos_set.update(range(cpos_list_lbound, cpos_list_rbound + 1))
lookups = lookups_by_cpos(subcorpus.collection.corpus, list(cpos_set))
return {'matches': matches, **lookups}
def export_subcorpus(
subcorpus: CQiSubcorpus,
context: int = 25,
cutoff: float = float('inf'),
offset: int = 0
) -> Dict:
if subcorpus.size == 0:
return {"matches": []}
first_match = max(0, offset)
last_match = min((offset + cutoff - 1), (subcorpus.size - 1))
match_boundaries = zip(
range(first_match, last_match + 1),
subcorpus.dump(subcorpus.fields['match'], first_match, last_match),
subcorpus.dump(subcorpus.fields['matchend'], first_match, last_match)
)
cpos_set = set()
matches = []
for match_num, match_start, match_end in match_boundaries:
c = (match_start, match_end)
if match_start == 0 or context == 0:
lc = None
cpos_list_lbound = match_start
else:
lc_lbound = max(0, (match_start - context))
lc_rbound = match_start - 1
lc = (lc_lbound, lc_rbound)
cpos_list_lbound = lc_lbound
if match_end == (subcorpus.collection.corpus.size - 1) or context == 0:
rc = None
cpos_list_rbound = match_end
else:
rc_lbound = match_end + 1
rc_rbound = min(
(match_end + context),
(subcorpus.collection.corpus.size - 1)
)
rc = (rc_lbound, rc_rbound)
cpos_list_rbound = rc_rbound
match = {'num': match_num, 'lc': lc, 'c': c, 'rc': rc}
matches.append(match)
cpos_set.update(range(cpos_list_lbound, cpos_list_rbound + 1))
lookups = lookups_by_cpos(subcorpus.collection.corpus, list(cpos_set))
return {'matches': matches, **lookups}

View File

@ -1,45 +0,0 @@
from flask_login import current_user
from flask_socketio import join_room
from app import hashids, socketio
from app.decorators import socketio_login_required
from app.models import Corpus
@socketio.on('GET /corpora/<corpus_id>')
@socketio_login_required
def get_corpus(corpus_hashid):
corpus_id = hashids.decode(corpus_hashid)
corpus = Corpus.query.get(corpus_id)
if corpus is None:
return {'options': {'status': 404, 'statusText': 'Not found'}}
if not (
corpus.is_public
or corpus.user == current_user
or current_user.is_administrator()
):
return {'options': {'status': 403, 'statusText': 'Forbidden'}}
return {
'body': corpus.to_json_serializable(),
'options': {
'status': 200,
'statusText': 'OK',
'headers': {'Content-Type: application/json'}
}
}
@socketio.on('SUBSCRIBE /corpora/<corpus_id>')
@socketio_login_required
def subscribe_corpus(corpus_hashid):
corpus_id = hashids.decode(corpus_hashid)
corpus = Corpus.query.get(corpus_id)
if corpus is None:
return {'options': {'status': 404, 'statusText': 'Not found'}}
if not (
corpus.is_public
or corpus.user == current_user
or current_user.is_administrator()
):
return {'options': {'status': 403, 'statusText': 'Forbidden'}}
join_room(f'/corpora/{corpus.hashid}')
return {'options': {'status': 200, 'statusText': 'OK'}}

View File

@ -1,2 +0,0 @@
from .. import bp
from . import json_routes, routes

View File

@ -1,15 +0,0 @@
from flask import request, url_for
from app.models import CorpusFile
from ..utils import corpus_endpoint_arguments_constructor as corpus_eac
def corpus_file_dynamic_list_constructor():
corpus_id = request.view_args['corpus_id']
corpus_file_id = request.view_args['corpus_file_id']
corpus_file = CorpusFile.query.filter_by(corpus_id=corpus_id, id=corpus_file_id).first_or_404()
return [
{
'text': f'{corpus_file.author}: {corpus_file.title} ({corpus_file.publishing_year})',
'url': url_for('.corpus_file', corpus_id=corpus_id, corpus_file_id=corpus_file_id)
}
]

