Files
mayan-edms/mayan/apps/document_parsing/forms.py
Roberto Rosario 0699ad0556 Add support for new document page structure
Documents now have their own dedicated DocumentPage
submodel. The old DocumentPage is now called DocumentVersionPage.
This allows mappings between document pages and document version
pages, allowing renumbering, appending pages.
DocumentPages have a content_object to map them to any other
object. For now they only map to DocumentVersionPages.
New option added to the version upload form to append the
pages of the new version.
A new view was added to just append new pages with wraps the
new document version upload form and hides the append pages
checkbox set to True.
Add a new action, reset_pages to reset the pages of the
document to those of the latest version.

Missing: appending tests, checks for proper content_object in OCR and
document parsing.

Author: Roberto Rosario <roberto.rosario@mayan-edms.com>
Date:   Thu Oct 11 12:00:25 2019 -0400
2019-10-10 11:55:42 -04:00

82 lines
2.6 KiB
Python

from __future__ import unicode_literals
from django import forms
from django.utils.encoding import force_text
from django.utils.html import conditional_escape
from django.utils.safestring import mark_safe
from django.utils.translation import ugettext_lazy as _, ugettext
from mayan.apps.common.widgets import TextAreaDiv
from .models import DocumentVersionPageContent
class DocumentContentForm(forms.Form):
"""
Form that concatenates all of a document pages' text content into a
single textarea widget
"""
def __init__(self, *args, **kwargs):
self.document = kwargs.pop('instance', None)
super(DocumentContentForm, self).__init__(*args, **kwargs)
content = []
self.fields['contents'].initial = ''
try:
document_pages = self.document.pages.all()
except AttributeError:
document_pages = []
for document_page in document_pages:
try:
page_content = document_page.content_object.content.content
except DocumentVersionPageContent.DoesNotExist:
pass
else:
content.append(conditional_escape(force_text(page_content)))
content.append(
'\n\n\n<hr/><div class="document-page-content-divider">- %s -</div><hr/>\n\n\n' % (
ugettext(
'Page %(page_number)d'
) % {'page_number': document_page.page_number}
)
)
self.fields['contents'].initial = mark_safe(''.join(content))
contents = forms.CharField(
label=_('Contents'),
widget=TextAreaDiv(
attrs={
'class': 'text_area_div full-height',
'data-height-difference': 360
}
)
)
class DocumentPageContentForm(forms.Form):
contents = forms.CharField(
label=_('Contents'),
widget=TextAreaDiv(
attrs={
'class': 'text_area_div full-height',
'data-height-difference': 360
}
)
)
def __init__(self, *args, **kwargs):
document_page = kwargs.pop('instance', None)
super(DocumentPageContentForm, self).__init__(*args, **kwargs)
content = ''
self.fields['contents'].initial = ''
try:
page_content = document_page.content_object.content.content
except DocumentVersionPageContent.DoesNotExist:
pass
else:
content = conditional_escape(force_text(page_content))
self.fields['contents'].initial = mark_safe(content)