Feature request: User specifiable dictionaries

[mortsynluap]

I prefer to keep data in YAML and then convert it to JSON for use in particular applications. Unfortunately the ordering of dictionary entries is not preserved.

I would like to be able to specify the class that is to be used in place of a dict, e.g., an OrderedDict or a Namespace (attached) that allows dotted notation for references.
namespace.zip

[ingydotnet]

The ordering of dict keys can be disabled in the next release. 5.1 should be out in a week or 2. 5.1b1 is already out and 5.1b2 (with windows wheels goes out tonight).

[cknoll]

Does this mean the sorting of maps is retained reliably (e.g. accross platforms) when converting from yaml to python?

Example :

yaml.__version__                                                                                                                                                        
# Out[6]: '5.3.1'

yaml.load("""
  a: 1
  b:
    x: 3
    d: 4
""", Loader=yaml.SafeLoader)
# Out[7]: {'a': 1, 'b': {'x': 3, 'd': 4}}

Can I rely on the order of the b-dict? This seems relevant due to the existence of the package oyaml which "is a drop-in replacement for PyYAML which preserves dict ordering".

Additionally, as the original poster suggested OrderedDict I would be interested for which python version this question was relevant as since python3.7 insertion order is maintained inside dict (GvR-Mail, SO-answer.

So maybe this Issue can be closed?

[mortsynluap]

I wanted to se my own dictionary class (Namespace) that provides "dotted" access to dictionaries, i.e., assuming your example was read into variable Y, I could access fields using Y.b.d vs. Y['b']['d'].

Re-reading the documentation and searched around some, I came up with the following that does what I was asking for:

yaml.add_constructor(yaml.resolver.BaseResolver.DEFAULT_MAPPING_TAG, lambda loader, node: Namespace(*loader.construct_pairs(node)))

Thanks

[cknoll]

So I guess this issue could be closed.

@mortsynluap By the way: The Namespace-type seems interesting. Where do you import it from?

[mortsynluap]

Taking care of my email backlog 🙁 import yaml from . import Namespace logcfg = Namespace.Namespace(yaml.load(open("logging.config.yaml", "rt", encoding="utf-8"), Loader=yaml.Loader)) logcfg.incremental = True logcfg.handlers.console.level = logging.getLevelName(max(40 - (10 * ARGS.verbose), 0))

…

________________________________ From: C. Knoll <notifications@github.com> Sent: Sunday, October 18, 2020 2:48 PM To: yaml/pyyaml <pyyaml@noreply.github.com> Cc: mortsynluap <pnystrom@outlook.com>; Mention <mention@noreply.github.com> Subject: Re: [yaml/pyyaml] Feature request: User specifiable dictionaries (#252) So I guess this issue could be closed. @mortsynluap<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fmortsynluap&data=04%7C01%7C%7Cec652e2818c145512ff008d873af9d33%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637386545370727708%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=XRAdVrp9I%2BWPUEttZjtTng0UZCCR5emUtCdEFnlYQd4%3D&reserved=0> By the way: The Namespace-type seems interesting. Where do you import it from? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fyaml%2Fpyyaml%2Fissues%2F252%23issuecomment-711428527&data=04%7C01%7C%7Cec652e2818c145512ff008d873af9d33%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637386545370737702%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=u28w8MDk%2FP3kZP9Nm9Z7XHDA1uTLa8CCxRJ0roJeJAM%3D&reserved=0>, or unsubscribe<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FALLQL3IX7ORCB2I2OQUEBTLSLNPERANCNFSM4GXSC2UA&data=04%7C01%7C%7Cec652e2818c145512ff008d873af9d33%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637386545370737702%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=31AC5inh4XGA22rq8P50fUQpyqObeIAqYP%2Fkc5P%2FTqo%3D&reserved=0>.

[barraponto]

@cknoll not clear to me whether we can rely on key order. is it confirmed? is there any setting we must enable?

[cknoll]

To the best of my knowledge since python3.7 the mapping order (i.e. key order) is reliably maintained by dict.
Sources:

• https://mail.python.org/pipermail/python-dev/2017-December/151283.html
• https://stackoverflow.com/questions/39980323/are-dictionaries-ordered-in-python-3-6/39980744

I code based on this assumption and have never experienced a counter example. However, I typically test on GNU Linux.