View File

@ -1,125 +0,0 @@
from datetime import datetime
from flask import abort, current_app, request, url_for
from flask_login import current_user
from threading import Thread
from app import db
from app.decorators import content_negotiation
from app.models import Corpus, CorpusFollowerRole
from . import bp
from .decorators import corpus_follower_permission_required, corpus_owner_or_admin_required
import nltk
from string import punctuation
@bp.route('/<hashid:corpus_id>', methods=['DELETE'])
@corpus_owner_or_admin_required
@content_negotiation(produces='application/json')
def delete_corpus(corpus_id):
def _delete_corpus(app, corpus_id):
with app.app_context():
corpus = Corpus.query.get(corpus_id)
corpus.delete()
db.session.commit()
corpus = Corpus.query.get_or_404(corpus_id)
thread = Thread(
target=_delete_corpus,
args=(current_app._get_current_object(), corpus.id)
)
thread.start()
response_data = {
'message': f'Corpus "{corpus.title}" marked for deletion',
'category': 'corpus'
}
return response_data, 200
@bp.route('/<hashid:corpus_id>/build', methods=['POST'])
@corpus_follower_permission_required('MANAGE_FILES')
@content_negotiation(produces='application/json')
def build_corpus(corpus_id):
def _build_corpus(app, corpus_id):
with app.app_context():
corpus = Corpus.query.get(corpus_id)
corpus.build()
db.session.commit()
corpus = Corpus.query.get_or_404(corpus_id)
if len(corpus.files.all()) == 0:
abort(409)
thread = Thread(
target=_build_corpus,
args=(current_app._get_current_object(), corpus_id)
)
thread.start()
response_data = {
'message': f'Corpus "{corpus.title}" marked for building',
'category': 'corpus'
}
return response_data, 202
@bp.route('/stopwords')
@content_negotiation(produces='application/json')
def get_stopwords():
nltk.download('stopwords', quiet=True)
languages = ["german", "english", "catalan", "greek", "spanish", "french", "italian", "russian", "chinese"]
stopwords = {}
for language in languages:
stopwords[language] = nltk.corpus.stopwords.words(language)
stopwords['punctuation'] = list(punctuation) + ['', '|', '', '', '', '--']
stopwords['user_stopwords'] = []
response_data = stopwords
return response_data, 202
@bp.route('/<hashid:corpus_id>/generate-share-link', methods=['POST'])
@corpus_follower_permission_required('MANAGE_FOLLOWERS')
@content_negotiation(consumes='application/json', produces='application/json')
def generate_corpus_share_link(corpus_id):
data = request.json
if not isinstance(data, dict):
abort(400)
expiration = data.get('expiration')
if not isinstance(expiration, str):
abort(400)
role_name = data.get('role')
if not isinstance(role_name, str):
abort(400)
expiration_date = datetime.strptime(expiration, '%b %d, %Y')
cfr = CorpusFollowerRole.query.filter_by(name=role_name).first()
if cfr is None:
abort(400)
corpus = Corpus.query.get_or_404(corpus_id)
token = current_user.generate_follow_corpus_token(corpus.hashid, role_name, expiration_date)
corpus_share_link = url_for(
'corpora.follow_corpus',
corpus_id=corpus_id,
token=token,
_external=True
)
response_data = {
'message': 'Corpus share link generated',
'category': 'corpus',
'corpusShareLink': corpus_share_link
}
return response_data, 200
@bp.route('/<hashid:corpus_id>/is_public', methods=['PUT'])
@corpus_owner_or_admin_required
@content_negotiation(consumes='application/json', produces='application/json')
def update_corpus_is_public(corpus_id):
is_public = request.json
if not isinstance(is_public, bool):
abort(400)
corpus = Corpus.query.get_or_404(corpus_id)
corpus.is_public = is_public
db.session.commit()
response_data = {
'message': (
f'Corpus "{corpus.title}" is now'
f' {"public" if is_public else "private"}'
),
'category': 'corpus'
}
return response_data, 200

View File

@ -1,120 +0,0 @@
from flask import abort, flash, redirect, render_template, url_for
from flask_breadcrumbs import register_breadcrumb
from flask_login import current_user
from app import db
from app.models import (
Corpus,
CorpusFollowerAssociation,
CorpusFollowerRole,
User
)
from . import bp
from .decorators import corpus_follower_permission_required
from .forms import CreateCorpusForm
from .utils import (
corpus_endpoint_arguments_constructor as corpus_eac,
corpus_dynamic_list_constructor as corpus_dlc
)
@bp.route('')
@register_breadcrumb(bp, '.', '<i class="nopaque-icons left">I</i>My Corpora')
def corpora():
return redirect(url_for('main.dashboard', _anchor='corpora'))
@bp.route('/create', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.create', 'Create')
def create_corpus():
form = CreateCorpusForm()
if form.validate_on_submit():
try:
corpus = Corpus.create(
title=form.title.data,
description=form.description.data,
user=current_user
)
except OSError:
abort(500)
db.session.commit()
flash(f'Corpus "{corpus.title}" created', 'corpus')
return redirect(corpus.url)
return render_template(
'corpora/create.html.j2',
title='Create corpus',
form=form
)
@bp.route('/<hashid:corpus_id>')
@register_breadcrumb(bp, '.entity', '', dynamic_list_constructor=corpus_dlc)
def corpus(corpus_id):
corpus = Corpus.query.get_or_404(corpus_id)
cfrs = CorpusFollowerRole.query.all()
# TODO: Better solution for filtering admin
users = User.query.filter(User.is_public == True, User.id != current_user.id, User.id != corpus.user.id, User.role_id < 4).all()
cfa = CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=current_user.id).first()
if cfa is None:
if corpus.user == current_user or current_user.is_administrator():
cfr = CorpusFollowerRole.query.filter_by(name='Administrator').first()
else:
cfr = CorpusFollowerRole.query.filter_by(name='Anonymous').first()
else:
cfr = cfa.role
if corpus.user == current_user or current_user.is_administrator():
return render_template(
'corpora/corpus.html.j2',
title=corpus.title,
corpus=corpus,
cfr=cfr,
cfrs=cfrs,
users=users
)
if (current_user.is_following_corpus(corpus) or corpus.is_public):
cfas = CorpusFollowerAssociation.query.filter(Corpus.id == corpus_id, CorpusFollowerAssociation.follower_id != corpus.user.id).all()
return render_template(
'corpora/public_corpus.html.j2',
title=corpus.title,
corpus=corpus,
cfrs=cfrs,
cfr=cfr,
cfas=cfas,
users=users
)
abort(403)
@bp.route('/<hashid:corpus_id>/analysis')
@corpus_follower_permission_required('VIEW')
@register_breadcrumb(bp, '.entity.analysis', 'Analysis', endpoint_arguments_constructor=corpus_eac)
def analysis(corpus_id):
corpus = Corpus.query.get_or_404(corpus_id)
return render_template(
'corpora/analysis.html.j2',
corpus=corpus,
title=f'Analyse Corpus {corpus.title}'
)
@bp.route('/<hashid:corpus_id>/follow/<token>')
def follow_corpus(corpus_id, token):
corpus = Corpus.query.get_or_404(corpus_id)
if current_user.follow_corpus_by_token(token):
db.session.commit()
flash(f'You are following "{corpus.title}" now', category='corpus')
return redirect(url_for('corpora.corpus', corpus_id=corpus_id))
abort(403)
@bp.route('/import', methods=['GET', 'POST'])
@register_breadcrumb(bp, '.import', 'Import')
def import_corpus():
abort(503)
@bp.route('/<hashid:corpus_id>/export')
@corpus_follower_permission_required('VIEW')
@register_breadcrumb(bp, '.entity.export', 'Export', endpoint_arguments_constructor=corpus_eac)
def export_corpus(corpus_id):
abort(503)

