nopaque/app/models/tesseract_ocr_pipeline_model.py

133 lines
5.1 KiB
Python
Raw Normal View History

2024-03-05 15:02:23 +00:00
from flask import current_app, url_for
from flask_hashids import HashidMixin
from tqdm import tqdm
2024-03-07 14:49:04 +00:00
from pathlib import Path
2024-03-05 15:02:23 +00:00
import requests
import yaml
from app import db
2024-11-07 07:35:02 +00:00
from app.extensions.nopaque_sqlalchemy_extras import ContainerColumn
2024-03-05 15:02:23 +00:00
from .file_mixin import FileMixin
from .user import User
class TesseractOCRPipelineModel(FileMixin, HashidMixin, db.Model):
__tablename__ = 'tesseract_ocr_pipeline_models'
# Primary key
id = db.Column(db.Integer, primary_key=True)
# Foreign keys
user_id = db.Column(db.Integer, db.ForeignKey('users.id'))
# Fields
title = db.Column(db.String(64))
description = db.Column(db.String(255))
version = db.Column(db.String(16))
compatible_service_versions = db.Column(ContainerColumn(list, 255))
publisher = db.Column(db.String(128))
publisher_url = db.Column(db.String(512))
publishing_url = db.Column(db.String(512))
publishing_year = db.Column(db.Integer)
is_public = db.Column(db.Boolean, default=False)
# Relationships
user = db.relationship('User', back_populates='tesseract_ocr_pipeline_models')
@property
2024-03-07 14:49:04 +00:00
def path(self) -> Path:
return self.user.path / 'tesseract_ocr_pipeline_models' / f'{self.id}'
2024-03-05 15:02:23 +00:00
@property
def jsonpatch_path(self):
return f'{self.user.jsonpatch_path}/tesseract_ocr_pipeline_models/{self.hashid}'
@property
def url(self):
return url_for(
2024-11-15 14:59:08 +00:00
'contributions.tesseract_ocr_pipeline_models.entity',
2024-03-05 15:02:23 +00:00
tesseract_ocr_pipeline_model_id=self.id
)
@property
def user_hashid(self):
return self.user.hashid
@staticmethod
def insert_defaults(force_download=False):
nopaque_user = User.query.filter_by(username='nopaque').first()
2024-03-07 14:49:04 +00:00
default_records_file = Path(__file__).parent / 'default_records' / 'tesseract_ocr_pipeline_model.yml'
with default_records_file.open('r') as f:
default_records = yaml.safe_load(f)
for m in default_records:
2024-03-05 15:02:23 +00:00
model = TesseractOCRPipelineModel.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.filename = f'{model.id}.traineddata'
model.publisher = m['publisher']
model.publisher_url = m['publisher_url']
model.publishing_url = m['publishing_url']
model.publishing_year = m['publishing_year']
model.is_public = True
model.title = m['title']
model.version = m['version']
else:
model = TesseractOCRPipelineModel(
compatible_service_versions=m['compatible_service_versions'],
description=m['description'],
publisher=m['publisher'],
publisher_url=m['publisher_url'],
publishing_url=m['publishing_url'],
publishing_year=m['publishing_year'],
is_public=True,
title=m['title'],
user=nopaque_user,
version=m['version']
)
db.session.add(model)
db.session.flush(objects=[model])
db.session.refresh(model)
model.filename = f'{model.id}.traineddata'
2024-03-07 14:49:04 +00:00
if not model.path.exists() or force_download:
2024-03-05 15:02:23 +00:00
r = requests.get(m['url'], stream=True)
pbar = tqdm(
desc=f'{model.title} ({model.filename})',
unit="B",
unit_scale=True,
unit_divisor=1024,
total=int(r.headers['Content-Length'])
)
pbar.clear()
with open(model.path, 'wb') as f:
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()
def delete(self):
try:
2024-03-07 14:49:04 +00:00
self.path.unlink(missing_ok=True)
2024-03-05 15:02:23 +00:00
except OSError as e:
current_app.logger.error(e)
2024-03-07 14:49:04 +00:00
raise
2024-03-05 15:02:23 +00:00
db.session.delete(self)
def to_json_serializeable(self, backrefs=False, relationships=False):
json_serializeable = {
'id': self.hashid,
'compatible_service_versions': self.compatible_service_versions,
'description': self.description,
'publisher': self.publisher,
'publisher_url': self.publisher_url,
'publishing_url': self.publishing_url,
'publishing_year': self.publishing_year,
'is_public': self.is_public,
'title': self.title,
'version': self.version,
**self.file_mixin_to_json_serializeable()
}
if backrefs:
json_serializeable['user'] = \
self.user.to_json_serializeable(backrefs=True)
if relationships:
pass
return json_serializeable