Contributions update revised

This commit is contained in:
Patrick Jentsch
2022-11-15 15:11:16 +01:00
parent df5ae19e68
commit cd72614c0f
18 changed files with 481 additions and 444 deletions

View File

@ -12,6 +12,7 @@ from wtforms import (
from wtforms.validators import InputRequired, Length
from app.services import SERVICES
class CreateContributionBaseForm(FlaskForm):
title = StringField(
'Title',
@ -46,31 +47,8 @@ class CreateContributionBaseForm(FlaskForm):
)
submit = SubmitField()
class EditForm(CreateContributionBaseForm):
def prefill(self, model_file):
''' Pre-fill the form with data of an exististing corpus file '''
self.title.data = model_file.title
self.description.data = model_file.description
self.publisher.data = model_file.publisher
self.publishing_year.data = model_file.publishing_year
self.publisher_url.data = model_file.publisher_url
self.publishing_url.data = model_file.publishing_url
self.version.data = model_file.version
self.shared.data = model_file.shared
class EditTesseractOCRModelForm(EditForm):
pass
class EditSpaCyNLPPipelineModelForm(EditForm):
pipeline_name = StringField(
'Pipeline name',
validators=[InputRequired(), Length(max=64)]
)
def prefill(self, model_file):
super().prefill(model_file)
self.pipeline_name.data = model_file.pipeline_name
class TesseractOCRModelContributionForm(CreateContributionBaseForm):
class CreateTesseractOCRPipelineModelForm(CreateContributionBaseForm):
tesseract_model_file = FileField(
'File',
validators=[FileRequired()]
@ -78,6 +56,7 @@ class TesseractOCRModelContributionForm(CreateContributionBaseForm):
compatible_service_versions = SelectMultipleField(
'Compatible service versions'
)
def validate_tesseract_model_file(self, field):
current_app.logger.warning(field.data.filename)
if not field.data.filename.lower().endswith('.traineddata'):
@ -92,7 +71,8 @@ class TesseractOCRModelContributionForm(CreateContributionBaseForm):
]
self.compatible_service_versions.default = ''
class SpacyNLPModelContributionForm(CreateContributionBaseForm):
class CreateSpaCyNLPPipelineModelForm(CreateContributionBaseForm):
spacy_model_file = FileField(
'File',
validators=[FileRequired()]
@ -104,16 +84,45 @@ class SpacyNLPModelContributionForm(CreateContributionBaseForm):
'Pipeline name',
validators=[InputRequired(), Length(max=64)]
)
def validate_spacy_model_file(self, field):
current_app.logger.warning(field.data.filename)
if not field.data.filename.lower().endswith('.tar.gz'):
raise ValidationError('.tar.gz files only!')
def __init__(self, *args, **kwargs):
service_manifest = SERVICES['spacy-nlp-pipeline']
super().__init__(*args, **kwargs)
service_manifest = SERVICES['spacy-nlp-pipeline']
self.compatible_service_versions.choices = [('', 'Choose your option')]
self.compatible_service_versions.choices += [
(x, x) for x in service_manifest['versions'].keys()
]
self.compatible_service_versions.default = ''
class EditContributionBaseForm(CreateContributionBaseForm):
def prefill(self, model_file):
''' Pre-fill the form with data of an exististing corpus file '''
self.title.data = model_file.title
self.description.data = model_file.description
self.publisher.data = model_file.publisher
self.publishing_year.data = model_file.publishing_year
self.publisher_url.data = model_file.publisher_url
self.publishing_url.data = model_file.publishing_url
self.version.data = model_file.version
self.shared.data = model_file.shared
class EditTesseractOCRPipelineModelForm(EditContributionBaseForm):
pass
class EditSpaCyNLPPipelineModelForm(EditContributionBaseForm):
pipeline_name = StringField(
'Pipeline name',
validators=[InputRequired(), Length(max=64)]
)
def prefill(self, model_file):
super().prefill(model_file)
