Add changes for new OCR and NLP Pipelines

app/jobs/forms.py

@@ -21,9 +21,7 @@ class AddNLPJobForm(FlaskForm):
                            validators=[DataRequired()])
     submit = SubmitField()
     title = StringField('Title', validators=[DataRequired(), Length(1, 32)])
-    version = SelectField('Version',
-                          choices=[('2.2.0', 'Latest (2.2.0)'),
-                                   ('2.2.0', '2.2.0')],
+    version = SelectField('Version', choices=[('latest', 'Latest')],
                           validators=[DataRequired()])
     check_encoding = BooleanField('Check encoding')
 
@@ -54,15 +52,14 @@ class AddOCRJobForm(FlaskForm):
     split = BooleanField('Split')
     submit = SubmitField()
     title = StringField('Title', validators=[DataRequired(), Length(1, 32)])
-    version = SelectField('Version',
-                          choices=[('latest', 'Latest')],
+    version = SelectField('Version', choices=[('latest', 'Latest')],
                           validators=[DataRequired()])
 
     def validate_files(form, field):
         for file in field.data:
-            if not file.filename.lower().endswith(('.pdf', '.tif', '.tiff')):
+            if not file.filename.lower().endswith('.pdf'):
                 raise ValidationError('File does not have an approved '
-                                      'extension: .pdf | .tif | .tiff')
+                                      'extension: .pdf')
 
 
 class AddSetupFilesJobForm(FlaskForm):
@@ -71,8 +68,7 @@ class AddSetupFilesJobForm(FlaskForm):
     submit = SubmitField()
     title = StringField('Title', validators=[DataRequired(), Length(1, 32)])
     files = MultipleFileField('Files', validators=[DataRequired()])
-    version = SelectField('Version',
-                          choices=[('latest', 'Latest')],
+    version = SelectField('Version', choices=[('latest', 'Latest')],
                           validators=[DataRequired()])
 
     def validate_files(form, field):

app/services/views.py

@@ -41,8 +41,8 @@ def service(service):
                 service_args.append('--check-encoding')
         if service == 'ocr':
             service_args.append('-l {}'.format(add_job_form.language.data))
-            if not add_job_form.binarization.data:
-                service_args.append('--skip-binarisation')
+            if add_job_form.binarization.data:
+                service_args.append('--binarize')
         job = Job(creator=current_user,
                   description=add_job_form.description.data,
                   mem_mb=SERVICES[service]['resources']['mem_mb'],

app/templates/services/ocr.html.j2

@@ -53,7 +53,7 @@
             <div class="file-field input-field">
               <div class="btn">
                 <span>{{ add_job_form.files.label.text }}</span>
-                {{ add_job_form.files(accept='application/pdf, image/tiff') }}
+                {{ add_job_form.files(accept='application/pdf') }}
               </div>
               <div class="file-path-wrapper">
                 <input class="file-path validate" type="text">