Even more code clean up and test coverage.

.coveragerc

@@ -13,4 +13,5 @@ exclude_lines =
     from
     import
     logger
+    LOGGER
     pragma: no cover

apps/iiif/annotations/admin.py

@@ -1,4 +1,4 @@
-"""Django admin module for `apps.iiif.annotations`"""
+"""Django admin module for :class:`apps.iiif.annotations`"""
 from django.contrib import admin
 from import_export import resources, fields
 from import_export.admin import ImportExportModelAdmin

apps/iiif/annotations/apps.py

@@ -1,4 +1,4 @@
-"""Configuration for `apps.iiif.annotations`"""
+"""Configuration for :class:`apps.iiif.annotations`"""
 from django.apps import AppConfig
 
 class AnnotationsConfig(AppConfig):

apps/iiif/annotations/fixtures/annotations.json

@@ -6,7 +6,7 @@
       "y": 928,
       "w": 22,
       "h": 22,
-      "order": 54,
+      "order": 1,
       "content": "a",
       "resource_type": "cnt:ContentAsText",
       "motivation": "sc:painting",
@@ -147,4 +147,29 @@
       },
       "svg": "<span id='f846587c-1e1c-44d3-b1ce-20c0f7104dc5' style='font-family: monospace; height: 28px; width: 17px; font-size: 17.305315203955498px'>,</span>"
   }
+},
+{
+  "model": "annotations.annotation",
+  "pk": "f846588c-1e1c-44d3-b1ce-20c0f6109dc5",
+  "fields": {
+      "x": 1146,
+      "y": 928,
+      "w": 22,
+      "h": 22,
+      "order": 1,
+      "content": ",",
+      "resource_type": "cnt:ContentAsText",
+      "motivation": "sc:painting",
+      "format": "text/plain",
+      "canvas": "a7f1bd69-766c-4dd4-ab66-f4051fdd4cff",
+      "language": "en",
+      "owner": null,
+      "oa_annotation": {
+          "annotatedBy": {
+              "name": "ocr"
+          },
+          "@id": "f846587c-1e1c-44d3-b1ce-20c0f7104dc5"
+      },
+      "svg": "<span id='f846588c-1e1c-44d3-b1ce-20c0f6109dc5' style='font-family: monospace; height: 28px; width: 17px; font-size: 17.305315203955498px'>stankonia</span>"
+  }
 }]

apps/iiif/annotations/models.py

@@ -1,4 +1,4 @@
-"""Django models for `apps.iiif.annotations`"""
+"""Django models for :class:`apps.iiif.annotations`"""
 from django.contrib.postgres.fields import JSONField
 from django.db import models, IntegrityError
 from django.conf import settings

apps/iiif/annotations/tests/tests.py

@@ -1,5 +1,5 @@
 # pylint: disable = missing-function-docstring, invalid-name, line-too-long
-"""Test cases for `apps.iiif.annotations`."""
+"""Test cases for :class:`apps.iiif.annotations`."""
 from django.test import TestCase, Client
 from django.test import RequestFactory
 from django.conf import settings

apps/iiif/annotations/urls.py

@@ -1,4 +1,4 @@
-"""Url patterns for `apps.iiif.annotations`"""
+"""Url patterns for :class:`apps.iiif.annotations`"""
 from django.urls import path
 from . import views
 

apps/iiif/annotations/views.py

@@ -1,4 +1,4 @@
-"""Django views for `apps.iiif.annotations`"""
+"""Django views for :class:`apps.iiif.annotations`"""
 import json
 from django.views import View
 from django.core.serializers import serialize

apps/iiif/canvases/management/commands/rebuild_ocr.py

@@ -49,7 +49,7 @@ def handle(self, *args, **options):
                 )
         elif options['canvas']:
             try:
-                canvas = Canvas.objects.get(pid=options['canvas'])                    
+                canvas = Canvas.objects.get(pid=options['canvas'])
 
                 self.__rebuild(canvas, options['testing'])
                 self.stdout.write(
@@ -107,7 +107,8 @@ def __rebuild(self, canvas, testing=False):
                             y=word['y'],
                             canvas=canvas,
                             owner=USER.objects.get(username='ocr'),
-                            resource_type=Annotation.OCR
+                            resource_type=Annotation.OCR,
+                            order=word_order
                         )
                     word_order += 1
                     anno.content = word['content']

apps/iiif/canvases/services.py

@@ -28,6 +28,44 @@ def get_fake_canvas_info(canvas):
     response = fetch_url(canvas.service_id, timeout=settings.HTTP_REQUEST_TIMEOUT, format='json')
     return response
 
