Catalog

easysnowdata.catalog

The product catalog: one declarative entry per product (§3.3).

import easysnowdata as esd
esd.catalog.list(theme="snow")          # DataFrame of products
esd.catalog.describe("copernicus-dem")  # sources, credentials, resolution, citation
esd.catalog.search("swe")               # free-text search
esd.catalog.get("era5").sources         # the Source dataclasses

The same registry drives esd.catalog.health (the weekly probes and the README status table) and, in Phase 4, the generated docs pages.

Variable	One data variable of a product.
Probe	A minimal live check of one access route (§8).
Source	One access route to a product.
Product	One dataset, with one or more sources (the first is the default).
register	Add products to the registry (duplicate ids raise unless replace).
products	Read-only view of every registered product, keyed by id.
get	Return one product by id.
list	Return a DataFrame of products, optionally filtered.
search	Case-insensitive search over id, title, description, tags, variables and sources.
describe	Return a Markdown description of a product: sources, variables, citation.
themes
validate	Return the list of problems with product (empty when valid).
validate_all	Validate every registered product; returns all problems found.

easysnowdata.catalog.health

Live health probes for every catalog source (§8).

Each Probe performs the minimal request that confirms an access route is alive: a GET of the first byte for static file hosts (GET first — GRDC answers HEAD with 400; only a final 200/206 counts, figshare’s bot-challenge page answers 202), a STAC search, a Zarr metadata read, an Earth Engine getInfo() or an earthaccess search. Credentials come from easysnowdata.auth, so a probe whose source requires a provider that is not configured is reported as skip rather than fail.

Alongside those, §8’s two measuring probes:

kind="latency"

the newest datetime a time-series route serves, so a product that quietly stops being archived shows up as a date that stops moving rather than as nothing at all.

kind="virtualization"

whether a NetCDF/HDF product has DMR++ sidecars, and the parser earthaccess.virtualize() falls back to without them (§4.9). Both go through CMR’s metadata search, which needs no Earthdata Login, so they answer in a run with no credentials.

python -m easysnowdata.catalog.health (or the thin scripts/check_data_sources.py wrapper) runs everything and appends the results to data_status/history.json.

http_first_byte	Raise unless a GET of the first byte of url returns 200 or 206.
stac_search	Raise unless a STAC search of collection returns at least one item.
zarr_metadata	Raise unless the Zarr store at url opens with at least one variable.
gee_asset	Raise unless an Earth Engine asset answers getInfo().
earthdata_search	Raise unless earthaccess.search_data finds a granule.
stac_latest	The datetime of the newest item in collection.
cmr_latest	The start time of the newest granule of short_name, from CMR.
zarr_latest	The newest data the Zarr store at url actually serves.
gee_latest	The acquisition time of the newest image in an Earth Engine collection.
latest_available_day	Walk back day by day until one answers, for archives with no index.
dmrpp_status	Whether NASA publishes a DMR++ sidecar for short_name's granules.
run	Run every probe (or those of product_ids / kinds) and return the results.
update_history	Prepend results to the rolling history JSON (at most keep runs).
summarize
main	CLI: run the probes, print a summary, append to the history file.