View File

@ -1,17 +0,0 @@
from flask import request, url_for
from app.models import Corpus
def corpus_endpoint_arguments_constructor():
return {'corpus_id': request.view_args['corpus_id']}
def corpus_dynamic_list_constructor():
corpus_id = request.view_args['corpus_id']
corpus = Corpus.query.get_or_404(corpus_id)
return [
{
'text': f'<i class="material-icons left">book</i>{corpus.title}',
'url': url_for('.corpus', corpus_id=corpus_id)
}
]

View File

@ -1,11 +0,0 @@
from app import db
from flask import Flask
from .corpus_utils import check_corpora
from .job_utils import check_jobs
def daemon(app: Flask):
with app.app_context():
check_corpora()
check_jobs()
db.session.commit()

View File

@ -1,8 +1,7 @@
from flask import abort, current_app, request from flask import abort, request
from flask_login import current_user from flask_login import current_user
from functools import wraps from functools import wraps
from threading import Thread from typing import Optional
from typing import List, Union
from werkzeug.exceptions import NotAcceptable from werkzeug.exceptions import NotAcceptable
from app.models import Permission from app.models import Permission
@ -24,22 +23,21 @@ def admin_required(f):
def socketio_login_required(f): def socketio_login_required(f):
@wraps(f) @wraps(f)
def decorated_function(*args, **kwargs): def wrapper(*args, **kwargs):
if current_user.is_authenticated: if current_user.is_authenticated:
return f(*args, **kwargs) return f(*args, **kwargs)
else: return {'status': 401, 'statusText': 'Unauthorized'}
return {'code': 401, 'msg': 'Unauthorized'} return wrapper
return decorated_function
def socketio_permission_required(permission): def socketio_permission_required(permission):
def decorator(f): def decorator(f):
@wraps(f) @wraps(f)
def decorated_function(*args, **kwargs): def wrapper(*args, **kwargs):
if not current_user.can(permission): if not current_user.can(permission):
return {'code': 403, 'msg': 'Forbidden'} return {'status': 403, 'statusText': 'Forbidden'}
return f(*args, **kwargs) return f(*args, **kwargs)
return decorated_function return wrapper
return decorator return decorator
@ -47,27 +45,9 @@ def socketio_admin_required(f):
return socketio_permission_required(Permission.ADMINISTRATE)(f) return socketio_permission_required(Permission.ADMINISTRATE)(f)
def background(f):
'''
' This decorator executes a function in a Thread.
' Decorated functions need to be executed within a code block where an
' app context exists.
'
' NOTE: An app object is passed as a keyword argument to the decorated
' function.
'''
@wraps(f)
def wrapped(*args, **kwargs):
kwargs['app'] = current_app._get_current_object()
thread = Thread(target=f, args=args, kwargs=kwargs)
thread.start()
return thread
return wrapped
def content_negotiation( def content_negotiation(
produces: Union[str, List[str], None] = None, produces: Optional[str | list[str]] = None,
consumes: Union[str, List[str], None] = None consumes: Optional[str | list[str]] = None
): ):
def decorator(f): def decorator(f):
@wraps(f) @wraps(f)

View File

@ -1,25 +1,32 @@
from flask import current_app, render_template from flask import current_app, Flask, render_template
from flask_mail import Message from flask_mail import Message
from threading import Thread from threading import Thread
from app import mail from app import mail
def create_message(recipient, subject, template, **kwargs): def create_message(
subject_prefix: str = current_app.config['NOPAQUE_MAIL_SUBJECT_PREFIX'] recipient: str,
msg: Message = Message( subject: str,
body=render_template(f'{template}.txt.j2', **kwargs), template: str,
html=render_template(f'{template}.html.j2', **kwargs), **context
) -> Message:
message = Message(
body=render_template(f'{template}.txt.j2', **context),
html=render_template(f'{template}.html.j2', **context),
recipients=[recipient], recipients=[recipient],
subject=f'{subject_prefix} {subject}' subject=f'[nopaque] {subject}'
) )
return msg return message
def send(msg, *args, **kwargs): def send(message: Message) -> Thread:
def _send(app, msg): def _send(app: Flask, message: Message):
with app.app_context(): with app.app_context():
mail.send(msg) mail.send(message)
thread = Thread(target=_send, args=[current_app._get_current_object(), msg]) thread = Thread(
target=_send,
args=[current_app._get_current_object(), message]
)
thread.start() thread.start()
return thread return thread

View File

@ -0,0 +1,20 @@
from flask import abort
from flask_admin import (
AdminIndexView as _AdminIndexView,
expose
)
from flask_admin.contrib.sqla import ModelView as _ModelView
from flask_login import current_user
class AdminIndexView(_AdminIndexView):
@expose('/')
def index(self):
if not current_user.is_administrator:
abort(403)
return super().index()
class ModelView(_ModelView):
def is_accessible(self):
return current_user.is_administrator

