nopaque/app/services/views.py

from datetime import datetime
from flask import current_app, flash, redirect, render_template, url_for
from . import services
from flask_login import current_user, login_required
from .forms import NewOCRJobForm, NewNLPJobForm
from ..models import Job
from ..import swarm
from .. import db
from threading import Thread
import os
import json


@services.route('/ocr', methods=['GET', 'POST'])
@login_required
def ocr():
    new_ocr_job_form = NewOCRJobForm()
    if new_ocr_job_form.validate_on_submit():
        app = current_app._get_current_object()
        ocr_job = Job(creator=current_user._get_current_object(),
                      description=new_ocr_job_form.description.data,
                      service="ocr",
                      ressources=json.dumps({"n_cores": 2,
                                             "mem_mb": 4096}),
                      service_args=json.dumps({"args": ["--keep-intermediates",
                                                        "--skip-binarisation"],
                                               "lang": new_ocr_job_form.language.data,
                                               "version": new_ocr_job_form.version.data}),
                      status="queued",
                      title=new_ocr_job_form.title.data)
        db.session.add(ocr_job)
        db.session.commit()

        dir = os.path.join(app.config['OPAQUE_STORAGE'],
                           str(ocr_job.user_id),
                           'jobs',
                           str(ocr_job.id))

        try:
            os.makedirs(dir)
        except OSError:
            flash('OSError!')
        else:
            for file in new_ocr_job_form.files.data:
                file.save(os.path.join(dir, file.filename))
            '''
            ' TODO: Let the scheduler run this job in the background.
            '
            ' NOTE: Using self created threads is just for testing purpose as
            '       there is no scheduler available.
            '''
            thread = Thread(target=swarm.run, args=(ocr_job,))
            thread.start()
            flash('Job created!')
        return redirect(url_for('services.ocr'))

    return render_template(
        'services/ocr.html.j2',
        title='Optical Character Recognition',
        new_ocr_job_form=new_ocr_job_form
    )


@services.route('/nlp', methods=['GET', 'POST'])
@login_required
def nlp():
    new_nlp_job_form = NewNLPJobForm()
    if new_nlp_job_form.validate_on_submit():
        app = current_app._get_current_object()
        nlp_job = Job(creator=current_user._get_current_object(),
                      description=new_nlp_job_form.description.data,
                      service="nlp",
                      ressources=json.dumps({"n_cores": 1,
                                             "mem_mb": 2048}),
                      service_args=json.dumps({"args": [],
                                               "lang": new_nlp_job_form.language.data,
                                               "version": new_nlp_job_form.version.data}),
                      status="queued",
                      title=new_nlp_job_form.title.data)
        db.session.add(nlp_job)
        db.session.commit()

        dir = os.path.join(app.config['OPAQUE_STORAGE'],
                           str(nlp_job.user_id),
                           'jobs',
                           str(nlp_job.id))

        try:
            os.makedirs(dir)
        except OSError:
            flash('OSError!')
        else:
            for file in new_nlp_job_form.files.data:
                file.save(os.path.join(dir, file.filename))
            '''
            ' TODO: Let the scheduler run this job in the background.
            '
            ' NOTE: Using self created threads is just for testing purpose as
            '       there is no scheduler available.
            '''
            thread = Thread(target=swarm.run, args=(nlp_job,))
            thread.start()
            flash('Job created!')
        return redirect(url_for('services.nlp'))

    return render_template(
        'services/nlp.html.j2',
        title='Natrual Language Processing',
        new_nlp_job_form=new_nlp_job_form
    )