AttributeError: '_io.TextIOWrapper' object has no attribute 'nameofColumn' datatable

Question

I have a large file that contains a table. I am going to open it with the datatable package in python. Consider the following code

import datatable as dt
from datatable import f
df = dt.open('Mybigfile.nff')
df = df[(f.c1 > 0) & (f.c1 < 100),:]
print(df.shape)

when I run it like python code.py where code.py contains the above code, everything is fine.

But when I put the code in an flask-api I get the error

AttributeError: '_io.TextIOWrapper' object has no attribute 'c1'

from flask import Flask, request
from flask_restful import Resource, Api
from flask_cors import CORS
from json import dumps
import datatable as dt
from datatable import f

app = Flask(__name__)
CORS(app)
api = Api(app) 

class test(Resource):
    def get(self):
        df = dt.open('Mybigfile.nff')
        df = df[(f.c1 > 0) & (f.c1 < 100),:]
        print(df.shape)
        return 1

api.add_resource(test, '/gettest')
if __name__ == '__main__':
    app.run(host= '0.0.0.0',port=12345)

Answer 1

It seems the error comes from expression (f.c1 > 0). For some reason (and I don't know how it happens), your f variable is not the object imported from datatable, but rather an object of class _io.TextIOWrapper.

The problem can be resolved, however, with one of these workarounds:

1. Import f under a different name, for example from datatable import f as F. Then you'll write df[(F.c1 > 0) & (F.c1 < 100), :].

2. Use f directly from the dt namespace: df[(dt.f.c1 > 0) & (dt.f.c1 < 100), :].

Finally, you can try to find out how f became a TextIOWrapper in the first place. You probably have a code fragment somewhere that reads with open(...) as f: .... When that piece is executed, f becomes redefined.