View File

@ -0,0 +1,42 @@
import json
from app import db
class ContainerColumn(db.TypeDecorator):
impl = db.String
def __init__(self, container_type, *args, **kwargs):
super().__init__(*args, **kwargs)
self.container_type = container_type
def process_bind_param(self, value, dialect):
if isinstance(value, self.container_type):
return json.dumps(value)
elif isinstance(value, str) and isinstance(json.loads(value), self.container_type):
return value
else:
return TypeError()
def process_result_value(self, value, dialect):
return json.loads(value)
class IntEnumColumn(db.TypeDecorator):
impl = db.Integer
def __init__(self, enum_type, *args, **kwargs):
super().__init__(*args, **kwargs)
self.enum_type = enum_type
def process_bind_param(self, value, dialect):
if isinstance(value, self.enum_type) and isinstance(value.value, int):
return value.value
elif isinstance(value, int):
return self.enum_type(value).value
elif isinstance(value, str):
return self.enum_type[value].value
else:
return TypeError()
def process_result_value(self, value, dialect):
return self.enum_type(value)

View File

@ -1,18 +1,2 @@
from flask import Blueprint from .handle_corpora import handle_corpora
from flask_login import login_required from .handle_jobs import handle_jobs
bp = Blueprint('jobs', __name__)
@bp.before_request
@login_required
def before_request():
'''
Ensures that the routes in this package can only be visited by users that
are logged in.
'''
pass
from . import routes, json_routes

View File

@ -1,12 +1,16 @@
from app import docker_client
from app.models import Corpus, CorpusStatus
from flask import current_app from flask import current_app
import docker import docker
import os import os
import shutil import shutil
from app import db, docker_client, scheduler
from app.models import Corpus, CorpusStatus
def check_corpora(): def handle_corpora():
with scheduler.app.app_context():
_handle_corpora()
def _handle_corpora():
corpora = Corpus.query.all() corpora = Corpus.query.all()
for corpus in [x for x in corpora if x.status == CorpusStatus.SUBMITTED]: for corpus in [x for x in corpora if x.status == CorpusStatus.SUBMITTED]:
_create_build_corpus_service(corpus) _create_build_corpus_service(corpus)
@ -17,13 +21,14 @@ def check_corpora():
for corpus in [x for x in corpora if x.status == CorpusStatus.RUNNING_ANALYSIS_SESSION and x.num_analysis_sessions == 0]: for corpus in [x for x in corpora if x.status == CorpusStatus.RUNNING_ANALYSIS_SESSION and x.num_analysis_sessions == 0]:
corpus.status = CorpusStatus.CANCELING_ANALYSIS_SESSION corpus.status = CorpusStatus.CANCELING_ANALYSIS_SESSION
for corpus in [x for x in corpora if x.status == CorpusStatus.RUNNING_ANALYSIS_SESSION]: for corpus in [x for x in corpora if x.status == CorpusStatus.RUNNING_ANALYSIS_SESSION]:
_checkout_analysing_corpus_container(corpus) _checkout_cqpserver_container(corpus)
for corpus in [x for x in corpora if x.status == CorpusStatus.STARTING_ANALYSIS_SESSION]: for corpus in [x for x in corpora if x.status == CorpusStatus.STARTING_ANALYSIS_SESSION]:
_create_cqpserver_container(corpus) _create_cqpserver_container(corpus)
for corpus in [x for x in corpora if x.status == CorpusStatus.CANCELING_ANALYSIS_SESSION]: for corpus in [x for x in corpora if x.status == CorpusStatus.CANCELING_ANALYSIS_SESSION]:
_remove_cqpserver_container(corpus) _remove_cqpserver_container(corpus)
db.session.commit()
def _create_build_corpus_service(corpus): def _create_build_corpus_service(corpus: Corpus):
''' # Docker service settings # ''' ''' # Docker service settings # '''
''' ## Command ## ''' ''' ## Command ## '''
command = ['bash', '-c'] command = ['bash', '-c']
@ -45,12 +50,10 @@ def _create_build_corpus_service(corpus):
''' ## Constraints ## ''' ''' ## Constraints ## '''
constraints = ['node.role==worker'] constraints = ['node.role==worker']
''' ## Image ## ''' ''' ## Image ## '''
image = f'{current_app.config["NOPAQUE_DOCKER_IMAGE_PREFIX"]}cwb:r1879' image = f'{current_app.config["NOPAQUE_DOCKER_IMAGE_PREFIX"]}cwb:r1887'
''' ## Labels ## ''' ''' ## Labels ## '''
labels = { labels = {
'origin': current_app.config['SERVER_NAME'], 'nopaque.server_name': current_app.config['SERVER_NAME']
'type': 'corpus.build',
'corpus_id': str(corpus.id)
} }
''' ## Mounts ## ''' ''' ## Mounts ## '''
mounts = [] mounts = []
@ -95,7 +98,7 @@ def _create_build_corpus_service(corpus):
return return
corpus.status = CorpusStatus.QUEUED corpus.status = CorpusStatus.QUEUED
def _checkout_build_corpus_service(corpus): def _checkout_build_corpus_service(corpus: Corpus):
service_name = f'build-corpus_{corpus.id}' service_name = f'build-corpus_{corpus.id}'
try: try:
service = docker_client.services.get(service_name) service = docker_client.services.get(service_name)
@ -123,8 +126,7 @@ def _checkout_build_corpus_service(corpus):
except docker.errors.DockerException as e: except docker.errors.DockerException as e:
current_app.logger.error(f'Remove service "{service_name}" failed: {e}') current_app.logger.error(f'Remove service "{service_name}" failed: {e}')
def _create_cqpserver_container(corpus): def _create_cqpserver_container(corpus: Corpus):
''' # Docker container settings # '''
''' ## Command ## ''' ''' ## Command ## '''
command = [] command = []
command.append( command.append(
@ -139,9 +141,9 @@ def _create_cqpserver_container(corpus):
''' ## Entrypoint ## ''' ''' ## Entrypoint ## '''
entrypoint = ['bash', '-c'] entrypoint = ['bash', '-c']
''' ## Image ## ''' ''' ## Image ## '''
image = f'{current_app.config["NOPAQUE_DOCKER_IMAGE_PREFIX"]}cwb:r1879' image = f'{current_app.config["NOPAQUE_DOCKER_IMAGE_PREFIX"]}cwb:r1887'
''' ## Name ## ''' ''' ## Name ## '''
name = f'cqpserver_{corpus.id}' name = f'nopaque-cqpserver-{corpus.id}'
''' ## Network ## ''' ''' ## Network ## '''
network = f'{current_app.config["NOPAQUE_DOCKER_NETWORK_NAME"]}' network = f'{current_app.config["NOPAQUE_DOCKER_NETWORK_NAME"]}'
''' ## Volumes ## ''' ''' ## Volumes ## '''
@ -198,8 +200,8 @@ def _create_cqpserver_container(corpus):
return return
corpus.status = CorpusStatus.RUNNING_ANALYSIS_SESSION corpus.status = CorpusStatus.RUNNING_ANALYSIS_SESSION
def _checkout_analysing_corpus_container(corpus): def _checkout_cqpserver_container(corpus: Corpus):
container_name = f'cqpserver_{corpus.id}' container_name = f'nopaque-cqpserver-{corpus.id}'
try: try:
docker_client.containers.get(container_name) docker_client.containers.get(container_name)
except docker.errors.NotFound as e: except docker.errors.NotFound as e:
@ -209,8 +211,8 @@ def _checkout_analysing_corpus_container(corpus):
except docker.errors.DockerException as e: except docker.errors.DockerException as e:
current_app.logger.error(f'Get container "{container_name}" failed: {e}') current_app.logger.error(f'Get container "{container_name}" failed: {e}')
def _remove_cqpserver_container(corpus): def _remove_cqpserver_container(corpus: Corpus):
container_name = f'cqpserver_{corpus.id}' container_name = f'nopaque-cqpserver-{corpus.id}'
try: try:
container = docker_client.containers.get(container_name) container = docker_client.containers.get(container_name)
except docker.errors.NotFound: except docker.errors.NotFound:

