Skip to content
Snippets Groups Projects
models.py 32.2 KiB
Newer Older
Patrick Jentsch's avatar
Patrick Jentsch committed
from app.converters.vrt import normalize_vrt_file
from datetime import datetime, timedelta
from flask import current_app, url_for
from flask_hashids import HashidMixin
from flask_login import UserMixin
Patrick Jentsch's avatar
Patrick Jentsch committed
from itsdangerous import BadSignature, TimedJSONWebSignatureSerializer
from werkzeug.security import generate_password_hash, check_password_hash
from . import db, login
TRANSKRIBUS_HTR_MODELS = \
    json.loads(requests.get('https://transkribus.eu/TrpServer/rest/models/text').content)['trpModelMetadata']  # noqa


class IntEnumColumn(db.TypeDecorator):
    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
    def process_result_value(self, value, dialect):
        return self.enum_type(value)
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):  # noqa
            return value
        else:
            return TypeError()

    def process_result_value(self, value, dialect):
        return json.loads(value)


class FileMixin:
    creation_date = db.Column(db.DateTime, default=datetime.utcnow)
    last_edited_date = db.Column(db.DateTime, default=datetime.utcnow)
    mimetype = db.Column(db.String(255))

    def file_mixin_to_dict(self, backrefs=False, relationships=False):
        return {
            'creation_date': self.creation_date.isoformat() + 'Z',
            'filename': self.filename,
            'last_edited_date': self.last_edited_date.isoformat() + 'Z',
            'mimetype': self.mimetype
        }


class Permission(IntEnum):
    '''
    Defines User permissions as integers by the power of 2. User permission
    can be evaluated using the bitwise operator &.
    '''
    ADMINISTRATE = 1
    CONTRIBUTE = 2
    USE_API = 4


class Role(HashidMixin, db.Model):
Patrick Jentsch's avatar
Patrick Jentsch committed
    # Primary key
    id = db.Column(db.Integer, primary_key=True)
Stephan Porada's avatar
Stephan Porada committed
    # Fields
    default = db.Column(db.Boolean, default=False, index=True)
Patrick Jentsch's avatar
Patrick Jentsch committed
    name = db.Column(db.String(64), unique=True)
Patrick Jentsch's avatar
Patrick Jentsch committed
    permissions = db.Column(db.Integer)
Patrick Jentsch's avatar
Patrick Jentsch committed
    # Relationships
    users = db.relationship('User', backref='role', lazy='dynamic')

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
        if self.permissions is None:
            self.permissions = 0
        return f'<Role {self.name}>'
    def add_permission(self, permission):
        if not self.has_permission(permission):
            self.permissions += permission
    def has_permission(self, permission):
        return self.permissions & permission == permission

    def remove_permission(self, permission):
        if self.has_permission(permission):
            self.permissions -= permission

    def reset_permissions(self):
        self.permissions = 0

    def to_dict(self, backrefs=False, relationships=False):
        dict_role = {
            'id': self.hashid,
            'default': self.default,
            'name': self.name,
            'permissions': self.permissions
        }
        if relationships:
                x.to_dict(backrefs=False, relationships=True)
                for x in self.users
            }
        return dict_role
        roles = {
            'User': [],
            'API user': [Permission.USE_API],
            'Contributor': [Permission.CONTRIBUTE],
            'Administrator': [
                Permission.ADMINISTRATE,
                Permission.CONTRIBUTE,
                Permission.USE_API
            ]
        }
        default_role_name = 'User'
        for role_name, permissions in roles.items():
            role = Role.query.filter_by(name=role_name).first()
                role = Role(name=role_name)
            for permission in permissions:
                role.add_permission(permission)
            role.default = role.name == default_role_name
class UserSettingJobStatusMailNotificationLevel(IntEnum):
    NONE = 1
    END = 2
    ALL = 3


