Bump nopaque version 1.1.0 -> 1.1.1

- breadcrumbs (!flask-breadcrumbs)
- manual modal button
- api blueprint (!flask-marshmallow/!marshmallow-sqlalchemy)

.dockerignore

@@ -5,8 +5,9 @@
 !app
 !migrations
 !tests
-!.flaskenv
 !boot.sh
 !config.py
-!nopaque.py
+!docker-nopaque-entrypoint.sh
 !requirements.txt
+!requirements.freezed.txt
+!wsgi.py

.env.tpl

@@ -1,204 +1,20 @@
-################################################################################
+##############################################################################
-# Docker                                                                       #
+# Environment variables used by Docker Compose config files.                 #
-################################################################################
+##############################################################################
-# DEFAULT: ./data
-# NOTE: Use `.` as <project-basedir>
-# HOST_DATA_DIR=
-
-# Example: 1000
 # HINT: Use this bash command `id -u`
+# NOTE: 0 (= root user) is not allowed
 HOST_UID=
 
-# Example: 1000
 # HINT: Use this bash command `id -g`
+# NOTE: 0 (= root group) is not allowed
 HOST_GID=
 
-# Example: 999
 # HINT: Use this bash command `getent group docker | cut -d: -f3`
 HOST_DOCKER_GID=
 
-# DEFAULT: ./logs
+# DEFAULT: nopaque
-# NOTES: Use `.` as <project-basedir>
+NOPAQUE_DOCKER_NETWORK_NAME=nopaque
-# HOST_LOG_DIR=
 
-# DEFAULT: nopaque_default
+# NOTE: This must be a network share and it must be available on all
-# DOCKER_NETWORK_NAME=
+#       Docker Swarm nodes, mounted to the same path.
-
+HOST_NOPAQUE_DATA_PATH=/mnt/nopaque
-################################################################################
-# Flask                                                                        #
-# https://flask.palletsprojects.com/en/1.1.x/config/                           #
-################################################################################
-# CHOOSE ONE: http, https
-# DEFAULT: http
-# PREFERRED_URL_SCHEME=
-
-# DEFAULT: hard to guess string
-# HINT: Use this bash command `python -c "import uuid; print(uuid.uuid4().hex)"`
-# SECRET_KEY=
-
-# DEFAULT: localhost:5000
-# Example: nopaque.example.com/nopaque.example.com:5000
-# HINT: If your instance is publicly available on a different Port then 80/443,
-#       you will have to add this to the server name
-# SERVER_NAME=
-
-# CHOOSE ONE: False, True
-# DEFAULT: False
-# HINT: Set to true if you redirect http to https
-# SESSION_COOKIE_SECURE=
-
-
-################################################################################
-# Flask-Assets                                                                 #
-# https://webassets.readthedocs.io/en/latest/                                  #
-################################################################################
-# CHOOSE ONE: False, True
-# DEFAULT: False
-# ASSETS_DEBUG=
-
-
-################################################################################
-# Flask-Hashids                                                                #
-# https://github.com/Pevtrick/Flask-Hashids                                    #
-################################################################################
-# DEFAULT: 16
-# HASHIDS_MIN_LENGTH=
-
-# NOTE: Use this bash command `python -c "import uuid; print(uuid.uuid4().hex)"`
-#       It is strongly recommended that this is NEVER the same as the SECRET_KEY
-HASHIDS_SALT=
-
-
-################################################################################
-# Flask-Login                                                                  #
-# https://flask-login.readthedocs.io/en/latest/                                #
-################################################################################
-# CHOOSE ONE: False, True
-# DEFAULT: False
-# HINT: Set to true if you redirect http to https
-# REMEMBER_COOKIE_SECURE=
-
-
-################################################################################
-# Flask-Mail                                                                   #
-# https://pythonhosted.org/Flask-Mail/                                         #
-################################################################################
-# EXAMPLE: nopaque Admin <nopaque@example.com>
-MAIL_DEFAULT_SENDER=
-
-MAIL_PASSWORD=
-
-# EXAMPLE: smtp.example.com
-MAIL_SERVER=
-
-# EXAMPLE: 587
-MAIL_PORT=
-
-# CHOOSE ONE: False, True
-# DEFAULT: False
-# MAIL_USE_SSL=
-
-# CHOOSE ONE: False, True
-# DEFAULT: False
-# MAIL_USE_TLS=
-
-# EXAMPLE: nopaque@example.com
-MAIL_USERNAME=
-
-
-################################################################################
-# Flask-SQLAlchemy                                                             #
-# https://flask-sqlalchemy.palletsprojects.com/en/2.x/config/                  #
-################################################################################
-# DEFAULT: 'sqlite:///<nopaque-basedir>/data.sqlite'
-# NOTE: Use `.` as <nopaque-basedir>,
-#       Don't use a SQLite database when using Docker
-# SQLALCHEMY_DATABASE_URI=
-
-
-################################################################################
-# nopaque                                                                      #
-################################################################################
-# An account is registered with this email adress gets automatically assigned
-# the administrator role.
-# EXAMPLE: admin.nopaque@example.com
-NOPAQUE_ADMIN=
-
-# DEFAULT: /mnt/nopaque
-# NOTE: This must be a network share and it must be available on all Docker
-#       Swarm nodes
-# NOPAQUE_DATA_DIR=
-
-# CHOOSE ONE: False, True
-# DEFAULT: True
-# NOPAQUE_IS_PRIMARY_INSTANCE=
-
-# transport://[userid:password]@hostname[:port]/[virtual_host]
-NOPAQUE_SOCKETIO_MESSAGE_QUEUE_URI=
-
-# NOTE: Get these from the nopaque development team
-NOPAQUE_DOCKER_REGISTRY_USERNAME=
-NOPAQUE_DOCKER_REGISTRY_PASSWORD=
-
-# DEFAULT: %Y-%m-%d %H:%M:%S
-# NOPAQUE_LOG_DATE_FORMAT=
-
-# DEFAULT: [%(asctime)s] %(levelname)s in %(pathname)s (function: %(funcName)s, line: %(lineno)d): %(message)s
-# NOPAQUE_LOG_FORMAT=
-
-# DEFAULT: INFO
-# CHOOSE ONE: CRITICAL, ERROR, WARNING, INFO, DEBUG
-# NOPAQUE_LOG_LEVEL=
-
-# CHOOSE ONE: False, True
-# DEFAULT: True
-# NOPAQUE_LOG_FILE_ENABLED=
-
-# DEFAULT: <nopaque-basedir>/logs
-# NOTE: Use `.` as <nopaque-basedir>
-# NOPAQUE_LOG_FILE_DIR=
-
-# DEFAULT: NOPAQUE_LOG_LEVEL
-# CHOOSE ONE: CRITICAL, ERROR, WARNING, INFO, DEBUG
-# NOPAQUE_LOG_FILE_LEVEL=
-
-# CHOOSE ONE: False, True
-# DEFAULT: False
-# NOPAQUE_LOG_STDERR_ENABLED=
-
-# CHOOSE ONE: CRITICAL, ERROR, WARNING, INFO, DEBUG
-# DEFAULT: NOPAQUE_LOG_LEVEL
-# NOPAQUE_LOG_STDERR_LEVEL=
-
-# CHOOSE ONE: False, True
-# DEFAULT: False
-# HINT: Set this to True only if you are using a proxy in front of nopaque
-# NOPAQUE_PROXY_FIX_ENABLED=
-
-# DEFAULT: 0
-# Number of values to trust for X-Forwarded-For
-# NOPAQUE_PROXY_FIX_X_FOR=
-
-# DEFAULT: 0
-# Number of values to trust for X-Forwarded-Host
-# NOPAQUE_PROXY_FIX_X_HOST=
-
-# DEFAULT: 0
-# Number of values to trust for X-Forwarded-Port
-# NOPAQUE_PROXY_FIX_X_PORT=
-
-# DEFAULT: 0
-# Number of values to trust for X-Forwarded-Prefix
-# NOPAQUE_PROXY_FIX_X_PREFIX=
-
-# DEFAULT: 0
-# Number of values to trust for X-Forwarded-Proto
-# NOPAQUE_PROXY_FIX_X_PROTO=
-
-# CHOOSE ONE: False, True
-# DEFAULT: False
-# NOPAQUE_TRANSKRIBUS_ENABLED=
-
-# READ-COOP account data: https://readcoop.eu/
-# NOPAQUE_READCOOP_USERNAME=
-# NOPAQUE_READCOOP_PASSWORD=

.flaskenv

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

.gitignore

@@ -1,9 +1,7 @@
 # nopaque specifics
 app/static/gen/
-data/
+volumes/
 docker-compose.override.yml
-logs/
-!logs/dummy
 *.env
 
 *.pjentsch-testing

.gitlab-ci.yml

@@ -0,0 +1,84 @@
+include:
+  - template: Security/Container-Scanning.gitlab-ci.yml
+
+##############################################################################
+# Pipeline stages in order of execution                                      #
+##############################################################################
+stages:
+  - build
+  - publish
+  - sca
+
+##############################################################################
+# Pipeline behavior                                                          #
+##############################################################################
+workflow:
+  rules:
+    # Run the pipeline on commits to the default branch
+    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH
+      variables:
+        # Set the Docker image tag to `latest`
+        DOCKER_IMAGE: $CI_REGISTRY_IMAGE:latest
+      when: always
+    # Run the pipeline on tag creation
+    - if: $CI_COMMIT_TAG
+      variables:
+        # Set the Docker image tag to the Git tag name
+        DOCKER_IMAGE: $CI_REGISTRY_IMAGE:$CI_COMMIT_REF_NAME
+      when: always
+    # Don't run the pipeline on all other occasions
+    - when: never
+
+##############################################################################
+# Default values for pipeline jobs                                           #
+##############################################################################
+default:
+  image: docker:24.0.6
+  services:
+    - docker:24.0.6-dind
+  tags:
+    - docker
+
+##############################################################################
+# CI/CD variables for all jobs in the pipeline                               #
+##############################################################################
+variables:
+  DOCKER_TLS_CERTDIR: /certs
+  DOCKER_BUILD_PATH: .
+  DOCKERFILE: Dockerfile
+
+##############################################################################
+# Pipeline jobs                                                              #
+##############################################################################
+build:
+  stage: build
+  script:
+    - docker build --tag $DOCKER_IMAGE --file $DOCKERFILE $DOCKER_BUILD_PATH
+    - docker save $DOCKER_IMAGE > docker_image.tar
+  artifacts:
+    paths:
+      - docker_image.tar
+
+publish:
+  stage: publish
+  before_script:
+    - docker login --username gitlab-ci-token --password $CI_JOB_TOKEN $CI_REGISTRY
+  script:
+    - docker load --input docker_image.tar
+    - docker push $DOCKER_IMAGE
+  after_script:
+    - docker logout $CI_REGISTRY
+
+container_scanning:
+  stage: sca
+  rules:
+    # Run the job on commits to the default branch
+    - if: $CI_COMMIT_BRANCH == $CI_DEFAULT_BRANCH
+      when: always
+    # Run the job on tag creation
+    - if: $CI_COMMIT_TAG
+      when: always
+    # Don't run the job on all other occasions
+    - when: never
+  variables:
+    CS_IMAGE: $DOCKER_IMAGE