View File

@ -1,11 +1,3 @@
from app import db, docker_client, hashids
from app.models import (
Job,
JobResult,
JobStatus,
TesseractOCRPipelineModel,
SpaCyNLPPipelineModel
)
from datetime import datetime from datetime import datetime
from flask import current_app from flask import current_app
from werkzeug.utils import secure_filename from werkzeug.utils import secure_filename
@ -13,9 +5,21 @@ import docker
import json import json
import os import os
import shutil import shutil
from app import db, docker_client, hashids, scheduler
from app.models import (
Job,
JobResult,
JobStatus,
TesseractOCRPipelineModel,
SpaCyNLPPipelineModel
)
def check_jobs(): def handle_jobs():
with scheduler.app.app_context():
_handle_jobs()
def _handle_jobs():
jobs = Job.query.all() jobs = Job.query.all()
for job in [x for x in jobs if x.status == JobStatus.SUBMITTED]: for job in [x for x in jobs if x.status == JobStatus.SUBMITTED]:
_create_job_service(job) _create_job_service(job)
@ -23,8 +27,9 @@ def check_jobs():
_checkout_job_service(job) _checkout_job_service(job)
for job in [x for x in jobs if x.status == JobStatus.CANCELING]: for job in [x for x in jobs if x.status == JobStatus.CANCELING]:
_remove_job_service(job) _remove_job_service(job)
db.session.commit()
def _create_job_service(job): def _create_job_service(job: Job):
''' # Docker service settings # ''' ''' # Docker service settings # '''
''' ## Service specific settings ## ''' ''' ## Service specific settings ## '''
if job.service == 'file-setup-pipeline': if job.service == 'file-setup-pipeline':
@ -81,9 +86,7 @@ def _create_job_service(job):
constraints = ['node.role==worker'] constraints = ['node.role==worker']
''' ## Labels ## ''' ''' ## Labels ## '''
labels = { labels = {
'origin': current_app.config['SERVER_NAME'], 'origin': current_app.config['SERVER_NAME']
'type': 'job',
'job_id': str(job.id)
} }
''' ## Mounts ## ''' ''' ## Mounts ## '''
mounts = [] mounts = []
@ -164,7 +167,7 @@ def _create_job_service(job):
return return
job.status = JobStatus.QUEUED job.status = JobStatus.QUEUED
def _checkout_job_service(job): def _checkout_job_service(job: Job):
service_name = f'job_{job.id}' service_name = f'job_{job.id}'
try: try:
service = docker_client.services.get(service_name) service = docker_client.services.get(service_name)
@ -213,7 +216,7 @@ def _checkout_job_service(job):
except docker.errors.DockerException as e: except docker.errors.DockerException as e:
current_app.logger.error(f'Remove service "{service_name}" failed: {e}') current_app.logger.error(f'Remove service "{service_name}" failed: {e}')
def _remove_job_service(job): def _remove_job_service(job: Job):
service_name = f'job_{job.id}' service_name = f'job_{job.id}'
try: try:
service = docker_client.services.get(service_name) service = docker_client.services.get(service_name)

