extract_relations_between_entities.py

import os
import stanza
import chardet
from stanza.server import CoreNLPClient
from stanza.server.client import PermanentlyFailedException

# Download the Stanford CoreNLP package and English models
stanza.install_corenlp()
stanza.download_corenlp_models(model='english-kbp', version='4.2.0')

# Create a client object that uses the CoreNLP server and includes the kbp processor
client = CoreNLPClient(annotators="tokenize,pos,lemma,depparse,sentiment,ner,kbp".split(), kbp_model_name='english-kbp', timeout=600000, memory='6G')

# Create the output directory if it doesn't exist
os.makedirs("C:\\python\\autoindex\\relation_extraction\\", exist_ok=True)

# Iterate through the txt files in C:\python\autoindex\txt_output.
for filename in os.listdir("C:\\python\\autoindex\\txt_output"):
    # Detect the file encoding.
    with open(f"C:\\python\\autoindex\\txt_output\\{filename}", "rb") as f:
        file_bytes = f.read()
        detected = chardet.detect(file_bytes)
        encoding = detected['encoding']

    # Read the file using the detected encoding.
    with open(f"C:\\python\\autoindex\\txt_output\\{filename}", "r", encoding=encoding, errors="replace") as f:
        document = f.read()

    # Try to annotate the document with the client
    try:
        doc = client.annotate(document)
    except PermanentlyFailedException as e:
        # If the server cannot start because of port conflict, try to stop the previous server and start a new one
        if "unable to start the CoreNLP server on port" in str(e):
            print("Trying to stop the previous server and start a new one...")
            client.stop()
            client.start()
            doc = client.annotate(document)
        else:
            # If the error is not related to port conflict, raise it
            raise e

    # Extract the relations from the document.
    relations = []
    for sentence in doc.sentence:
        relations.extend(sentence.relation)

    # Split the filename and the extension
    filename, extension = os.path.splitext(filename)

    # Write the relations to a file using the filename without the extension
    with open(f"C:\\python\\autoindex\\relation_extraction\\{filename}_relations.txt", "w") as f:
        for relation in relations:
            f.write(str(relation) + "\n")

# Close the client when done
client.close()