.vscode/extensions.json

@@ -1,7 +1,8 @@
 {
     "recommendations": [
-        "samuelcolvin.jinjahtml",
+        "irongeek.vscode-env",
         "ms-azuretools.vscode-docker",
-        "ms-python.python"
+        "ms-python.python",
+        "samuelcolvin.jinjahtml"
     ]
 }

.vscode/settings.json

@@ -1,13 +1,17 @@
 {
     "editor.rulers": [79],
+    "editor.tabSize": 4,
+    "emmet.includeLanguages": {
+        "jinja-html": "html"
+    },
+    "files.associations": {
+        ".flaskenv": "env",
+        "*.env.tpl": "env",
+        "*.txt.j2": "jinja"
+    },
     "files.insertFinalNewline": true,
-    "python.terminal.activateEnvironment": false,
+    "files.trimFinalNewlines": true,
-    "[css]": {
+    "files.trimTrailingWhitespace": true,
-        "editor.tabSize": 2
-    },
-    "[scss]": {
-        "editor.tabSize": 2
-    },
     "[html]": {
         "editor.tabSize": 2
     },
@@ -16,8 +20,5 @@
     },
     "[jinja-html]": {
         "editor.tabSize": 2
-    },
-    "[jinja-js]": {
-        "editor.tabSize": 2
     }
 }

Dockerfile

@@ -1,50 +1,57 @@
-FROM python:3.8.10-slim-buster
+FROM python:3.10.13-slim-bookworm
 
 
 LABEL authors="Patrick Jentsch <p.jentsch@uni-bielefeld.de>"
 
 
-ARG DOCKER_GID
+# Set environment variables
-ARG UID
-ARG GID
-
-
 ENV LANG="C.UTF-8"
 ENV PYTHONDONTWRITEBYTECODE="1"
 ENV PYTHONUNBUFFERED="1"
 
 
+# Install system dependencies
 RUN apt-get update \
  && apt-get install --no-install-recommends --yes \
       build-essential \
+      gosu \
       libpq-dev \
  && rm --recursive /var/lib/apt/lists/*
 
 
-RUN groupadd --gid "${DOCKER_GID}" docker \
+# Create a non-root user
- && groupadd --gid "${GID}" nopaque \
+RUN useradd --create-home --no-log-init nopaque \
- && useradd --create-home --gid nopaque --groups "${DOCKER_GID}" --no-log-init --uid "${UID}" nopaque
+ && groupadd docker \
+ && usermod --append --groups docker nopaque
+
 USER nopaque
 WORKDIR /home/nopaque
 
 
-ENV PYTHON3_VENV_PATH="/home/nopaque/venv"
+# Create a Python virtual environment
-RUN python3 -m venv "${PYTHON3_VENV_PATH}"
+ENV NOPAQUE_PYTHON3_VENV_PATH="/home/nopaque/.venv"
-ENV PATH="${PYTHON3_VENV_PATH}/bin:${PATH}"
+RUN python3 -m venv "${NOPAQUE_PYTHON3_VENV_PATH}"
+ENV PATH="${NOPAQUE_PYTHON3_VENV_PATH}/bin:${PATH}"
 
 
-COPY --chown=nopaque:nopaque requirements.txt .
+# Install Python dependencies
-RUN python3 -m pip install --requirement requirements.txt \
+COPY --chown=nopaque:nopaque requirements.freezed.txt requirements.freezed.txt
- && rm requirements.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 migrations migrations
 COPY --chown=nopaque:nopaque tests tests
-COPY --chown=nopaque:nopaque .flaskenv boot.sh config.py nopaque.py ./
+COPY --chown=nopaque:nopaque boot.sh config.py wsgi.py ./
 
 
 EXPOSE 5000
 
 
-ENTRYPOINT ["./boot.sh"]
+USER root
+
+
+ENTRYPOINT ["docker-nopaque-entrypoint.sh"]

README.md

@@ -1,5 +1,8 @@
 # nopaque
 
+![release badge](https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque/-/badges/release.svg)
+![pipeline badge](https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque/badges/master/pipeline.svg?ignore_skipped=true)
+
 nopaque bundles various tools and services that provide humanities scholars with DH methods and thus can support their various individual research processes. Using nopaque, researchers can subject digitized sources to Optical Character Recognition (OCR). The resulting text files can then be used as a data basis for Natural Language Processing (NLP). The texts are automatically subjected to various linguistic annotations. The data processed via NLP can then be summarized in the web application as corpora and analyzed by means of an information retrieval system through complex search queries. The range of functions of the web application will be successively extended according to the needs of the researchers.
 
 ## Prerequisites and requirements
@@ -32,7 +35,7 @@ username@hostname:~$ sudo mount --types cifs --options gid=${USER},password=nopa
 # Clone the nopaque repository
 username@hostname:~$ git clone https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque.git
 # 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 .env.tpl .env
 # Fill out the variables within these files.

app/__init__.py

@@ -2,9 +2,10 @@ from apifairy import APIFairy
 from config import Config
 from docker import DockerClient
 from flask import Flask
+from flask.logging import default_handler
+from flask_admin import Admin
 from flask_apscheduler import APScheduler
 from flask_assets import Environment
-from flask_breadcrumbs import Breadcrumbs, default_breadcrumb_root
 from flask_login import LoginManager
 from flask_mail import Mail
 from flask_marshmallow import Marshmallow
@@ -13,91 +14,143 @@ from flask_paranoid import Paranoid
 from flask_socketio import SocketIO
 from flask_sqlalchemy import SQLAlchemy
 from flask_hashids import Hashids
-from werkzeug.exceptions import HTTPException
+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()
 assets = Environment()
-breadcrumbs = Breadcrumbs()
 db = SQLAlchemy()
-docker_client = DockerClient()
 hashids = Hashids()
 login = LoginManager()
-login.login_view = 'auth.login'
-login.login_message = 'Please log in to access this page.'
 ma = Marshmallow()
 mail = Mail()
 migrate = Migrate(compare_type=True)
 paranoid = Paranoid()
-paranoid.redirect_view = '/'
 scheduler = APScheduler()
 socketio = SocketIO()
 
 
 def create_app(config: Config = Config) -> Flask:
-    ''' Creates an initialized Flask (WSGI Application) object. '''
+    ''' Creates an initialized Flask object. '''
+
     app = Flask(__name__)
     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(
         app.config['NOPAQUE_DOCKER_REGISTRY_USERNAME'],
         password=app.config['NOPAQUE_DOCKER_REGISTRY_PASSWORD'],
         registry=app.config['NOPAQUE_DOCKER_REGISTRY']
     )
 
+    from .models import AnonymousUser, User
+
+    admin.init_app(app, index_view=AdminIndexView())
     apifairy.init_app(app)
     assets.init_app(app)
-    breadcrumbs.init_app(app)
     db.init_app(app)
     hashids.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)
     mail.init_app(app)
     migrate.init_app(app, db)
     paranoid.init_app(app)
+    paranoid.redirect_view = '/'
     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')
 
-    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)
 
-    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')
 
-    from .corpora import bp as corpora_blueprint
+    from .blueprints.corpora import bp as corpora_blueprint
-    default_breadcrumb_root(corpora_blueprint, '.corpora')
     app.register_blueprint(corpora_blueprint, cli_group='corpus', url_prefix='/corpora')
 
-    from .errors import bp as errors_bp
+    from .blueprints.errors import bp as 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')
 
-    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)
 
-    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')
 
-    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')
 
-    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, url_prefix='/users')
+
+    from .blueprints.workshops import bp as workshops_blueprint
+    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

app/admin/__init__.py

@@ -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

app/admin/forms.py

@@ -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()]

app/admin/json_routes.py

@@ -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

app/admin/routes.py

@@ -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
-    )

app/blueprints/api/jobs.py

@@ -5,8 +5,8 @@ from flask import abort, Blueprint
 from werkzeug.exceptions import InternalServerError
 from app import db, hashids
 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 .schemas import EmptySchema, JobSchema, SpaCyNLPPipelineJobSchema, TesseractOCRPipelineJobSchema, TesseractOCRPipelineModelSchema
 
 
 bp = Blueprint('jobs', __name__)
@@ -77,7 +77,7 @@ def delete_job(job_id):
     job = Job.query.get(job_id)
     if job is None:
         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)
     try:
         job.delete()
@@ -97,6 +97,6 @@ def get_job(job_id):
     job = Job.query.get(job_id)
     if job is None:
         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)
     return job

app/blueprints/api/schemas.py

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

app/blueprints/api/users.py

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

app/blueprints/auth/__init__.py

@@ -0,0 +1,27 @@
+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'
+        and request.endpoint != 'main.accept_terms_of_use'
+    ):
+        return redirect(url_for('auth.unconfirmed'))
+
+
+from . import routes

app/blueprints/auth/forms.py

@@ -60,7 +60,11 @@ class RegistrationForm(FlaskForm):
 
     def validate_username(self, field):
         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):

app/blueprints/auth/routes.py

@@ -1,5 +1,4 @@
 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 app import db
 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'])
-@register_breadcrumb(bp, '.register', 'Register')
 def register():
     if current_user.is_authenticated:
         return redirect(url_for('main.dashboard'))
@@ -67,7 +49,6 @@ def register():
 
 
 @bp.route('/login', methods=['GET', 'POST'])
-@register_breadcrumb(bp, '.login', 'Login')
 def login():
     if current_user.is_authenticated:
         return redirect(url_for('main.dashboard'))
@@ -98,7 +79,6 @@ def logout():
 
 
 @bp.route('/unconfirmed')
-@register_breadcrumb(bp, '.unconfirmed', 'Unconfirmed')
 @login_required
 def unconfirmed():
     if current_user.confirmed:
@@ -141,7 +121,6 @@ def confirm(token):
 
 
 @bp.route('/reset-password-request', methods=['GET', 'POST'])
-@register_breadcrumb(bp, '.reset_password_request', 'Password Reset')
 def reset_password_request():
     if current_user.is_authenticated:
         return redirect(url_for('main.dashboard'))
@@ -171,7 +150,6 @@ def reset_password_request():
 
