Create utils/config_utils.py
Browse files- utils/config_utils.py +37 -0
utils/config_utils.py
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import logging
|
| 3 |
+
|
| 4 |
+
def set_tokenizer_parallelism(enabled=False):
|
| 5 |
+
"""
|
| 6 |
+
Configure tokenizer parallelism to avoid fork-related warnings.
|
| 7 |
+
"""
|
| 8 |
+
os.environ["TOKENIZERS_PARALLELISM"] = str(enabled).lower()
|
| 9 |
+
|
| 10 |
+
logger = logging.getLogger("ConfigUtils")
|
| 11 |
+
if not logger.handlers:
|
| 12 |
+
logging.basicConfig(level=logging.INFO)
|
| 13 |
+
logger.info(f"Tokenizers parallelism set to: {enabled}")
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
def load_urls_from_file(file_path, logger_name="ConfigUtils"):
|
| 17 |
+
"""
|
| 18 |
+
Load URLs from a text file, ignoring empty lines and comments.
|
| 19 |
+
"""
|
| 20 |
+
logger = logging.getLogger(logger_name)
|
| 21 |
+
if not logger.handlers:
|
| 22 |
+
logging.basicConfig(level=logging.INFO)
|
| 23 |
+
|
| 24 |
+
urls = []
|
| 25 |
+
try:
|
| 26 |
+
with open(file_path, 'r') as f:
|
| 27 |
+
for line in f:
|
| 28 |
+
line = line.strip()
|
| 29 |
+
if line and not line.startswith('#'):
|
| 30 |
+
urls.append(line)
|
| 31 |
+
logger.info(f"Loaded {len(urls)} URLs from {file_path}")
|
| 32 |
+
for i, url in enumerate(urls):
|
| 33 |
+
logger.debug(f" URL {i+1}: {url}")
|
| 34 |
+
return urls
|
| 35 |
+
except Exception as e:
|
| 36 |
+
logger.error(f"Error loading URLs from {file_path}: {str(e)}")
|
| 37 |
+
return []
|