matthias/mayan-edms
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Code cleanup

Browse Source
This commit is contained in:
Roberto Rosario
2011-07-21 11:46:15 -04:00
parent 46aa89d924
commit 90e876ca93
36 changed files with 125 additions and 169 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
apps/ocr/api.py
View File

@@ -56,7 +56,7 @@ def run_tesseract(input_filename, lang=None):
os.close(fd)
ocr_output = os.extsep.join([filepath, u'txt'])
command = [unicode(TESSERACT_PATH), unicode(input_filename), unicode(filepath)]
# TODO: Tesseract 3.0 segfaults
#if lang is not None:
# command.extend([u'-l', lang])
@@ -68,15 +68,15 @@ def run_tesseract(input_filename, lang=None):
cleanup(filepath)
cleanup(ocr_output)
raise TesseractError(error_text)
fd = codecs.open(ocr_output, 'r', 'utf-8')
text = fd.read().strip()
fd.close()
os.unlink(filepath)
os.unlink(filepath)
return text
def do_document_ocr(queue_document):
"""
@@ -91,12 +91,12 @@ def do_document_ocr(queue_document):
except (ParserError, ParserUnknownFile):
# Fall back to doing visual OCR
##ocr_transformations, warnings = queue_document.get_transformation_list()
document_filepath = document_page.document.get_image_cache_name(page=document_page.page_number)
unpaper_output_filename = u'%s_unpaper_out_page_%s%s%s' % (document_page.document.uuid, document_page.page_number, os.extsep, UNPAPER_FILE_FORMAT)
unpaper_output_filepath = os.path.join(TEMPORARY_DIRECTORY, unpaper_output_filename)
unpaper_input=convert(document_filepath, file_format=UNPAPER_FILE_FORMAT)
unpaper_input = convert(document_filepath, file_format=UNPAPER_FILE_FORMAT)
execute_unpaper(input_filepath=unpaper_input, output_filepath=unpaper_output_filepath)
#from PIL import Image, ImageOps
@@ -108,7 +108,7 @@ def do_document_ocr(queue_document):
#im.save(unpaper_output_filepath)
# Convert to TIFF
pre_ocr_filepath = output_filepath=convert(input_filepath=unpaper_output_filepath, file_format=DEFAULT_OCR_FILE_FORMAT)
pre_ocr_filepath = convert(input_filepath=unpaper_output_filepath, file_format=DEFAULT_OCR_FILE_FORMAT)
# Tesseract needs an explicit file extension
pre_ocr_filepath_w_ext = os.extsep.join([pre_ocr_filepath, DEFAULT_OCR_FILE_EXTENSION])
os.rename(pre_ocr_filepath, pre_ocr_filepath_w_ext)

2
apps/ocr/forms.py
View File

@@ -1,6 +1,4 @@
from django import forms
from django.utils.translation import ugettext_lazy as _
from django.utils.translation import ugettext
from ocr.models import QueueTransformation

2
apps/ocr/managers.py
View File

@@ -37,5 +37,5 @@ class QueueTransformationManager(models.Manager):
)
except Exception, e:
warnings.append(e)
return transformations, warnings

2
apps/ocr/models.py
View File

@@ -48,7 +48,7 @@ class QueueDocument(models.Model):
ordering = ('datetime_submitted',)
verbose_name = _(u'queue document')
verbose_name_plural = _(u'queue documents')
def get_transformation_list(self):
return QueueTransformation.objects.get_for_object_as_list(self)

14
apps/ocr/parsers/__init__.py
View File

@@ -1,9 +1,3 @@
import codecs
import os
import subprocess
import tempfile
import sys
import slate
from django.utils.translation import ugettext as _
@@ -21,14 +15,14 @@ def pdf_parser(document_page):
fd = document_page.document.open()
pdf_pages = slate.PDF(fd)
fd.close()
if pdf_pages[document_page.page_number - 1] == '\x0c':
raise ParserError
document_page.content = pdf_pages[document_page.page_number - 1]
document_page.page_label = _(u'Text extracted from PDF')
document_page.save()
def parse_document_page(document_page):
try:
@@ -36,5 +30,5 @@ def parse_document_page(document_page):
except KeyError:
raise ParserUnknownFile
register_parser('application/pdf', pdf_parser)

