Move corpora views, and forms in package.

This commit is contained in:
Patrick Jentsch
2019-10-31 10:25:48 +01:00
parent 8fb52fcd31
commit 3135eb0897
8 changed files with 179 additions and 194 deletions

6
app/corpora/__init__.py Normal file
View File

@@ -0,0 +1,6 @@
from flask import Blueprint
corpora = Blueprint('corpora', __name__)
from . import views

65
app/corpora/forms.py Normal file
View File

@@ -0,0 +1,65 @@
from flask_wtf import FlaskForm
from wtforms import (FileField, StringField, SubmitField,
ValidationError, IntegerField, SelectField, TextAreaField)
from wtforms.validators import DataRequired, Length
class AddCorpusFileForm(FlaskForm):
author = StringField('Author', validators=[DataRequired(), Length(1, 64)])
file = FileField('File', validators=[DataRequired()])
publishing_year = IntegerField('Publishing year',
validators=[DataRequired()])
submit = SubmitField()
title = StringField('Title', validators=[DataRequired(), Length(1, 64)])
def validate_file(form, field):
if not field.data.filename.lower().endswith('.vrt'):
raise ValidationError('File does not have an approved extension: '
'.vrt')
class AddCorpusForm(FlaskForm):
description = StringField('Description',
validators=[DataRequired(), Length(1, 64)])
submit = SubmitField()
title = StringField('Title', validators=[DataRequired(), Length(1, 32)])
class EditCorpusFileForm(FlaskForm):
author = StringField('Author', validators=[DataRequired(), Length(1, 64)])
corpus_file_id = IntegerField('', validators=[DataRequired()])
publishing_year = IntegerField('Publishing year',
validators=[DataRequired()])
submit = SubmitField()
title = StringField('Title', validators=[DataRequired(), Length(1, 64)])
class QueryForm(FlaskForm):
query = TextAreaField('CQP Query', validators=[DataRequired(),
(Length(1, 1024))])
hits_per_page = SelectField('Hits per page',
choices=[('', 'Nr. of hits per page'),
('10', '10'),
('20', '20'),
('30', '30'),
('40', '40'),
('50', '50')],
validators=[DataRequired()])
context = SelectField('Context',
choices=[('', 'Words of context around hit'),
('5', '5'),
('10', '10'),
('15', '15'),
('20', '20'),
('25', '25')],
validators=[DataRequired()])
submit = SubmitField('Start Query')
class QueryDownloadForm(FlaskForm):
file_type = SelectField('File type',
choices=[('', 'Choose file type'),
('csv', 'csv'),
('json', 'json'),
('excel', 'excel')],
validators=[DataRequired()])

