Fix the bug of reading string NA as NaN in the function exists_qlib_data.

[OzzyXu]

Fix the bug of reading string NA as NaN of exists_qlib_data in /qlib/utils/__init__.py.

Description

Nano Labs Ltd is a new Nasdaq-listing company with the ticker name NA from August 1, 2022. The default na_value list of pd.read_csv includes "NA". Changed the default reading behavior of pd.read_csv in exists_qlib_data by adding keep_default_na=False. Removed two values ("NA" and "NULL") from the default NA list while reading the first column of "all.txt", which normally are all strings.

Motivation and Context

To fix the bug in #1720.

How Has This Been Tested?

• Pass the test by running: pytest qlib/tests/test_all_pipeline.py under upper directory of qlib.
• If you are adding a new feature, test on your own test scripts.

Screenshots of Test Results (if appropriate):

1. Pipeline test:
   

2. Your own tests:
   
   Place the attached file all.txt under \qlib_data\us_data_made\instruments and test with the following code

import qlib
import pandas as pd
import sys, site
from pathlib import Path
from qlib.utils import exists_qlib_data
from qlib.constant import REG_US

scripts_dir = Path.cwd().parent.joinpath("scripts")
provider_uri = "./qlib_data/us_data_made"  # target_dir
if not exists_qlib_data(provider_uri):
    print(f"Qlib data is not found in {provider_uri}")
    sys.path.append(str(scripts_dir))
    from get_data import GetData

    GetData().qlib_data(target_dir=provider_uri, region=REG_US)
qlib.init(provider_uri=provider_uri, region=REG_US)

Types of changes

• Fix bugs
• Add new feature
• Update documentation

[OzzyXu]

@microsoft-github-policy-service agree

[OzzyXu]

@SunsetWolf Hey, can I ask why all tests from sources other than slow failed? Do I need to take care of this? Thank you.