 
 @bp.route('/reset-password/<token>', methods=['GET', 'POST'])
-@register_breadcrumb(bp, '.reset_password', 'Password Reset')
 def reset_password(token):
     if current_user.is_authenticated:
         return redirect(url_for('main.dashboard'))

app/blueprints/contributions/__init__.py

@@ -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')

app/blueprints/contributions/routes.py

@@ -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')

app/blueprints/contributions/spacy_nlp_pipeline_models/__init__.py

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

app/blueprints/contributions/spacy_nlp_pipeline_models/forms.py

@@ -1,7 +1,7 @@
 from flask_wtf.file import FileField, FileRequired
 from wtforms import StringField, ValidationError
 from wtforms.validators import InputRequired, Length
-from app.services import SERVICES
+from app.blueprints.services import SERVICES
 from ..forms import ContributionBaseForm, UpdateContributionBaseForm
 
 
@@ -16,8 +16,8 @@ class CreateSpaCyNLPPipelineModelForm(ContributionBaseForm):
     )
 
     def validate_spacy_model_file(self, field):
-        if not field.data.filename.lower().endswith('.tar.gz'):
+        if not field.data.filename.lower().endswith(('.tar.gz', ('.whl'))):
-            raise ValidationError('.tar.gz files only!')
+            raise ValidationError('.tar.gz or .whl files only!')
 
     def __init__(self, *args, **kwargs):
         if 'prefix' not in kwargs:

app/blueprints/contributions/spacy_nlp_pipeline_models/json_routes.py

@@ -1,13 +1,14 @@
 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 app import db
 from app.decorators import content_negotiation, permission_required
 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')
 def delete_spacy_model(spacy_nlp_pipeline_model_id):
     def _delete_spacy_model(app, spacy_nlp_pipeline_model_id):
@@ -17,7 +18,7 @@ def delete_spacy_model(spacy_nlp_pipeline_model_id):
             db.session.commit()
 
     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)
     thread = Thread(
         target=_delete_spacy_model,
@@ -31,7 +32,7 @@ def delete_spacy_model(spacy_nlp_pipeline_model_id):
     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')
 @content_negotiation(consumes='application/json', produces='application/json')
 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):
         abort(400)
     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)
     snpm.is_public = is_public
     db.session.commit()

app/blueprints/contributions/spacy_nlp_pipeline_models/routes.py

@@ -1,6 +1,5 @@
 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.models import SpaCyNLPPipelineModel
 from . import bp
@@ -8,23 +7,17 @@ from .forms import (
     CreateSpaCyNLPPipelineModelForm,
     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()
     if form.is_submitted():
         if not form.validate():
@@ -48,7 +41,7 @@ def create_spacy_nlp_pipeline_model():
             abort(500)
         db.session.commit()
         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(
         'contributions/spacy_nlp_pipeline_models/create.html.j2',
         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)
-    if not (snpm.user == current_user or current_user.is_administrator()):
+    if not (snpm.user == current_user or current_user.is_administrator):
         abort(403)
     form = UpdateSpaCyNLPPipelineModelForm(data=snpm.to_json_serializeable())
     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):
             flash(f'SpaCy NLP Pipeline model "{snpm.title}" updated')
             db.session.commit()
-        return redirect(url_for('.spacy_nlp_pipeline_models'))
+        return redirect(url_for('.index'))
     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}',
         form=form,
         spacy_nlp_pipeline_model=snpm

app/blueprints/contributions/tesseract_ocr_pipeline_models/__init__.py

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

app/blueprints/contributions/tesseract_ocr_pipeline_models/forms.py

@@ -1,6 +1,6 @@
 from flask_wtf.file import FileField, FileRequired
 from wtforms import ValidationError
-from app.services import SERVICES
+from app.blueprints.services import SERVICES
 from ..forms import ContributionBaseForm, UpdateContributionBaseForm
 
 

app/blueprints/contributions/tesseract_ocr_pipeline_models/json_routes.py

@@ -7,7 +7,7 @@ from app.models import TesseractOCRPipelineModel
 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')
 def delete_tesseract_model(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()
 
     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)
     thread = Thread(
         target=_delete_tesseract_ocr_pipeline_model,
@@ -31,7 +31,7 @@ def delete_tesseract_model(tesseract_ocr_pipeline_model_id):
     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')
 @content_negotiation(consumes='application/json', produces='application/json')
 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):
         abort(400)
     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)
     topm.is_public = is_public
     db.session.commit()

app/blueprints/contributions/tesseract_ocr_pipeline_models/routes.py

@@ -1,5 +1,4 @@
 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 TesseractOCRPipelineModel
@@ -8,23 +7,15 @@ from .forms import (
     CreateTesseractOCRPipelineModelForm,
     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()
     if form.is_submitted():
         if not form.validate():
@@ -47,7 +38,7 @@ def create_tesseract_ocr_pipeline_model():
             abort(500)
         db.session.commit()
         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(
         'contributions/tesseract_ocr_pipeline_models/create.html.j2',
         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)
-    if not (topm.user == current_user or current_user.is_administrator()):
+    if not (topm.user == current_user or current_user.is_administrator):
         abort(403)
     form = UpdateTesseractOCRPipelineModelForm(data=topm.to_json_serializeable())
     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):
             flash(f'Tesseract OCR Pipeline model "{topm.title}" updated')
             db.session.commit()
-        return redirect(url_for('.tesseract_ocr_pipeline_models'))
+        return redirect(url_for('.index'))
     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}',
         form=form,
         tesseract_ocr_pipeline_model=topm

app/blueprints/corpora/__init__.py

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

app/blueprints/corpora/cli.py

@@ -1,7 +1,7 @@
-from app.models import Corpus, CorpusStatus
+from flask import current_app
-import os
 import shutil
 from app import db
+from app.models import Corpus, CorpusStatus
 from . import bp
 
 
@@ -18,10 +18,17 @@ def reset():
     ]
     for corpus in [x for x in Corpus.query.all() if x.status in status]:
         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.num_analysis_sessions = 0
     db.session.commit()

app/blueprints/corpora/decorators.py

@@ -10,7 +10,7 @@ def corpus_follower_permission_required(*permissions):
         def decorated_function(*args, **kwargs):
             corpus_id = kwargs.get('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()
                 if cfa is None:
                     abort(403)
@@ -26,7 +26,7 @@ def corpus_owner_or_admin_required(f):
     def decorated_function(*args, **kwargs):
         corpus_id = kwargs.get('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)
         return f(*args, **kwargs)
     return decorated_function

app/blueprints/corpora/files/json_routes.py

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

app/blueprints/corpora/files/routes.py

@@ -6,25 +6,19 @@ from flask import (
     send_from_directory,
     url_for
 )
-from flask_breadcrumbs import register_breadcrumb
-import os
 from app import db
 from app.models import Corpus, CorpusFile, CorpusStatus
 from ..decorators import corpus_follower_permission_required
-from ..utils import corpus_endpoint_arguments_constructor as corpus_eac
 from . import bp
 from .forms import CreateCorpusFileForm, UpdateCorpusFileForm
-from .utils import corpus_file_dynamic_list_constructor as corpus_file_dlc
 
 
 @bp.route('/<hashid:corpus_id>/files')
-@register_breadcrumb(bp, '.entity.files', 'Files', endpoint_arguments_constructor=corpus_eac)
 def corpus_files(corpus_id):
     return redirect(url_for('.corpus', _anchor='files', corpus_id=corpus_id))
 
 
 @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')
 def create_corpus_file(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'])
-@register_breadcrumb(bp, '.entity.files.entity', '', dynamic_list_constructor=corpus_file_dlc)
 @corpus_follower_permission_required('MANAGE_FILES')
 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()
@@ -92,9 +85,9 @@ def 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()
     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,
-        attachment_filename=corpus_file.filename,
+        download_name=corpus_file.filename,
         mimetype=corpus_file.mimetype
     )

app/blueprints/corpora/followers/json_routes.py

@@ -0,0 +1,76 @@
+from flask import abort, flash, jsonify, make_response, request
+from flask_login import current_user
+from app import db
+from app.decorators import content_negotiation
+from app.models import (
+    Corpus,
+    CorpusFollowerAssociation,
+    CorpusFollowerRole,
+    User
+)
+from ..decorators import corpus_follower_permission_required
+from . import bp
+
+
+@bp.route('/<hashid:corpus_id>/followers', methods=['POST'])
+@corpus_follower_permission_required('MANAGE_FOLLOWERS')
+@content_negotiation(consumes='application/json', produces='application/json')
+def create_corpus_followers(corpus_id):
+    usernames = request.json
+    if not (isinstance(usernames, list) or all(isinstance(u, str) for u in usernames)):
+        abort(400)
+    corpus = Corpus.query.get_or_404(corpus_id)
+    for username in usernames:
+        user = User.query.filter_by(username=username, is_public=True).first_or_404()
+        user.follow_corpus(corpus)
+    db.session.commit()
+    response_data = {
+        'message': f'Users are now following "{corpus.title}"',
+        'category': 'corpus'
+    }
+    return response_data, 200
+
+
+@bp.route('/<hashid:corpus_id>/followers/<hashid:follower_id>/role', methods=['PUT'])
+@corpus_follower_permission_required('MANAGE_FOLLOWERS')
+@content_negotiation(consumes='application/json', produces='application/json')
+def update_corpus_follower_role(corpus_id, follower_id):
+    role_name = request.json
+    if not isinstance(role_name, str):
+        abort(400)
+    cfr = CorpusFollowerRole.query.filter_by(name=role_name).first()
+    if cfr is None:
+        abort(400)
+    cfa = CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=follower_id).first_or_404()
+    cfa.role = cfr
+    db.session.commit()
+    response_data = {
+        'message': f'User "{cfa.follower.username}" is now {cfa.role.name}',
+        'category': 'corpus'
+    }
+    return response_data, 200
+
+
+@bp.route('/<hashid:corpus_id>/followers/<hashid:follower_id>', methods=['DELETE'])
+def delete_corpus_follower(corpus_id, follower_id):
+    cfa = CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=follower_id).first_or_404()
+    if not (
+        current_user.id == follower_id
+        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 current_user.is_administrator):
+        abort(403)
+    if current_user.id == follower_id:
+        flash(f'You are no longer following "{cfa.corpus.title}"', 'corpus')
+        response = make_response()
+        response.status_code = 204
+    else:
+        response_data = {
+            'message': f'"{cfa.follower.username}" is not following "{cfa.corpus.title}" anymore',
+            'category': 'corpus'
+        }
+        response = jsonify(response_data)
+        response.status_code = 200
+    cfa.follower.unfollow_corpus(cfa.corpus)
+    db.session.commit()
+    return response

app/blueprints/corpora/routes.py

@@ -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

app/blueprints/errors/handlers.py

@@ -4,11 +4,17 @@ from . import bp
 
 
 @bp.app_errorhandler(HTTPException)
-def handle_http_exception(error):
+def handle_http_exception(e: HTTPException):
     ''' Generic HTTP exception handler '''
     accept_json = request.accept_mimetypes.accept_json
     accept_html = request.accept_mimetypes.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

app/blueprints/jobs/__init__.py

@@ -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')

app/blueprints/jobs/inputs/__init__.py

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

app/blueprints/jobs/inputs/routes.py

@@ -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
+    )

app/blueprints/jobs/results/__init__.py

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

app/blueprints/jobs/results/routes.py

@@ -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
+    )

app/blueprints/jobs/routes.py

@@ -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

app/blueprints/main/cli.py

@@ -1,7 +1,9 @@
 from flask import current_app
 from flask_migrate import upgrade
-import os
+from pathlib import Path
+from app import db
 from app.models import (
+    Corpus,
     CorpusFollowerRole,
     Role,
     SpaCyNLPPipelineModel,
@@ -14,25 +16,22 @@ from . import bp
 @bp.cli.command('deploy')
 def deploy():
     ''' Run deployment tasks. '''
-    # Make default directories
+
     print('Make default directories')
     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')
     upgrade()
 
