使用ruamel.yaml修改YAML后输出丢失注释的问题

用ruamel.yaml加载带注释的YAML文件，改完字段值再写回去，结果注释全没了——字段更新、引号缩进都正常，就是注释丢了，这情况怎么处理？

你的原始代码 #

import ruamel.yaml

source = open("source.yaml", "r")
target = open("target.yaml", "r")

sourceValues = ruamel.yaml.load(source, ruamel.yaml.RoundTripLoader, preserve_quotes=True)
targetValues = ruamel.yaml.load(target, ruamel.yaml.RoundTripLoader, preserve_quotes=True)

source.close()
target.close()

# 修改target的属性
targetValues['test']['something'] = sourceValues['test']['something']

with open('target.yaml', 'w') as conf:
  ruamel.yaml.dump(targetValues, conf, ruamel.yaml.RoundTripDumper)

原始带注释的YAML（target.yaml） #

test:
  # some comment
  something: "something" # another comment
  else: 123

执行后的输出 #

test:
  something: "something-updated"
  else: 123456

为啥会丢注释？ #

你直接用targetValues['test']['something'] = ...赋值的时候，是把原来的节点对象整个替换掉了。ruamel.yaml的RoundTripLoader加载出来的节点里是带着注释信息的，一替换，这些注释数据就被扔了，自然写回去的时候没注释。

怎么解决？ #

别直接替换节点，只修改原有节点的value属性就行，这样注释会跟着保留。另外建议用with语句管文件，不用手动close，更省心。

修改后的代码：

import ruamel.yaml

# 用with自动处理文件关闭
with open("source.yaml", "r") as source, open("target.yaml", "r") as target:
    sourceValues = ruamel.yaml.load(source, ruamel.yaml.RoundTripLoader, preserve_quotes=True)
    targetValues = ruamel.yaml.load(target, ruamel.yaml.RoundTripLoader, preserve_quotes=True)

# 只改节点的值，原节点的注释不动
targetValues['test']['something'].value = sourceValues['test']['something'].value

with open('target.yaml', 'w') as conf:
    # dump的时候加上preserve_quotes，确保引号格式不变
    ruamel.yaml.dump(targetValues, conf, ruamel.yaml.RoundTripDumper, preserve_quotes=True)

额外补充 #

如果要手动加注释或者调整注释，可以用ruamel.yaml的专用API：

# 在something字段前面加注释
targetValues['test'].yaml_set_comment_before_key('something', '# some comment')
# 给something字段加行尾注释
targetValues['test']['something'].yaml_add_eol_comment('another comment', column=20)

内容的提问来源于stack exchange，提问作者Ventrue