+def get_fake_ocr():
+    return [
+        {
+            "h": 22,
+            "w": 22,
+            "x": 1146,
+            "y": 928,
+            "content": "Dope"
+        },
+        {
+            "h": 222,
+            "w": 222,
+            "x": 11462,
+            "y": 9282,
+            "content": ""
+        },
+        {
+            "h": 21,
+            "w": 21,
+            "x": 1141,
+            "y": 9281,
+            "content": "southernplayalisticadillacmuzik"
+        },
+        {
+            "h": 213,
+            "w": 213,
+            "x": 11413,
+            "y": 92813
+        },
+        {
+            "h": 214,
+            "w": 214,
+            "x": 11414,
+            "y": 92814,
+            "content": " "
+        }
+    ]
+
 def get_ocr(canvas):
     """Function to determine method for fetching OCR for a canvas.
 
@@ -36,6 +74,8 @@ def get_ocr(canvas):
     :return: List of dicts of parsed OCR data.
     :rtype: list
     """
+    if 'fake.info' in canvas.IIIF_IMAGE_SERVER_BASE.IIIF_IMAGE_SERVER_BASE:
+        return get_fake_ocr()
     if canvas.default_ocr == "line":
         result = fetch_alto_ocr(canvas)
         return add_alto_ocr(result)
@@ -178,8 +218,8 @@ def add_alto_ocr(result):
     for zones in surface:
         if 'zone' in zones.tag:
             for line in zones:
-                if line[-1].text is None:
-                    continue
+                # if line[-1].text is None:
+                #     continue
                 ocr.append({
                     'content': line[-1].text,
                     'h': int(line.attrib['lry']) - int(line.attrib['uly']),

apps/iiif/canvases/tests/tests.py

@@ -1,9 +1,10 @@
 """
-Test cases for `apps.iiif.canvases`
+Test cases for :class:`apps.iiif.canvases`
 """
 import json
 from io import StringIO
 import httpretty
+from bs4 import BeautifulSoup
 from django.test import TestCase, Client
 from django.urls import reverse
 from django.core.management import call_command
@@ -29,7 +30,7 @@ def setUp(self):
     def test_default_iiif_image_server_url(self):
         i_server = IServer()
         assert i_server.IIIF_IMAGE_SERVER_BASE == settings.IIIF_IMAGE_SERVER_BASE
-    
+
     def test_app_config(self):
         assert CanvasesConfig.verbose_name == 'Canvases'
         assert CanvasesConfig.name == 'apps.iiif.canvases'
@@ -94,7 +95,7 @@ def test_ia_ocr_creation(self):
 
     def test_fedora_ocr_creation(self):
         valid_fedora_positional_response = """523\t 116\t 151\t  45\tDistillery\r\n 704\t 117\t 148\t  52\tplaid,"\r\n""".encode('UTF-8-sig')
-        
+
         ocr = services.add_positional_ocr(self.canvas, valid_fedora_positional_response)
         assert len(ocr) == 2
         for word in ocr:
@@ -145,6 +146,9 @@ def test_line_by_line_from_alto(self):
         assert ocr.x == 916
         assert ocr.y == 0
 
+        for num, anno in enumerate(updated_canvas.annotation_set.all(), start=1):
+            assert anno.order == num
+
     @httpretty.activate
     def test_ocr_from_tsv(self):
         tsv = """content\tx\ty\tw\th\nJordan\t459\t391\t89\t43\t\n\t453\t397\t397\t3\n \t1\t2\t3\t4\n"""
@@ -173,7 +177,7 @@ def test_from_bad_alto(self):
         assert ocr is None
 
