API Reference

Public

SpaceDataModel.AbstractDataVariable — Type

A variable v of a type derived from AbstractDataVariable should at least implement:

  • Base.parent(v): the parent array of the variable

Optional:

  • tdimnum(v): the index of the time dimension (timedim, times derive from it)
  • units(v): the units of the variable
  • getmeta(v): the metadata of the variable
  • name(v): the name of the variable
  • dims(v, i): the coordinate of the i-th dimension of the variable
  • dims(v, name): the dimension named name of the variable
  • coords(v): further coordinates, axes => values (e.g. labels of a dimension)
source
SpaceDataModel.Archive — Type
Archive(pattern, reader=(paths, t0, t1) -> paths)

A source mirroring a URL-per-step archive: remotefiles enumerates, localize caches, reader(paths, t0, t1) opens and returns data over [t0, t1).

Unpublished steps are dropped, a gap being normal; a range publishing nothing throws, since no reader can open an empty file list.

source
SpaceDataModel.Dataset — Method
Dataset(name, source; selectors=(;), metadata=NoMetadata(), kw...)

A named dataset with domains for its selectors and a source that materializes it over a time range via getdata.

selectors maps each selector name to its domain: a single value, a collection, or Any for open. Values are spelled as strings.

name may carry {selector} placeholders, filled when selection pins open domains; a source templates itself the same way.

source
SpaceDataModel.FilePattern — Type
FilePattern(pattern; cadence=Day(1), kw...)

A template for the names of an archive's files, one per cadence.

Text in {} is a placeholder: {t:fmt} renders the step's start through the Dates format fmt ({t:mm-dd}, {t:yyyyjjj} with j for day of year) and {t1:fmt} the last instant it covers, for archives naming both ends. A name without a format is a keyword; {name|U} and {name|L} render its value upper- or lowercased. Everything else is literal.

Keywords may be filled at construction, when specializing, or at the call; every one must be filled before a name is rendered. Filling a keyword with "*" leaves a wildcard for resolve_url to match against the archive's listing.

julia> using Dates

julia> FilePattern("g13_xrs_1m_{t:yyyymm}01_{t1:yyyymmdd}.nc"; cadence=Month(1))(Date(2012, 2))
"g13_xrs_1m_20120201_20120229.nc"
source
SpaceDataModel.OverlayDict — Type
OverlayDict(base, overlay)
OverlayDict{K, V}(base)

Dictionary view where overlay entries shadow base; writes update only overlay.

source
SpaceDataModel.Product — Type
Product(dataset, variable; metadata=NoMetadata(), kw...)

A variable of a dataset: with optional layered metadata (e.g. a plot label override).

source
SpaceDataModel.Registry — Type
Registry(name, datasets; defaults=(;), metadata=NoMetadata(), kw...)

A relation of Datasets: rows sharing a selector vocabulary. A mission, an instrument, or any other grouping is a Registry.

source
SpaceDataModel.get_schema — Method
get_schema(data)

Get the metadata schema for data. Returns the tagged schema if metadata is a SchemaDict, otherwise infers from content ("CATDESC" ⇒ ISTPSchema), falling back to DefaultSchema.

source
SpaceDataModel.getmeta — Function
getmeta(x, key, default=nothing)

Get metadata value associated with key for object x, or default if key is not present.

source
SpaceDataModel.localize — Method
localize(url; dir=datadir(), update=false) -> String
localize(urls; ntasks=4, kw...) -> Vector{String}

Return a local path for each URL, downloading files missing from the cache.

Set the SPACE_DATA_DIR environment variable to override the default cache directory.

Set update = true to force re-download, or a Period to refetch if the local copy is older.

source
SpaceDataModel.remotefiles — Method
remotefiles(p::FilePattern, t0, t1; refresh=false, version="*", dir=datadir(), kw...) :: Vector{String}

URLs of the files over [t0, t1). Steps with no published file are dropped rather than an error, since a gap in an archive is normal.

Keywords fill p's placeholders; version defaults to the * wildcard, picking the highest listed. refresh=true bypasses memoized directory listings.

source
SpaceDataModel.resolve_url — Method
resolve_url(url; refresh=false, dir=datadir())

Concrete URL for url, where each * in the file name is replaced with the highest version field published in the directory, or nothing if the directory lists no such file. When the directory cannot be listed (e.g., the server is down), the highest version cached under dir stands in.

A field matches any width and any separator (e.g., v03, v03.06, v03_01, v1.1.10). A name may carry more than one, as MAVEN's …_v02_r00.cdf does; fields rank left to right.

source
SpaceDataModel.setmeta — Function
setmeta(x, key => value, ...; symbolkey => value2, ...)
setmeta(x, dict::AbstractDict)

Update metadata for object x for key key to have value value and return x.

source
SpaceDataModel.setmeta! — Function
setmeta!(x, key => value, ...; symbolkey => value2, ...)
setmeta!(x, dict::AbstractDict)

Update metadata for object x in-place and return x. The metadata container must be mutable.

The arguments could be multiple key-value pairs or a dictionary of metadata; keyword arguments are also accepted.

Examples

setmeta!(x, :units => "m/s", :source => "sensor")
setmeta!(x, Dict(:units => "m/s", :quality => "good"))
setmeta!(x; units="m/s", calibrated=true)

Throws an error if the metadata is not mutable. Use setmeta for immutable metadata.

source

Private

Base.filter — Method
filter(reg::Registry; selectors...)

The rows of reg every supplied selector matches, each with the matched columns pinned.

source
SpaceDataModel.coords — Method
coords(x)

Coordinates of x besides its dims, axes => values with axes a tuple, e.g. the labels of a dimension that has values.

source
SpaceDataModel.dims — Function
dims(x)
dims(x, i)
dims(x, name)

The dimensions of object x, its i-th dimension, or the dimension associated with the given name. A dimension without coordinates is nothing, the default (see AbstractDataVariable for its dims field default).

A dimension may be time-varying or dependent on other dimensions; in such cases, the effective size of the corresponding array dimension ndims can be greater than 1.

source
SpaceDataModel.mask_invalid — Method
mask_invalid(A; fillval, validmin, validmax, dims = nothing)
mask_invalid(A, checks::ValidityChecks, dims = nothing)

Copy of A with elements equal to fillval or outside [validmin, validmax] replaced by NaN; see ValidityChecks for their defaults and forms. Checks with several components apply along dimensions dims, an integer or a range; by default depend_1, the first non-time dimension.

Mask before arithmetic or unit conversion: metadata describes stored values.

Int8/UInt8, Int16/UInt16 and Float16 promote to Float32, wider integers to Float64, so Int64/UInt64 beyond 2^53 compare after rounding. Non-Real arrays are returned unchanged by the keyword form. A PermutedDimsArray gives one with the same permutation.

source
SpaceDataModel.pin — Method
pin(ds, values; complete=false)

Pin the open domains of ds named in values and rebuild its name and source with the pinned spellings; untouched domains stay open unless complete=true, which demands every one resolves.

source
SpaceDataModel.resolve — Method
resolve(data, lookup)

Resolve metadata value from data using lookup.

Patterns

  • "key" / :key: direct metadata lookup
  • ("k1", "k2", ...): priority lookup, first hit wins
  • Via(f, lookup): f(data) then resolve lookup on the result
  • lookup => default: use default (or default(data)) if lookup misses
  • f::Function: call f(data)
source
SpaceDataModel.select — Method
select(reg; selectors...)

The one dataset of reg every supplied selector matches, with omitted selectors filled from reg.defaults where the dataset carries them.

source