Refactor the OCR app API

This refactor adds two new endpoints to view the OCR
content of versions and documents.

Signed-off-by: Roberto Rosario <Roberto.Rosario.Gonzalez@mayan-edms.com>
This commit is contained in:
Roberto Rosario
2019-02-19 03:21:36 -04:00
parent e03f017e7f
commit c2dd01d51e
7 changed files with 327 additions and 175 deletions

View File

@@ -9,48 +9,45 @@ from ..permissions import (
permission_ocr_document, permission_ocr_content_view,
)
TEST_DOCUMENT_CONTENT = 'Mayan EDMS Documentation'
from .literals import TEST_DOCUMENT_CONTENT
class OCRAPITestCase(DocumentTestMixin, BaseAPITestCase):
"""
Test the OCR app API endpoints
"""
def setUp(self):
super(OCRAPITestCase, self).setUp()
self.login_user()
def _request_document_ocr_submit_view(self):
return self.post(
viewname='rest_api:document-ocr-submit-view',
viewname='rest_api:document-ocr-submit',
kwargs={'document_id': self.document.pk}
)
def test_submit_document_no_access(self):
def test_submit_document_no_permission(self):
response = self._request_document_ocr_submit_view()
self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
self.assertFalse(hasattr(self.document.pages.first(), 'ocr_content'))
#TODO: mock OCR here
def test_submit_document_with_access(self):
self.grant_access(
permission=permission_ocr_document, obj=self.document
)
response = self._request_document_ocr_submit_view()
self.assertEqual(response.status_code, status.HTTP_202_ACCEPTED)
self.assertTrue(hasattr(self.document.pages.first(), 'ocr_content'))
def _request_document_version_ocr_submit_view(self):
return self.post(
viewname='rest_api:document-version-ocr-submit-view',
viewname='rest_api:document_version-ocr-submit',
kwargs={
'document_id': self.document.pk,
'document_version_id': self.document.latest_version.pk
}
)
def test_submit_document_version_no_access(self):
def test_submit_document_version_no_permission(self):
response = self._request_document_version_ocr_submit_view()
self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
self.assertFalse(hasattr(self.document.pages.first(), 'ocr_content'))
def test_submit_document_version_with_access(self):
@@ -59,29 +56,84 @@ class OCRAPITestCase(DocumentTestMixin, BaseAPITestCase):
)
response = self._request_document_version_ocr_submit_view()
self.assertEqual(response.status_code, status.HTTP_202_ACCEPTED)
self.assertTrue(hasattr(self.document.pages.first(), 'ocr_content'))
def _request_document_page_content_view(self):
def _request_document_content_view(self):
return self.get(
viewname='rest_api:document-page-ocr-content-view',
viewname='rest_api:document-ocr-content',
kwargs={
'document_id': self.document.pk,
'document_version_id': self.document.latest_version.pk,
'document_page_id': self.document.latest_version.pages.first().pk
'document_id': self.test_document.pk,
}
)
def test_get_document_version_page_content_no_access(self):
def test_get_document_content_no_permission(self):
response = self._request_document_content_view()
self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
def test_get_document_content_with_access(self):
self.document.submit_for_ocr()
self.grant_access(
permission=permission_ocr_content_view, obj=self.document
)
response = self._request_document_content_view()
self.assertEqual(response.status_code, status.HTTP_200_OK)
self.assertTrue(
TEST_DOCUMENT_CONTENT in response.data['text']
)
def _request_document_page_content_view(self):
latest_version = self.test_document.latest_version
return self.get(
viewname='rest_api:document_page-ocr-content',
kwargs={
'document_id': self.test_document.pk,
'document_version_id': latest_version.pk,
'document_page_id': latest_version.pages.first().pk
}
)
def test_get_document_version_page_content_no_permission(self):
response = self._request_document_page_content_view()
self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
def test_get_document_version_page_content_with_access(self):
self.document.submit_for_ocr()
self.grant_access(
permission=permission_ocr_content_view, obj=self.document
)
response = self._request_document_page_content_view()
self.assertEqual(response.status_code, status.HTTP_200_OK)
self.assertTrue(
TEST_DOCUMENT_CONTENT in response.data['content']
TEST_DOCUMENT_CONTENT in response.data['text']
)
def _request_document_version_content_view(self):
latest_version = self.test_document.latest_version
return self.get(
viewname='rest_api:document_version-ocr-content',
kwargs={
'document_id': self.test_document.pk,
'document_version_id': latest_version.pk,
}
)
def test_get_document_version_version_content_no_permission(self):
response = self._request_document_version_content_view()
self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
def test_get_document_version_version_content_with_access(self):
self.document.submit_for_ocr()
self.grant_access(
permission=permission_ocr_content_view, obj=self.document
)
response = self._request_document_version_content_view()
self.assertEqual(response.status_code, status.HTTP_200_OK)
self.assertTrue(
TEST_DOCUMENT_CONTENT in response.data['text']
)