from app.utils import background_prepare_corpus_file 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, EditCorpusFileForm, 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=['GET', 'POST']) @login_required def add_corpus(): add_corpus_form = AddCorpusForm() if add_corpus_form.validate_on_submit(): corpus = Corpus(creator=current_user, description=add_corpus_form.description.data, status='unprepared', 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 added!') return redirect(url_for('corpora.corpus', corpus_id=corpus.id)) return render_template('corpora/add_corpus.html.j2', add_corpus_form=add_corpus_form, title='Add corpus') @corpora.route('/') @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', corpus=corpus, title='Corpus') @corpora.route('//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_id=corpus_id, form=form, dl_form=dl_form, title='Corpus: {}'.format(corpus.title)) @corpora.route('//prepare') @login_required def prepare_corpus(corpus_id): corpus = Corpus.query.get_or_404(corpus_id) if not (corpus.creator == current_user or current_user.is_administrator()): abort(403) if len(corpus.files.all()) > 0: corpus.status = 'preparable' db.session.commit() flash('Corpus marked for preparation!') else: flash('Can not prepare corpus, please add corpus file(s).') return redirect(url_for('corpora.corpus', corpus_id=corpus_id)) @corpora.route('//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('//files/add', methods=['GET', '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 add_corpus_file_form.validate_on_submit(): 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)) corpus_file = CorpusFile(author=add_corpus_file_form.author.data, corpus=corpus, dir=dir, filename=filename, publishing_year=add_corpus_file_form.publishing_year.data, title=add_corpus_file_form.title.data) db.session.add(corpus_file) db.session.commit() background_thread = threading.Thread( args=(current_app._get_current_object(), corpus_file.id), target=background_prepare_corpus_file ) background_thread.start() flash('Corpus file added!') return redirect(url_for('corpora.corpus', corpus_id=corpus_id)) return render_template('corpora/add_corpus_file.html.j2', add_corpus_file_form=add_corpus_file_form, corpus=corpus, title='Add corpus file') @corpora.route('//files//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('//files//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) @corpora.route('//files//edit', methods=['GET', 'POST']) @login_required def edit_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) edit_corpus_file_form = EditCorpusFileForm() if edit_corpus_file_form.validate_on_submit(): corpus_file.author = edit_corpus_file_form.author.data corpus_file.publishing_year = edit_corpus_file_form.publishing_year.data corpus_file.title = edit_corpus_file_form.title.data corpus_file.insert_metadata() db.session.commit() flash('Corpus file edited!') return redirect(url_for('corpora.corpus', corpus_id=corpus_id)) return render_template('corpora/edit_corpus_file.html.j2', edit_corpus_file_form=edit_corpus_file_form, corpus_file=corpus_file, title='Edit corpus file')