def test_multi_index(self):
"""
Test that documents can be stored in multiple indexes.
"""
self.index.delete_instance()
indexes = [Index.create(name='idx-%s' % i) for i in range(3)]
document = Document.create(content='hueybear')
for index in indexes:
index.index(document.content, document)
self.assertEqual(Document.select().count(), 1)
self.assertEqual(Index.select().count(), 3)
self.assertEqual(IndexDocument.select().count(), 3)
query = (IndexDocument.select(
Index.name,
IndexDocument.document).join(Index).order_by(Index.name).dicts())
idx_doc_data = [idx_doc for idx_doc in query]
self.assertEqual(idx_doc_data, [
{
'document': document.get_id(),
'name': 'idx-0'
},
{
'document': document.get_id(),
'name': 'idx-1'
},
{
'document': document.get_id(),
'name': 'idx-2'
},
])
def test_reindex(self):
"""
Test that an existing document can be re-indexed, updating the
content and metadata in the process.
"""
doc = self.index.index('test doc', foo='bar', baze='nug')
doc_db = (Document.select(Document._meta.primary_key,
Document.content).get())
self.assertTrue(doc_db.get_id() is not None)
self.assertEqual(doc_db.content, 'test doc')
self.assertEqual(doc_db.metadata, {'foo': 'bar', 'baze': 'nug'})
updated_doc = self.index.index('updated doc',
document=doc,
foo='bazz',
nug='x')
self.assertEqual(Document.select().count(), 1)
self.assertEqual(updated_doc.metadata, {'foo': 'bazz', 'nug': 'x'})
u_doc_db = (Document.select(Document._meta.primary_key,
Document.content).get())
self.assertEqual(u_doc_db.content, 'updated doc')
self.assertEqual(u_doc_db.get_id(), doc_db.get_id())
self.assertEqual(u_doc_db.metadata, {'foo': 'bazz', 'nug': 'x'})
# Verify through relationship properties.
self.assertEqual(IndexDocument.select().count(), 1)
idx_doc = IndexDocument.get()
self.assertEqual(idx_doc.__data__['document'], u_doc_db.get_id())
self.assertEqual(idx_doc.__data__['index'], self.index.id)
def _get_document(self, pk):
if isinstance(pk, int) or (pk and pk.isdigit()):
query = Document.all().where(Document._meta.primary_key == pk)
try:
return query.get()
except Document.DoesNotExist:
pass
return get_object_or_404(Document.all(), Document.identifier == pk)
def test_document_detail_delete(self):
idx = Index.create(name='idx')
alt_idx = Index.create(name='alt-idx')
d1 = idx.index('doc 1', k1='v1', k2='v2')
d2 = idx.index('doc 2', k3='v3')
d2.attach('foo.jpg', 'bar')
alt_idx.add_to_index(d1)
alt_idx.add_to_index(d2)
self.assertEqual(Metadata.select().count(), 3)
self.assertEqual(Attachment.select().count(), 1)
response = self.app.delete('/documents/%s/' % d2.get_id())
data = json_load(response.data)
self.assertEqual(data, {'success': True})
self.assertEqual(Metadata.select().count(), 2)
self.assertEqual(Attachment.select().count(), 0)
response = self.app.delete('/documents/%s/' % d2.get_id())
self.assertEqual(response.status_code, 404)
self.assertEqual(Document.select().count(), 1)
self.assertEqual(IndexDocument.select().count(), 2)
self.assertEqual([d.get_id() for d in idx.documents], [d1.get_id()])
self.assertEqual([d.get_id() for d in alt_idx.documents],
[d1.get_id()])
def test_index_document_validation(self):
idx = Index.create(name='idx')
response = self.post_json('/documents/', {'content': 'foo'})
self.assertEqual(response['error'],
'You must specify either an "index" or "indexes".')
response = self.post_json('/documents/', {'content': 'x', 'index': ''})
self.assertEqual(response['error'],
'You must specify either an "index" or "indexes".')
response = self.post_json('/documents/', {
'content': 'foo',
'index': 'missing'
})
self.assertEqual(response['error'],
'The following indexes were not found: missing.')