class User(HashidMixin, UserMixin, db.Model):
Patrick Jentsch's avatar
Patrick Jentsch committed
    # Primary key
    id = db.Column(db.Integer, primary_key=True)
    # Foreign keys
    role_id = db.Column(db.Integer, db.ForeignKey('roles.id'))
Stephan Porada's avatar
Stephan Porada committed
    # Fields
Patrick Jentsch's avatar
Patrick Jentsch committed
    confirmed = db.Column(db.Boolean, default=False)
    email = db.Column(db.String(254), unique=True, index=True)
    last_seen = db.Column(db.DateTime(), default=datetime.utcnow)
    member_since = db.Column(db.DateTime(), default=datetime.utcnow)
    password_hash = db.Column(db.String(128))
    token = db.Column(db.String(32), index=True, unique=True)
    token_expiration = db.Column(db.DateTime)
    username = db.Column(db.String(64), unique=True, index=True)
    setting_dark_mode = db.Column(db.Boolean, default=False)
    setting_job_status_mail_notification_level = db.Column(
        IntEnumColumn(UserSettingJobStatusMailNotificationLevel),
        default=UserSettingJobStatusMailNotificationLevel.END
    # Backrefs: role: Role
Patrick Jentsch's avatar
Patrick Jentsch committed
    # Relationships
    tesseract_ocr_models = db.relationship(
        'TesseractOCRModel',
        backref='user',
        cascade='all, delete-orphan',
        lazy='dynamic'
    )
    transkribus_htr_models = db.relationship(
        'TranskribusHTRModel',
        backref='user',
        cascade='all, delete-orphan',
        lazy='dynamic'
    )
    corpora = db.relationship(
        'Corpus',
        backref='user',
        cascade='all, delete-orphan',
        lazy='dynamic'
    )
    jobs = db.relationship(
        'Job',
        backref='user',
        cascade='all, delete-orphan',
        lazy='dynamic'
    )

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
        if self.role is not None:
            return
        if self.email == current_app.config['NOPAQUE_ADMIN']:
            self.role = Role.query.filter_by(name='Administrator').first()
        else:
            self.role = Role.query.filter_by(default=True).first()

    def __repr__(self):
        return f'<User {self.username}>'
    def jsonpatch_path(self):
        return f'/users/{self.hashid}'

    @property
    def password(self):
        raise AttributeError('password is not a readable attribute')

    @password.setter
    def password(self, password):
        self.password_hash = generate_password_hash(password)

    @property
    def path(self):
        return os.path.join(
            current_app.config.get('NOPAQUE_DATA_DIR'), 'users', str(self.id))

    def can(self, permission):
        return self.role.has_permission(permission)
Stephan Porada's avatar
Stephan Porada committed
    def confirm(self, token):
Patrick Jentsch's avatar
Patrick Jentsch committed
        s = TimedJSONWebSignatureSerializer(current_app.config['SECRET_KEY'])
Stephan Porada's avatar
Stephan Porada committed
        try:
            data = s.loads(token.encode('utf-8'))
Patrick Jentsch's avatar
Patrick Jentsch committed
        except BadSignature:
Stephan Porada's avatar
Stephan Porada committed
            return False
        if data.get('confirm') != self.hashid:
Stephan Porada's avatar
Stephan Porada committed
            return False
        self.confirmed = True
        db.session.add(self)
        return True

Patrick Jentsch's avatar
Patrick Jentsch committed
    def delete(self):
        shutil.rmtree(self.path, ignore_errors=True)
        db.session.delete(self)
    def generate_confirmation_token(self, expiration=3600):
        s = TimedJSONWebSignatureSerializer(
            current_app.config['SECRET_KEY'], expiration)
        return s.dumps({'confirm': self.hashid}).decode('utf-8')

    def generate_reset_token(self, expiration=3600):
        s = TimedJSONWebSignatureSerializer(
            current_app.config['SECRET_KEY'], expiration)
        return s.dumps({'reset': self.hashid}).decode('utf-8')