-    # Insert/Update default database values
     print('Insert/Update default Roles')
     Role.insert_defaults()
     print('Insert/Update default Users')
@@ -43,3 +42,10 @@ def deploy():
     SpaCyNLPPipelineModel.insert_defaults()
     print('Insert/Update default TesseractOCRPipelineModels')
     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

app/blueprints/main/routes.py

@@ -0,0 +1,96 @@
+from flask import abort, flash, jsonify, redirect, render_template, url_for
+from flask_login import current_user, login_required, login_user
+from app.blueprints.auth.forms import LoginForm
+from app.models import Corpus, User
+from . import bp
+from app import db
+
+
+@bp.route('/', methods=['GET', 'POST'])
+def index():
+    form = LoginForm()
+    if form.validate_on_submit():
+        user = User.query.filter((User.email == form.user.data.lower()) | (User.username == form.user.data)).first()
+        if user and user.verify_password(form.password.data):
+            login_user(user, form.remember_me.data)
+            flash('You have been logged in')
+            return redirect(url_for('.dashboard'))
+        flash('Invalid email/username or password', category='error')
+        redirect(url_for('.index'))
+    return render_template(
+        'main/index.html.j2',
+        title='nopaque',
+        form=form
+    )
+
+
+@bp.route('/faq')
+def faq():
+    return render_template(
+        'main/faq.html.j2',
+        title='Frequently Asked Questions'
+    )
+
+
+@bp.route('/dashboard')
+@login_required
+def dashboard():
+    return render_template(
+        'main/dashboard.html.j2',
+        title='Dashboard'
+    )
+
+
+@bp.route('/manual')
+def manual():
+    return render_template(
+        'main/manual.html.j2',
+        title='Manual'
+    )
+
+
+@bp.route('/news')
+def news():
+    return render_template(
+        'main/news.html.j2',
+        title='News'
+    )
+
+
+@bp.route('/privacy-policy')
+def privacy_policy():
+    return render_template(
+        'main/privacy_policy.html.j2',
+        title='Privacy statement (GDPR)'
+    )
+
+
+@bp.route('/terms-of-use')
+def terms_of_use():
+    return render_template(
+        'main/terms_of_use.html.j2',
+        title='Terms of use'
+    )
+
+
+@bp.route('/accept-terms-of-use', methods=['POST'])
+@login_required
+def accept_terms_of_use():
+    current_user.terms_of_use_accepted = True
+    db.session.commit()
+
+    return jsonify('You accepted the terms of use'), 202
+
+
+
+@bp.route('/social')
+@login_required
+def social():
+    corpora = Corpus.query.filter(Corpus.is_public == True, Corpus.user != current_user).all()
+    users = User.query.filter(User.is_public == True, User.id != current_user.id).all()
+    return render_template(
+        'main/social.html.j2',
+        title='Social',
+        corpora=corpora,
+        users=users
+    )

app/blueprints/services/__init__.py

@@ -1,12 +1,11 @@
 from flask import Blueprint
 from flask_login import login_required
-import os
+from pathlib import Path
 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)
 
 bp = Blueprint('services', __name__)

app/blueprints/services/forms.py

@@ -167,7 +167,6 @@ class CreateSpacyNLPPipelineJobForm(CreateJobBaseForm):
         version = kwargs.pop('version', service_manifest['latest_version'])
         super().__init__(*args, **kwargs)
         service_info = service_manifest['versions'][version]
-        print(service_info)
         if self.encoding_detection.render_kw is None:
             self.encoding_detection.render_kw = {}
         self.encoding_detection.render_kw['disabled'] = True

app/blueprints/services/routes.py

@@ -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
 import requests
 from app import db, hashids
@@ -20,13 +19,11 @@ from .forms import (
 
 
 @bp.route('/services')
-@register_breadcrumb(bp, '.', 'Services')
 def services():
     return redirect(url_for('main.dashboard'))
 
 
 @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():
     service = 'file-setup-pipeline'
     service_manifest = SERVICES[service]
@@ -56,7 +53,7 @@ def file_setup_pipeline():
                 abort(500)
         job.status = JobStatus.SUBMITTED
         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')
         return {}, 201, {'Location': job.url}
     return render_template(
@@ -67,7 +64,6 @@ def file_setup_pipeline():
 
 
 @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():
     service_name = 'tesseract-ocr-pipeline'
     service_manifest = SERVICES[service_name]
@@ -100,7 +96,7 @@ def tesseract_ocr_pipeline():
             abort(500)
         job.status = JobStatus.SUBMITTED
         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')
         return {}, 201, {'Location': job.url}
     tesseract_ocr_pipeline_models = [
@@ -118,7 +114,6 @@ def tesseract_ocr_pipeline():
 
 
 @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():
     if not current_app.config.get('NOPAQUE_TRANSKRIBUS_ENABLED'):
         abort(404)
@@ -164,7 +159,7 @@ def transkribus_htr_pipeline():
             abort(500)
         job.status = JobStatus.SUBMITTED
         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')
         return {}, 201, {'Location': job.url}
     return render_template(
@@ -176,7 +171,6 @@ def transkribus_htr_pipeline():
 
 
 @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():
     service = 'spacy-nlp-pipeline'
     service_manifest = SERVICES[service]
@@ -210,7 +204,7 @@ def spacy_nlp_pipeline():
             abort(500)
         job.status = JobStatus.SUBMITTED
         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')
         return {}, 201, {'Location': job.url}
     return render_template(
@@ -223,7 +217,6 @@ def spacy_nlp_pipeline():
 
 
 @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():
     return render_template(
         'services/corpus_analysis.html.j2',

app/blueprints/services/services.yml

@@ -10,7 +10,7 @@ file-setup-pipeline:
 tesseract-ocr-pipeline:
   name: 'Tesseract OCR Pipeline'
   publisher: 'Bielefeld University - CRC 1288 - INF'
-  latest_version: '0.1.1'
+  latest_version: '0.1.2'
   versions:
     0.1.0:
       methods:
@@ -23,6 +23,12 @@ tesseract-ocr-pipeline:
         - 'ocropus_nlbin_threshold'
       publishing_year: 2022
       url: 'https://gitlab.ub.uni-bielefeld.de/sfb1288inf/tesseract-ocr-pipeline/-/releases/v0.1.1'
+    0.1.2:
+      methods:
+        - 'binarization'
+        - 'ocropus_nlbin_threshold'
+      publishing_year: 2023
+      url: 'https://gitlab.ub.uni-bielefeld.de/sfb1288inf/tesseract-ocr-pipeline/-/releases/v0.1.2'
 transkribus-htr-pipeline:
   name: 'Transkribus HTR Pipeline'
   publisher: 'Bielefeld University - CRC 1288 - INF'
@@ -41,7 +47,7 @@ transkribus-htr-pipeline:
 spacy-nlp-pipeline:
   name: 'SpaCy NLP Pipeline'
   publisher: 'Bielefeld University - CRC 1288 - INF'
-  latest_version: '0.1.2'
+  latest_version: '0.1.1'
   versions:
     0.1.0:
       methods:
@@ -56,5 +62,5 @@ spacy-nlp-pipeline:
     0.1.2:
       methods:
         - 'encoding_detection'
-      publishing_year: 2022
+      publishing_year: 2024
       url: 'https://gitlab.ub.uni-bielefeld.de/sfb1288inf/spacy-nlp-pipeline/-/releases/v0.1.2'

app/blueprints/settings/__init__.py

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

app/blueprints/settings/forms.py

@@ -1,6 +1,5 @@
-from flask_login import current_user
 from flask_wtf import FlaskForm
-from flask_wtf.file import FileField, FileRequired
+from flask_wtf.file import FileField, FileRequired, FileSize
 from wtforms import (
     PasswordField,
     SelectField,
@@ -17,7 +16,6 @@ from wtforms.validators import (
     Regexp
 )
 from app.models import User, UserSettingJobStatusMailNotificationLevel
-from app.wtforms.validators import FileSize
 
 
 class UpdateAccountInformationForm(FlaskForm):
@@ -41,7 +39,7 @@ class UpdateAccountInformationForm(FlaskForm):
     )
     submit = SubmitField()
 
-    def __init__(self, user, *args, **kwargs):
+    def __init__(self, user: User, *args, **kwargs):
         if 'data' not in kwargs:
             kwargs['data'] = user.to_json_serializeable()
         if 'prefix' not in kwargs:
@@ -91,7 +89,7 @@ class UpdateProfileInformationForm(FlaskForm):
     )
     submit = SubmitField()
 
-    def __init__(self, user, *args, **kwargs):
+    def __init__(self, user: User, *args, **kwargs):
         if 'data' not in kwargs:
             kwargs['data'] = user.to_json_serializeable()
         if 'prefix' not in kwargs:
@@ -100,7 +98,7 @@ class UpdateProfileInformationForm(FlaskForm):
 
 
 class UpdateAvatarForm(FlaskForm):
-    avatar = FileField('File', validators=[FileRequired(), FileSize(2)])
+    avatar = FileField('File', validators=[FileRequired(), FileSize(2_000_000)])
     submit = SubmitField()
 
     def validate_avatar(self, field):
@@ -132,7 +130,7 @@ class UpdatePasswordForm(FlaskForm):
     )
     submit = SubmitField()
 