     def test_canvas_detail(self):
-        kwargs = { 'manifest': self.manifest.pid, 'pid': self.canvas.pid }
+        kwargs = {'manifest': self.manifest.pid, 'pid': self.canvas.pid}
         url = reverse('RenderCanvasDetail', kwargs=kwargs)
         response = self.client.get(url)
         serialized_canvas = json.loads(response.content.decode('UTF-8-sig'))
@@ -211,7 +215,7 @@ def test_wide_image_crops(self):
         assert canvas.thumbnail_crop_landscape == "%s/%s/pct:25,0,50,100/,250/0/default.jpg" % (canvas.IIIF_IMAGE_SERVER_BASE, pid)
         assert canvas.thumbnail_crop_tallwide == "%s/%s/pct:5,5,90,90/250,/0/default.jpg" % (canvas.IIIF_IMAGE_SERVER_BASE, pid)
         assert canvas.thumbnail_crop_volume == "%s/%s/pct:25,15,50,85/,600/0/default.jpg" % (canvas.IIIF_IMAGE_SERVER_BASE, pid)
-    
+
     def test_result_property(self):
         assert self.canvas.result == "a retto , dio Quef\u00eca de'"
 
@@ -221,12 +225,18 @@ def test_get_image_info(self):
         updated_canvas = Canvas.objects.get(pk=self.canvas.pk)
         assert updated_canvas.image_info['height'] == 3000
         assert updated_canvas.image_info['width'] == 3000
-    
+
     def test_command_output_rebuild_canvas(self):
         out = StringIO()
         call_command('rebuild_ocr', canvas=Canvas.objects.all().first().pid, stdout=out)
         assert 'OCR rebuilt for canvas' in out.getvalue()
 
+    def test_command_output_rebuild_canvas_with_no_existing_annotations(self):
+        canvas = CanvasFactory.create(manifest=self.manifest)
+        out = StringIO()
+        call_command('rebuild_ocr', canvas=canvas.pid, stdout=out)
+        assert 'OCR rebuilt for canvas' in out.getvalue()
+
     def test_command_output_rebuild_manifest(self):
         out = StringIO()
         call_command('rebuild_ocr', manifest=Manifest.objects.all().first().pid, stdout=out)
@@ -247,22 +257,69 @@ def test_command_output_rebuild_pid_not_given(self):
         call_command('rebuild_ocr', stdout=out)
         assert 'ERROR: your must provide a manifest or canvas pid' in out.getvalue()
 