    def get_token(self, expires_in=3600):
        now = datetime.utcnow()
        if self.token and self.token_expiration > now + timedelta(seconds=60):
            return self.token
        self.token = base64.b64encode(os.urandom(24)).decode('utf-8')
        self.token_expiration = now + timedelta(seconds=expires_in)
        db.session.add(self)
        return self.token

    def is_administrator(self):
        return self.can(Permission.ADMINISTRATE)

    def makedirs(self):
        os.mkdir(self.path)
        os.mkdir(os.path.join(self.path, 'tesseract_ocr_models'))
        os.mkdir(os.path.join(self.path, 'corpora'))
        os.mkdir(os.path.join(self.path, 'jobs'))

    def revoke_token(self):
        self.token_expiration = datetime.utcnow() - timedelta(seconds=1)

    def to_dict(self, backrefs=False, relationships=False):
        dict_user = {
            'id': self.hashid,
            'role_id': self.role.hashid,
            'confirmed': self.confirmed,
            'email': self.email,
            'last_seen': self.last_seen.isoformat() + 'Z',
            'member_since': self.member_since.isoformat() + 'Z',
            'username': self.username,
            'settings': {
                'dark_mode': self.setting_dark_mode,
                'job_status_mail_notification_level':
                    self.setting_job_status_mail_notification_level.name
            }
        }
        if backrefs:
            dict_user['role'] = self.role.to_dict(
                backrefs=True, relationships=False)
        if relationships:
            dict_user['corpora'] = {
                x.hashid: x.to_dict(backrefs=False, relationships=True)
                for x in self.corpora
            }
            dict_user['jobs'] = {
                x.hashid: x.to_dict(backrefs=False, relationships=True)
                for x in self.jobs
            }
                x.hashid: x.to_dict(backrefs=False, relationships=True)
            }
        return dict_user

    def verify_password(self, password):
        return check_password_hash(self.password_hash, password)

    @staticmethod
    def check_token(token):
        user = User.query.filter_by(token=token).first()
        if user is None or user.token_expiration < datetime.utcnow():
            return None
        return user

    @staticmethod
    def insert_defaults():
        if User.query.filter_by(username='nopaque').first() is not None:
            return
        user = User(username='nopaque')
        db.session.add(user)
        db.session.flush(objects=[user])
        db.session.refresh(user)
        try:
            user.makedirs()
        except OSError as e:
            current_app.logger.error(e)
            db.session.rollback()
        db.session.commit()

    @staticmethod
    def reset_password(token, new_password):
        s = TimedJSONWebSignatureSerializer(current_app.config['SECRET_KEY'])
        try:
            data = s.loads(token.encode('utf-8'))
        except BadSignature:
            return False
        user = User.query.get(data.get('reset'))
        if user is None:
            return False
        user.password = new_password
        db.session.add(user)
        return True
class TesseractOCRModel(FileMixin, HashidMixin, db.Model):
    __tablename__ = 'tesseract_ocr_models'
    # Primary key
    id = db.Column(db.Integer, primary_key=True)
    # Foreign keys
    user_id = db.Column(db.Integer, db.ForeignKey('users.id'))
    # Fields
    compatible_service_versions = db.Column(ContainerColumn(list, 255))
    description = db.Column(db.String(255))
    publisher = db.Column(db.String(128))
    publisher_url = db.Column(db.String(512))
    publishing_url = db.Column(db.String(512))
    shared = db.Column(db.Boolean, default=False)
    title = db.Column(db.String(64))
    version = db.Column(db.String(16))
    # Backrefs: user: User

    @property
    def path(self):
        return os.path.join(
            self.user.path,
            'tesseract_ocr_models',
            str(self.id)
        )

    def to_dict(self, backrefs=False, relationships=False):
        dict_tesseract_ocr_model = {
            'id': self.hashid,
            'user_id': self.user.hashid,
            'compatible_service_versions': self.compatible_service_versions,
            'description': self.description,
            'publisher': self.publisher,
            'publisher_url': self.publisher_url,
            'publishing_url': self.publishing_url,
            'shared': self.shared,
            'title': self.title,
            **self.file_mixin_to_dict()
        }
        if backrefs:
            dict_tesseract_ocr_model['user'] = self.user.to_dict(
                backrefs=True, relationships=False)
        if relationships:
            pass
        return dict_tesseract_ocr_model

