Try to fix encoding error
This commit is contained in:
@@ -7,10 +7,10 @@ from typing import Optional
|
|||||||
import click
|
import click
|
||||||
from tqdm import tqdm
|
from tqdm import tqdm
|
||||||
|
|
||||||
|
from common.log import logger
|
||||||
|
from common.stdout_wrapper import SAFE_STDOUT
|
||||||
from config import config
|
from config import config
|
||||||
from text.cleaner import clean_text
|
from text.cleaner import clean_text
|
||||||
from common.stdout_wrapper import SAFE_STDOUT
|
|
||||||
from common.log import logger
|
|
||||||
|
|
||||||
preprocess_text_config = config.preprocess_text_config
|
preprocess_text_config = config.preprocess_text_config
|
||||||
|
|
||||||
@@ -89,7 +89,10 @@ def preprocess(
|
|||||||
)
|
)
|
||||||
)
|
)
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
logger.error(f"An error occurred at line:\n{line.strip()}\n{e}")
|
logger.error(
|
||||||
|
f"An error occurred at line:\n{line.strip()}\n{e}",
|
||||||
|
encoding="utf-8",
|
||||||
|
)
|
||||||
with open(error_log_path, "a", encoding="utf-8") as error_log:
|
with open(error_log_path, "a", encoding="utf-8") as error_log:
|
||||||
error_log.write(f"{line.strip()}\n{e}\n\n")
|
error_log.write(f"{line.strip()}\n{e}\n\n")
|
||||||
error_count += 1
|
error_count += 1
|
||||||
@@ -172,8 +175,9 @@ def preprocess(
|
|||||||
f"An error occurred in {error_count} lines. Please check {error_log_path} for details."
|
f"An error occurred in {error_count} lines. Please check {error_log_path} for details."
|
||||||
)
|
)
|
||||||
raise Exception(
|
raise Exception(
|
||||||
f"An error occurred in {error_count} lines. Please check {error_log_path} for details."
|
f"An error occurred in {error_count} lines. Please check `Data/you_model_name/text_error.log` file for details."
|
||||||
)
|
)
|
||||||
|
# 何故か{error_log_path}をraiseすると文字コードエラーが起きるので上のように書いている
|
||||||
else:
|
else:
|
||||||
logger.info(
|
logger.info(
|
||||||
"Training set and validation set generation from texts is complete!"
|
"Training set and validation set generation from texts is complete!"
|
||||||
|
|||||||
Reference in New Issue
Block a user