如何修改Python logging模块的路径分隔符？兼谈跨OS适配问题

问题描述 #

想用Python的logging模块生成自定义JSON格式日志，需要记录pathname变量，但Windows下logging输出的路径用反斜杠\，导致JSON解析时出现Invalid escape character错误。

问题一：路径分隔符是否依赖操作系统？ #

是的，路径分隔符完全由操作系统决定：

• Windows系统默认使用**反斜杠\**作为路径分隔符
• Unix类系统（Linux、macOS等）默认使用**正斜杠/**作为路径分隔符
  代码部署到Unix系统后，pathname会自动使用正斜杠，不会出现JSON转义问题，但Windows环境下需要手动处理。

问题二：如何修改分隔符解决JSON错误？ #

可以通过两种方式处理，目标是将路径中的反斜杠替换为正斜杠，或者转义为双反斜杠（符合JSON语法要求）：

测试代码 #

config.json #

{
    "version": 1,
    "disable_existing_loggers": true,
    "formatters": {
        "standard": {
            "format": "{\"timestamp\":%(created)f,\"time\":\"%(asctime)s\",\"path\":\"%(pathname)s\",\"file\":\"%(filename)s\",\"line\":%(lineno)d,\"level\":\"%(levelname)s\",\"text\":\"%(message)s\"}",
            "datefmt": "%Y/%m/%d-%H:%M:%S"
        }
    },
    "handlers": {
        "console": {
            "class": "logging.StreamHandler",
            "formatter": "standard",
            "level": "DEBUG",
            "stream": "ext://sys.stdout"
        }
    },
    "loggers": {
        "": {
            "handlers": ["console"],
            "level": "DEBUG",
            "propagate": false
        }
    }
}

example.py #

import logging
import json
import logging.config

with open("./config.json", "r", encoding="UTF_8") as file:
    config = json.load(file)
logging.config.dictConfig(config)
logger = logging.getLogger(__name__)
logger.info("This is an info log.")

当前输出（Windows下） #

{"timestamp":1726603464.416349,"time":"2024/09/17-23:34:24","path":"d:\python\project\example.py","file":"example.py","line":9,"level":"INFO","text":"This is an info log."}

期望输出 #

# 方式一：转义成双反斜杠
{"timestamp":1726603464.416349,"time":"2024/09/17-23:34:24","path":"d:\\python\\project\\example.py","file":"example.py","line":9,"level":"INFO","text":"This is an info log."}

# 方式二：替换为正斜杠
{"timestamp":1726603464.416349,"time":"2024/09/17-23:34:24","path":"d:/python/project/example.py","file":"example.py","line":9,"level":"INFO","text":"This is an info log."}

解决方案 #

方案1：自定义Formatter类处理路径分隔符 #

创建自定义Formatter，在格式化日志时替换路径中的反斜杠：

import logging
import json
import logging.config

class JsonFormatter(logging.Formatter):
    def format(self, record):
        # 替换反斜杠为正斜杠，若要转成双反斜杠则用replace("\\", "\\\\")
        record.pathname = record.pathname.replace("\\", "/")
        return super().format(record)

# 加载配置并替换formatter
with open("./config.json", "r", encoding="UTF_8") as file:
    config = json.load(file)

logging.config.dictConfig(config)
# 为所有handler设置自定义Formatter
for handler in logging.getLogger().handlers:
    handler.setFormatter(JsonFormatter(
        fmt=config["formatters"]["standard"]["format"],
        datefmt=config["formatters"]["standard"]["datefmt"]
    ))

logger = logging.getLogger(__name__)
logger.info("This is an info log.")

方案2：用json.dumps生成合法JSON（推荐） #

避免手动拼接JSON字符串，直接构造字典后用json.dumps序列化，自动处理转义：

import logging
import json
import logging.config

class JsonFormatter(logging.Formatter):
    def format(self, record):
        log_dict = {
            "timestamp": record.created,
            "time": self.formatTime(record, self.datefmt),
            "path": record.pathname.replace("\\", "/"),  # 替换为正斜杠，去掉该行则json.dumps自动转义反斜杠
            "file": record.filename,
            "line": record.lineno,
            "level": record.levelname,
            "text": record.getMessage()
        }
        return json.dumps(log_dict)

# 简化配置，无需在config中写JSON字符串
config = {
    "version": 1,
    "disable_existing_loggers": True,
    "handlers": {
        "console": {
            "class": "logging.StreamHandler",
            "level": "DEBUG",
            "stream": "ext://sys.stdout"
        }
    },
    "loggers": {
        "": {
            "handlers": ["console"],
            "level": "DEBUG",
            "propagate": False
        }
    }
}

logging.config.dictConfig(config)
for handler in logging.getLogger().handlers:
    handler.setFormatter(JsonFormatter(datefmt="%Y/%m/%d-%H:%M:%S"))

logger = logging.getLogger(__name__)
logger.info("This is an info log.")

该方案避免了手动拼接JSON的转义错误，稳定性更高。

内容的提问来源于stack exchange，提问作者Aria AAs