View File

@ -1,73 +0,0 @@
from flask import abort, current_app
from flask_login import current_user
from threading import Thread
import os
from app import db
from app.decorators import admin_required, content_negotiation
from app.models import Job, JobStatus
from . import bp
@bp.route('/<hashid:job_id>', methods=['DELETE'])
@content_negotiation(produces='application/json')
def delete_job(job_id):
def _delete_job(app, job_id):
with app.app_context():
job = Job.query.get(job_id)
job.delete()
db.session.commit()
job = Job.query.get_or_404(job_id)
if not (job.user == current_user or current_user.is_administrator()):
abort(403)
thread = Thread(
target=_delete_job,
args=(current_app._get_current_object(), job_id)
)
thread.start()
response_data = {
'message': f'Job "{job.title}" marked for deletion'
}
return response_data, 202
@bp.route('/<hashid:job_id>/log')
@admin_required
@content_negotiation(produces='application/json')
def job_log(job_id):
job = Job.query.get_or_404(job_id)
if job.status not in [JobStatus.COMPLETED, JobStatus.FAILED]:
response = {'errors': {'message': 'Job status is not completed or failed'}}
return response, 409
with open(os.path.join(job.path, 'pipeline_data', 'logs', 'pyflow_log.txt')) as log_file:
log = log_file.read()
response_data = {
'jobLog': log
}
return response_data, 200
@bp.route('/<hashid:job_id>/restart', methods=['POST'])
@content_negotiation(produces='application/json')
def restart_job(job_id):
def _restart_job(app, job_id):
with app.app_context():
job = Job.query.get(job_id)
job.restart()
db.session.commit()
job = Job.query.get_or_404(job_id)
if not (job.user == current_user or current_user.is_administrator()):
abort(403)
if job.status == JobStatus.FAILED:
response = {'errors': {'message': 'Job status is not "failed"'}}
return response, 409
thread = Thread(
target=_restart_job,
args=(current_app._get_current_object(), job_id)
)
thread.start()
response_data = {
'message': f'Job "{job.title}" marked for restarting'
}
return response_data, 202

View File

@ -1,60 +0,0 @@
from flask import (
abort,
redirect,
render_template,
send_from_directory,
url_for
)
from flask_breadcrumbs import register_breadcrumb
from flask_login import current_user
import os
from app.models import Job, JobInput, JobResult
from . import bp
from .utils import job_dynamic_list_constructor as job_dlc
@bp.route('')
@register_breadcrumb(bp, '.', '<i class="nopaque-icons left">J</i>My Jobs')
def corpora():
return redirect(url_for('main.dashboard', _anchor='jobs'))
@bp.route('/<hashid:job_id>')
@register_breadcrumb(bp, '.entity', '', dynamic_list_constructor=job_dlc)
def job(job_id):
job = Job.query.get_or_404(job_id)
if not (job.user == current_user or current_user.is_administrator()):
abort(403)
return render_template(
'jobs/job.html.j2',
title='Job',
job=job
)
@bp.route('/<hashid:job_id>/inputs/<hashid:job_input_id>/download')
def download_job_input(job_id, job_input_id):
job_input = JobInput.query.filter_by(job_id=job_id, id=job_input_id).first_or_404()
if not (job_input.job.user == current_user or current_user.is_administrator()):
abort(403)
return send_from_directory(
os.path.dirname(job_input.path),
os.path.basename(job_input.path),
as_attachment=True,
attachment_filename=job_input.filename,
mimetype=job_input.mimetype
)
@bp.route('/<hashid:job_id>/results/<hashid:job_result_id>/download')
def download_job_result(job_id, job_result_id):
job_result = JobResult.query.filter_by(job_id=job_id, id=job_result_id).first_or_404()
if not (job_result.job.user == current_user or current_user.is_administrator()):
abort(403)
return send_from_directory(
os.path.dirname(job_result.path),
os.path.basename(job_result.path),
as_attachment=True,
attachment_filename=job_result.filename,
mimetype=job_result.mimetype
)

View File

@ -1,13 +0,0 @@
from flask import request, url_for
from app.models import Job
def job_dynamic_list_constructor():
job_id = request.view_args['job_id']
job = Job.query.get_or_404(job_id)
return [
{
'text': f'<i class="nopaque-icons left service-icons" data-service="{job.service}"></i>{job.title}',
'url': url_for('.job', job_id=job_id)
}
]

File diff suppressed because it is too large Load Diff

45
app/models/__init__.py Normal file
View File

@ -0,0 +1,45 @@
from .anonymous_user import AnonymousUser
from .avatar import Avatar
from .corpus_file import CorpusFile
from .corpus_follower_association import CorpusFollowerAssociation
from .corpus_follower_role import CorpusFollowerPermission, CorpusFollowerRole
from .corpus import CorpusStatus, Corpus
from .job_input import JobInput
from .job_result import JobResult
from .job import JobStatus, Job
from .role import Permission, Role
from .spacy_nlp_pipeline_model import SpaCyNLPPipelineModel
from .tesseract_ocr_pipeline_model import TesseractOCRPipelineModel
from .token import Token
from .user import (
ProfilePrivacySettings,
UserSettingJobStatusMailNotificationLevel,
User
)
_models = [
Avatar,
CorpusFile,
CorpusFollowerAssociation,
CorpusFollowerRole,
Corpus,
JobInput,
JobResult,
Job,
Role,
SpaCyNLPPipelineModel,
TesseractOCRPipelineModel,
Token,
User
]
_enums = [
CorpusFollowerPermission,
CorpusStatus,
JobStatus,
Permission,
ProfilePrivacySettings,
UserSettingJobStatusMailNotificationLevel
]

