intention_relations_count.py

import os
from tqdm import tqdm
from collections import Counter

def file_len_with_relation(filename):
    sync_count = 0
    async_count = 0
    contingency_count = 0

    with open(filename) as f:
        for i, line in enumerate(f):

            if "and simultaneously," in line:
                sync_count += 1
            
            if "mainly because" in line:
                contingency_count += 1
                
            
            if "usually after" in line:
                async_count += 1
    return sync_count, async_count, contingency_count
          
                


total_sync = 0
total_async = 0
total_contingency = 0

for file in tqdm(os.listdir("/data/jbai/session_data/vera_relations_finetuned")):
    if file.endswith(".csv"):
        sync, async_, contingency = file_len_with_relation(f"/data/jbai/session_data/vera_relations_finetuned/{file}")
        total_sync += sync
        total_async += async_
        total_contingency += contingency

    print("Total sync: ", total_sync)
    print("Total async: ", total_async)
    print("Total contingency: ", total_contingency)

    print("Total relations: ", total_sync + total_async + total_contingency)