response = self.post_json('/documents/', {
'content': 'foo',
'indexes': ['missing', 'idx', 'blah']
})
self.assertEqual(
response['error'],
'The following indexes were not found: missing, blah.')
self.assertEqual(Document.select().count(), 0)
def create(self):
data = validator.parse_post(
['content'], ['identifier', 'index', 'indexes', 'metadata'])
indexes = validator.validate_indexes(data)
if indexes is None:
error('You must specify either an "index" or "indexes".')
if data.get('identifier'):
try:
document = self._get_document(data['identifier'])
except NotFound:
pass
else:
return self.update(data['identifier'])
document = Document.create(content=data['content'],
identifier=data.get('identifier'))
if data.get('metadata'):
document.metadata = data['metadata']
logger.info('Created document with id=%s', document.get_id())
for index in indexes:
index.add_to_index(document)
logger.info('Added document %s to index %s', document.get_id(),
index.name)
if len(request.files):
self.attach_files(document)
return self.detail(document.get_id())
def list_view(self):
# Allow filtering by index.
idx_list = request.args.getlist('index')
if idx_list:
indexes = Index.select(Index.id).where(Index.name << idx_list)
else:
indexes = None
document_count = Document.select().count()
return jsonify(self._search_response(indexes, True, document_count))
def attachment_download(document_id, pk):
document = get_object_or_404(Document.all(),
Document._meta.primary_key == document_id)
attachment = get_object_or_404(document.attachments,
Attachment.filename == pk)
response = make_response(attachment.blob.data)
response.headers['Content-Type'] = attachment.mimetype
response.headers['Content-Length'] = attachment.length
response.headers['Content-Disposition'] = 'inline; filename=%s' % (
attachment.filename)
return response
def test_document_detail_post(self):
idx = Index.create(name='idx')
alt_idx = Index.create(name='alt-idx')
doc = idx.index('test doc', foo='bar', nug='baze')
alt_doc = idx.index('alt doc')
url = '/documents/%s/' % doc.get_id()
def assertDoc(doc, content, metadata=None, indexes=None):
doc_db = self.refresh_doc(doc)
self.assertEqual(doc_db.content, content)
self.assertEqual([idx.name for idx in doc_db.get_indexes()],
indexes or [])
self.assertEqual(doc_db.metadata, metadata or {})
# Update the content.
response = self.post_json(url, {'content': 'updated'})
assertDoc(doc, 'updated', {'foo': 'bar', 'nug': 'baze'}, ['idx'])
# Test updating metadata.
response = self.post_json(
url, {'metadata': dict(doc.metadata, nug='baz', herp='derp')})
assertDoc(doc, 'updated', {
'foo': 'bar',
'nug': 'baz',
'herp': 'derp'
}, ['idx'])
# Test clearing metadata.
response = self.post_json(url, {'metadata': None})
assertDoc(doc, 'updated', {}, ['idx'])
# Test updating indexes.
response = self.post_json(url, {'indexes': ['idx', 'alt-idx']})
assertDoc(doc, 'updated', {}, ['alt-idx', 'idx'])
# Test clearing indexes.
response = self.post_json(url, {'indexes': []})
assertDoc(doc, 'updated', {}, [])
# Ensure alt_doc has not been affected.
assertDoc(alt_doc, 'alt doc', {}, ['idx'])
# Sanity check.
self.assertEqual(Document.select().count(), 2)
def test_index_update_delete(self):
idx = Index.create(name='idx')
alt_idx = Index.create(name='alt-idx')
doc = idx.index(content='foo')
alt_idx.index(doc.content, doc)
idx.index('idx only')
alt_idx.index('alt only')
response = self.post_json('/idx/', {'name': 'idx-updated'})
self.assertEqual(response['id'], idx.id)
self.assertEqual(response['name'], 'idx-updated')
self.assertEqual([doc['content'] for doc in response['documents']],
['foo', 'idx only'])
response = self.app.delete('/idx-updated/')
data = json_load(response.data)
self.assertEqual(data, {'success': True})
self.assertEqual(Document.select().count(), 3)
self.assertEqual(IndexDocument.select().count(), 2)
self.assertEqual(Index.select().count(), 1)
def refresh_doc(self, doc):
return (Document.all().where(
Document._meta.primary_key == doc.get_id()).get())