self.pipeline_name.data = model_file.pipeline_name

View File

@ -2,10 +2,19 @@ from flask import abort, current_app, flash, Markup, render_template, url_for
from flask_login import login_required, current_user
from threading import Thread
from app import db
from app.decorators import admin_required, permission_required
from app.models import Permission, SpaCyNLPPipelineModel, TesseractOCRPipelineModel
from app.decorators import permission_required
from app.models import (
Permission,
SpaCyNLPPipelineModel,
TesseractOCRPipelineModel
)
from . import bp
from .forms import TesseractOCRModelContributionForm, EditSpaCyNLPPipelineModelForm, EditTesseractOCRModelForm, SpacyNLPModelContributionForm
from .forms import (
CreateSpaCyNLPPipelineModelForm,
CreateTesseractOCRPipelineModelForm,
EditSpaCyNLPPipelineModelForm,
EditTesseractOCRPipelineModelForm
)
@bp.before_request
@ -16,30 +25,17 @@ def before_request():
@bp.route('/')
@login_required
@admin_required
def contributions():
tesseract_ocr_user_models = [
x for x in current_user.tesseract_ocr_pipeline_models
]
spacy_nlp_user_models = [
x for x in current_user.spacy_nlp_pipeline_models
]
return render_template(
'contributions/contribution_overview.html.j2',
tesseract_ocr_user_models=tesseract_ocr_user_models,
spacy_nlp_user_models=spacy_nlp_user_models,
userId = current_user.hashid,
title='Contribution Overview'
'contributions/contributions.html.j2',
title='Contributions'
)
@bp.route('/edit-tesseract-model/<hashid:tesseract_ocr_pipeline_model_id>', methods=['GET', 'POST'])
@login_required
@bp.route('/tesseract_ocr_pipeline_models/<hashid:tesseract_ocr_pipeline_model_id>', methods=['GET', 'POST'])
def tesseract_ocr_pipeline_model(tesseract_ocr_pipeline_model_id):
tesseract_ocr_pipeline_model = TesseractOCRPipelineModel.query.get_or_404(
tesseract_ocr_pipeline_model_id
)
form = EditTesseractOCRModelForm(prefix='tesseract-ocr-model-edit-form')
tesseract_ocr_pipeline_model = TesseractOCRPipelineModel.query.get_or_404(tesseract_ocr_pipeline_model_id)
form = EditTesseractOCRPipelineModelForm(prefix='edit-tesseract-ocr-pipeline-model-form')
if form.validate_on_submit():
if tesseract_ocr_pipeline_model.title != form.title.data:
tesseract_ocr_pipeline_model.title = form.title.data
@ -58,47 +54,50 @@ def tesseract_ocr_pipeline_model(tesseract_ocr_pipeline_model_id):
if tesseract_ocr_pipeline_model.shared != form.shared.data:
tesseract_ocr_pipeline_model.shared = form.shared.data
db.session.commit()
message = Markup(f'Model "<a href="contribute/{tesseract_ocr_pipeline_model.hashid}">{tesseract_ocr_pipeline_model.title}</a>" updated')
flash(message, category='corpus')
return {}, 201, {'Location': url_for('contributions.contributions')}
tesseract_ocr_pipeline_model_url = url_for(
'.tesseract_ocr_pipeline_model',
tesseract_ocr_pipeline_model_id=tesseract_ocr_pipeline_model.id
)
message = Markup(f'Tesseract OCR Pipeline model "<a href="{tesseract_ocr_pipeline_model_url}">{tesseract_ocr_pipeline_model.title}</a>" updated')
flash(message)
return {}, 201, {'Location': tesseract_ocr_pipeline_model_url}
form.prefill(tesseract_ocr_pipeline_model)
return render_template(
'contributions/tesseract_ocr_pipeline_model.html.j2',
tesseract_ocr_pipeline_model=tesseract_ocr_pipeline_model,
form=form,
title='Edit your Tesseract OCR model'
tesseract_ocr_pipeline_model=tesseract_ocr_pipeline_model,
title='Edit Tesseract OCR Pipeline Model'
)
@bp.route('/edit-tesseract-model/<hashid:tesseract_ocr_pipeline_model_id>', methods=['DELETE'])
@login_required
@bp.route('/tesseract_ocr_pipeline_models/<hashid:tesseract_ocr_pipeline_model_id>', methods=['DELETE'])
def delete_tesseract_model(tesseract_ocr_pipeline_model_id):
def _delete_tesseract_model(app, tesseract_ocr_pipeline_model_id):
