Use literal style for just multiline strings in ruamel.yaml

Question

I would like to have a custom ruamel.yaml dumper that uses Literal style for all multiline strings and the default style otherwise. For example:

import sys
import ruamel.yaml

data = {"a": "hello", "b": "hello\nthere\nworld"}

print("Default style")
yaml = ruamel.yaml.YAML()
yaml.dump(data, sys.stdout)

print()
print("style='|'")
yaml = ruamel.yaml.YAML()
yaml.default_style = "|"
yaml.dump(data, sys.stdout)

This produces:

Default style
a: hello
b: "hello\nthere\nworld"

style='|'
"a": |-
  hello
"b": |-
  hello
  there
  world

My desired output is:

a: hello
b: |-
  hello
  there
  world

Answer 1

There are multiple ways to achieve what you want. If you have control over building up the data structure, it is often easiest to add a LiteralScalarString if appropriate:

import sys
import ruamel.yaml

def lim(s):  # literal if multi-line
    if '\n' in s:
        return ruamel.yaml.scalarstring.LiteralScalarString(s)
    return s

data = {'a': lim('hello'), 'b': lim('hello\nthere\nworld')}

    
yaml = ruamel.yaml.YAML()
yaml.dump(data, sys.stdout)

which gives:

a: hello
b: |-
  hello
  there
  world

This gives you easy fine control over what gets dumped as literal style.

If you don't add all the data individually (but e.g. read them from a JSON file), you can walk over your data structure after it is fully constructed and update it in place:

import sys
import ruamel.yaml

def tmtl(d):
    """translate multi-line to literal,
       only acts on dict values and sequence items, not on keys
    """
    if isinstance(d, dict):
        for k, v in d.items():
            if isinstance(v, str) and '\n' in v:
                d[k] = ruamel.yaml.scalarstring.LiteralScalarString(v)
            else:
                tmtl(v)
    elif isinstance(d, list):
        for idx, item in enumerate(d):
            if isinstance(item, str) and '\n' in item:
                d[idx] = ruamel.yaml.scalarstring.LiteralScalarString(item)

data = {'a': 'hello', 'b': 'hello\nthere\nworld'}
tmtl(data)
    
yaml = ruamel.yaml.YAML()
yaml.dump(data, sys.stdout)

which gives:

a: hello
b: |-
  hello
  there
  world

If you cannot update your data, you could rewrite tmtl in the program above so it builds a new data structure and returns that, but at that point it is IMO easier to change the representer:

import sys
import ruamel.yaml

CKS = ruamel.yaml.comments.CommentedKeySeq  # so you can have sequences as keys in a mapping

class MyRepresenter(ruamel.yaml.representer.RoundTripRepresenter):
    def represent_str(self, s):
        if '\n' in s:
            return self.represent_scalar('tag:yaml.org,2002:str', s, style='|')
        return self.represent_scalar('tag:yaml.org,2002:str', s)

MyRepresenter.add_representer(str, MyRepresenter.represent_str)

data = {'a': 'hello', 'b': 'hello\nthere\nworld', CKS((1, 2)): ['nested works\nas well\n\n']}
    
yaml = ruamel.yaml.YAML()
yaml.Representer = MyRepresenter
yaml.dump(data, sys.stdout)

which gives:

a: hello
b: |-
  hello
  there
  world
[1, 2]:
- |+
  nested works
  as well

...

As you can see the trailing newlines of the final literal style scalar automatically causes the chomping indicator to change from strip (-) to keep (+) and the explicit document end marker (...) to appear.