    @staticmethod
    def insert_defaults():
        user = User.query.filter_by(username='nopaque').first()
        defaults_file = os.path.join(
            os.path.dirname(os.path.abspath(__file__)),
            'TesseractOCRModel.defaults.yml'
        )
        with open(defaults_file, 'r') as f:
            defaults = yaml.safe_load(f)
        for m in defaults:
            model = TesseractOCRModel.query.filter_by(title=m['title'], version=m['version']).first()  # noqa
            if model is not None:
                model.compatible_service_versions = m['compatible_service_versions']
                model.description = m['description']
                model.publisher = m['publisher']
                model.publisher_url = m['publisher_url']
                model.publishing_url = m['publishing_url']
                model.publishing_year = m['publishing_year']
                model.shared = True
                model.title = m['title']
                model.version = m['version']
            model = TesseractOCRModel(
                compatible_service_versions=m['compatible_service_versions'],
                description=m['description'],
                publisher=m['publisher'],
                publisher_url=m['publisher_url'],
                publishing_url=m['publishing_url'],
            db.session.add(model)
            db.session.flush(objects=[model])
            db.session.refresh(model)
            model.filename = f'{model.id}.traineddata'
                desc=f'{model.title} ({model.filename})',
                unit="B",
                unit_scale=True,
                unit_divisor=1024,
                total=int(r.headers['Content-Length'])
            )
            pbar.clear()
                for chunk in r.iter_content(chunk_size=1024):
                    if chunk:  # filter out keep-alive new chunks
                        pbar.update(len(chunk))
                        f.write(chunk)
                pbar.close()
        db.session.commit()


class TranskribusHTRModel(HashidMixin, db.Model):
    __tablename__ = 'transkribus_htr_models'
    # Primary key
    id = db.Column(db.Integer, primary_key=True)
    # Foreign keys
    user_id = db.Column(db.Integer, db.ForeignKey('users.id'))
    # Fields
    shared = db.Column(db.Boolean, default=False)
    transkribus_model_id = db.Column(db.Integer)
    transkribus_name = db.Column(db.String(64))
    # Backrefs: user: User

    def to_dict(self, backrefs=False, relationships=False):
        dict_tesseract_ocr_model = {
            'id': self.hashid,
            'user_id': self.user.hashid,
            'shared': self.shared,
            'transkribus_model_id': self.transkribus_model_id,
            'transkribus_name': self.transkribus_name
        }
        if backrefs:
            dict_tesseract_ocr_model['user'] = \
                self.user.to_dict(backrefs=True, relationships=False)
        if relationships:
            pass
        return dict_tesseract_ocr_model

    @staticmethod
    def insert_defaults():
        user = User.query.filter_by(username='nopaque').first()
        models = [
            m for m in TRANSKRIBUS_HTR_MODELS if True
            and 'creator' in m and m['creator'] == 'Transkribus Team'
            and 'docType' in m and m['docType'] == 'handwritten'
        ]
        for m in models:
            model = TranskribusHTRModel.query.filter_by(transkribus_model_id=m['modelId']).first()  # noqa
            if model is not None:
                model.shared = True
                model.transkribus_model_id = m['modelId']
                model.transkribus_name = m['name']
                continue
            model = TranskribusHTRModel(
                shared=True,
                transkribus_name=m['name'],
                transkribus_model_id=m['modelId'],
                user=user,
            )
            db.session.add(model)
        db.session.commit()