View File

@ -0,0 +1,10 @@
from flask_login import AnonymousUserMixin
class AnonymousUser(AnonymousUserMixin):
def can(self, permissions):
return False
@property
def is_administrator(self):
return False

40
app/models/avatar.py Normal file
View File

@ -0,0 +1,40 @@
from flask import current_app
from flask_hashids import HashidMixin
from pathlib import Path
from app import db
from .file_mixin import FileMixin
class Avatar(HashidMixin, FileMixin, db.Model):
__tablename__ = 'avatars'
# Primary key
id = db.Column(db.Integer, primary_key=True)
# Foreign keys
user_id = db.Column(db.Integer, db.ForeignKey('users.id'))
# Relationships
user = db.relationship('User', back_populates='avatar')
@property
def path(self) -> Path:
return self.user.path / 'avatar'
# return os.path.join(self.user.path, 'avatar')
def delete(self):
try:
self.path.unlink(missing_ok=True)
except OSError as e:
current_app.logger.error(e)
raise
db.session.delete(self)
def to_json_serializeable(self, backrefs=False, relationships=False):
json_serializeable = {
'id': self.hashid,
**self.file_mixin_to_json_serializeable()
}
if backrefs:
json_serializeable['user'] = \
self.user.to_json_serializeable(backrefs=True)
if relationships:
pass
return json_serializeable

199
app/models/corpus.py Normal file
View File

@ -0,0 +1,199 @@
from datetime import datetime
from enum import IntEnum
from flask import current_app, url_for
from flask_hashids import HashidMixin
from sqlalchemy.ext.associationproxy import association_proxy
from pathlib import Path
import shutil
import xml.etree.ElementTree as ET
from app import db
from app.converters.vrt import normalize_vrt_file
from app.extensions.nopaque_sqlalchemy_type_decorators import IntEnumColumn
from .corpus_follower_association import CorpusFollowerAssociation
class CorpusStatus(IntEnum):
UNPREPARED = 1
SUBMITTED = 2
QUEUED = 3
BUILDING = 4
BUILT = 5
FAILED = 6
STARTING_ANALYSIS_SESSION = 7
RUNNING_ANALYSIS_SESSION = 8
CANCELING_ANALYSIS_SESSION = 9
@staticmethod
def get(corpus_status: 'CorpusStatus | int | str') -> 'CorpusStatus':
if isinstance(corpus_status, CorpusStatus):
return corpus_status
if isinstance(corpus_status, int):
return CorpusStatus(corpus_status)
if isinstance(corpus_status, str):
return CorpusStatus[corpus_status]
raise TypeError('corpus_status must be CorpusStatus, int, or str')
class Corpus(HashidMixin, db.Model):
'''
Class to define a corpus.
'''
__tablename__ = 'corpora'
# Primary key
id = db.Column(db.Integer, primary_key=True)
# Foreign keys
user_id = db.Column(db.Integer, db.ForeignKey('users.id'))
# Fields
creation_date = db.Column(db.DateTime(), default=datetime.utcnow)
description = db.Column(db.String(255))
status = db.Column(
IntEnumColumn(CorpusStatus),
default=CorpusStatus.UNPREPARED
)
title = db.Column(db.String(32))
num_analysis_sessions = db.Column(db.Integer, default=0)
num_tokens = db.Column(db.Integer, default=0)
is_public = db.Column(db.Boolean, default=False)
# Relationships
files = db.relationship(
'CorpusFile',
back_populates='corpus',
lazy='dynamic',
cascade='all, delete-orphan'
)
corpus_follower_associations = db.relationship(
'CorpusFollowerAssociation',
back_populates='corpus',
cascade='all, delete-orphan'
)
followers = association_proxy(
'corpus_follower_associations',
'follower',
creator=lambda u: CorpusFollowerAssociation(follower=u)
)
user = db.relationship('User', back_populates='corpora')
# "static" attributes
max_num_tokens = 2_147_483_647
def __repr__(self):
return f'<Corpus {self.title}>'
@property
def analysis_url(self):
return url_for('corpora.analysis', corpus_id=self.id)
@property
def jsonpatch_path(self):
return f'{self.user.jsonpatch_path}/corpora/{self.hashid}'
@property
def path(self) -> Path:
return self.user.path / 'corpora' / f'{self.id}'
@property
def url(self):
return url_for('corpora.corpus', corpus_id=self.id)
@property
def user_hashid(self):
return self.user.hashid
@staticmethod
def create(**kwargs):
corpus = Corpus(**kwargs)
db.session.add(corpus)
db.session.flush(objects=[corpus])
db.session.refresh(corpus)
corpus_files_dir = corpus.path / 'files'
corpus_cwb_dir = corpus.path / 'cwb'
corpus_cwb_data_dir = corpus_cwb_dir / 'data'
corpus_cwb_registry_dir = corpus_cwb_dir / 'registry'
try:
corpus.path.mkdir()
corpus_files_dir.mkdir()
corpus_cwb_dir.mkdir()
corpus_cwb_data_dir.mkdir()
corpus_cwb_registry_dir.mkdir()
except OSError as e:
# TODO: Potential leftover cleanup
current_app.logger.error(e)
db.session.rollback()
raise
return corpus
def build(self):
corpus_cwb_dir = self.path / 'cwb'
corpus_cwb_data_dir = corpus_cwb_dir / 'data'
corpus_cwb_registry_dir = corpus_cwb_dir / 'registry'
try:
shutil.rmtree(corpus_cwb_dir, ignore_errors=True)
corpus_cwb_dir.mkdir()
corpus_cwb_data_dir.mkdir()
corpus_cwb_registry_dir.mkdir()
except OSError as e:
current_app.logger.error(e)
self.status = CorpusStatus.FAILED
raise
corpus_element = ET.fromstring('<corpus>\n</corpus>')
for corpus_file in self.files:
normalized_vrt_path = corpus_cwb_dir / f'{corpus_file.id}.norm.vrt'
try:
normalize_vrt_file(corpus_file.path, normalized_vrt_path)
except:
self.status = CorpusStatus.FAILED
return
element_tree = ET.parse(normalized_vrt_path)
text_element = element_tree.getroot()
text_element.set('author', corpus_file.author)
text_element.set('title', corpus_file.title)
text_element.set(
'publishing_year',
f'{corpus_file.publishing_year}'
)
text_element.set('address', corpus_file.address or 'NULL')
text_element.set('booktitle', corpus_file.booktitle or 'NULL')
text_element.set('chapter', corpus_file.chapter or 'NULL')
text_element.set('editor', corpus_file.editor or 'NULL')
text_element.set('institution', corpus_file.institution or 'NULL')
text_element.set('journal', corpus_file.journal or 'NULL')
text_element.set('pages', f'{corpus_file.pages}' or 'NULL')
text_element.set('publisher', corpus_file.publisher or 'NULL')
text_element.set('school', corpus_file.school or 'NULL')
text_element.tail = '\n'
# corpus_element.insert(1, text_element)
corpus_element.append(text_element)
ET.ElementTree(corpus_element).write(
corpus_cwb_dir / 'corpus.vrt',
encoding='utf-8'
)
self.status = CorpusStatus.SUBMITTED
def delete(self):
shutil.rmtree(self.path, ignore_errors=True)
db.session.delete(self)
def to_json_serializeable(self, backrefs=False, relationships=False):
json_serializeable = {
'id': self.hashid,
'creation_date': f'{self.creation_date.isoformat()}Z',
'description': self.description,
'max_num_tokens': self.max_num_tokens,
'num_analysis_sessions': self.num_analysis_sessions,
'num_tokens': self.num_tokens,
'status': self.status.name,
'title': self.title,
'is_public': self.is_public
}
if backrefs:
json_serializeable['user'] = \
self.user.to_json_serializeable(backrefs=True)
if relationships:
json_serializeable['corpus_follower_associations'] = {
x.hashid: x.to_json_serializeable()
for x in self.corpus_follower_associations
}
json_serializeable['files'] = {
x.hashid: x.to_json_serializeable(relationships=True)
for x in self.files
}
return json_serializeable