-    # def test_command_rebuild_ocr(self):
-    #     iiif_server = IServer.objects.get(IIIF_IMAGE_SERVER_BASE='https://images.readux.ecds.emory/')
-    #     self.canvas.IIIF_IMAGE_SERVER_BASE = iiif_server
-    #     self.canvas.save()
-    #     out = StringIO()
-    #     self.canvas.label = 'karl'
-    #     call_command('rebuild_ocr', canvas=self.canvas.pid, testing=True, stdout=out)
-    #     assert 'yup' in out.getvalue()
-    #     # ocr = canvas.annotation_set.all().first()
-    #     # assert ocr.h == 43
-    #     # assert ocr.w == 89
-    #     # assert ocr.x == 459
-    #     # assert ocr.y == 391
-    #     # assert 'Jordan' in ocr.content
-    #     # assert len(canvas.annotation_set.all()) == 1
-
+    def test_command_rebuild_ocr_canvas(self):
+        original_anno_count = self.canvas.annotation_set.all().count()
+        # Check the OCR attributes before rebuilding.
+        first_anno = self.canvas.annotation_set.all().first()
+        assert first_anno.h == 22
+        assert first_anno.w == 22
+        assert first_anno.x == 1146
+        assert first_anno.y == 928
+        original_span = BeautifulSoup(first_anno.content, 'html.parser')
+        assert 'Dope' not in original_span.string
+        assert original_span.span is not None
+        assert original_span.span.span is None
+        self.canvas.IIIF_IMAGE_SERVER_BASE = IServer.objects.get(
+            IIIF_IMAGE_SERVER_BASE='http://fake.info'
+        )
+        self.canvas.save()
+        out = StringIO()
+        call_command('rebuild_ocr', canvas=self.canvas.pid, testing=True, stdout=out)
+        assert 'OCR rebuilt for canvas' in out.getvalue()
+        ocr = self.canvas.annotation_set.all().first()
+        assert ocr.h == 22
+        assert ocr.w == 22
+        assert ocr.x == 1146
+        assert ocr.y == 928
+        new_span = BeautifulSoup(ocr.content, 'html.parser')
+        assert 'Dope' in new_span.string
+        assert original_span.string not in new_span.string
+        assert new_span.span is not None
+        assert new_span.span.span is None
+        assert len(self.canvas.annotation_set.all()) == original_anno_count + 1
+
+    def test_command_rebuild_ocr_manifest(self):
+        canvas = Canvas.objects.get(pk='a7f1bd69-766c-4dd4-ab66-f4051fdd4cff')
+        original_anno_count = canvas.annotation_set.all().count()
+        # Check the OCR attributes before rebuilding.
+        first_anno = canvas.annotation_set.all().first()
+        assert first_anno.h == 22
+        assert first_anno.w == 22
+        assert first_anno.x == 1146
+        assert first_anno.y == 928
+        original_span = BeautifulSoup(first_anno.content, 'html.parser')
+        assert 'southernplayalisticadillacmuzik' not in original_span.string
+        assert original_span.span is not None
+        assert original_span.span.span is None
+        canvas.IIIF_IMAGE_SERVER_BASE = IServer.objects.get(
+            IIIF_IMAGE_SERVER_BASE='http://fake.info'
+        )
+        canvas.save()
+        out = StringIO()
+        call_command('rebuild_ocr', manifest=canvas.manifest.pid, testing=True, stdout=out)
+        assert 'OCR rebuilt for manifest' in out.getvalue()
+        ocr = canvas.annotation_set.all().first()
+        assert ocr.h == 22
+        assert ocr.w == 22
+        assert ocr.x == 1146
+        assert ocr.y == 928
+        new_span = BeautifulSoup(ocr.content, 'html.parser')
+        assert 'Dope' in new_span.string
+        assert original_span.string not in new_span.string
+        assert new_span.span is not None
+        assert new_span.span.span is None
+        assert len(canvas.annotation_set.all()) == original_anno_count + 1
+
     def test_no_alto_for_internet_archive(self):
         iiif_server = IServer.objects.get(IIIF_IMAGE_SERVER_BASE='https://iiif.archivelab.org/iiif/')
         canvas = CanvasFactory(IIIF_IMAGE_SERVER_BASE=iiif_server, manifest=self.canvas.manifest)

apps/iiif/canvases/urls.py

@@ -1,5 +1,5 @@
 """
-URL patterns for `apps.iiif.canvases`
+URL patterns for :class:`apps.iiif.canvases`
 """
 from django.urls import path
 from .views import IIIFV2Detail, IIIFV2List

apps/iiif/canvases/views.py

@@ -70,5 +70,6 @@ def get(self, request, *args, **kwargs): # pylint: disable = unused-argument
                     'canvas',
                     self.get_queryset()
                 )
-            )
+            ),
+            safe=False
         )

apps/iiif/kollections/tests/factories.py

@@ -8,7 +8,7 @@
 
 class CollectionFactory(DjangoModelFactory):
     """
-    Factory for mocking `apps.iiif.kollections.models.Collection` objects.
+    Factory for mocking :class:`apps.iiif.kollections.models.Collection` objects.
     """
     pid = str(random.randrange(2000, 5000))
     label = Faker("name")

apps/iiif/kollections/tests/tests.py

@@ -11,6 +11,7 @@
 import config.settings.local as settings
 from ..views import CollectionSitemap
 from ..models import Collection
+from ..admin import CollectionAdmin, ManifestInline
 from ...manifests.models import Manifest
 
 class KollectionTests(TestCase):
@@ -168,3 +169,11 @@ def test_serialize_single_object(self):
         collection = json.loads(serialize('kollection', [Collection.objects.all().first()]))
         assert collection['@type'] == 'sc:Collection'
         assert isinstance(collection, dict)
+
+    def test_collection_admin_inlines(self):
+        pid = Manifest.collections.through.objects.all().first().manifest.pid
+        admin_pid = CollectionAdmin.inlines[0].manifest_pid(
+            ManifestInline,
+            Manifest.collections.through.objects.all().first()
+        )
+        assert pid == admin_pid