def test_migrate_backend(self):
# verify: attachment is in couch and migration not complete
with maybe_not_found():
s3_blobs = sum(1 for b in self.db.new_db._s3_bucket().objects.all())
self.assertEqual(s3_blobs, 0)
with tempdir() as tmp:
filename = join(tmp, "file.txt")
# do migration
migrated, skipped = mod.MIGRATIONS[self.slug].migrate(filename)
self.assertGreaterEqual(migrated, self.test_size)
# verify: migration state recorded
mod.BlobMigrationState.objects.get(slug=self.slug)
# verify: migrated data was written to the file
with open(filename) as fh:
lines = list(fh)
ids = {d._id for d in self.migrate_docs}
migrated = {d["_id"] for d in (json.loads(x) for x in lines)}
self.assertEqual(len(ids.intersection(migrated)), self.test_size)
# verify: attachment was copied to new blob db
for doc in self.migrate_docs:
exp = SavedBasicExport.get(doc._id)
self.assertEqual(exp._rev, doc._rev) # rev should not change
self.assertTrue(doc.blobs)
bucket = doc._blobdb_bucket()
for meta in doc.blobs.values():
content = self.db.new_db.get(meta.id, bucket)
self.assertEqual(len(content.read()), meta.content_length)
def test_migrate_backend(self):
# verify: attachment is in couch and migration not complete
with maybe_not_found():
s3_blobs = sum(1
for b in self.db.new_db._s3_bucket().objects.all())
self.assertEqual(s3_blobs, 0)
with tempdir() as tmp:
filename = join(tmp, "file.txt")
# do migration
migrated, skipped = mod.MIGRATIONS[self.slug].migrate(filename)
self.assertGreaterEqual(migrated, self.test_size)
# verify: migration state recorded
mod.BlobMigrationState.objects.get(slug=self.slug)
# verify: migrated data was written to the file
with open(filename) as fh:
lines = list(fh)
ids = {d._id for d in self.migrate_docs}
migrated = {d["_id"] for d in (json.loads(x) for x in lines)}
self.assertEqual(len(ids.intersection(migrated)), self.test_size)
# verify: attachment was copied to new blob db
for doc in self.migrate_docs:
exp = SavedBasicExport.get(doc._id)
self.assertEqual(exp._rev, doc._rev) # rev should not change
self.assertTrue(doc.blobs)
bucket = doc._blobdb_bucket()
for meta in doc.blobs.values():
content = self.db.new_db.get(meta.id, bucket)
self.assertEqual(len(content.read()), meta.content_length)
def download_saved_export(request, export_id):
export = SavedBasicExport.get(export_id)
attach = export._attachments[export.configuration.filename]
response = HttpResponse(mimetype=attach["content_type"])
response.write(export.fetch_attachment(export.configuration.filename))
response['Content-Disposition'] = 'attachment; filename=%s' % export.configuration.filename
return response
def modify_doc_and_print_status(num, total):
if not modified:
# do concurrent modification
doc = SavedBasicExport.get(saved._id)
doc.set_payload(new_payload)
doc.save()
modified.append(True)
print_status(num, total)
def test_migrate_happy_path(self):
saved = SavedBasicExport(configuration=_mk_config())
saved.save()
payload = b'binary data not valid utf-8 \xe4\x94'
name = saved.get_attachment_name()
super(BlobMixin, saved).put_attachment(payload, name)
saved.save()
self.do_migration([saved])
exp = SavedBasicExport.get(saved._id)
self.assertEqual(exp.get_payload(), payload)
def test_migrate_happy_path(self):
saved = SavedBasicExport(configuration=_mk_config())
saved.save()
payload = b'binary data not valid utf-8 \xe4\x94'
name = saved.get_attachment_name()
super(BlobMixin, saved).put_attachment(payload, name)
saved.save()
self.do_migration([saved])
exp = SavedBasicExport.get(saved._id)
self.assertEqual(exp.get_payload(), payload)
def download_saved_export(request, export_id):
export = SavedBasicExport.get(export_id)
attach = export._attachments[export.get_attachment_name()]
response = HttpResponse(mimetype=attach["content_type"])
response.write(export.get_payload())
# ht: http://stackoverflow.com/questions/1207457/convert-unicode-to-string-in-python-containing-extra-symbols
normalized_filename = unicodedata.normalize(
'NFKD', unicode(export.configuration.filename),
).encode('ascii','ignore')
response['Content-Disposition'] = 'attachment; filename=%s' % normalized_filename
return response
def download_saved_export(request, export_id):
export = SavedBasicExport.get(export_id)
content_type = Format.from_format(export.configuration.format).mimetype
payload = export.get_payload(stream=True)
response = StreamingHttpResponse(FileWrapper(payload), content_type=content_type)
if export.configuration.format != 'html':
# ht: http://stackoverflow.com/questions/1207457/convert-unicode-to-string-in-python-containing-extra-symbols
normalized_filename = unicodedata.normalize(
'NFKD', unicode(export.configuration.filename),
).encode('ascii', 'ignore')
response['Content-Disposition'] = 'attachment; filename="%s"' % normalized_filename
return response
def download_saved_export(request, export_id):
export = SavedBasicExport.get(export_id)
attach = export._attachments[export.get_attachment_name()]
response = HttpResponse(mimetype=attach["content_type"])
response.write(export.get_payload())
# ht: http://stackoverflow.com/questions/1207457/convert-unicode-to-string-in-python-containing-extra-symbols
normalized_filename = unicodedata.normalize(
