FIX: Fixed information schema.

Title added. Descriptions added.

schemas/information_validator.json

@@ -1,25 +1,86 @@
 {
 	"bsonType": "object",
-	"required": ["chunk_id", "content", "embedding", "metadata", "embedding_model", "embedding_dim"],
+	"title": "Information validator",
+	"required": [
+		"chunk_id",
+		"content",
+		"embedding",
+		"embedding_model",
+		"embedding_dim",
+		"metadata"
+	],
 	"properties": {
-		"chunk_id": {"bsonType": "string"},
-		"content": {"bsonType": "string"},
-		"embedding": {"bsonType": "array", "items": {"bsonType": "double"}},
-		"embedding_model": {"bsonType": "string"},
-		"embedding_dim": {"bsonType": "int", "minimum": 1},
+		"chunk_id": {
+			"bsonType": "string",
+			"description": "'chunk_id' must be a string and is required."
+		},
+		"content": {
+			"bsonType": "string",
+			"description": "'content' must be a string and is required."
+		},
+		"embedding": {
+			"bsonType": "array",
+			"items": { "bsonType": "double" },
+			"description": "'embedding' must be an array of floats and is required."
+		},
+		"embedding_model": {
+			"bsonType": "string",
+			"description": "'embedding_model' must be a string and is required."
+		},
+		"embedding_dim": {
+			"bsonType": "int",
+			"minimum": 1,
+			"description": "'embedding_dim' must be an int greater than 1 and is required."
+		},
 		"metadata": {
 			"bsonType": "object",
-			"required": ["parent_id", "source", "task", "sequence", "total_chunks", "content_type"],
+			"title": "Metadata validator",
+			"required": [
+				"parent_id",
+				"source",
+				"task",
+				"sequence",
+				"total_chunks",
+				"content_type"
+			],
 			"properties": {
-				"parent_id": {"bsonType": "string"},
-				"source": {"bsonType": "string"},
-				"task": {"bsonType": "string"},
-				"sequence": {"bsonType": "int"},
-				"total_chunks": {"bsonType": "int"},
-				"content_type": {"bsonType": "string"},
-				"related_chunks": {"bsonType": "array", "items": {"bsonType": "string"}},
-				"chunk_length": {"bsonType": "int"},
-				"created_timestamp": {"bsonType": "date"}
+				"parent_id": {
+					"bsonType": "string",
+					"description": "'parent_id' must be a string and is required."
+				},
+				"source": {
+					"bsonType": "string",
+					"description": "'source' must be a string and is required."
+				},
+				"task": {
+					"bsonType": "string",
+					"description": "'task' must be a string and is required."
+				},
+				"sequence": {
+					"bsonType": "int",
+					"description": "'sequence' must be an int and is required."
+				},
+				"total_chunks": {
+					"bsonType": "int",
+					"description": "'total_chunks' must be an int and is required."
+				},
+				"content_type": {
+					"bsonType": "string",
+					"description": "'content_type' must be a string and is required."
+				},
+				"related_chunks": {
+					"bsonType": "array",
+					"items": { "bsonType": "string" },
+					"description": "'related_chunks' must be an array of strings and is optional."
+				},
+				"chunk_length": {
+					"bsonType": "int",
+					"description": "'chunk_length' must be an int and is optional."
+				},
+				"created_timestamp": {
+					"bsonType": "date",
+					"description": "'created_timestamp' must be a date and is optional."
+				}
 			}
 		}
 	}