schema/aggregate.schema.json
{
"$schema": "https://json-schema.org/draft/2020-12/schema",
"$id": "emergence/aggregate.schema.json",
"title": "Day aggregates (fractal topics, GIS, day stats)",
"type": "object",
"defs": {
"essence_ref": {
"type": "object",
"required": ["id", "essence", "source_id", "url"],
"properties": {
"id": { "type": "string" },
"essence": { "type": "string" },
"source_id": { "type": "string" },
"url": { "type": "string" },
"title": { "type": "string" }
}
}
},
"properties": {
"events": {
"type": "object",
"description": "the event layer output (events.json): every ok essence belongs to exactly one event",
"required": ["date", "generated_at", "clustering", "events"],
"properties": {
"date": { "type": "string" },
"generated_at": { "type": "string" },
"model": { "type": "string" },
"clustering": {
"type": "object",
"required": ["candidates", "ai_calls", "fallbacks"],
"properties": {
"candidates": { "type": "integer" },
"ai_calls": { "type": "integer" },
"fallbacks": { "type": "integer" }
}
},
"events": {
"type": "array",
"items": {
"type": "object",
"required": ["id", "members", "sources", "topic", "subtopic", "facets", "reports", "weight"],
"properties": {
"id": { "type": "string", "pattern": "^evt_[a-f0-9]{12}$" },
"seed": { "type": "string" },
"members": { "type": "array", "minItems": 1, "items": { "type": "string" } },
"sources": { "type": "array", "items": { "type": "string" } },
"topic": { "type": "string" },
"subtopic": { "type": "string" },
"facets": {
"type": "array",
"items": {
"type": "object",
"required": ["id", "label", "members"],
"properties": {
"id": { "type": "string", "pattern": "^fct_[a-f0-9]{12}$" },
"label": { "type": "string" },
"members": { "type": "array", "items": { "type": "string" } }
}
}
},
"locations": {
"type": "array",
"items": {
"type": "object",
"required": ["country_code", "name"],
"properties": { "country_code": { "type": "string" }, "name": { "type": "string" } }
}
},
"reports": { "type": "integer", "minimum": 1 },
"weight": { "type": "number", "description": "corroboration x complexity; feeds the day-portrait ranking" }
}
}
}
}
},
"day": {
"type": "object",
"required": ["date", "generated_at", "counts", "sources", "spend"],
"properties": {
"date": { "type": "string", "pattern": "^\\d{4}-\\d{2}-\\d{2}$" },
"generated_at": { "type": "string" },
"counts": {
"type": "object",
"required": ["articles", "essences", "failures"],
"properties": {
"articles": { "type": "integer" },
"essences": { "type": "integer" },
"failures": { "type": "integer" },
"events": { "type": ["integer", "null"] }
}
},
"sources": {
"type": "object",
"additionalProperties": {
"type": "object",
"required": ["status"],
"properties": {
"status": { "enum": ["ok", "failed"] },
"http": { "type": ["integer", "null"] },
"fetched": { "type": "integer" },
"new": { "type": "integer" }
}
}
},
"spend": {
"type": "object",
"required": ["tokens_in", "tokens_out", "usd"],
"properties": {
"tokens_in": { "type": "integer" },
"tokens_out": { "type": "integer" },
"usd": { "type": "number" }
}
}
}
},
"topics": {
"type": "object",
"required": ["date", "tree"],
"properties": {
"date": { "type": "string" },
"tree": {
"type": "array",
"items": {
"type": "object",
"required": ["topic", "name", "count", "subtopics"],
"properties": {
"topic": { "type": "string" },
"name": { "type": "string" },
"count": { "type": "integer" },
"subtopics": {
"type": "array",
"items": {
"type": "object",
"required": ["subtopic", "name", "count", "essences"],
"properties": {
"subtopic": { "type": "string" },
"name": { "type": "string" },
"count": { "type": "integer" },
"essences": { "type": "array", "items": { "$ref": "#/defs/essence_ref" } }
}
}
}
}
}
}
}
},
"gis": {
"type": "object",
"required": ["date", "countries"],
"properties": {
"date": { "type": "string" },
"resolution": { "const": "country" },
"countries": {
"type": "array",
"items": {
"type": "object",
"required": ["country_code", "name", "lat", "lon", "count", "essences"],
"properties": {
"country_code": { "type": "string", "pattern": "^[A-Z]{2}$" },
"name": { "type": "string" },
"lat": { "type": "number" },
"lon": { "type": "number" },
"count": { "type": "integer" },
"essences": { "type": "array", "items": { "$ref": "#/defs/essence_ref" } }
}
}
}
}
}
}
}