def _delete_tesseract_ocr_pipeline_model(app, tesseract_ocr_pipeline_model_id):
with app.app_context():
model = TesseractOCRPipelineModel.query.get(tesseract_ocr_pipeline_model_id)
model.delete()
tesseract_ocr_pipeline_model = TesseractOCRPipelineModel.query.get(tesseract_ocr_pipeline_model_id)
tesseract_ocr_pipeline_model.delete()
db.session.commit()
model = TesseractOCRPipelineModel.query.get_or_404(tesseract_ocr_pipeline_model_id)
if not (model.user == current_user or current_user.is_administrator()):
tesseract_ocr_pipeline_model = TesseractOCRPipelineModel.query.get_or_404(tesseract_ocr_pipeline_model_id)
if not (tesseract_ocr_pipeline_model.user == current_user or current_user.is_administrator()):
abort(403)
thread = Thread(
target=_delete_tesseract_model,
target=_delete_tesseract_ocr_pipeline_model,
args=(current_app._get_current_object(), tesseract_ocr_pipeline_model_id)
)
thread.start()
return {}, 202
@bp.route('/add-tesseract-ocr-pipeline-model', methods=['GET', 'POST'])
def add_tesseract_ocr_pipeline_model():
form = TesseractOCRModelContributionForm(
prefix='contribute-tesseract-ocr-pipeline-model-form'
)
@bp.route('/tesseract_ocr_pipeline_models/create', methods=['GET', 'POST'])
def create_tesseract_ocr_pipeline_model():
form = CreateTesseractOCRPipelineModelForm(prefix='create-tesseract-ocr-pipeline-model-form')
if form.is_submitted():
if not form.validate():
response = {'errors': form.errors}
return response, 400
try:
tesseract_ocr_model = TesseractOCRPipelineModel.create(
tesseract_ocr_pipeline_model = TesseractOCRPipelineModel.create(
form.tesseract_model_file.data,
compatible_service_versions=form.compatible_service_versions.data,
description=form.description.data,
@ -114,27 +113,24 @@ def add_tesseract_ocr_pipeline_model():
except OSError:
abort(500)
db.session.commit()
message = Markup(f'Model "{tesseract_ocr_model.title}" created')
tesseract_ocr_pipeline_model_url = url_for(
'.tesseract_ocr_pipeline_model',
tesseract_ocr_pipeline_model_id=tesseract_ocr_pipeline_model.id
)
message = Markup(f'Tesseract OCR Pipeline model "<a href="{tesseract_ocr_pipeline_model_url}">{tesseract_ocr_pipeline_model.title}</a>" created')
flash(message)
return {}, 201, {'Location': url_for('contributions.contributions')}
tesseract_ocr_pipeline_models = [
x for x in TesseractOCRPipelineModel.query.all()
]
return {}, 201, {'Location': tesseract_ocr_pipeline_model_url}
return render_template(
'contributions/contribute_tesseract_ocr_models.html.j2',
'contributions/create_tesseract_ocr_pipeline_model.html.j2',
form=form,
tesseract_ocr_pipeline_models=tesseract_ocr_pipeline_models,
title='Tesseract OCR Model Contribution'
title='Create Tesseract OCR Pipeline Model'
)
@bp.route('/edit-spacy-model//<hashid:spacy_nlp_pipeline_model_id>', methods=['GET', 'POST'])
@login_required
@bp.route('/spacy-nlp-pipeline-models/<hashid:spacy_nlp_pipeline_model_id>', methods=['GET', 'POST'])
def spacy_nlp_pipeline_model(spacy_nlp_pipeline_model_id):
spacy_nlp_pipeline_model = SpaCyNLPPipelineModel.query.get_or_404(
spacy_nlp_pipeline_model_id
)
form = EditSpaCyNLPPipelineModelForm(prefix='spacy-nlp-model-edit-form')
spacy_nlp_pipeline_model = SpaCyNLPPipelineModel.query.get_or_404(spacy_nlp_pipeline_model_id)
form = EditSpaCyNLPPipelineModelForm(prefix='edit-spacy-nlp-pipeline-model-form')
if form.validate_on_submit():
if spacy_nlp_pipeline_model.title != form.title.data:
spacy_nlp_pipeline_model.title = form.title.data
@ -154,30 +150,33 @@ def spacy_nlp_pipeline_model(spacy_nlp_pipeline_model_id):
spacy_nlp_pipeline_model.version = form.version.data
if spacy_nlp_pipeline_model.shared != form.shared.data:
spacy_nlp_pipeline_model.shared = form.shared.data
