Contribution Package Spacy NLP

This commit is contained in:
Inga Kirschnick
2022-11-07 09:15:38 +01:00
parent 46ba14b923
commit 05340ea7ff
13 changed files with 564 additions and 57 deletions

View File

@ -46,6 +46,18 @@ 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 TesseractOCRModelContributionForm(CreateContributionBaseForm):
tesseract_model_file = FileField(
'File',
@ -67,16 +79,23 @@ class TesseractOCRModelContributionForm(CreateContributionBaseForm):
]
self.compatible_service_versions.default = ''
class TesseractOCRModelEditForm(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 SpacyNLPModelContributionForm(CreateContributionBaseForm):
spacy_model_file = FileField(
'File',
validators=[FileRequired()]
)
compatible_service_versions = SelectMultipleField(
'Compatible service versions'
)
def validate_spacy(self, field):
if field.data.mimetype != '.tar.gz':
raise ValidationError('.tar.gz files only!')
def __init__(self, *args, **kwargs):
service_manifest = SERVICES['spacy-nlp-pipeline']
super().__init__(*args, **kwargs)
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 = ''

View File

@ -1,11 +1,11 @@
from flask import abort, current_app, flash, Markup, redirect, render_template, url_for
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 TesseractOCRPipelineModel, Permission
from app.models import Permission, SpaCyNLPPipelineModel, TesseractOCRPipelineModel
from . import bp
from .forms import TesseractOCRModelContributionForm, TesseractOCRModelEditForm
from .forms import TesseractOCRModelContributionForm, EditForm, SpacyNLPModelContributionForm
@bp.before_request
@ -22,20 +22,26 @@ 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
]
spacy_models = SpaCyNLPPipelineModel.query.all()
print(spacy_models)
return render_template(
'contributions/contribution_overview.html.j2',
tesseractOCRUserModels=tesseract_ocr_user_models,
tesseract_ocr_user_models=tesseract_ocr_user_models,
spacy_nlp_user_models=spacy_nlp_user_models,
userId = current_user.hashid,
title='Contribution Overview'
)
@bp.route('/<hashid:tesseract_ocr_pipeline_model_id>', methods=['GET', 'POST'])
@bp.route('/edit-tesseract-model/<hashid:tesseract_ocr_pipeline_model_id>', methods=['GET', 'POST'])
@login_required
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 = TesseractOCRModelEditForm(prefix='tesseract-ocr-model-edit-form')
form = EditForm(prefix='tesseract-ocr-model-edit-form')
if form.validate_on_submit():
if tesseract_ocr_pipeline_model.title != form.title.data:
tesseract_ocr_pipeline_model.title = form.title.data
@ -65,7 +71,7 @@ def tesseract_ocr_pipeline_model(tesseract_ocr_pipeline_model_id):
title='Edit your Tesseract OCR model'
)
@bp.route('/<hashid:tesseract_ocr_pipeline_model_id>', methods=['DELETE'])
@bp.route('/edit-tesseract-model/<hashid:tesseract_ocr_pipeline_model_id>', methods=['DELETE'])
@login_required
def delete_tesseract_model(tesseract_ocr_pipeline_model_id):
def _delete_tesseract_model(app, tesseract_ocr_pipeline_model_id):
@ -123,3 +129,95 @@ def add_tesseract_ocr_pipeline_model():
tesseract_ocr_pipeline_models=tesseract_ocr_pipeline_models,
title='Tesseract OCR Model Contribution'
)
@bp.route('/edit-spacy-model//<hashid:spacy_nlp_pipeline_model_id>', methods=['GET', 'POST'])
@login_required
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 = EditForm(prefix='spacy-nlp-model-edit-form')
if form.validate_on_submit():
if spacy_nlp_pipeline_model.title != form.title.data:
spacy_nlp_pipeline_model.title = form.title.data
if spacy_nlp_pipeline_model.description != form.description.data:
spacy_nlp_pipeline_model.description = form.description.data
if spacy_nlp_pipeline_model.publisher != form.publisher.data:
spacy_nlp_pipeline_model.publisher = form.publisher.data
if spacy_nlp_pipeline_model.publishing_year != form.publishing_year.data:
spacy_nlp_pipeline_model.publishing_year = form.publishing_year.data
if spacy_nlp_pipeline_model.publisher_url != form.publisher_url.data:
spacy_nlp_pipeline_model.publisher_url = form.publisher_url.data
if spacy_nlp_pipeline_model.publishing_url != form.publishing_url.data:
spacy_nlp_pipeline_model.publishing_url = form.publishing_url.data
if spacy_nlp_pipeline_model.version != form.version.data:
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
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')}
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'
)
@bp.route('/edit-spacy-model/<hashid:spacy_nlp_pipeline_model_id>', methods=['DELETE'])
@login_required
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()
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()):
abort(403)
thread = Thread(
target=_delete_spacy_model,
args=(current_app._get_current_object(), 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')
if form.is_submitted():
if not form.validate():
response = {'errors': form.errors}
return response, 400
try:
spacy_nlp_model = SpaCyNLPPipelineModel.create(
form.spacy_model_file.data,
compatible_service_versions=form.compatible_service_versions.data,
description=form.description.data,
publisher=form.publisher.data,
publisher_url=form.publisher_url.data,
publishing_url=form.publishing_url.data,
publishing_year=form.publishing_year.data,
shared=form.shared.data,
title=form.title.data,
version=form.version.data,
user=current_user
)
except OSError:
abort(500)
db.session.commit()
message = Markup(f'Model "{spacy_nlp_model.title}" created')
flash(message)
return {}, 201, {'Location': url_for('contributions.contributions')}
spacy_nlp_pipeline_models = [
x for x in SpaCyNLPPipelineModel.query.all()
]
return render_template(
'contributions/contribute_spacy_nlp_models.html.j2',
form=form,
spacy_nlp_pipeline_models=spacy_nlp_pipeline_models,
title='spaCy NLP Model Contribution'
)