SubsamplingInfo

class PlasmaCalcs.dimensions.subsampling.SubsamplingInfo(json, *, meta=UNSET, _json_indent=4, _init_appliers=True)

Bases: object

information about subsampling

json: dict of subsampling info: {

‘forall’: { # optional; if not provided use vals from byvars.

‘snaps’: [mode, info], # optional, can provide here or in byvars but not both.

‘array_dims’: [list of array dim names in order], # provide here or in byvars but not both.

},

‘byvars’: [ # optional; if not provided do snaps slicing only

{

‘snaps’: [mode, info], # optional, can provide here or in forall but not both.

‘array_dims’: [list of array dim names in order], # provide here or in forall but not both.

‘exact’: { # optional

var1: [mode1, info1],

var2: [mode2, info2], # optional

…,

},

‘regex’: { # optional

re_var3: [mode3, info3],

…,

},

},

…, # optionally can provide more dicts like above

]

}

where:

snaps = how to subsample snapshots (use dim=’snap’ for these). Example:

‘snaps’: [‘slice’, {‘snap’: [start, stop, step]}]

exact = dict of exact names; subsampling for var with this name

regex = dict of regex patterns; subsampling for var if re.fullmatch(pattern, var)

mode = mode for subsampling. e.g. ‘slice’

info = info for subsampling, format depends on mode.

e.g. for ‘slice’, info = dict of {dim: [start, stop, step]}.

Note: if multiple matches for any var, always use the earliest one;

i.e. earliest dict in list, and always exact match before regex match.

meta: dict, None, or UNSET

metadata about the subsampling info. Must be in json format, if provided.

None –> no metadata

UNSET –> use code_snapshot_info() to get metadata when this object is created.

_json_indent: int

number of spaces to indent when saving json file.

_init_appliers: bool

whether to self.init_appliers() during __init__.

Usually should be True, but if debugging might want to use False.

Methods

get_applier(var, *[, missing_ok])	returns a SubsamplingMode instance appropriate for var based on self.
get_var_info(var[, default])	returns the (mode, info, array_dims, None_or_snaps_mode_and_info) for var within self.
has_byvars_snaps()	return True if 'snaps' in any byvars, else False.
has_forall_snaps()	returns True if 'snaps' in 'forall', else False.
implied_from(json_or_filename_or_info, **kw_cls)	return cls object from json, filename, or SubsamplingInfo object.
init_appliers()	initialize all subsampling appliers for this SubsamplingInfo object.
init_checks()	do some sanity checks about this SubsamplingInfo object.
init_meta()	initialize self.meta.
load([file_or_dir])	load based on subsampling_info.txt.
register_applier_mode(cls_applier_to_register)	register cls_applier_to_register as a known applier mode for subsampling.
save([dir])	saves subsampling info.
update_meta(d)	self.meta.update(d).

Attributes

MODES

get_applier(var, *, missing_ok=True)

returns a SubsamplingMode instance appropriate for var based on self.

Assumes self.init_appliers() was run and is up-to-date.

var: str

variable name to get applier for, or string ‘snaps’.

‘snaps’ –> get ‘snaps’ applier from forall, else crash.

else –> get first match from byvars.

missing_ok: bool

When var has no match in self:

if missing_ok: return SubsamplingIdentity instance.

otherwise: raise SubsamplingNotFoundError.

get_var_info(var, default=UNSET)

returns the (mode, info, array_dims, None_or_snaps_mode_and_info) for var within self.

default: UNSET or any object

if var not found, return default if provided, else raise SubsamplingNotFoundError

has_byvars_snaps()

return True if ‘snaps’ in any byvars, else False.

has_forall_snaps()

returns True if ‘snaps’ in ‘forall’, else False.

classmethod implied_from(json_or_filename_or_info, **kw_cls)

return cls object from json, filename, or SubsamplingInfo object.

init_appliers()

initialize all subsampling appliers for this SubsamplingInfo object.

init_checks()

do some sanity checks about this SubsamplingInfo object.

Raise SubsamplingFormatError if any checks fail.

Here, checks:

- uniqueness of var keys: expect to not have exact same var key appear in multiple places.

- snaps can appear in forall or byvars, but not both.

- array_dims can appear in forall or byvars, but not both.

- array_dims must be provided (possibly via forall) for any byvars dict with at least 1 var.

init_meta()

initialize self.meta. None remains None. UNSET becomes code_snapshot_info().

classmethod load(file_or_dir=None, **kw_cls)

load based on subsampling_info.txt.

file_or_dir: None or str

where to load from.

None –> load from current directory

directory –> load from ‘subsampling_info.txt’ here,

or from ‘subsampling_info/subsampling_info.txt’ here.

file –> load from this file.

static register_applier_mode(cls_applier_to_register, mode=None)

register cls_applier_to_register as a known applier mode for subsampling.

save(dir='.')

saves subsampling info. Return SubsamplingInfoPathManager instance with relevant paths.

Crashes if this would overwrite any existing files.

dir: str

save to ‘subsampling_info/subsampling_info.txt’ within this directory,

or to ‘subsampling_info.txt’ within this dir if dir basename == ‘subsampling_info’.

Also saves ‘subsampling_meta.txt’ if relevant, at same level as subsampling_info.txt.

subsampling_path_manager_cls

alias of SubsamplingInfoPathManager

update_meta(d)

self.meta.update(d). If self.meta is None, keep it as None.