Add input options for ocr job creation.

This commit is contained in:
Patrick Jentsch 2019-08-01 08:22:17 +02:00
parent 009e188924
commit 78040536f2
3 changed files with 110 additions and 27 deletions

View File

@ -1,6 +1,35 @@
from flask_wtf import FlaskForm
from wtforms import SubmitField
from flask_wtf.file import FileAllowed, FileRequired
from wtforms import FileField, SelectField, StringField, SubmitField
from wtforms.validators import DataRequired, Length
class OCRJobForm(FlaskForm):
description = StringField(
'Description',
validators=[DataRequired(), Length(1, 64)]
)
file = FileField(
'File',
validators=[
FileAllowed(['pdf', 'tif', 'tiff']),
FileRequired()
]
)
language = SelectField(
'Language',
choices=[
('', 'Choose your option'),
('eng', 'English'),
('enm', 'English, Middle (1100-1500)'),
('fra', 'French'),
('frm', 'French, Middle (ca. 1400-1600)'),
('deu', 'German'),
('frk', 'German Fraktur'),
('ita', 'Italian'),
('por', 'Portuguese'),
('spa', 'Spanish; Castilian')
],
validators=[DataRequired()]
)
submit = SubmitField('Submit')

View File

@ -1,9 +1,12 @@
from flask import redirect, render_template, url_for
from datetime import datetime
from flask import current_app, flash, redirect, render_template, url_for, request
from . import services
from flask_login import current_user, login_required
from .forms import OCRJobForm
from ..import swarm
from threading import Thread
import hashlib
import os
@services.route('/ocr', methods=['GET', 'POST'])
@ -11,29 +14,44 @@ from threading import Thread
def ocr():
ocr_job_form = OCRJobForm()
if ocr_job_form.validate_on_submit():
'''
' TODO: Implement a Job class. For now a dictionary representation is
' enough.
'''
job = {'worker': None,
'creator': current_user.id,
'id': '5fd40cb0cadef3ab5676c4968fc3d748',
'requested_cpus': 2,
'requested_memory': 2048,
'service': 'ocr',
'service_args': {'lang': 'eng',
'version': 'latest'
},
'status': 'queued'
}
'''
' 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=(job,))
thread.start()
app = current_app._get_current_object()
id = hashlib.md5(
(current_user.username + '_' + datetime.now().isoformat()).encode()
).hexdigest()
dir = os.path.join(app.config['OPAQUE_UPLOAD_DIRECTORY'], id)
try:
os.mkdir(dir)
except FileExistsError:
# Possible MD5 hash collision occurred.
flash('Internal error occurred, please try again!')
else:
file = ocr_job_form.file.data
file.save(os.path.join(dir, file.filename))
'''
' TODO: Implement a Job class. For now a dictionary representation
' is enough.
'''
job = {'worker': None,
'creator': current_user.id,
'id': id,
'requested_cpus': 2,
'requested_memory': 2048,
'service': 'ocr',
'service_args': {'lang': ocr_job_form.language.data,
'version': 'latest'
},
'status': 'queued'
}
'''
' 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=(job,))
thread.start()
return redirect(url_for('services.ocr'))
return render_template(

View File

@ -76,10 +76,46 @@
</div>
<div class="col s12">
<div class="card large">
<div class="card">
<div class="card-content">
<form method="POST">
<form method="POST" enctype="multipart/form-data">
{{ ocr_job_form.hidden_tag() }}
<div class="row">
<div class="col s8">
<div class="input-field">
<i class="material-icons prefix">description</i>
{{ ocr_job_form.description(placeholder='Job description') }}
{{ ocr_job_form.description.label }}
{% for error in ocr_job_form.description.errors %}
<span class="helper-text red-text">{{ error }}</span>
{% endfor %}
</div>
</div>
<div class="col s4">
<div class="input-field">
<i class="material-icons prefix">language</i>
{{ ocr_job_form.language() }}
{{ ocr_job_form.language.label }}
{% for error in ocr_job_form.language.errors %}
<span class="helper-text red-text">{{ error }}</span>
{% endfor %}
</div>
</div>
<div class="col s8">
<div class="file-field input-field">
<div class="btn">
<span>{{ ocr_job_form.file.label.text }}</span>
{{ ocr_job_form.file() }}
</div>
<div class="file-path-wrapper">
<input class="file-path validate" type="text">
</div>
{% for error in ocr_job_form.file.errors %}
<span class="helper-text red-text">{{ error }}</span>
{% endfor %}
</div>
</div>
</div>
{{ ocr_job_form.submit(class='btn') }}
</form>
</div>