-    def __init__(self, user, *args, **kwargs):
+    def __init__(self, user: User, *args, **kwargs):
         if 'prefix' not in kwargs:
             kwargs['prefix'] = 'update-password-form'
         super().__init__(*args, **kwargs)
@@ -154,7 +152,7 @@ class UpdateNotificationsForm(FlaskForm):
     )
     submit = SubmitField()
 
-    def __init__(self, user, *args, **kwargs):
+    def __init__(self, user: User, *args, **kwargs):
         if 'data' not in kwargs:
             kwargs['data'] = user.to_json_serializeable()
         if 'prefix' not in kwargs:

app/blueprints/settings/routes.py

@@ -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

app/blueprints/users/__init__.py

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

app/blueprints/users/cli.py

@@ -0,0 +1,12 @@
+from app.models import User
+from app import db
+from . import bp
+
+
+@bp.cli.command('reset')
+def reset():
+    ''' Reset terms of use accept '''
+    for user in [x for x in User.query.all() if x.terms_of_use_accepted]:
+        print(f'Resetting user {user.username}')
+        user.terms_of_use_accepted = False
+    db.session.commit()

app/blueprints/users/events.py

@@ -0,0 +1,91 @@
+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 {
+            'code': 400,
+            'name': 'Bad Request',
+            'description': 'Invalid User ID.'
+        }
+
+    user_id = hashids.decode(user_hashid)
+
+    if not isinstance(user_id, int):
+        return {
+            'code': 400,
+            'name': 'Bad Request',
+            'description': 'Invalid User ID.'
+        }
+
+    user = User.query.get(user_id)
+
+    if user is None:
+        return {
+            'code': 404,
+            'name': 'Not Found',
+            'description': 'User not found.'
+        }
+
+    if not (
+        user == current_user
+        or current_user.is_administrator
+    ):
+        return {
+            'code': 403,
+            'name': 'Forbidden',
+            'description': 'Not allowed to subscribe to this user.'
+        }
+
+    join_room(f'/users/{user.hashid}')
+
+    return {'code': 204, 'name': 'No Content'}
+
+
+@socketio.on('UNSUBSCRIBE User')
+@socketio_login_required
+def unsubscribe(user_hashid: str) -> dict:
+    if not isinstance(user_hashid, str):
+        return {
+            'code': 400,
+            'name': 'Bad Request',
+            'description': 'Invalid User ID.'
+        }
+
+    user_id = hashids.decode(user_hashid)
+
+    if not isinstance(user_id, int):
+        return {
+            'code': 400,
+            'name': 'Bad Request',
+            'description': 'Invalid User ID.'
+        }
+
+    user = User.query.get(user_id)
+
+    if user is None:
+        return {
+            'code': 404,
+            'name': 'Not Found',
+            'description': 'User not found.'
+        }
+
+    if not (
+        user == current_user
+        or current_user.is_administrator
+    ):
+        return {
+            'code': 403,
+            'name': 'Forbidden',
+            'description': 'Not allowed to unsubscribe from this user.'
+        }
+
+    leave_room(f'/users/{user.hashid}')
+
+    return {'code': 204, 'name': 'No Content'}

app/blueprints/users/routes.py

@@ -0,0 +1,134 @@
+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

app/blueprints/workshops/__init__.py

@@ -0,0 +1,5 @@
+from flask import Blueprint
+
+
+bp = Blueprint('workshops', __name__)
+from . import routes

app/blueprints/workshops/routes.py

@@ -0,0 +1,15 @@
+from flask import redirect, render_template, url_for
+from . import bp
+
+
+@bp.route('')
+def workshops():
+    return redirect(url_for('main.dashboard'))
+
+
+@bp.route('/fgho_sommerschule_2023')
+def fgho_sommerschule_2023():
+    return render_template(
+        'workshops/fgho_sommerschule_2023.html.j2',
+        title='FGHO Sommerschule 2023',
+    )

app/contributions/__init__.py

@@ -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
-)

app/contributions/routes.py

@@ -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'))

app/contributions/spacy_nlp_pipeline_models/utils.py

@@ -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)
-        }
-    ]

app/contributions/tesseract_ocr_pipeline_models/__init__.py

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

app/contributions/tesseract_ocr_pipeline_models/utils.py

@@ -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)
-        }
-    ]

app/contributions/transkribus_htr_pipeline_models/__init__.py

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

app/contributions/transkribus_htr_pipeline_models/routes.py

@@ -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)

app/converters/sandpaper.py

@@ -1,33 +1,34 @@
+from datetime import datetime
 from flask import current_app
+from pathlib import Path
+import json
+import shutil
 from app import db
 from app.models import User, Corpus, CorpusFile
-from datetime import datetime
-import json
-import os
-import shutil
 
 
 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.data_dir = data_dir
 
     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:
             if not json_user['confirmed']:
                 current_app.logger.info(f'Skip unconfirmed user {json_user["username"]}')
                 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)
             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"]}...')
-        user = User(
+        try:
+            user = User.create(
                 confirmed=json_user['confirmed'],
                 email=json_user['email'],
                 last_seen=datetime.fromtimestamp(json_user['last_seen']),
@@ -35,47 +36,34 @@ class SandpaperConverter:
                 password_hash=json_user['password_hash'],  # TODO: Needs to be added manually
                 username=json_user['username']
             )
-        db.session.add(user)
+        except OSError:
-        db.session.flush(objects=[user])
-        db.session.refresh(user)
-        try:
-            user.makedirs()
-        except OSError as e:
-            current_app.logger.error(e)
-            db.session.rollback()
             raise Exception('Internal Server Error')
         for json_corpus in json_user['corpora'].values():
             if not json_corpus['files'].values():
                 current_app.logger.info(f'Skip empty corpus {json_corpus["title"]}')
                 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)
         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"]}...')
-        corpus = Corpus(
+        try:
+            corpus = Corpus.create(
                 user=user,
                 creation_date=datetime.fromtimestamp(json_corpus['creation_date']),
                 description=json_corpus['description'],
                 title=json_corpus['title']
             )
-        db.session.add(corpus)
+        except OSError:
-        db.session.flush(objects=[corpus])
-        db.session.refresh(corpus)
-        try:
-            corpus.makedirs()
-        except OSError as e:
-            current_app.logger.error(e)
-            db.session.rollback()
             raise Exception('Internal Server Error')
         for json_corpus_file in json_corpus['files'].values():
             self.convert_corpus_file(json_corpus_file, corpus, corpus_dir)
         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"]}...')
         corpus_file = CorpusFile(
             corpus=corpus,
@@ -99,13 +87,13 @@ class SandpaperConverter:
         db.session.refresh(corpus_file)
         try:
             shutil.copy2(
-                os.path.join(corpus_dir, json_corpus_file['filename']),
+                corpus_dir / json_corpus_file['filename'],
                 corpus_file.path
             )
         except:
             current_app.logger.warning(
                 '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}'
             )

app/converters/vrt.py

@@ -1,69 +1,25 @@
 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}...')
 
-    with open(input_file) as f:
+    with input_file.open() as f:
         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 has_ent_as_s_attr: {has_ent_as_s_attr}')
 
     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']:
-        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']:
-        pos_attrs_to_string_function = pos_attrs_to_string_2
+        pos_attrs_to_string_function = _pos_attrs_to_string_2
     else:
         raise Exception('Can not handle format')
 
@@ -113,5 +69,49 @@ def normalize_vrt_file(input_file, output_file):
                     current_ent = pos_attrs[4]
         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)
+
+
+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'

app/corpora/cqi_over_sio/__init__.py

@@ -1,113 +0,0 @@
-from cqi import CQiClient
-from cqi.errors import CQiException
-from flask import session
-from flask_login import current_user
-from flask_socketio import ConnectionRefusedError
-from threading import Lock
-from app import db, hashids, socketio
-from app.decorators import socketio_login_required
-from app.models import Corpus, CorpusStatus
-import math
-
-
-'''
-This package tunnels the Corpus Query interface (CQi) protocol through
-Socket.IO (SIO) by wrapping each CQi function in a seperate SIO event.
-
-This module only handles the SIO connect/disconnect, which handles the setup
-and teardown of necessary ressources for later use. Each CQi function has a
-corresponding SIO event. The event handlers are spread across the different
-modules within this package.
-
-Basic concept:
-1. A client connects to the SIO namespace and provides the id of a corpus to be
-   analysed.
-     1.1 The analysis session counter of the corpus is incremented.
-     1.2 A CQiClient and a (Mutex) Lock belonging to it is created.
-     1.3 Wait until the CQP server is running.
-     1.4 Connect the CQiClient to the server.
-     1.5 Save the CQiClient and the Lock in the session for subsequential use.
-2. A client emits an event and may provide a single json object with necessary
-   arguments for the targeted CQi function.
-3. A SIO event handler (decorated with cqi_over_socketio) gets executed.
-     - The event handler function defines all arguments. Hence the client
-       is sent as a single json object, the decorator decomposes it to fit
-       the functions signature. This also includes type checking and proper
-       use of the lock (acquire/release) mechanism.
-4. Wait for more events
-5. The client disconnects from the SIO namespace
-     1.1 The analysis session counter of the corpus is decremented.
-     1.2 The CQiClient and (Mutex) Lock belonging to it are teared down.
-'''
-
-
-NAMESPACE = '/cqi_over_sio'
-
-
-from .cqi import *  # noqa
-
-
-@socketio.on('connect', namespace=NAMESPACE)
-@socketio_login_required
-def connect(auth):
-    # the auth variable is used in a hacky way. It contains the corpus id for
-    # which a corpus analysis session should be started.
-    corpus_id = hashids.decode(auth['corpus_id'])
-    corpus = Corpus.query.get(corpus_id)
-    if corpus is None:
-        # return {'code': 404, 'msg': 'Not Found'}
-        raise ConnectionRefusedError('Not Found')
-    if not (corpus.user == current_user
-            or current_user.is_following_corpus(corpus)
-            or current_user.is_administrator()):
-        # return {'code': 403, 'msg': 'Forbidden'}
-        raise ConnectionRefusedError('Forbidden')
-    if corpus.status not in [
-        CorpusStatus.BUILT,
-        CorpusStatus.STARTING_ANALYSIS_SESSION,
-        CorpusStatus.RUNNING_ANALYSIS_SESSION,
-        CorpusStatus.CANCELING_ANALYSIS_SESSION
-    ]:
-        # return {'code': 424, 'msg': 'Failed Dependency'}
-        raise ConnectionRefusedError('Failed Dependency')
-    if corpus.num_analysis_sessions is None:
-        corpus.num_analysis_sessions = 0
-        db.session.commit()
-    corpus.num_analysis_sessions = Corpus.num_analysis_sessions + 1
-    db.session.commit()
-    retry_counter = 20
-    while corpus.status != CorpusStatus.RUNNING_ANALYSIS_SESSION:
-        if retry_counter == 0:
-            corpus.num_analysis_sessions = Corpus.num_analysis_sessions - 1
-            db.session.commit()
-            return {'code': 408, 'msg': 'Request Timeout'}
-        socketio.sleep(3)
-        retry_counter -= 1
-        db.session.refresh(corpus)
-    cqi_client = CQiClient(f'cqpserver_{corpus_id}', timeout=math.inf)
-    session['cqi_over_sio'] = {
-        'corpus_id': corpus_id,
-        'cqi_client': cqi_client,
-        'cqi_client_lock': Lock(),
-    }
-    # return {'code': 200, 'msg': 'OK'}
-
-
-@socketio.on('disconnect', namespace=NAMESPACE)
-def disconnect():
-    try:
-        cqi_client: CQiClient = session['cqi_over_sio']['cqi_client']
-        cqi_client_lock: Lock = session['cqi_over_sio']['cqi_client_lock']
-    except KeyError:
-        return
-    cqi_client_lock.acquire()
-    try:
-        cqi_client.api.ctrl_bye()
-    except (BrokenPipeError, CQiException):
-        pass
-    cqi_client_lock.release()
-    corpus = Corpus.query.get(session['cqi_over_sio']['corpus_id'])
-    corpus.num_analysis_sessions = Corpus.num_analysis_sessions - 1
-    db.session.commit()
-    session.pop('cqi_over_sio')
-    # return {'code': 200, 'msg': 'OK'}

