음성 수집 데이터 Pair(WAV/JSON) 일괄 체크 스크립트

dir_pair_check.py

from collections import defaultdict
import os
import time
import re
from datetime import datetime

### 디렉터리 경로
# 역슬래시는 이중으로
# 예시) = "C:\\NIA_COMPLETE_labeling\\"
sourcePath = "/Users/paragonnov/Documents/org/submit"

# 시작 시간 기록
startTime = time.time()

exceptionLogPath = datetime.now().strftime("exception-%Y-%m-%d_%H.%M.%S") + ".log"
exceptionLog = open(exceptionLogPath, "a", encoding="utf-8")

beforePath = "" 

for path, dirs, files in os.walk(sourcePath):

    print(path)

    checkList = defaultdict(int)

    for file in files:
        try:
            regexResult = re.search("(.*)(\..\w*$)", file)

            if not regexResult:
                raise Exception(path + "/" + file + "\n    No RegEx result" + "\n")

            fileNameNoExt = regexResult.group(1)

            if not fileNameNoExt:
                continue
                
            checkList[fileNameNoExt] += 1

        except Exception as e:
            # 예외 로그에 기록
            exceptionLog.write(path + "\n    err: " + str(e) + "\n")

    for key in checkList.keys():
        if checkList[key] and checkList[key] < 2:
            exceptionLog.write("\"" + key + "\" in \"" + path + "\"\n    WARN: file is not a pair\n")

exceptionLog.close()

print("✅ Job is done.\n")
print("🔴 Running Time:", round(time.time() - startTime, 2), "\bs")