def test_queue_failed_task(app, http_client, base_url):
task_id = 'my_task'
task = dict(state='FAILED')
app.service_manager.pending_tasks[task_id] = task
rsp = yield http_client.fetch(base_url + '/queue/{}'.format(task_id))
assert rsp.code == 200
assert json.loads(to_string(rsp.body))['state'] == 'FAILED'
def poll():
while True:
rsp2 = yield http_client.fetch(
'{}{}'.format(base_url, rsp. headers['Content-Location']),
method='GET',
follow_redirects=False)
assert rsp2.code != 303
state = json.loads(to_string(rsp2.body))['state']
if state == 'FAILED':
break
def test_get_existing_service(app, http_client, base_url):
app.service_manager.db.execute("DELETE FROM entities; "
"DELETE FROM index_url; "
"DELETE FROM index_service_id;")
row = app.service_manager.db.put(dict(
url='url',
state='PENDING'
))
query = "select added_id from entities where HEX(id)='%s';" % (
to_string(row['id']))
added_id = app.service_manager.db.query(query)
req = HTTPRequest(url=base_url+'/api/v1/service/{}'.format(
added_id[0]['added_id']))
rsp = yield http_client.fetch(req)
assert rsp.code == 200
assert json.loads(to_string(rsp.body)) == \
{"state": "PENDING", "url": "url"}
def test_get_services(app, http_client, base_url):
app.service_manager.db.execute("DELETE FROM entities; "
"DELETE FROM index_url; "
"DELETE FROM index_service_id;")
app.service_manager.db.put(dict(
url='url',
state='PENDING'
))
req = HTTPRequest(url=base_url+'/api/v1/services')
rsp = yield http_client.fetch(req)
assert rsp.code == 200
assert json.loads(to_string(rsp.body)) == \
{"services": [{"url": "url",
"state": "PENDING"}]}
def create_service(service, config):
logging.debug('service: {} config: {}'.format(service, config))
# sanity checks
def check_field_present(field):
if field not in service:
raise ValueError("Request doesn't have the {} field".format(field))
check_field_present('url')
logging.debug('Adding service: {}'.format(service))
tmpdir = tempfile.mkdtemp()
def run_helper(cmd, env=None):
try:
logging.debug('Running cmd: {}'.format(cmd))
return run_cmd(cmd,
env=env,
stdout_callback=None,
stderr_callback=None)
except CmdException:
raise
# download file
run_helper('cd {}; curl -O {}'.format(tmpdir, service['url']))
filename = tmpdir + '/' + service['url'].split('/')[-1]
filetype = magic.from_file(filename)
filebase = filename.split('.')[-1]
clean_filetype = to_string(filetype).lower()
if 'pdf' in clean_filetype:
run_helper('cd {}; convert {} {}'.format(tmpdir,
filename,
filebase + '.png'))
filename = filebase + '.png'
elif 'jpeg' in clean_filetype or 'png' in clean_filetype:
pass
else:
raise ValueError('File {} unsupported'.format(filename))
# run tesseract
output = tmpdir + '/' + str(uuid.uuid4())
tesseract_cmd = \
'cd {}; tesseract {} {}'.format(tmpdir, filename, output)
run_helper('cd {}; {}'.format(tmpdir, tesseract_cmd))
# return recognized text
return open(output + '.txt').read()
def create_service(service, config):
logging.debug('service: {} config: {}'.format(service, config))
# sanity checks
def check_field_present(field):
if field not in service:
raise ValueError("Request doesn't have the {} field".format(field))
check_field_present('url')
logging.debug('Adding service: {}'.format(service))
tmpdir = tempfile.mkdtemp()
def run_helper(cmd, env=None):
try:
logging.debug('Running cmd: {}'.format(cmd))
return run_cmd(cmd,
env=env,
stdout_callback=None,
stderr_callback=None)
except CmdException:
raise
# download file
run_helper('cd {}; curl -O {}'.format(tmpdir, service['url']))
filename = tmpdir + '/' + service['url'].split('/')[-1]
filetype = magic.from_file(filename)
filebase = filename.split('.')[-1]
clean_filetype = to_string(filetype).lower()
if 'pdf' in clean_filetype:
run_helper('cd {}; convert {} {}'.format(tmpdir, filename,
filebase + '.png'))
filename = filebase + '.png'
elif 'jpeg' in clean_filetype or 'png' in clean_filetype:
pass
else:
raise ValueError('File {} unsupported'.format(filename))
# run tesseract
output = tmpdir + '/' + str(uuid.uuid4())
tesseract_cmd = \
'cd {}; tesseract {} {}'.format(tmpdir, filename, output)
run_helper('cd {}; {}'.format(tmpdir, tesseract_cmd))
# return recognized text
return open(output + '.txt').read()