class JobInput(FileMixin, HashidMixin, db.Model):
Patrick Jentsch's avatar
Patrick Jentsch committed
    __tablename__ = 'job_inputs'
    # Primary key
    id = db.Column(db.Integer, primary_key=True)
    # Foreign keys
    job_id = db.Column(db.Integer, db.ForeignKey('jobs.id'))
    # Backrefs: job: Job

    def __repr__(self):
        return f'<JobInput {self.filename}>'
    @property
    def download_url(self):
        return url_for(
            'jobs.download_job_input',
            job_id=self.job.id,
            job_input_id=self.id
        )
        return f'{self.job.jsonpatch_path}/inputs/{self.hashid}'
        return os.path.join(self.job.path, 'inputs', str(self.id))
    def to_dict(self, backrefs=False, relationships=False):
        dict_job_input = {
            'id': self.hashid,
            'job_id': self.job.hashid,
            'download_url': self.download_url,
            'url': self.url,
            **self.file_mixin_to_dict()
        }
        if backrefs:
            dict_job_input['job'] = self.job.to_dict(
                backrefs=True, relationships=False)
        return dict_job_input

    @property
    def url(self):
        return url_for(
            'jobs.job',
            job_id=self.job_id,
            _anchor=f'job-{self.job.hashid}-input-{self.hashid}'
        )

    @property
    def user_hashid(self):
        return self.job.user.hashid
    @property
    def user_id(self):
        return self.job.user_id

class JobResult(FileMixin, HashidMixin, db.Model):
Patrick Jentsch's avatar
Patrick Jentsch committed
    __tablename__ = 'job_results'
    # Primary key
    id = db.Column(db.Integer, primary_key=True)
    # Foreign keys
    job_id = db.Column(db.Integer, db.ForeignKey('jobs.id'))
    # Backrefs: job: Job

    def __repr__(self):
        return f'<JobResult {self.filename}>'
    @property
    def download_url(self):
        return url_for(
            'jobs.download_job_result',
            job_id=self.job_id,
            job_result_id=self.id
        )
        return f'{self.job.jsonpatch_path}/results/{self.hashid}'
        return os.path.join(self.job.path, 'results', str(self.id))
    def to_dict(self, backrefs=False, relationships=False):
        dict_job_result = {
            'id': self.hashid,
            'job_id': self.job.hashid,
            'download_url': self.download_url,
            'url': self.url,
            **self.file_mixin_to_dict(
                backrefs=backrefs, relationships=relationships)
        }
        if backrefs:
            dict_job_result['job'] = self.job.to_dict(
                backrefs=True, relationships=False)
        return dict_job_result

    @property
    def url(self):
        return url_for(
            'jobs.job',
            job_id=self.job_id,
            _anchor=f'job-{self.job.hashid}-result-{self.hashid}'
        )

    @property
    def user_hashid(self):
        return self.job.user.hashid
    @property
    def user_id(self):
        return self.job.user_id

class JobStatus(IntEnum):
    INITIALIZING = 1
    SUBMITTED = 2
    QUEUED = 3
    RUNNING = 4
    CANCELING = 5
    CANCELED = 6
    COMPLETED = 7
    FAILED = 8


class Job(HashidMixin, db.Model):
Stephan Porada's avatar
Stephan Porada committed
    Class to define Jobs.
Stephan Porada's avatar
Stephan Porada committed
    __tablename__ = 'jobs'
Patrick Jentsch's avatar
Patrick Jentsch committed
    # Primary key
Stephan Porada's avatar
Stephan Porada committed
    id = db.Column(db.Integer, primary_key=True)
    # Foreign keys
    user_id = db.Column(db.Integer, db.ForeignKey('users.id'))
Stephan Porada's avatar
Stephan Porada committed
    # Fields
    creation_date = db.Column(db.DateTime(), default=datetime.utcnow)
Patrick Jentsch's avatar
Patrick Jentsch committed
    description = db.Column(db.String(255))
    end_date = db.Column(db.DateTime())
Patrick Jentsch's avatar
Patrick Jentsch committed
    service = db.Column(db.String(64))
    service_args = db.Column(ContainerColumn(dict, 255))
Patrick Jentsch's avatar
Patrick Jentsch committed
    service_version = db.Column(db.String(16))
        IntEnumColumn(JobStatus),
        default=JobStatus.INITIALIZING
    )
Patrick Jentsch's avatar
Patrick Jentsch committed
    title = db.Column(db.String(32))
    # Backrefs: user: User
Patrick Jentsch's avatar
Patrick Jentsch committed
    # Relationships
    inputs = db.relationship(
        'JobInput',
        backref='job',
        cascade='all, delete-orphan',
        lazy='dynamic'
    )
    results = db.relationship(
        'JobResult',
        backref='job',
        cascade='all, delete-orphan',
        lazy='dynamic'
    )