app/corpora/cqi_over_sio/cqi.py

@@ -1,113 +0,0 @@
-from cqi import CQiClient
-from cqi.errors import CQiException
-from cqi.status import CQiStatus
-from flask import session
-from inspect import signature
-from threading import Lock
-from typing import Callable, Dict, List
-from app import socketio
-from app.decorators import socketio_login_required
-from . import NAMESPACE as ns
-from .extensions import CQI_EXTENSION_FUNCTION_NAMES
-from . import extensions as extensions_module
-
-
-CQI_FUNCTION_NAMES: List[str] = [
-    'ask_feature_cl_2_3',
-    'ask_feature_cqi_1_0',
-    'ask_feature_cqp_2_3',
-    'cl_alg2cpos',
-    'cl_attribute_size',
-    'cl_cpos2alg',
-    'cl_cpos2id',
-    'cl_cpos2lbound',
-    'cl_cpos2rbound',
-    'cl_cpos2str',
-    'cl_cpos2struc',
-    'cl_drop_attribute',
-    'cl_id2cpos',
-    'cl_id2freq',
-    'cl_id2str',
-    'cl_idlist2cpos',
-    'cl_lexicon_size',
-    'cl_regex2id',
-    'cl_str2id',
-    'cl_struc2cpos',
-    'cl_struc2str',
-    'corpus_alignment_attributes',
-    'corpus_charset',
-    'corpus_drop_corpus',
-    'corpus_full_name',
-    'corpus_info',
-    'corpus_list_corpora',
-    'corpus_positional_attributes',
-    'corpus_properties',
-    'corpus_structural_attribute_has_values',
-    'corpus_structural_attributes',
-    'cqp_drop_subcorpus',
-    'cqp_dump_subcorpus',
-    'cqp_fdist_1',
-    'cqp_fdist_2',
-    'cqp_list_subcorpora',
-    'cqp_query',
-    'cqp_subcorpus_has_field',
-    'cqp_subcorpus_size',
-    'ctrl_bye',
-    'ctrl_connect',
-    'ctrl_last_general_error',
-    'ctrl_ping',
-    'ctrl_user_abort'
-]
-
-
-@socketio.on('cqi', namespace=ns)
-@socketio_login_required
-def cqi_over_sio(fn_name: str, fn_args: Dict = {}):
-    try:
-        cqi_client: CQiClient = session['cqi_over_sio']['cqi_client']
-        cqi_client_lock: Lock = session['cqi_over_sio']['cqi_client_lock']
-    except KeyError:
-        return {'code': 424, 'msg': 'Failed Dependency'}
-    if fn_name in CQI_FUNCTION_NAMES:
-        fn: Callable = getattr(cqi_client.api, fn_name)
-    elif fn_name in CQI_EXTENSION_FUNCTION_NAMES:
-        fn: Callable = getattr(extensions_module, fn_name)
-    else:
-        return {'code': 400, 'msg': 'Bad Request'}
-    for param in signature(fn).parameters.values():
-        if param.default is param.empty:
-            if param.name not in fn_args:
-                return {'code': 400, 'msg': 'Bad Request'}
-        else:
-            if param.name not in fn_args:
-                continue
-        if type(fn_args[param.name]) is not param.annotation:
-            return {'code': 400, 'msg': 'Bad Request'}
-    cqi_client_lock.acquire()
-    try:
-        fn_return_value = fn(**fn_args)
-    except BrokenPipeError:
-        fn_return_value = {
-            'code': 500,
-            'msg': 'Internal Server Error'
-        }
-    except CQiException as e:
-        return {
-            'code': 502,
-            'msg': 'Bad Gateway',
-            'payload': {
-                'code': e.code,
-                'desc': e.description,
-                'msg': e.__class__.__name__
-            }
-        }
-    finally:
-        cqi_client_lock.release()
-    if isinstance(fn_return_value, CQiStatus):
-        payload = {
-            'code': fn_return_value.code,
-            'msg': fn_return_value.__class__.__name__
-        }
-    else:
-        payload = fn_return_value
-    return {'code': 200, 'msg': 'OK', 'payload': payload}

