fix: node label and object type duplication detect

.vscode/settings.json

@@ -0,0 +1,21 @@
+{
+    "editor.formatOnSave": true,
+    "[javascript]": {
+        "editor.defaultFormatter": "esbenp.prettier-vscode"
+      },
+      "[typescript]": {
+        "editor.defaultFormatter": "esbenp.prettier-vscode"
+      },
+      "[javascriptreact]": {
+        "editor.defaultFormatter": "esbenp.prettier-vscode"
+      },
+      "[typescriptreact]": {
+        "editor.defaultFormatter": "esbenp.prettier-vscode"
+      },
+      "[json]": {
+        "editor.defaultFormatter": "esbenp.prettier-vscode"
+      },
+    "[jsonc]": {
+        "editor.defaultFormatter": "esbenp.prettier-vscode"
+    }
+}

packages/graph-schema-utils/src/formatters/json/extensions.test.ts

@@ -0,0 +1,75 @@
+import { describe, test } from "vitest";
+import { readFile } from "../../../test/fs.utils.js";
+import path from "path";
+import { fromJson } from "./index.js";
+import { strict as assert } from "node:assert";
+import {
+  hasDuplicateNodeLabelIds,
+  hasDuplicateNodeObjectTypeIds,
+} from "./extensions.js";
+import { RootSchemaJsonStruct } from "./types.js";
+
+describe("JSON formatter", () => {
+  describe("hasDuplicateNodeLabelIds", () => {
+    test("Identifies duplicated node labels", () => {
+      const schema = readFile(
+        path.resolve(__dirname, "./test-schemas/duplicated-nodeLabel-ids.json")
+      );
+      const schemaJson = JSON.parse(schema) as RootSchemaJsonStruct;
+
+      assert.strictEqual(
+        hasDuplicateNodeLabelIds(
+          schemaJson.graphSchemaRepresentation.graphSchema
+        ),
+        true
+      );
+    });
+  });
+  describe("hasDuplicateNodeObjectTypeIds", () => {
+    test("Identifies duplicated node object types", () => {
+      const schema = readFile(
+        path.resolve(
+          __dirname,
+          "./test-schemas/duplicated-nodeObjectType-ids.json"
+        )
+      );
+      const schemaJson = JSON.parse(schema) as RootSchemaJsonStruct;
+
+      assert.strictEqual(
+        hasDuplicateNodeObjectTypeIds(
+          schemaJson.graphSchemaRepresentation.graphSchema
+        ),
+        true
+      );
+    });
+  });
+  describe("fromJson", () => {
+    const schemaWithDuplicatedNodeLabels = readFile(
+      path.resolve(__dirname, "./test-schemas/duplicated-nodeLabel-ids.json")
+    );
+    test("Identifies duplicated node labels adn throws an error", () => {
+      assert.throws(() => fromJson(schemaWithDuplicatedNodeLabels), {
+        message: "Duplicate node label IDs found in schema",
+      });
+    });
+
+    test("Identifies duplicated node object types and throws an error", () => {
+      const schema = readFile(
+        path.resolve(
+          __dirname,
+          "./test-schemas/duplicated-nodeObjectType-ids.json"
+        )
+      );
+      assert.throws(() => fromJson(schema), {
+        message: "Duplicate node object type IDs found in schema",
+      });
+    });
+
+    test("Does not throw an error if there are no duplicated ids in node labels or node object types", () => {
+      const schema = readFile(
+        path.resolve(__dirname, "./test-schemas/full.json")
+      );
+      assert.doesNotThrow(() => fromJson(schema));
+    });
+  });
+});

packages/graph-schema-utils/src/formatters/json/extensions.ts

@@ -23,6 +23,7 @@ import {
 } from "../../model/index.js";
 import {
   ConstraintJsonStruct,
+  GraphSchemaJsonStruct,
   IndexJsonStruct,
   isLookupIndexJsonStruct,
   isNodeLabelConstraintJsonStruct,
@@ -107,12 +108,48 @@ export function fromJson(schema: string): GraphSchema {
   return fromJsonStruct(schemaJson);
 }
 