    def __repr__(self):
        return f'<Job {self.title}>'
        return f'{self.user.jsonpatch_path}/jobs/{self.hashid}'
        return os.path.join(self.user.path, 'jobs', str(self.id))
Stephan Porada's avatar
Stephan Porada committed

    @property
Patrick Jentsch's avatar
Patrick Jentsch committed
    def url(self):
        return url_for('jobs.job', job_id=self.id)
    @property
    def user_hashid(self):
        return self.user.hashid
Stephan Porada's avatar
Stephan Porada committed

Patrick Jentsch's avatar
Patrick Jentsch committed
    def delete(self):
        Delete the job and its inputs and results from the database.
        if self.status not in [JobStatus.COMPLETED, JobStatus.FAILED]:  # noqa
            self.status = JobStatus.CANCELING
                # In case the daemon handled a job in any way
                if self.status != JobStatus.CANCELING:
                    self.status = JobStatus.CANCELING
                    db.session.commit()
                sleep(1)
                db.session.refresh(self)
        shutil.rmtree(self.path, ignore_errors=True)
Patrick Jentsch's avatar
Patrick Jentsch committed
        db.session.delete(self)

    def makedirs(self):
        os.mkdir(self.path)
        os.mkdir(os.path.join(self.path, 'inputs'))
        os.mkdir(os.path.join(self.path, 'pipeline_data'))
        os.mkdir(os.path.join(self.path, 'results'))

    def restart(self):
        '''
        Restart a job - only if the status is complete or failed
        if self.status not in [JobStatus.COMPLETED, JobStatus.FAILED]:  # noqa
            raise Exception('Could not restart job: status is not "completed/failed"')  # noqa
        shutil.rmtree(os.path.join(self.path, 'results'), ignore_errors=True)
        shutil.rmtree(os.path.join(self.path, 'pyflow.data'), ignore_errors=True)  # noqa
        for result in self.results:
            db.session.delete(result)
        self.end_date = None
    def to_dict(self, backrefs=False, relationships=False):
            'id': self.hashid,
            'user_id': self.user.hashid,
            'creation_date': self.creation_date.isoformat() + 'Z',
            'description': self.description,
            'end_date': None if self.end_date is None else f'{self.end_date.isoformat()}Z',  # noqa
            'service_version': self.service_version,
            'url': self.url
        if backrefs:
            dict_job['user'] = self.user.to_dict(
                backrefs=True, relationships=False)
        if relationships:
            dict_job['inputs'] = {
                x.hashid: x.to_dict(backrefs=False, relationships=True)
                for x in self.inputs
            }
            dict_job['results'] = {
                x.hashid: x.to_dict(backrefs=False, relationships=True)
                for x in self.results
            }
Stephan Porada's avatar
Stephan Porada committed