app/corpora/cqi_over_sio/extensions.py

@@ -1,314 +0,0 @@
-from collections import Counter
-from cqi import CQiClient
-from cqi.models.corpora import Corpus as CQiCorpus
-from cqi.models.attributes import (
-    PositionalAttribute as CQiPositionalAttribute,
-    StructuralAttribute as CQiStructuralAttribute
-)
-from cqi.status import StatusOk as CQiStatusOk
-from flask import session
-from typing import Dict, List
-import gzip
-import json
-import math
-import os
-from app import db
-from app.models import Corpus
-from .utils import lookups_by_cpos, partial_export_subcorpus, export_subcorpus
-
-
-CQI_EXTENSION_FUNCTION_NAMES: List[str] = [
-    'ext_corpus_update_db',
-    'ext_corpus_static_data',
-    'ext_corpus_paginate_corpus',
-    'ext_cqp_paginate_subcorpus',
-    'ext_cqp_partial_export_subcorpus',
-    'ext_cqp_export_subcorpus',
-]
-
-
-def ext_corpus_update_db(corpus: str) -> CQiStatusOk:
-    db_corpus = Corpus.query.get(session['cqi_over_sio']['corpus_id'])
-    cqi_client: CQiClient = session['cqi_over_sio']['cqi_client']
-    cqi_corpus: CQiCorpus = cqi_client.corpora.get(corpus)
-    db_corpus.num_tokens = cqi_corpus.size
-    db.session.commit()
-    return CQiStatusOk()
-
-
-def ext_corpus_static_data(corpus: str) -> Dict:
-    db_corpus = Corpus.query.get(session['cqi_over_sio']['corpus_id'])
-    static_corpus_data_file = os.path.join(db_corpus.path, 'cwb', 'static.json.gz')
-    if os.path.exists(static_corpus_data_file):
-        with open(static_corpus_data_file, 'rb') as f:
-            return f.read()
-    cqi_client: CQiClient = session['cqi_over_sio']['cqi_client']
-    cqi_corpus: CQiCorpus = cqi_client.corpora.get(corpus)
-    cqi_p_attrs: Dict[str, CQiPositionalAttribute] = {
-        p_attr.name: p_attr
-        for p_attr in cqi_corpus.positional_attributes.list()
-    }
-    cqi_s_attrs: Dict[str, CQiStructuralAttribute] = {
-        s_attr.name: s_attr
-        for s_attr in cqi_corpus.structural_attributes.list()
-    }
-    static_corpus_data = {
-        'corpus': {
-            'bounds': [0, cqi_corpus.size - 1],
-            'counts': {
-                'token': cqi_corpus.size
-            },
-            'freqs': {}
-        },
-        'p_attrs': {},
-        's_attrs': {},
-        'values': {'p_attrs': {}, 's_attrs': {}}
-    }
-    for p_attr in cqi_p_attrs.values():
-        static_corpus_data['corpus']['freqs'][p_attr.name] = {}
-        chunk_size = 10000
-        p_attr_id_list = list(range(p_attr.lexicon_size))
-        chunks = [p_attr_id_list[i:i+chunk_size] for i in range(0, len(p_attr_id_list), chunk_size)]
-        del p_attr_id_list
-        for chunk in chunks:
-            # print(f'corpus.freqs.{p_attr.name}: {chunk[0]} - {chunk[-1]}')
-            static_corpus_data['corpus']['freqs'][p_attr.name].update(
-                dict(zip(chunk, p_attr.freqs_by_ids(chunk)))
-            )
-        del chunks
-        static_corpus_data['p_attrs'][p_attr.name] = {}
-        cpos_list = list(range(cqi_corpus.size))
-        chunks = [cpos_list[i:i+chunk_size] for i in range(0, len(cpos_list), chunk_size)]
-        del cpos_list
-        for chunk in chunks:
-            # print(f'p_attrs.{p_attr.name}: {chunk[0]} - {chunk[-1]}')
-            static_corpus_data['p_attrs'][p_attr.name].update(
-                dict(zip(chunk, p_attr.ids_by_cpos(chunk)))
-            )
-        del chunks
-        static_corpus_data['values']['p_attrs'][p_attr.name] = {}
-        p_attr_id_list = list(range(p_attr.lexicon_size))
-        chunks = [p_attr_id_list[i:i+chunk_size] for i in range(0, len(p_attr_id_list), chunk_size)]
-        del p_attr_id_list
-        for chunk in chunks:
-            # print(f'values.p_attrs.{p_attr.name}: {chunk[0]} - {chunk[-1]}')
-            static_corpus_data['values']['p_attrs'][p_attr.name].update(
-                dict(zip(chunk, p_attr.values_by_ids(chunk)))
-            )
-        del chunks
-    for s_attr in cqi_s_attrs.values():
-        if s_attr.has_values:
-            continue
-        static_corpus_data['corpus']['counts'][s_attr.name] = s_attr.size
-        static_corpus_data['s_attrs'][s_attr.name] = {'lexicon': {}, 'values': None}
-        static_corpus_data['values']['s_attrs'][s_attr.name] = {}
-        ##########################################################################
-        # A faster way to get cpos boundaries for smaller s_attrs                #
-        ##########################################################################
-        # if s_attr.name in ['s', 'ent']:
-        #     cqi_corpus.query('Last', f'<{s_attr.name}> []* </{s_attr.name}>;')
-        #     cqi_subcorpus = cqi_corpus.subcorpora.get('Last')
-        #     first_match = 0
-        #     last_match = cqi_subcorpus.size - 1
-        #     match_boundaries = zip(
-        #         range(first_match, last_match + 1),
-        #         cqi_subcorpus.dump(cqi_subcorpus.fields['match'], first_match, last_match),
-        #         cqi_subcorpus.dump(cqi_subcorpus.fields['matchend'], first_match, last_match)
-        #     )
-        #     for id, lbound, rbound in match_boundaries:
-        #         static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id] = {}
-        #         static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['bounds'] = [lbound, rbound]
-        #         static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['counts'] = {}
-        #         static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['counts']['token'] = rbound - lbound + 1
-        #     cqi_subcorpus.drop()
-        for id in range(0, s_attr.size):
-            # print(f's_attrs.{s_attr.name}.lexicon.{id}')
-            static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id] = {
-                'bounds': None,
-                'counts': None,
-                'freqs': None
-            }
-            if s_attr.name != 'text':
-                continue
-            lbound, rbound = s_attr.cpos_by_id(id)
-            # print(f's_attrs.{s_attr.name}.lexicon.{id}.bounds')
-            static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['bounds'] = [lbound, rbound]
-            # print(f's_attrs.{s_attr.name}.lexicon.{id}.counts')
-            static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['counts'] = {}
-            static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['counts']['token'] = rbound - lbound + 1
-            cpos_list = list(range(lbound, rbound + 1))
-            chunks = [cpos_list[i:i+chunk_size] for i in range(0, len(cpos_list), chunk_size)]
-            del cpos_list
-            ent_ids = set()
-            for chunk in chunks:
-                # print(f'Gather ent_ids from cpos: {chunk[0]} - {chunk[-1]}')
-                ent_ids.update({x for x in cqi_s_attrs['ent'].ids_by_cpos(chunk) if x != -1})
-            static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['counts']['ent'] = len(ent_ids)
-            del ent_ids
-            s_ids = set()
-            for chunk in chunks:
-                # print(f'Gather s_ids from cpos: {chunk[0]} - {chunk[-1]}')
-                s_ids.update({x for x in cqi_s_attrs['s'].ids_by_cpos(chunk) if x != -1})
-            static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['counts']['s'] = len(s_ids)
-            del s_ids
-            # print(f's_attrs.{s_attr.name}.lexicon.{id}.freqs')
-            static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['freqs'] = {}
-            for p_attr in cqi_p_attrs.values():
-                p_attr_ids = []
-                for chunk in chunks:
-                    # print(f'Gather p_attr_ids from cpos: {chunk[0]} - {chunk[-1]}')
-                    p_attr_ids.extend(p_attr.ids_by_cpos(chunk))
-                static_corpus_data['s_attrs'][s_attr.name]['lexicon'][id]['freqs'][p_attr.name] = dict(Counter(p_attr_ids))
-                del p_attr_ids
-            del chunks
-        sub_s_attrs = cqi_corpus.structural_attributes.list(filters={'part_of': s_attr})
-        s_attr_value_names: List[str] = [
-            sub_s_attr.name[(len(s_attr.name) + 1):]
-            for sub_s_attr in sub_s_attrs
-        ]
-        s_attr_id_list = list(range(s_attr.size))
-        chunks = [s_attr_id_list[i:i+chunk_size] for i in range(0, len(s_attr_id_list), chunk_size)]
-        del s_attr_id_list
-        sub_s_attr_values = []
-        for sub_s_attr in sub_s_attrs:
-            tmp = []
-            for chunk in chunks:
-                tmp.extend(sub_s_attr.values_by_ids(chunk))
-            sub_s_attr_values.append(tmp)
-            del tmp
-        del chunks
-        # print(f's_attrs.{s_attr.name}.values')
-        static_corpus_data['s_attrs'][s_attr.name]['values'] = s_attr_value_names
-        # print(f'values.s_attrs.{s_attr.name}')
-        static_corpus_data['values']['s_attrs'][s_attr.name] = {
-            s_attr_id: {
-                s_attr_value_name: sub_s_attr_values[s_attr_value_name_idx][s_attr_id_idx]
-                for s_attr_value_name_idx, s_attr_value_name in enumerate(
-                    static_corpus_data['s_attrs'][s_attr.name]['values']
-                )
-            } for s_attr_id_idx, s_attr_id in enumerate(range(0, s_attr.size))
-        }
-        del sub_s_attr_values
-    with gzip.open(static_corpus_data_file, 'wt') as f:
-        json.dump(static_corpus_data, f)
-    del static_corpus_data
-    with open(static_corpus_data_file, 'rb') as f:
-        return f.read()
-
-
-def ext_corpus_paginate_corpus(
-    corpus: str,
-    page: int = 1,
-    per_page: int = 20
-) -> Dict:
-    cqi_client: CQiClient = session['cqi_over_sio']['cqi_client']
-    cqi_corpus = cqi_client.corpora.get(corpus)
-    # Sanity checks
-    if (
-        per_page < 1
-        or page < 1
-        or (
-            cqi_corpus.size > 0
-            and page > math.ceil(cqi_corpus.size / per_page)
-        )
-    ):
-        return {'code': 416, 'msg': 'Range Not Satisfiable'}
-    first_cpos = (page - 1) * per_page
-    last_cpos = min(cqi_corpus.size, first_cpos + per_page)
-    cpos_list = [*range(first_cpos, last_cpos)]
-    lookups = lookups_by_cpos(cqi_corpus, cpos_list)
-    payload = {}
-    # the items for the current page
-    payload['items'] = [cpos_list]
-    # the lookups for the items
-    payload['lookups'] = lookups
-    # the total number of items matching the query
-    payload['total'] = cqi_corpus.size
-    # the number of items to be displayed on a page.
-    payload['per_page'] = per_page
-    # The total number of pages
-    payload['pages'] = math.ceil(payload['total'] / payload['per_page'])
-    # the current page number (1 indexed)
-    payload['page'] = page if payload['pages'] > 0 else None
-    # True if a previous page exists
-    payload['has_prev'] = payload['page'] > 1 if payload['page'] else False
-    # True if a next page exists.
-    payload['has_next'] = payload['page'] < payload['pages'] if payload['page'] else False  # noqa
-    # Number of the previous page.
-    payload['prev_num'] = payload['page'] - 1 if payload['has_prev'] else None
-    # Number of the next page
-    payload['next_num'] = payload['page'] + 1 if payload['has_next'] else None
-    return payload
-
-
-def ext_cqp_paginate_subcorpus(
-    subcorpus: str,
-    context: int = 50,
-    page: int = 1,
-    per_page: int = 20
-) -> Dict:
-    corpus_name, subcorpus_name = subcorpus.split(':', 1)
-    cqi_client: CQiClient = session['cqi_over_sio']['cqi_client']
-    cqi_corpus = cqi_client.corpora.get(corpus_name)
-    cqi_subcorpus = cqi_corpus.subcorpora.get(subcorpus_name)
-    # Sanity checks
-    if (
-        per_page < 1
-        or page < 1
-        or (
-            cqi_subcorpus.size > 0
-            and page > math.ceil(cqi_subcorpus.size / per_page)
-        )
-    ):
-        return {'code': 416, 'msg': 'Range Not Satisfiable'}
-    offset = (page - 1) * per_page
-    cutoff = per_page
-    cqi_results_export = export_subcorpus(
-        cqi_subcorpus, context=context, cutoff=cutoff, offset=offset)
-    payload = {}
-    # the items for the current page
-    payload['items'] = cqi_results_export.pop('matches')
-    # the lookups for the items
-    payload['lookups'] = cqi_results_export
-    # the total number of items matching the query
-    payload['total'] = cqi_subcorpus.size
-    # the number of items to be displayed on a page.
-    payload['per_page'] = per_page
-    # The total number of pages
-    payload['pages'] = math.ceil(payload['total'] / payload['per_page'])
-    # the current page number (1 indexed)
-    payload['page'] = page if payload['pages'] > 0 else None
-    # True if a previous page exists
-    payload['has_prev'] = payload['page'] > 1 if payload['page'] else False
-    # True if a next page exists.
-    payload['has_next'] = payload['page'] < payload['pages'] if payload['page'] else False  # noqa
-    # Number of the previous page.
-    payload['prev_num'] = payload['page'] - 1 if payload['has_prev'] else None
-    # Number of the next page
-    payload['next_num'] = payload['page'] + 1 if payload['has_next'] else None
-    return payload
-
-
-def ext_cqp_partial_export_subcorpus(
-    subcorpus: str,
-    match_id_list: list,
-    context: int = 50
-) -> Dict:
-    corpus_name, subcorpus_name = subcorpus.split(':', 1)
-    cqi_client: CQiClient = session['cqi_over_sio']['cqi_client']
-    cqi_corpus = cqi_client.corpora.get(corpus_name)
-    cqi_subcorpus = cqi_corpus.subcorpora.get(subcorpus_name)
-    cqi_subcorpus_partial_export = partial_export_subcorpus(cqi_subcorpus, match_id_list, context=context)
-    return cqi_subcorpus_partial_export
-
-
-def ext_cqp_export_subcorpus(
-    subcorpus: str,
-    context: int = 50
-) -> Dict:
-    corpus_name, subcorpus_name = subcorpus.split(':', 1)
-    cqi_client: CQiClient = session['cqi_over_sio']['cqi_client']
-    cqi_corpus = cqi_client.corpora.get(corpus_name)
-    cqi_subcorpus = cqi_corpus.subcorpora.get(subcorpus_name)
-    cqi_subcorpus_export = export_subcorpus(cqi_subcorpus, context=context)
-    return cqi_subcorpus_export