'NFKD',
unicode(export.configuration.filename),
).encode('ascii', 'ignore')
response[
'Content-Disposition'] = 'attachment; filename=%s' % normalized_filename
return response
def test_migrate_with_concurrent_modification(self):
# setup data
saved = SavedBasicExport(configuration=_mk_config())
saved.save()
name = saved.get_attachment_name()
new_payload = 'something new'
old_payload = 'something old'
super(BlobMixin, saved).put_attachment(old_payload, name)
super(BlobMixin, saved).put_attachment(old_payload, "other")
saved.save()
# verify: attachments are in couch
self.assertEqual(len(saved._attachments), 2)
self.assertEqual(len(saved.external_blobs), 0)
modified = []
print_status = mod.print_status
# setup concurrent modification
def modify_doc_and_print_status(num, total):
if not modified:
# do concurrent modification
doc = SavedBasicExport.get(saved._id)
doc.set_payload(new_payload)
doc.save()
modified.append(True)
print_status(num, total)
# hook print_status() call to simulate concurrent modification
with replattr(mod, "print_status", modify_doc_and_print_status):
# do migration
migrated, skipped = mod.MIGRATIONS[self.slug].migrate()
self.assertGreaterEqual(skipped, 1)
# verify: migration state not set when docs are skipped
with self.assertRaises(mod.BlobMigrationState.DoesNotExist):
mod.BlobMigrationState.objects.get(slug=self.slug)
# verify: attachments were not migrated
exp = SavedBasicExport.get(saved._id)
self.assertEqual(len(exp._attachments), 1, exp._attachments)
self.assertEqual(len(exp.external_blobs), 1, exp.external_blobs)
self.assertEqual(exp.get_payload(), new_payload)
self.assertEqual(exp.fetch_attachment("other"), old_payload)
def test_migrate_with_concurrent_modification(self):
saved = SavedBasicExport(configuration=_mk_config())
saved.save()
name = saved.get_attachment_name()
new_payload = 'something new'
old_payload = 'something old'
super(BlobMixin, saved).put_attachment(old_payload, name)
super(BlobMixin, saved).put_attachment(old_payload, "other")
saved.save()
self.assertEqual(len(saved._attachments), 2)
def modify(doc):
doc = SavedBasicExport.get(doc._id)
doc.set_payload(new_payload)
doc.save()
self.do_failed_migration({saved: (1, 1)}, modify)
exp = SavedBasicExport.get(saved._id)
self.assertEqual(exp.get_payload(), new_payload)
self.assertEqual(exp.fetch_attachment("other"), old_payload)
def test_migrate_with_concurrent_modification(self):
saved = SavedBasicExport(configuration=_mk_config())
saved.save()
name = saved.get_attachment_name()
new_payload = 'something new'
old_payload = 'something old'
super(BlobMixin, saved).put_attachment(old_payload, name)
super(BlobMixin, saved).put_attachment(old_payload, "other")
saved.save()
self.assertEqual(len(saved._attachments), 2)
def modify(doc):
doc = SavedBasicExport.get(doc._id)
doc.set_payload(new_payload)
doc.save()
self.do_failed_migration({saved: (1, 1)}, modify)
exp = SavedBasicExport.get(saved._id)
self.assertEqual(exp.get_payload(), new_payload)
self.assertEqual(exp.fetch_attachment("other"), old_payload)
def test_migrate_saved_exports(self):
# setup data
saved = SavedBasicExport(configuration=_mk_config())
saved.save()
payload = 'something small and simple'
name = saved.get_attachment_name()
super(BlobMixin, saved).put_attachment(payload, name)
saved.save()
# verify: attachment is in couch and migration not complete
self.assertEqual(len(saved._attachments), 1)
self.assertEqual(len(saved.external_blobs), 0)
with tempdir() as tmp, replattr(SavedBasicExport, "migrating_blobs_from_couch", True):
filename = join(tmp, "file.txt")
# do migration
migrated, skipped = mod.MIGRATIONS[self.slug].migrate(filename)
self.assertGreaterEqual(migrated, 1)
# verify: migration state recorded
mod.BlobMigrationState.objects.get(slug=self.slug)
# verify: migrated data was written to the file
with open(filename) as fh:
lines = list(fh)
doc = {d["_id"]: d for d in (json.loads(x) for x in lines)}[saved._id]
self.assertEqual(doc["_rev"], saved._rev)
self.assertEqual(len(lines), migrated, lines)
# verify: attachment was moved to blob db
exp = SavedBasicExport.get(saved._id)
self.assertNotEqual(exp._rev, saved._rev)
self.assertEqual(len(exp.blobs), 1, repr(exp.blobs))
self.assertFalse(exp._attachments, exp._attachments)
self.assertEqual(len(exp.external_blobs), 1)
self.assertEqual(exp.get_payload(), payload)
def hq_download_saved_export(req, domain, export_id):
export = SavedBasicExport.get(export_id)
# quasi-security hack: the first key of the index is always assumed
# to be the domain
assert domain == export.configuration.index[0]
return couchexport_views.download_saved_export(req, export_id)
def modify(doc):
doc = SavedBasicExport.get(doc._id)
doc.set_payload(new_payload)
doc.save()
def hq_download_saved_export(req, domain, export_id):
export = SavedBasicExport.get(export_id)
# quasi-security hack: the first key of the index is always assumed
# to be the domain
assert domain == export.configuration.index[0]
return couchexport_views.download_saved_export(req, export_id)
def modify(doc):
doc = SavedBasicExport.get(doc._id)
doc.set_payload(new_payload)
doc.save()