class CorpusFile(FileMixin, HashidMixin, db.Model):
Patrick Jentsch's avatar
Patrick Jentsch committed
    __tablename__ = 'corpus_files'
    # Primary key
    id = db.Column(db.Integer, primary_key=True)
    # Foreign keys
    corpus_id = db.Column(db.Integer, db.ForeignKey('corpora.id'))
Stephan Porada's avatar
Stephan Porada committed
    # Fields
    address = db.Column(db.String(255))
    author = db.Column(db.String(255))
    booktitle = db.Column(db.String(255))
    chapter = db.Column(db.String(255))
    editor = db.Column(db.String(255))
    institution = db.Column(db.String(255))
    journal = db.Column(db.String(255))
    pages = db.Column(db.String(255))
    publisher = db.Column(db.String(255))
    publishing_year = db.Column(db.Integer)
    school = db.Column(db.String(255))
    title = db.Column(db.String(255))
    # Backrefs: corpus: Corpus
    @property
    def download_url(self):
        return url_for(
            'corpora.download_corpus_file',
            corpus_id=self.corpus_id,
            corpus_file_id=self.id
        )
Patrick Jentsch's avatar
Patrick Jentsch committed
        return f'{self.corpus.jsonpatch_path}/files/{self.hashid}'
        return os.path.join(self.corpus.path, 'files', str(self.id))
    @property
    def url(self):
        return url_for(
            'corpora.corpus_file',
            corpus_id=self.corpus_id,
            corpus_file_id=self.id
        )

    @property
    def user_hashid(self):
        return self.corpus.user.hashid
    @property
    def user_id(self):
        return self.corpus.user_id

    def delete(self):
        except OSError:
            current_app.logger.error(
                f'Removing {self.path} led to an OSError!'
        db.session.delete(self)
        self.corpus.status = CorpusStatus.UNPREPARED
    def to_dict(self, backrefs=False, relationships=False):
        dict_corpus_file = {
            'id': self.hashid,
            'corpus_id': self.corpus.hashid,
            'download_url': self.download_url,
            'url': self.url,
            'address': self.address,
            'author': self.author,
            'booktitle': self.booktitle,
            'chapter': self.chapter,
            'editor': self.editor,
            'institution': self.institution,
            'journal': self.journal,
            'pages': self.pages,
            'publisher': self.publisher,
            'publishing_year': self.publishing_year,
            'school': self.school,
            'title': self.title,
            **self.file_mixin_to_dict(
                backrefs=backrefs, relationships=relationships)
        }
        if backrefs:
            dict_corpus_file['corpus'] = self.corpus.to_dict(
                backrefs=True, relationships=False)
Patrick Jentsch's avatar
Patrick Jentsch committed
        return dict_corpus_file
class CorpusStatus(IntEnum):
    UNPREPARED = 1
    SUBMITTED = 2
    QUEUED = 3
    BUILDING = 4
    BUILT = 5
    FAILED = 6
    STARTING_ANALYSIS_SESSION = 7
    RUNNING_ANALYSIS_SESSION = 8
    CANCELING_ANALYSIS_SESSION = 9


class Corpus(HashidMixin, db.Model):
Patrick Jentsch's avatar
Patrick Jentsch committed
    Class to define a corpus.
Patrick Jentsch's avatar
Patrick Jentsch committed
    __tablename__ = 'corpora'
    # Primary key
    id = db.Column(db.Integer, primary_key=True)
    # Foreign keys
    user_id = db.Column(db.Integer, db.ForeignKey('users.id'))
Stephan Porada's avatar
Stephan Porada committed
    # Fields
    creation_date = db.Column(db.DateTime(), default=datetime.utcnow)
    description = db.Column(db.String(255))
    last_edited_date = db.Column(db.DateTime(), default=datetime.utcnow)
        IntEnumColumn(CorpusStatus),
        default=CorpusStatus.UNPREPARED
    )
Patrick Jentsch's avatar
Patrick Jentsch committed
    title = db.Column(db.String(32))
    num_analysis_sessions = db.Column(db.Integer, default=0)
    num_tokens = db.Column(db.Integer, default=0)
    # Backrefs: user: User
Patrick Jentsch's avatar
Patrick Jentsch committed
    # Relationships
    files = db.relationship(
        'CorpusFile',
        backref='corpus',
        lazy='dynamic',
        cascade='all, delete-orphan'
    )
    max_num_tokens = 2147483647
Patrick Jentsch's avatar
Patrick Jentsch committed

    def __repr__(self):
        return f'<Corpus {self.title}>'