app/corpora/cqi_over_sio/utils.py

@@ -1,133 +0,0 @@
-from cqi.models.corpora import Corpus
-from cqi.models.subcorpora import Subcorpus
-from typing import Dict, List
-from app.models import Corpus
-
-
-def lookups_by_cpos(corpus: Corpus, 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 = attr.values_by_cpos(cpos_list)
-        for i, cpos in enumerate(cpos_list):
-            lookups['cpos_lookup'][cpos][attr.attrs['name']] = \
-                cpos_attr_values[i]
-    for attr in corpus.structural_attributes.list():
-        # We only want to iterate over non subattributes, identifiable by
-        # attr.attrs['has_values'] == False
-        if attr.attrs['has_values']:
-            continue
-        cpos_attr_ids = 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.attrs['name']] = cpos_attr_ids[i]
-        occured_attr_ids = [x for x in set(cpos_attr_ids) if x != -1]
-        if not occured_attr_ids:
-            continue
-        subattrs = corpus.structural_attributes.list(filters={'part_of': attr})
-        if not subattrs:
-            continue
-        lookup_name = f'{attr.attrs["name"]}_lookup'
-        lookups[lookup_name] = {}
-        for attr_id in occured_attr_ids:
-            lookups[lookup_name][attr_id] = {}
-        for subattr in subattrs:
-            subattr_name = subattr.attrs['name'][(len(attr.attrs['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: Subcorpus,
-    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: Subcorpus,
-    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}

app/corpora/events.py

@@ -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'}}

app/corpora/files/__init__.py

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

app/corpora/files/utils.py

@@ -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)
-        }
-    ]

app/corpora/followers/json_routes.py

@@ -1,76 +0,0 @@
-from flask import abort, flash, jsonify, make_response, request
-from flask_login import current_user
-from app import db
-from app.decorators import content_negotiation
-from app.models import (
-    Corpus,
-    CorpusFollowerAssociation,
-    CorpusFollowerRole,
-    User
-)
-from ..decorators import corpus_follower_permission_required
-from . import bp
-
-
-# @bp.route('/<hashid:corpus_id>/followers', methods=['POST'])
-# @corpus_follower_permission_required('MANAGE_FOLLOWERS')
-# @content_negotiation(consumes='application/json', produces='application/json')
-# def create_corpus_followers(corpus_id):
-#     usernames = request.json
-#     if not (isinstance(usernames, list) or all(isinstance(u, str) for u in usernames)):
-#         abort(400)
-#     corpus = Corpus.query.get_or_404(corpus_id)
-#     for username in usernames:
-#         user = User.query.filter_by(username=username, is_public=True).first_or_404()
-#         user.follow_corpus(corpus)
-#     db.session.commit()
-#     response_data = {
-#         'message': f'Users are now following "{corpus.title}"',
-#         'category': 'corpus'
-#     }
-#     return response_data, 200
-
-
-# @bp.route('/<hashid:corpus_id>/followers/<hashid:follower_id>/role', methods=['PUT'])
-# @corpus_follower_permission_required('MANAGE_FOLLOWERS')
-# @content_negotiation(consumes='application/json', produces='application/json')
-# def update_corpus_follower_role(corpus_id, follower_id):
-#     role_name = request.json
-#     if not isinstance(role_name, str):
-#         abort(400)
-#     cfr = CorpusFollowerRole.query.filter_by(name=role_name).first()
-#     if cfr is None:
-#         abort(400)
-#     cfa = CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=follower_id).first_or_404()
-#     cfa.role = cfr
-#     db.session.commit()
-#     response_data = {
-#         'message': f'User "{cfa.follower.username}" is now {cfa.role.name}',
-#         'category': 'corpus'
-#     }
-#     return response_data, 200
-
-
-# @bp.route('/<hashid:corpus_id>/followers/<hashid:follower_id>', methods=['DELETE'])
-# def delete_corpus_follower(corpus_id, follower_id):
-#     cfa = CorpusFollowerAssociation.query.filter_by(corpus_id=corpus_id, follower_id=follower_id).first_or_404()
-#     if not (
-#         current_user.id == follower_id
-#         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 current_user.is_administrator()):
-#         abort(403)
-#     if current_user.id == follower_id:
-#         flash(f'You are no longer following "{cfa.corpus.title}"', 'corpus')
-#         response = make_response()
-#         response.status_code = 204
-#     else:
-#         response_data = {
-#             'message': f'"{cfa.follower.username}" is not following "{cfa.corpus.title}" anymore',
-#             'category': 'corpus'
-#         }
-#         response = jsonify(response_data)
-#         response.status_code = 200
-#     cfa.follower.unfollow_corpus(cfa.corpus)
-#     db.session.commit()
-#     return response

app/corpora/json_routes.py

@@ -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')
-    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

app/corpora/routes.py

@@ -1,121 +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):
-        abort(404)
-        # 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)

app/corpora/utils.py

@@ -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)
-        }
-    ]

app/daemon/__init__.py

@@ -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()

app/decorators.py

@@ -1,8 +1,7 @@
-from flask import abort, current_app, request
+from flask import abort, request
 from flask_login import current_user
 from functools import wraps
-from threading import Thread
+from typing import Optional
-from typing import List, Union
 from werkzeug.exceptions import NotAcceptable
 from app.models import Permission
 
@@ -24,22 +23,21 @@ def admin_required(f):
 
 def socketio_login_required(f):
     @wraps(f)
-    def decorated_function(*args, **kwargs):
+    def wrapper(*args, **kwargs):
         if current_user.is_authenticated:
             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 decorator(f):
         @wraps(f)
-        def decorated_function(*args, **kwargs):
+        def wrapper(*args, **kwargs):
             if not current_user.can(permission):
-                return {'code': 403, 'msg': 'Forbidden'}
+                return {'status': 403, 'statusText': 'Forbidden'}
             return f(*args, **kwargs)
-        return decorated_function
+        return wrapper
     return decorator
 
 
@@ -47,27 +45,9 @@ def socketio_admin_required(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(
-    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):
         @wraps(f)

app/email.py

@@ -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 threading import Thread
 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],
-        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():
-            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()
     return thread

app/extensions/nopaque_flask_admin_views.py

@@ -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

app/extensions/nopaque_sqlalchemy_type_decorators.py

@@ -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)

app/jobs/__init__.py

@@ -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

app/jobs/handle_corpora.py

@@ -1,12 +1,16 @@
-from app import docker_client
-from app.models import Corpus, CorpusStatus
 from flask import current_app
 import docker
 import os
 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()
     for corpus in [x for x in corpora if x.status == CorpusStatus.SUBMITTED]:
         _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]:
         corpus.status = CorpusStatus.CANCELING_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]:
         _create_cqpserver_container(corpus)
     for corpus in [x for x in corpora if x.status == CorpusStatus.CANCELING_ANALYSIS_SESSION]:
         _remove_cqpserver_container(corpus)
+    db.session.commit()
 
-def _create_build_corpus_service(corpus):
+def _create_build_corpus_service(corpus: Corpus):
     ''' # Docker service settings # '''
     ''' ## Command ## '''
     command = ['bash', '-c']
@@ -45,12 +50,10 @@ def _create_build_corpus_service(corpus):
     ''' ## Constraints ## '''
     constraints = ['node.role==worker']
     ''' ## Image ## '''
-    image = f'{current_app.config["NOPAQUE_DOCKER_IMAGE_PREFIX"]}cwb:r1702'
+    image = f'{current_app.config["NOPAQUE_DOCKER_IMAGE_PREFIX"]}cwb:r1887'
     ''' ## 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 = []
@@ -95,7 +98,7 @@ def _create_build_corpus_service(corpus):
         return
     corpus.status = CorpusStatus.QUEUED
 
-def _checkout_build_corpus_service(corpus):
+def _checkout_build_corpus_service(corpus: Corpus):
     service_name = f'build-corpus_{corpus.id}'
     try:
         service = docker_client.services.get(service_name)
@@ -123,8 +126,7 @@ def _checkout_build_corpus_service(corpus):
     except docker.errors.DockerException as 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.append(
@@ -139,11 +141,11 @@ def _create_cqpserver_container(corpus):
     ''' ## Entrypoint ## '''
     entrypoint = ['bash', '-c']
     ''' ## Image ## '''
-    image = f'{current_app.config["NOPAQUE_DOCKER_IMAGE_PREFIX"]}cwb:r1702'
+    image = f'{current_app.config["NOPAQUE_DOCKER_IMAGE_PREFIX"]}cwb:r1887'
     ''' ## Name ## '''
-    name = f'cqpserver_{corpus.id}'
+    name = f'nopaque-cqpserver-{corpus.id}'
     ''' ## Network ## '''
-    network = f'{current_app.config["DOCKER_NETWORK_NAME"]}'
+    network = f'{current_app.config["NOPAQUE_DOCKER_NETWORK_NAME"]}'
     ''' ## Volumes ## '''
     volumes = []
     ''' ### Corpus data volume ### '''
@@ -198,8 +200,8 @@ def _create_cqpserver_container(corpus):
         return
     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:
         docker_client.containers.get(container_name)
     except docker.errors.NotFound as e:
@@ -209,8 +211,8 @@ def _checkout_analysing_corpus_container(corpus):
     except docker.errors.DockerException as 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:
         container = docker_client.containers.get(container_name)
     except docker.errors.NotFound:

app/jobs/handle_jobs.py

@@ -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 flask import current_app
 from werkzeug.utils import secure_filename
@@ -13,9 +5,21 @@ import docker
 import json
 import os
 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()
     for job in [x for x in jobs if x.status == JobStatus.SUBMITTED]:
         _create_job_service(job)
@@ -23,8 +27,9 @@ def check_jobs():
         _checkout_job_service(job)
     for job in [x for x in jobs if x.status == JobStatus.CANCELING]:
         _remove_job_service(job)
+    db.session.commit()
 
-def _create_job_service(job):
+def _create_job_service(job: Job):
     ''' # Docker service settings # '''
     ''' ## Service specific settings ## '''
     if job.service == 'file-setup-pipeline':
@@ -81,9 +86,7 @@ def _create_job_service(job):
     constraints = ['node.role==worker']
     ''' ## Labels ## '''
     labels = {
-        'origin': current_app.config['SERVER_NAME'],
+        'origin': current_app.config['SERVER_NAME']
-        'type': 'job',
-        'job_id': str(job.id)
     }
     ''' ## Mounts ## '''
     mounts = []
@@ -164,7 +167,7 @@ def _create_job_service(job):
         return
     job.status = JobStatus.QUEUED
 
-def _checkout_job_service(job):
+def _checkout_job_service(job: Job):
     service_name = f'job_{job.id}'
     try:
         service = docker_client.services.get(service_name)
@@ -213,7 +216,7 @@ def _checkout_job_service(job):
     except docker.errors.DockerException as 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}'
     try:
         service = docker_client.services.get(service_name)

app/jobs/json_routes.py

@@ -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

app/jobs/routes.py

@@ -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
-    )

app/jobs/utils.py

@@ -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)
-        }
-    ]