Use `yaml` for `.info` files.

[scarlehoff]

LHAPDF uses yaml for .info files so we should do the same everywhere.

Some places where we are just reading them and looking at them like some old fortran programmer are:

nnpdf/validphys2/src/validphys/lhio.py

Line 248 in c4084de

with original_info.open() as orig_file, new_info.open('w') as new_file:

nnpdf/validphys2/src/validphys/lhio.py

Line 305 in c4084de

with open(base/f'{pdf}.info', 'r') as inn, \

nnpdf/validphys2/src/validphys/scripts/postfit.py

Line 59 in c4084de

with open(info_path, 'r+') as f:

Some checks that we need to perform:

• Can all info files for all PDFs in the LHAPDF server be read as yaml files?
• If they are dumped by vp as yaml files, can they be read back by LHAPDF?

Some caveats that I would add:

• The output of lists should styled as: Flavors: [-5, -4, -3, -2, -1, 21, 1, 2, 3, 4, 5]
• The way to solve this issue is having a function write_info in lhaindex or lhio that writes the yaml file with the right options.

Continuation of the discussion of #1551

[alecandido]

• The output of lists should styled as: Flavors: [-5, -4, -3, -2, -1, 21, 1, 2, 3, 4, 5]

In order to get something like this with pyyaml, the only simple option you have is using something like yadism is doing

    return yaml.dump(out, stream, default_flow_style=None)

and hope for the best.

https://github.com/N3PDF/yadism/blob/1be26b69c011352b716298819678e1dce8a902d3/src/yadism/output.py#L290

(the alternative being writing your custom representer, that is painful and I'm not sure it can bring you to an end)

[Zaharid]

It is a bit silly but

In [1]: from reportengine.compat import yaml

In [2]: res = yaml.round_trip_load("""
   ...: a: [x, y , z]
   ...: b:
   ...:   - 1
   ...:   - 2
   ...:   - 4
   ...: """)

In [3]: proto = yaml.round_trip_load("[patata]")

In [4]: proto[:] = ["actual", "values"]

In [5]: res["flavour"] = proto

In [6]: yaml.round_trip_dump(res)
Out[6]: 'a: [x, y, z]\nb:\n- 1\n- 2\n- 4\nflavour: [actual, values]\n'

In [7]: print(_)
a: [x, y, z]
b:
- 1
- 2
- 4
flavour: [actual, values]

[alecandido]

What kind of object is res? I guess not a dict...

Either that or proto is not a list. Or maybe both.

[Zaharid]

They are wrappers over dict and list with additional information on formatting. proto = yaml.round_trip_load("[patata]") is a way to tell it I want a plain formatted list rather than a line by line one.