Patrick Jentsch's avatar
Patrick Jentsch committed
    @property
    def analysis_url(self):
        return url_for('corpora.analyse_corpus', corpus_id=self.id)

        return f'{self.user.jsonpatch_path}/corpora/{self.hashid}'
        return os.path.join(self.user.path, 'corpora', str(self.id))
    @property
    def url(self):
        return url_for('corpora.corpus', corpus_id=self.id)

    @property
    def user_hashid(self):
        return self.user.hashid
Patrick Jentsch's avatar
Patrick Jentsch committed

    def build(self):
        corpus_element = ET.fromstring('<corpus>\n</corpus>')
        for corpus_file in self.files:
Patrick Jentsch's avatar
Patrick Jentsch committed
            normalized_vrt_path = os.path.join(self.path, 'cwb', f'{corpus_file.id}.norm.vrt')
            try:
                normalize_vrt_file(corpus_file.path, normalized_vrt_path)
            except:
                self.status = CorpusStatus.FAILED
                return
            element_tree = ET.parse(normalized_vrt_path)
            text_element = element_tree.getroot()
            text_element.set('address', corpus_file.address or 'NULL')
            text_element.set('author', corpus_file.author)
            text_element.set('booktitle', corpus_file.booktitle or 'NULL')
            text_element.set('chapter', corpus_file.chapter or 'NULL')
            text_element.set('editor', corpus_file.editor or 'NULL')
            text_element.set('institution', corpus_file.institution or 'NULL')
            text_element.set('journal', corpus_file.journal or 'NULL')
            text_element.set('pages', corpus_file.pages or 'NULL')
            text_element.set('publisher', corpus_file.publisher or 'NULL')
            text_element.set('publishing_year', str(corpus_file.publishing_year))  # noqa
            text_element.set('school', corpus_file.school or 'NULL')
            text_element.set('title', corpus_file.title)
            text_element.tail = '\n'
            # corpus_element.insert(1, text_element)
            corpus_element.append(text_element)
        ET.ElementTree(corpus_element).write(
            os.path.join(self.path, 'cwb', 'corpus.vrt'),
            encoding='utf-8'
        )
        self.last_edited_date = datetime.utcnow()
    def delete(self):
        shutil.rmtree(self.path, ignore_errors=True)
    def makedirs(self):
        os.mkdir(self.path)
        os.mkdir(os.path.join(self.path, 'files'))
        os.mkdir(os.path.join(self.path, 'cwb'))
        os.mkdir(os.path.join(self.path, 'cwb', 'data'))
        os.mkdir(os.path.join(self.path, 'cwb', 'registry'))

    def to_dict(self, backrefs=False, relationships=False):
        dict_corpus = {
            'id': self.hashid,
            'user_id': self.user.hashid,
            'analysis_url': self.analysis_url,
            'url': self.url,
            'creation_date': self.creation_date.isoformat() + 'Z',
            'description': self.description,
            'max_num_tokens': self.max_num_tokens,
            'num_analysis_sessions': self.num_analysis_sessions,
            'num_tokens': self.num_tokens,
            'last_edited_date': self.last_edited_date.isoformat() + 'Z',
            'title': self.title
        }
        if backrefs:
            dict_corpus['user'] = self.user.to_dict(
                backrefs=True,
                relationships=False
            )
        if relationships:
            dict_corpus['files'] = {
Patrick Jentsch's avatar
Patrick Jentsch committed
                x.hashid: x.to_dict(backrefs=False, relationships=True)
                for x in self.files
            }
        return dict_corpus
Patrick Jentsch's avatar
Patrick Jentsch committed

Patrick Jentsch's avatar
Patrick Jentsch committed

def load_user(user_id):
    return User.query.get(int(user_id))