1{
2  "$schema": "https://json-schema.org/draft/2020-12/schema",
3  "title": "Source",
4  "description": "A research source with stable identity. source_id = sha256(canonical_locator)[:16].",
5  "type": "object",
6  "required": ["source_id", "canonical_locator", "raw_url", "title", "source_type", "metadata_status", "registered_at"],
7  "properties": {
8    "source_id": {
9      "type": "string",
10      "pattern": "^[0-9a-f]{16}$",
11      "description": "sha256(canonical_locator)[:16] — stable across edits and continuation"
12    },
13    "canonical_locator": {
14      "type": "string",
15      "description": "Canonical identifier: doi:10.1038/..., arxiv:2305.14251, or normalized URL (scheme+host+path, no fragment/tracking params)"
16    },
17    "raw_url": {
18      "type": "string",
19      "description": "Original URL as retrieved, before normalization"
20    },
21    "title": {
22      "type": "string"
23    },
24    "authors": {
25      "type": ["array", "null"],
26      "items": { "type": "string" },
27      "default": null
28    },
29    "year": {
30      "type": ["string", "null"],
31      "default": null
32    },
33    "source_type": {
34      "type": "string",
35      "enum": ["web", "academic", "documentation", "code", "news", "government", "book"]
36    },
37    "metadata_status": {
38      "type": "string",
39      "enum": ["unverified", "doi_verified", "url_verified", "title_matched"],
40      "description": "How far metadata has been verified"
41    },
42    "registered_at": {
43      "type": "string",
44      "format": "date-time",
45      "description": "ISO 8601 timestamp when source was registered"
46    }
47  },
48  "additionalProperties": false
49}
50