current_app.logger.warning(db.session.dirty)
db.session.commit()
message = Markup(f'Model "<a href="contribute/{spacy_nlp_pipeline_model.hashid}">{spacy_nlp_pipeline_model.title}</a>" updated')
flash(message, category='corpus')
return {}, 201, {'Location': url_for('contributions.contributions')}
print(spacy_nlp_pipeline_model.to_json())
spacy_nlp_pipeline_model_url = url_for(
'.spacy_nlp_pipeline_model',
spacy_nlp_pipeline_model_id=spacy_nlp_pipeline_model.id
)
message = Markup(f'SpaCy NLP Pipeline model "<a href="{spacy_nlp_pipeline_model_url}">{spacy_nlp_pipeline_model.title}</a>" updated')
flash(message)
return {}, 201, {'Location': url_for('.contributions')}
form.prefill(spacy_nlp_pipeline_model)
return render_template(
'contributions/spacy_nlp_pipeline_model.html.j2',
spacy_nlp_pipeline_model=spacy_nlp_pipeline_model,
form=form,
title='Edit your spaCy NLP model'
spacy_nlp_pipeline_model=spacy_nlp_pipeline_model,
title=f'{spacy_nlp_pipeline_model.title} [{spacy_nlp_pipeline_model.version}]'
)
@bp.route('/edit-spacy-model/<hashid:spacy_nlp_pipeline_model_id>', methods=['DELETE'])
@login_required
@bp.route('/spacy-nlp-pipeline-models/<hashid:spacy_nlp_pipeline_model_id>', methods=['DELETE'])
def delete_spacy_model(spacy_nlp_pipeline_model_id):
def _delete_spacy_model(app, spacy_nlp_pipeline_model_id):
with app.app_context():
model = SpaCyNLPPipelineModel.query.get(spacy_nlp_pipeline_model_id)
model.delete()
spacy_nlp_pipeline_model = SpaCyNLPPipelineModel.query.get(spacy_nlp_pipeline_model_id)
spacy_nlp_pipeline_model.delete()
db.session.commit()
model = SpaCyNLPPipelineModel.query.get_or_404(spacy_nlp_pipeline_model_id)
if not (model.user == current_user or current_user.is_administrator()):
spacy_nlp_pipeline_model = SpaCyNLPPipelineModel.query.get_or_404(spacy_nlp_pipeline_model_id)
if not (spacy_nlp_pipeline_model.user == current_user or current_user.is_administrator()):
abort(403)
thread = Thread(
target=_delete_spacy_model,
@ -186,15 +185,16 @@ def delete_spacy_model(spacy_nlp_pipeline_model_id):
thread.start()
return {}, 202
@bp.route('/add-spacy-nlp-pipeline-model', methods=['GET', 'POST'])
def add_spacy_nlp_pipeline_model():
form = SpacyNLPModelContributionForm(prefix='contribute-spacy-nlp-pipeline-model-form')
@bp.route('/spacy-nlp-pipeline-models/create', methods=['GET', 'POST'])
def create_spacy_nlp_pipeline_model():
form = CreateSpaCyNLPPipelineModelForm(prefix='create-spacy-nlp-pipeline-model-form')
if form.is_submitted():
if not form.validate():
response = {'errors': form.errors}
return response, 400
try:
spacy_nlp_model = SpaCyNLPPipelineModel.create(
spacy_nlp_pipeline_model = SpaCyNLPPipelineModel.create(
form.spacy_model_file.data,
compatible_service_versions=form.compatible_service_versions.data,
description=form.description.data,
@ -211,15 +211,15 @@ def add_spacy_nlp_pipeline_model():
except OSError:
abort(500)
db.session.commit()
message = Markup(f'Model "{spacy_nlp_model.title}" created')
spacy_nlp_pipeline_model_url = url_for(
'.spacy_nlp_pipeline_model',
spacy_nlp_pipeline_model_id=spacy_nlp_pipeline_model.id
)
message = Markup(f'SpaCy NLP Pipeline model "<a href="{spacy_nlp_pipeline_model_url}">{spacy_nlp_pipeline_model.title}</a>" created')
flash(message)
return {}, 201, {'Location': url_for('contributions.contributions')}
spacy_nlp_pipeline_models = [
x for x in SpaCyNLPPipelineModel.query.all()
]
return {}, 201, {'Location': spacy_nlp_pipeline_model_url}
return render_template(
'contributions/contribute_spacy_nlp_models.html.j2',
'contributions/create_spacy_nlp_pipeline_model.html.j2',
form=form,
spacy_nlp_pipeline_models=spacy_nlp_pipeline_models,
title='spaCy NLP Model Contribution'
title='Create SpaCy NLP Pipeline Model'
)