如何避免使用日志记录将转义字符输出到日志文件？

Question

我正在尝试为 INFO、WARNING 等级别名称着色，但问题是当消息被记录到日志文件时 - 有奇怪的字符。但是，它可以完美地打印到控制台。

代码：

import logging
import sys

logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s [%(levelname)s] %(message)s",
    handlers=[
        logging.FileHandler("log.txt"),
        logging.StreamHandler(sys.stdout)
    ]
)

 logging.addLevelName(logging.INFO, "3[1;31m%s3[1;0m" % logging.getLevelName(logging.INFO))

logging.info("This is just an information for you")

这会用颜色打印级别名称“INFO”，但输出是：

2022-04-05 02:44:52,741 [[1;31mINFO[1;0m] This is just an information for you

如何解决这个问题？

编辑：

logging.addLevelName(logging.INFO, "3[1;94m%s3[1;0m" % logging.getLevelName(logging.INFO))
logging.addLevelName(logging.WARNING, "3[1;93m%s3[1;0m" % logging.getLevelName(logging.WARNING))
logging.addLevelName(logging.ERROR, "3[1;91m%s3[1;0m" % logging.getLevelName(logging.ERROR))

Answer 1

原始答案（不是 fully-functional；查看已接受的答案）：

这些字符出现在 gui 文本编辑器中（但不出现在命令行中——取决于所使用的 shell），因为它们不理解 ASCII 转义序列——除非它们被专门配置为所以。

您可以添加代码以在将这些字符写入 log 文件时删除这些字符：

import logging
import logging.config
import re
import sys

from argparse import Namespace
from pathlib import Path
from typing import Optional

class NoColourFormatter(logging.Formatter):

    """Log formatter that strips terminal colour escape codes from the log message."""

    ANSI_RE = re.compile(r"\x1b\[[0-9;]*m")

    def format(self, record):
        """Return logger message with terminal escapes removed."""
        return "[%s] [%s]: %s" % (
            record.levelname,
            record.name,
            re.sub(self.ANSI_RE, "", record.msg % record.args),
        )

并像这样使用它：

# same file as above

def config_logger(args: Optional[Namespace] = None) -> None:

    logger = logging.getLogger(__package__)

    # other code

    logformatter = NoColourFormatter()
    loghandler = logging.FileHandler(args.logfile, mode="w", encoding="utf8")

    loghandler.setFormatter(logformatter)
    logger.addHandler(loghandler)

Answer 2

更新为每个级别分配了单独的颜色。要更改颜色，只需将 set_colours() 中的相应数字替换为所需的值即可。

这里有一些东西可以直接用来生成你想要的日志：

import logging
import re
import time
import sys

def set_colour(level):
    """
    Sets colour of text for the level name in
    logging statements using a dispatcher.
    """
    escaped = "[3[1;%sm%s3[1;0m]"
    return {
        'INFO': lambda: logging.addLevelName(logging.INFO, escaped % ('94', level)),
        'WARNING': lambda: logging.addLevelName(logging.ERROR, escaped % ('93', level)),
        'ERROR': lambda: logging.addLevelName(logging.WARNING, escaped % ('91', level))
    }.get(level, lambda: None)()


class NoColorFormatter(logging.Formatter):
    """
    Log formatter that strips terminal colour
    escape codes from the log message.
    """

    # Regex for ANSI colour codes
    ANSI_RE = re.compile(r"\x1b\[[0-9;]*m")

    def format(self, record):
        """Return logger message with terminal escapes removed."""
        return "%s %s %s" % (
            time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()),
            re.sub(self.ANSI_RE, "", record.levelname),
            record.msg,
        )

# Create logger
logger = logging.getLogger(__package__)

# Create formatters
logformatter = NoColorFormatter()
colorformatter = logging.Formatter("%(asctime)s %(levelname)s %(message)s")

# Set logging colours
for level in 'INFO', 'ERROR', 'WARNING':
    set_colour(level)

# Set logging level
logger.setLevel(logging.INFO)

# Set log handlers
loghandler = logging.FileHandler("log.txt", mode="w", encoding="utf8")
streamhandler = logging.StreamHandler(sys.stdout)

# Set log formatters
loghandler.setFormatter(logformatter)
streamhandler.setFormatter(colorformatter)

# Attach log handlers to logger
logger.addHandler(loghandler)
logger.addHandler(streamhandler)

# Example logging statements
logging.info("This is just an information for you")
logging.warning("This is just an information for you")
logging.error("This is just an information for you")

输出到stdout时有颜色，但在日志文件中没有。此输出的日志语句与原始代码中的日志语句之间的唯一区别是 %(asctime)s 似乎在 logging.Formatter 对象之外不起作用，因此我使用了 time反而;输出是一样的，除了没有任何毫秒。

Answer 3

我不会在全局范围内覆盖关卡名称，而是在特定的格式化程序中应用着色，然后在需要着色的处理程序上设置该格式化程序。

class ColouredFormatter(logging.Formatter):
    _lookup = {
        logging.INFO: "3[1;94m%s3[1;0m",
        logging.WARNING: "3[1;93m%s3[1;0m",
        logging.ERROR: "3[1;91m%s3[1;0m",
    }

    def format(self, record):
        result = super().format(record)
        # Operate on the result rather than record attributes to
        # avoid corrupting output for subsequent loggers.
        formatted_levelname = self._lookup[record.levelno] % record.levelname
        result = re.sub(rf'\[({record.levelname})\]', formatted_levelname, result, count=1)
        return result


streamhandler = logging.StreamHandler(sys.stdout)
streamhandler.setFormatter(ColouredFormatter(fmt="%(asctime)s [%(levelname)s] %(message)s"))

logging.basicConfig(
    level=logging.INFO,
    format="%(asctime)s [%(levelname)s] %(message)s",
    handlers=[
        streamhandler,
        logging.FileHandler("so71744086log.txt"),
    ]
)

如何避免使用日志记录将转义字符输出到日志文件？

How to avoid escape characters outputted to log file with logging?

python

logging

原始答案（不是 fully-functional；查看已接受的答案）：