output gets generated to stdout

[yyellin]

During initialization of stanfordnlp.Pipeline() output gets generated to stdout - this makes it difficult to integrate within command line utilities that too send their output to stdout.
So for example, assuming a python script 'process_text' I would like for it to accept input from stdin, and sent output to stdout, so that I can run something like:
grep "something" input_file | process_text
Currently this won't work.

[dlazesz]

Very sophisticated workaround:

import os
import sys
my_stdout = sys.stdout  # Save STDOUT
sys.stdout = open(os.devnull, 'w', encoding='UTF-8')  # Overload STDOUT to /dev/null


import stanfordnlp

nlp = stanfordnlp.Pipeline(processors='tokenize,mwt,pos,lemma,depparse', lang='en', models_dir='.')  # Shut up!
doc = nlp('Your text here!')
sys.stdout = my_stdout  # Restore STDOUT
for i, sentence in enumerate(doc.sentences):
    print(f"Sentence {i}:")
    sentence.print_tokens()

Should be done internally with some config switch.

[qipeng]

Good catch! This is definitely an area of improvement for us. Will add this to our TODO list for future releases.

[qipeng]

Now all logging has been moved to Python's logging package and controlled at the Pipeline level, part of the new v1.0.0 release: https://github.com/stanfordnlp/stanza/releases/tag/v1.0.0

Closing now!

[vegai]

You should probably not define a global logger in __init__.py because that seems to overwrite whatever loggers are defined by an application when it runs import stanza