toml format

[andrewgsavage]

I have started writing a toml writer and parser. This is the format I have at present:

[prefix.micro]
value = "1e-6"
defined_symbol = "µ"
aliases = [
    "μ",
    "u",
    "mu",
    "mc",
]


[unit.meter]
defined_symbol = "m"
aliases = [
    "metre",
]
value = "[length]"


[unit.kelvin]
defined_symbol = "K"
aliases = [
    "degK",
    "°K",
    "degree_Kelvin",
    "degreeK",
]
value = "[temperature]; offset: 0"

[unit.radian]
defined_symbol = "rad"
value = "[]"


[unit.pi]
defined_symbol = "π"
value = "3.1415926535897932384626433832795028841971693993751"


[unit.magnetic_flux_quantum]
defined_symbol = "Φ_0"
aliases = [
    "Phi_0",
]
value = "ℎ / (2 * e)"


[unit.degree_Celsius]
defined_symbol = "°C"
aliases = [
    "celsius",
    "degC",
    "degreeC",
]
value = "kelvin; offset: 273.15"


[unit.decibelwatt]
defined_symbol = "dBW"
value = "watt; logbase: 10; logfactor: 10"


# note the header does not enclose the dimension in [], as it would then need quoting, ie [dimension."[density]"]
[dimension.density]
value = "[mass] / [volume]"



# groups have no other attributes than their unit definitions.
[group.USCSLengthInternational.definitions.thou]
defined_symbol = "th"
aliases = [
    "mil_length",
]
value = "1e-3 * inch"



[system.cgs]
using_group_names = [
    "international",
    "Gaussian",
    "ESU",
]
rules = [
    "centimeter",
    "gram",
    "second",
]

[system.atomic]
using_group_names = [
    "international",
]
rules = [
    "bohr: meter",
    "electron_mass: gram",
    "atomic_unit_of_time: second",
    "atomic_unit_of_current: ampere",
    "atomic_unit_of_temperature: kelvin",
]


[system.spectroscopy]
aliases = [
    "sp",
]
relations = [
    "[length] <-> [frequency]: speed_of_light / n / value",
    "[frequency] -> [energy]: planck_constant * value",
    "[energy] -> [frequency]: value / planck_constant",
    "[wavenumber] <-> [length]: 1 / value",
]

[system.spectroscopy.defaults]
n = 1


not got to contexts yet!

I have used the same attribute names as the defintion statements. Could/should these be renamed to something clearer?
I have quoted everything as strings. In some places numbers could be set to floats or ints, but I think for a first version it's best to stay similar to the .txt file and have them parsed the same way.

It's much longer as you'd expect. The 800 line file has become 3200 lines!

[hgrecco]

First ... awesome! This will be perfect for next release (In April) as Python 3.11+ (as given NEP29) supports TOML out of the box.

I like the way it looks. I wonder if we should write a converter script to help users migrate and allow Pint to drop support for the older parser and format.

[andrewgsavage]

yes, I have an (unpolished!) converter script that can be added to the module. It'll need tomli_w to write it though.

[andrewgsavage]

oh, I missed dropping support for the older parser. I am not sure on that, this format is much more verbose so less nice to write. We could have some releases compatible with both .toml and .txt.

[jules-ch]

I had worked on this back in the day here was the format with retrocompatibility with the string format:

#1561 (comment)

And you could also use the same keys as the Definition dataclasses as it would get the same key with the toml load command.