150
app/corpora/views.py Normal file
View File

@@ -0,0 +1,150 @@
from flask import (abort, current_app, flash, redirect, request,
render_template, url_for, send_from_directory)
from flask_login import current_user, login_required
from . import corpora
from .forms import (AddCorpusFileForm, AddCorpusForm, QueryDownloadForm,
QueryForm)
from .. import db
from ..models import Corpus, CorpusFile
from werkzeug.utils import secure_filename
import os
import threading
import logging
@corpora.route('/add', methods=['POST'])
@login_required
def add_corpus():
add_corpus_form = AddCorpusForm()
if not add_corpus_form.validate_on_submit():
abort(400)
corpus = Corpus(creator=current_user,
description=add_corpus_form.description.data,
title=add_corpus_form.title.data)
db.session.add(corpus)
db.session.commit()
dir = os.path.join(current_app.config['OPAQUE_STORAGE_DIRECTORY'],
str(corpus.user_id), 'corpora', str(corpus.id))
try:
os.makedirs(dir)
except OSError:
flash('OSError!')
db.session.remove(corpus)
db.session.commit()
flash('Corpus created!')
return redirect(url_for('corpora.corpus', corpus_id=corpus.id))
@corpora.route('/<int:corpus_id>')
@login_required
def corpus(corpus_id):
corpus = Corpus.query.get_or_404(corpus_id)
if not (corpus.creator == current_user or current_user.is_administrator()):
abort(403)
return render_template('corpora/corpus.html.j2',
add_corpus_file_form=AddCorpusFileForm(),
corpus=corpus, title='Corpus')
@corpora.route('/<int:corpus_id>/analysis', methods=['GET', 'POST'])
@login_required
def corpus_analysis(corpus_id):
logger = logging.getLogger(__name__)
corpus = Corpus.query.get_or_404(corpus_id)
query = request.args.get('query')
logger.warning('Query first: {}'.format(query))
hits_per_page = request.args.get('hits_per_page', 30)
context = request.args.get('context', 10)
dl_form = QueryDownloadForm()
form = QueryForm(hits_per_page=hits_per_page, context=context, query=query)
if form.validate_on_submit():
logger = logging.getLogger(__name__)
logger.warning('Data has been sent!')
logger.warning('Data labels: {data}'.format(data=[data for data in form.data]))
logger.warning('Query Second: {q}'.format(q=form.query.data))
logger.warning('Hits: {hits}'.format(hits=form.hits_per_page.data))
logger.warning('Context: {context}'.format(context=form.context.data))
flash('Query has been sent!')
query = form.query.data
hits_per_page = form.hits_per_page.data
context = form.context.data
logger.warning('Query Thrid: {sq}'.format(sq=query))
return redirect(url_for('corpora.corpus_analysis', corpus_id=corpus_id,
query=query, hits_per_page=hits_per_page,
context=context))
return render_template('corpora/corpus_analysis.html.j2', corpus=corpus,
form=form, dl_form=dl_form,
title='Corpus: {}'.format(corpus.title))
@corpora.route('/<int:corpus_id>/delete')
@login_required
def delete_corpus(corpus_id):
corpus = Corpus.query.get_or_404(corpus_id)
if not (corpus.creator == current_user or current_user.is_administrator()):
abort(403)
delete_thread = threading.Thread(corpus.delete())
delete_thread.start()
flash('Corpus deleted!')
return redirect(url_for('main.dashboard'))
@corpora.route('/<int:corpus_id>/files/add', methods=['POST'])
@login_required
def add_corpus_file(corpus_id):
corpus = Corpus.query.get_or_404(corpus_id)
if not (corpus.creator == current_user or current_user.is_administrator()):
abort(403)
add_corpus_file_form = AddCorpusFileForm()
if not add_corpus_file_form.validate_on_submit():
abort(400)
file = add_corpus_file_form.file.data
filename = secure_filename(file.filename)
for corpus_file in corpus.files:
if filename == corpus_file.filename:
flash('File already registered to this corpus.')
return redirect(url_for('corpora.corpus', corpus_id=corpus_id))
# Save the file
dir = os.path.join(str(corpus.user_id), 'corpora', str(corpus.id))
file.save(os.path.join(current_app.config['OPAQUE_STORAGE_DIRECTORY'], dir,
filename))
author = add_corpus_file_form.author.data
publishing_year = add_corpus_file_form.publishing_year.data
title = add_corpus_file_form.title.data
corpus_file = CorpusFile(author=author, corpus=corpus, dir=dir,
filename=filename,
publishing_year=publishing_year, title=title)
db.session.add(corpus_file)
db.session.commit()
flash('Corpus file added!')
return redirect(url_for('corpora.corpus', corpus_id=corpus_id))
@corpora.route('/<int:corpus_id>/files/<int:corpus_file_id>/delete')
@login_required
def delete_corpus_file(corpus_id, corpus_file_id):
corpus_file = CorpusFile.query.get_or_404(corpus_file_id)
if not corpus_file.corpus_id == corpus_id:
abort(404)
if not (corpus_file.corpus.creator == current_user
or current_user.is_administrator()):
abort(403)
delete_thread = threading.Thread(corpus_file.delete())
delete_thread.start()
flash('Corpus file deleted!')
return redirect(url_for('corpora.corpus', corpus_id=corpus_id))
@corpora.route('/<int:corpus_id>/files/<int:corpus_file_id>/download')
@login_required
def download_corpus_file(corpus_id, corpus_file_id):
corpus_file = CorpusFile.query.get_or_404(corpus_file_id)
if not corpus_file.corpus_id == corpus_id:
abort(404)
if not (corpus_file.corpus.creator == current_user
or current_user.is_administrator()):
abort(403)
dir = os.path.join(current_app.config['OPAQUE_STORAGE_DIRECTORY'],
corpus_file.dir)
return send_from_directory(as_attachment=True, directory=dir,
filename=corpus_file.filename)