102
app/models/corpus_file.py Normal file
View File

@ -0,0 +1,102 @@
from flask import current_app, url_for
from flask_hashids import HashidMixin
from pathlib import Path
from app import db
from .corpus import CorpusStatus
from .file_mixin import FileMixin
class CorpusFile(FileMixin, HashidMixin, db.Model):
__tablename__ = 'corpus_files'
# Primary key
id = db.Column(db.Integer, primary_key=True)
# Foreign keys
corpus_id = db.Column(db.Integer, db.ForeignKey('corpora.id'))
# Fields
author = db.Column(db.String(255))
description = db.Column(db.String(255))
publishing_year = db.Column(db.Integer)
title = db.Column(db.String(255))
address = db.Column(db.String(255))
booktitle = db.Column(db.String(255))
chapter = db.Column(db.String(255))
editor = db.Column(db.String(255))
institution = db.Column(db.String(255))
journal = db.Column(db.String(255))
pages = db.Column(db.String(255))
publisher = db.Column(db.String(255))
school = db.Column(db.String(255))
# Relationships
corpus = db.relationship(
'Corpus',
back_populates='files'
)
@property
def download_url(self):
return url_for(
'corpora.download_corpus_file',
corpus_id=self.corpus_id,
corpus_file_id=self.id
)
@property
def jsonpatch_path(self):
return f'{self.corpus.jsonpatch_path}/files/{self.hashid}'
@property
def path(self) -> Path:
return self.corpus.path / 'files' / f'{self.id}'
@property
def url(self):
return url_for(
'corpora.corpus_file',
corpus_id=self.corpus_id,
corpus_file_id=self.id
)
@property
def user_hashid(self):
return self.corpus.user.hashid
@property
def user_id(self):
return self.corpus.user_id
def delete(self):
try:
self.path.unlink(missing_ok=True)
except OSError as e:
current_app.logger.error(e)
raise
db.session.delete(self)
self.corpus.status = CorpusStatus.UNPREPARED
def to_json_serializeable(self, backrefs=False, relationships=False):
json_serializeable = {
'id': self.hashid,
'address': self.address,
'author': self.author,
'description': self.description,
'booktitle': self.booktitle,
'chapter': self.chapter,
'editor': self.editor,
'institution': self.institution,
'journal': self.journal,
'pages': self.pages,
'publisher': self.publisher,
'publishing_year': self.publishing_year,
'school': self.school,
'title': self.title,
**self.file_mixin_to_json_serializeable(
backrefs=backrefs,
relationships=relationships
)
}
if backrefs:
json_serializeable['corpus'] = \
self.corpus.to_json_serializeable(backrefs=True)
if relationships:
pass
return json_serializeable

Some files were not shown because too many files have changed in this diff Show More