- forge.yaml: the repository's host (Gitea at git.kenschiano.net, limited visibility); the "Report a run" collector, the page's links (data/public/links.json, filled in by site/build.js) and the dataset metadata read it - collection/etl/forge.py: issues, pull requests (merge when checks succeed) and labels for Gitea and GitHub; `python -m collection forge ...` - .gitea/workflows: CI and the daily update (pushes a branch, opens a pull request that merges itself, opens an issue when a source breaks); .github/workflows kept for GitHub - docs: Gitea setup (REPO_SETTINGS), the bot reading atlas files from the repository Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UCzTd31GphpMXDhQqWrCMQ
136 lines
4.4 KiB
Python
136 lines
4.4 KiB
Python
"""Write data/croissant.json (Croissant 1.0 JSON-LD) from data/dictionary.yaml.
|
|
|
|
Croissant is the MLCommons metadata format for datasets: it says what each file is, its checksum, and what each
|
|
column means, so tools (and people) can load and cite the data without reading this repo's code.
|
|
|
|
Written by `python -m collection build` (a test checks it is up to date).
|
|
"""
|
|
import json
|
|
|
|
from . import forge
|
|
from .common import ROOT, pipeline, read_yaml, sha256_file
|
|
|
|
OUT = "data/croissant.json"
|
|
|
|
CONTEXT = {
|
|
"@language": "en",
|
|
"@vocab": "https://schema.org/",
|
|
"citeAs": "cr:citeAs",
|
|
"column": "cr:column",
|
|
"conformsTo": "dct:conformsTo",
|
|
"cr": "http://mlcommons.org/croissant/",
|
|
"data": {"@id": "cr:data", "@type": "@json"},
|
|
"dataType": {"@id": "cr:dataType", "@type": "@vocab"},
|
|
"dct": "http://purl.org/dc/terms/",
|
|
"examples": {"@id": "cr:examples", "@type": "@json"},
|
|
"extract": "cr:extract",
|
|
"field": "cr:field",
|
|
"fileObject": "cr:fileObject",
|
|
"fileProperty": "cr:fileProperty",
|
|
"format": "cr:format",
|
|
"includes": "cr:includes",
|
|
"isLiveDataset": "cr:isLiveDataset",
|
|
"jsonPath": "cr:jsonPath",
|
|
"key": "cr:key",
|
|
"md5": "cr:md5",
|
|
"parentField": "cr:parentField",
|
|
"path": "cr:path",
|
|
"recordSet": "cr:recordSet",
|
|
"references": "cr:references",
|
|
"regex": "cr:regex",
|
|
"repeated": "cr:repeated",
|
|
"replace": "cr:replace",
|
|
"sc": "https://schema.org/",
|
|
"separator": "cr:separator",
|
|
"source": "cr:source",
|
|
"subField": "cr:subField",
|
|
"transform": "cr:transform",
|
|
}
|
|
|
|
# Croissant has no boolean-as-text type; the CSV writes True/False, which sc:Boolean parses.
|
|
DATA_TYPES = {"text": "sc:Text", "integer": "sc:Integer", "boolean": "sc:Boolean", "date": "sc:Date"}
|
|
|
|
|
|
sha256 = sha256_file
|
|
|
|
|
|
def load_dictionary():
|
|
return read_yaml("data/dictionary.yaml")
|
|
|
|
|
|
def field_description(spec):
|
|
text = " ".join(spec["description"].split())
|
|
if spec.get("values"):
|
|
text += " Values: " + "; ".join(str(v) for v in spec["values"]) + "."
|
|
return text
|
|
|
|
|
|
def build(dictionary=None):
|
|
d = dictionary or load_dictionary()
|
|
ds = d["dataset"]
|
|
fg = forge.config()
|
|
sources = pipeline()["sources"]
|
|
|
|
distribution = []
|
|
for key, spec in d["files"].items():
|
|
path = ROOT / spec["path"]
|
|
entry = {
|
|
"@type": "cr:FileObject",
|
|
"@id": key,
|
|
"name": key,
|
|
"description": " ".join(spec["description"].split()),
|
|
"contentUrl": forge.file_url(spec["path"], fg),
|
|
"encodingFormat": spec["format"],
|
|
"sha256": sha256(path),
|
|
"contentSize": f"{path.stat().st_size} B",
|
|
}
|
|
distribution.append(entry)
|
|
|
|
fields = [
|
|
{
|
|
"@type": "cr:Field",
|
|
"@id": f"runs/{name}",
|
|
"name": name,
|
|
"description": field_description(spec),
|
|
"dataType": DATA_TYPES[spec["type"]],
|
|
"source": {"fileObject": {"@id": "runs_clean"}, "extract": {"column": name}},
|
|
}
|
|
for name, spec in d["columns"].items()
|
|
]
|
|
|
|
return {
|
|
"@context": CONTEXT,
|
|
"@type": "sc:Dataset",
|
|
"conformsTo": "http://mlcommons.org/croissant/1.0",
|
|
"name": ds["name"],
|
|
"description": " ".join(ds["description"].split()),
|
|
"url": fg.get("site") or forge.repo_url(fg),
|
|
"sameAs": forge.repo_url(fg),
|
|
"license": ds["license"],
|
|
"creator": {"@type": "sc:Person", "name": ds["creator"]},
|
|
"keywords": ds["keywords"],
|
|
"citeAs": ds["citation"].format(repository=forge.repo_url(fg)),
|
|
"isLiveDataset": True,
|
|
"isBasedOn": [
|
|
{"@type": "sc:CreativeWork", "name": s["label"], "url": next(iter(s["links"].values()))}
|
|
for s in sources.values()
|
|
# the route PDF is only fingerprinted, never redistributed; the bot's repo is private
|
|
if (s.get("privacy") or {}).get("redistribute", True) and s.get("links")
|
|
],
|
|
"distribution": distribution,
|
|
"recordSet": [
|
|
{
|
|
"@type": "cr:RecordSet",
|
|
"@id": "runs",
|
|
"name": "runs",
|
|
"description": "One coded threshing run per row, from every source (data/public/runs_clean.csv).",
|
|
"key": {"@id": "runs/run_id"},
|
|
"field": fields,
|
|
}
|
|
],
|
|
}
|
|
|
|
|
|
def render(doc):
|
|
return json.dumps(doc, indent=2, ensure_ascii=False) + "\n"
|