+export function hasDuplicateNodeLabelIds(
+  schema: GraphSchemaJsonStruct
+): boolean {
+  const ids = new Set<string>();
+
+  for (const nodeLabel of schema.nodeLabels) {
+    if (ids.has(nodeLabel.$id)) {
+      return true;
+    }
+    ids.add(nodeLabel.$id);
+  }
+
+  return false;
+}
+
+export function hasDuplicateNodeObjectTypeIds(
+  schema: GraphSchemaJsonStruct
+): boolean {
+  const ids = new Set<string>();
+
+  for (const nodeObjectType of schema.nodeObjectTypes) {
+    if (ids.has(nodeObjectType.$id)) {
+      return true;
+    }
+    ids.add(nodeObjectType.$id);
+  }
+
+  return false;
+}
+
 export function fromJsonStruct(schemaJson: RootSchemaJsonStruct): GraphSchema {
   const { graphSchema } = schemaJson.graphSchemaRepresentation;
+  if (hasDuplicateNodeLabelIds(graphSchema)) {
+    throw new Error("Duplicate node label IDs found in schema");
+  }
   const labels = graphSchema.nodeLabels.map(nodeLabel.create);
   const relationshipTypes = graphSchema.relationshipTypes.map(
     relationshipType.create
   );
+  if (hasDuplicateNodeObjectTypeIds(graphSchema)) {
+    throw new Error("Duplicate node object type IDs found in schema");
+  }
   const nodeObjectTypes = graphSchema.nodeObjectTypes.map(
     (nodeObjectTypeJson) =>
       nodeObjectType.create(nodeObjectTypeJson, (ref) => {

packages/graph-schema-utils/src/formatters/json/test-schemas/duplicated-nodeLabel-ids.json

@@ -0,0 +1,123 @@
+{
+  "graphSchemaRepresentation": {
+    "version": "1.0.0",
+    "graphSchema": {
+      "nodeLabels": [
+        {
+          "$id": "nl:0",
+          "token": "DOCS_CHUNKS_TABLE",
+          "properties": [
+            {
+              "$id": "p:0_0",
+              "token": "RELATIVE_PATH",
+              "type": {
+                "type": "string"
+              },
+              "nullable": false
+            },
+            {
+              "$id": "p:0_1",
+              "token": "SIZE",
+              "type": {
+                "type": "integer"
+              },
+              "nullable": false
+            },
+            {
+              "$id": "p:0_2",
+              "token": "FILE_URL",
+              "type": {
+                "type": "string"
+              },
+              "nullable": false
+            },
+            {
+              "$id": "p:0_3",
+              "token": "SCOPED_FILE_URL",
+              "type": {
+                "type": "string"
+              },
+              "nullable": false
+            },
+            {
+              "$id": "p:0_4",
+              "token": "CHUNK",
+              "type": {
+                "type": "string"
+              },
+              "nullable": false
+            }
+          ]
+        },
+        {
+          "$id": "nl:0",
+          "token": "DOCS_METADATA",
+          "properties": [
+            {
+              "$id": "p:0_0",
+              "token": "RELATIVE_PATH",
+              "type": {
+                "type": "string"
+              },
+              "nullable": false
+            },
+            {
+              "$id": "p:0_1",
+              "token": "MODEL_NAME",
+              "type": {
+                "type": "string"
+              },
+              "nullable": false
+            },
+            {
+              "$id": "p:0_2",
+              "token": "SKILLS",
+              "type": {
+                "type": "string"
+              },
+              "nullable": false
+            },
+            {
+              "$id": "p:0_3",
+              "token": "DOMAINS",
+              "type": {
+                "type": "string"
+              },
+              "nullable": false
+            },
+            {
+              "$id": "p:0_4",
+              "token": "SUMMARY",
+              "type": {
+                "type": "string"
+              },
+              "nullable": false
+            }
+          ]
+        }
+      ],
+      "relationshipTypes": [],
+      "nodeObjectTypes": [
+        {
+          "$id": "n:0",
+          "labels": [
+            {
+              "$ref": "#nl:0"
+            }
+          ]
+        },
+        {
+          "$id": "n:0",
+          "labels": [
+            {
+              "$ref": "#nl:0"
+            }
+          ]
+        }
+      ],
+      "relationshipObjectTypes": [],
+      "constraints": [],
+      "indexes": []
+    }
+  }
+}