yaml.dump and indentation:

[lal.george@…]

In python, when I have :

---

y = {'x' : 3, z : [{'val' : 3}, {'val' : 4}]} yaml.dump(y, default_flow_style=False)

---

The result is: 'x: 3\nz:\n- val: 3\n- val: 4\n'

I believe there should be a space before each hypen.

Best,

Lal

[xi]

The output is correct, see the production l-block-seq-entry(n,c) and Example 10.4:

 http://yaml.org/spec/cvs/current.html#l-block-seq-entry(n,c)

[lageorge@…]

Unfortunately, I don't understand the notation very well; but l-block-seq-entry(n,c) ::= s-indent(seq-spaces(n,c)) "-" ...

Since a new block is being started, won't there be an s-indent?

Best,

Lal

p.s. The java version of yaml does not like the dump produced by PyYaml?.

[xi]

Read the next paragraph after the production:

People read the “-” character as part of the indentation. Hence, block sequence entries require one less space of indentation, unless the block sequence is nested within another block sequence (hence the need for the block-in context and block-out context).

See also the example in the specification:

block:
- one
-
·- two

P.S. The YAML loader in Java must be broken then.

[anonymous]

Thanks for the quick response ...

I believe you and PyYaml? are correct. The Java implementation is returning null for z in:

'x: 3\nz:\n- val: 3\n- val: 4\n'

Do you know a work around I could use? Perhaps I could turn this feature off, by always having PyYaml? emit " -" wherever it emits "-".

Thanks for your help.

Best,

Lal

[xi]

You can use the following code:

import yaml

class MyDumper(yaml.Dumper):

    def increase_indent(self, flow=False, indentless=False):
        return super(MyDumper, self).increase_indent(flow, False)

y = {'x' : 3, 'z' : [{'val' : 3}, {'val' : 4}]}

print yaml.dump(y, Dumper=MyDumper, default_flow_style=False)