3
apps/ocr/urls.py
View File

@@ -13,10 +13,9 @@ urlpatterns = patterns('ocr.views',
url(r'^document/all/clean_up/$', 'all_document_ocr_cleanup', (), 'all_document_ocr_cleanup'),
url(r'^node/active/list/$', 'node_active_list', (), 'node_active_list'),
url(r'^queue/(?P<document_queue_id>\d+)/transformation/list/$', 'setup_queue_transformation_list', (), 'setup_queue_transformation_list'),
url(r'^queue/(?P<document_queue_id>\w+)/transformation/create/$', 'setup_queue_transformation_create', (), 'setup_queue_transformation_create'),
url(r'^queue/transformation/(?P<transformation_id>\w+)/edit/$', 'setup_queue_transformation_edit', (), 'setup_queue_transformation_edit'),
url(r'^queue/transformation/(?P<transformation_id>\w+)/delete/$', 'setup_queue_transformation_delete', (), 'setup_queue_transformation_delete'),
)

26
apps/ocr/views.py
View File

@@ -41,7 +41,7 @@ def queue_document_list(request, queue_name='default'):
'queue': document_queue,
'object_name': _(u'document queue'),
'navigation_object_name': 'queue',
'list_object_variable_name': 'queue_document',
'list_object_variable_name': 'queue_document',
'extra_columns': [
{'name': 'document', 'attribute': lambda x: document_link(x.document) if hasattr(x, 'document') else _(u'Missing document.')},
{'name': _(u'thumbnail'), 'attribute': lambda x: document_thumbnail(x.document)},
@@ -325,7 +325,7 @@ def node_active_list(request):
def setup_queue_transformation_list(request, document_queue_id):
#check_permissions(request.user, [PERMISSION_SOURCES_SETUP_EDIT])
document_queue = get_object_or_404(DocumentQueue, pk=document_queue_id)
context = {
@@ -336,7 +336,7 @@ def setup_queue_transformation_list(request, document_queue_id):
'queue': document_queue,
'object_name': _(u'document queue'),
'navigation_object_name': 'queue',
'list_object_variable_name': 'transformation',
'list_object_variable_name': 'transformation',
'extra_columns': [
{'name': _(u'order'), 'attribute': 'order'},
{'name': _(u'transformation'), 'attribute': lambda x: x.get_transformation_display()},
@@ -347,12 +347,12 @@ def setup_queue_transformation_list(request, document_queue_id):
}
return render_to_response('generic_list.html', context,
context_instance=RequestContext(request))
context_instance=RequestContext(request))
def setup_queue_transformation_edit(request, transformation_id):
#check_permissions(request.user, [PERMISSION_SOURCES_SETUP_EDIT])
transformation = get_object_or_404(QueueTransformation, pk=transformation_id)
redirect_view = reverse('setup_queue_transformation_list', args=[transformation.content_object.pk])
next = request.POST.get('next', request.GET.get('next', request.META.get('HTTP_REFERER', redirect_view)))
@@ -386,7 +386,7 @@ def setup_queue_transformation_edit(request, transformation_id):
],
'next': next,
},
context_instance=RequestContext(request))
context_instance=RequestContext(request))
def setup_queue_transformation_delete(request, transformation_id):
@@ -413,24 +413,23 @@ def setup_queue_transformation_delete(request, transformation_id):
'navigation_object_list': [
{'object': 'queue', 'name': _(u'document queue')},
{'object': 'transformation', 'name': _(u'transformation')}
],
],
'title': _(u'Are you sure you wish to delete queue transformation "%(transformation)s"') % {
'transformation': transformation.get_transformation_display(),
},
'previous': previous,
'form_icon': u'shape_square_delete.png',
},
context_instance=RequestContext(request))
context_instance=RequestContext(request))
def setup_queue_transformation_create(request, document_queue_id):
#check_permissions(request.user, [PERMISSION_SOURCES_SETUP_EDIT])
document_queue = get_object_or_404(DocumentQueue, pk=document_queue_id)
redirect_view = reverse('setup_queue_transformation_list', args=[document_queue.pk])
previous = request.POST.get('previous', request.GET.get('previous', request.META.get('HTTP_REFERER', redirect_view)))
if request.method == 'POST':
form = QueueTransformationForm_create(request.POST)
if form.is_valid():
@@ -439,7 +438,7 @@ def setup_queue_transformation_create(request, document_queue_id):
eval(form.cleaned_data['arguments'], {})
except:
messages.error(request, _(u'Queue transformation argument error.'))
else:
else:
try:
queue_tranformation = form.save(commit=False)
queue_tranformation.content_object = document_queue
@@ -450,7 +449,7 @@ def setup_queue_transformation_create(request, document_queue_id):
messages.error(request, _(u'Error creating queue transformation; %s') % e)
else:
form = QueueTransformationForm_create()
return render_to_response('generic_form.html', {
'form': form,
'queue': document_queue,
@@ -458,4 +457,3 @@ def setup_queue_transformation_create(request, document_queue_id):
'navigation_object_name': 'queue',
'title': _(u'Create new transformation for queue: %s') % document_queue,
}, context_instance=RequestContext(request))