nopaque/app/api/schemas.py

from apifairy.fields import FileField
from marshmallow import validate, validates, ValidationError
from marshmallow.decorators import post_dump
from app import ma
from app.auth import USERNAME_REGEX
from app.models import Job, JobStatus, TesseractOCRModel, Token, User, UserSettingJobStatusMailNotificationLevel
from app.services import SERVICES



class EmptySchema(ma.Schema):
    pass


class TokenSchema(ma.SQLAlchemySchema):
    class Meta:
        model = Token
        ordered = True

    access_token = ma.String(required=True)
    refresh_token = ma.String()


class TesseractOCRModelSchema(ma.SQLAlchemySchema):
    class Meta:
        model = TesseractOCRModel
        ordered = True

    hashid = ma.String(data_key='id', dump_only=True)
    user_hashid = ma.String(data_key='user_id', dump_only=True)
    title = ma.auto_field(
        required=True,
        validate=validate.Length(min=1, max=64)
    )
    description = ma.auto_field(
        required=True,
        validate=validate.Length(min=1, max=255)
    )
    version = ma.String(
        required=True,
        validate=validate.Length(min=1, max=16)
    )
    compatible_service_versions = ma.List(
        ma.String(required=True, validate=validate.Length(min=1, max=16)),
        required=True,
        validate=validate.Length(min=1, max=255)
    )
    publisher = ma.String(
        required=True,
        validate=validate.Length(min=1, max=128)
    )
    publisher_url = ma.String(
        validate=[validate.URL(), validate.Length(min=1, max=512)]
    )
    publishing_url = ma.String(
        required=True,
        validate=[validate.URL(), validate.Length(min=1, max=512)]
    )
    publishing_year = ma.Int(
        required=True
    )
    shared = ma.Boolean(required=True)


class JobSchema(ma.SQLAlchemySchema):
    class Meta:
        model = Job
        ordered = True

    hashid = ma.String(data_key='id', dump_only=True)
    user_hashid = ma.String(data_key='user_id', dump_only=True)
    title = ma.auto_field(
        required=True,
        validate=validate.Length(min=1, max=32)
    )
    description = ma.auto_field(
        required=True,
        validate=validate.Length(min=1, max=255)
    )
    creation_date = ma.auto_field(dump_only=True)
    end_date = ma.auto_field(dump_only=True)
    service = ma.String(
        dump_only=True,
        validate=validate.OneOf(SERVICES.keys())
    )
    service_args = ma.Dict(dump_only=True)
    service_version = ma.String(dump_only=True)
    status = ma.String(
        dump_only=True,
        validate=validate.OneOf(list(JobStatus.__members__.keys()))
    )

    @post_dump(pass_original=True)
    def post_dump(self, serialized_job, job, **kwargs):
        serialized_job['status'] = job.status.name
        return serialized_job


class TesseractOCRPipelineJobSchema(JobSchema):
    binarization = ma.Boolean(load_only=True, missing=False)
    model_id = ma.String(required=True, load_only=True)
    service_version = ma.auto_field(
        required=True,
        validate=[validate.Length(min=1, max=16), validate.OneOf(list(SERVICES['tesseract-ocr-pipeline']['versions'].keys()))]
    )
    pdf = FileField()

    @validates('pdf')
    def validate_pdf(self, value):
        if value.mimetype != 'application/pdf':
            raise ValidationError('PDF files only!')


class SpaCyNLPPipelineJobSchema(JobSchema):
    binarization = ma.Boolean(load_only=True, missing=False)
    model_id = ma.String(required=True, load_only=True)
    service_version = ma.auto_field(
        required=True,
        validate=[validate.Length(min=1, max=16), validate.OneOf(list(SERVICES['tesseract-ocr-pipeline']['versions'].keys()))]
    )
    txt = FileField(required=True)

    @validates('txt')
    def validate_txt(self, value):
        if value.mimetype != 'text/plain':
            raise ValidationError('Plain text files only!')


class UserSchema(ma.SQLAlchemySchema):
    class Meta:
        model = User
        ordered = True

    hashid = ma.String(data_key='id', dump_only=True)
    username = ma.auto_field(
        validate=[
            validate.Length(min=1, max=64),
            validate.Regexp(USERNAME_REGEX, error='Usernames must have only letters, numbers, dots or underscores')
        ]
    )
    email = ma.auto_field(validate=validate.Email())
    member_since = ma.auto_field(dump_only=True)
    last_seen = ma.auto_field(dump_only=True)
    password = ma.String(load_only=True)
    last_seen = ma.auto_field(dump_only=True)
    setting_dark_mode = ma.auto_field()
    setting_job_status_mail_notification_level = ma.String(
        validate=validate.OneOf(list(UserSettingJobStatusMailNotificationLevel.__members__.keys()))
    )

    @validates('email')
    def validate_email(self, email):
        if User.query.filter(User.email == email).first():
            raise ValidationError('Email already registered')

    @validates('username')
    def validate_username(self, username):
        if User.query.filter(User.username == username).first():
            raise ValidationError('Username already in use')

    @post_dump(pass_original=True)
    def post_dump(self, serialized_user, user, **kwargs):
        serialized_user['setting_job_status_mail_notification_level'] = \
            user.setting_job_status_mail_notification_level.name
        return serialized_user