search

openeventdata / mordecai

Full text geoparsing as a Python library
MIT License
742 stars 97 forks source link

Mordecai with Flask not working #79

Closed monalipawar closed 4 years ago

monalipawar commented 4 years ago

I am able to get mordecai result when used a standalone Python. But, when I integrate with Flask, I get no result

See below for sample code: from mordecai import Geoparser geo = Geoparser(es_hosts='***',es_port=80,es_ssl=False,country_threshold = 0.5) geo.geoparse("Wuhan")

Expected result below if ran as standalone:

[{'word': 'Wuhan', 'spans': [{'start': 0, 'end': 5}], 'country_predicted': 'CHN', 'country_conf': 0.9881995, 'geo': {'admin1': 'Hubei', 'lat': '30.58333', 'lon': '114.26667', 'country_code3': 'CHN', 'geonameid': '1791247', 'place_name': 'Wuhan', 'feature_class': 'P', 'feature_code': 'PPLA'}}]

If Ran from Flask
[{'word': 'Wuhan', 'spans': [{'start': 0, 'end': 5}], 'country_predicted': '', 'country_conf': 0}] or I get the error tensorflow.python.framework.errors_impl.InvalidArgumentError: Tensor dense_112_input:0, specified in either feed_devices or fetch_devices was not found in the Graph

I looked for this error in forums, where they suggest to use below line graph = tf.get_default_graph(). But, still I am not able to resolve it. Need help ! Thanks in advance

Below forum I am referring https://kobkrit.com/tensor-something-is-not-an-element-of-this-graph-error-in-keras-on-flask-web-server-4173a8fe15e1

monalipawar commented 4 years ago

This issue is been fixed by upgrading Tensorflow from 1.14 to 2.2. Also, added "threaded=False" when calling
app.run(host="0.0.0.0", port= 1234, debug=False, threaded=False)

Closing this. Thank you