feat(compass-collection): Process schema into format for LLM submission for Mock Data Generator – CLOUDP-337090 (#7205)

* WIP

* WIP

* Re order buttons

* WIP

* Rename datatest-id

* Move state to redux

* Update tests per comments

* Fix import

* Test file

* Tests cleanup

* Add tests

* Rename actions; Remove action wrapper

* Address comments

* WIP

* WIP

* Process schema

* WIP

* WIP

* WIP

* WIP

* WIP

* WIP

* Add more test

* Use type predicate validators

* Constants validator

* Create constant value

* Package lock update

* Types

* Package Lock

* Address comment

Author: jcobis

package-lock.json

@@ -66,7 +66,8 @@
     "react": "^17.0.2",
     "react-redux": "^8.1.3",
     "redux": "^4.2.1",
-    "redux-thunk": "^2.4.2"
+    "redux-thunk": "^2.4.2",
+    "bson": "^6.10.1"
   },
   "devDependencies": {
     "@mongodb-js/eslint-config-compass": "^1.4.6",

packages/compass-collection/package.json

@@ -1,5 +1,5 @@
 import type { Reducer, AnyAction, Action } from 'redux';
-import { analyzeDocuments, type Schema } from 'mongodb-schema';
+import { analyzeDocuments } from 'mongodb-schema';
 
 import type { CollectionMetadata } from 'mongodb-collection-model';
 import type { ThunkAction } from 'redux-thunk';
@@ -19,8 +19,10 @@ import {
   SCHEMA_ANALYSIS_STATE_INITIAL,
   type SchemaAnalysisError,
   type SchemaAnalysisState,
+  type FieldInfo,
 } from '../schema-analysis-types';
 import { calculateSchemaDepth } from '../calculate-schema-depth';
+import { processSchema } from '../transform-schema-to-field-info';
 import type { Document, MongoError } from 'mongodb';
 
 const DEFAULT_SAMPLE_SIZE = 100;
@@ -106,7 +108,7 @@ interface SchemaAnalysisStartedAction {
 
 interface SchemaAnalysisFinishedAction {
   type: CollectionActions.SchemaAnalysisFinished;
-  schema: Schema;
+  processedSchema: Record<string, FieldInfo>;
   sampleDocument: Document;
   schemaMetadata: {
     maxNestingDepth: number;
@@ -201,7 +203,7 @@ const reducer: Reducer<CollectionState, Action> = (
       ...state,
       schemaAnalysis: {
         status: SCHEMA_ANALYSIS_STATE_COMPLETE,
-        schema: action.schema,
+        processedSchema: action.processedSchema,
         sampleDocument: action.sampleDocument,
         schemaMetadata: action.schemaMetadata,
       },
@@ -420,7 +422,9 @@ export const analyzeCollectionSchema = (): CollectionThunkAction<
       schema.fields = schema.fields.filter(
         ({ path }) => !isInternalFieldPath(path[0])
       );
-      // TODO: Transform schema to structure that will be used by the LLM.
+
+      // Transform schema to structure that will be used by the LLM
+      const processedSchema = processSchema(schema);
 
       const maxNestingDepth = await calculateSchemaDepth(schema);
       const { database, collection } = toNS(namespace);
@@ -432,7 +436,7 @@ export const analyzeCollectionSchema = (): CollectionThunkAction<
       };
       dispatch({
         type: CollectionActions.SchemaAnalysisFinished,
-        schema,
+        processedSchema,
         sampleDocument: sampleDocuments[0],
         schemaMetadata,
       });

packages/compass-collection/src/modules/collection-tab.ts

@@ -1,5 +1,5 @@
 import type { Document } from 'mongodb';
-import { type Schema } from 'mongodb-schema';
+import type { PrimitiveSchemaType } from 'mongodb-schema';
 
 export const SCHEMA_ANALYSIS_STATE_INITIAL = 'initial';
 export const SCHEMA_ANALYSIS_STATE_ANALYZING = 'analyzing';
@@ -30,9 +30,35 @@ export type SchemaAnalysisErrorState = {
   error: SchemaAnalysisError;
 };
 
+/**
+ * MongoDB schema type
+ */
+export type MongoDBFieldType = PrimitiveSchemaType['name'];
+
+/**
+ * Primitive values that can appear in sample_values after BSON-to-primitive conversion.
+ * These are the JavaScript primitive equivalents of BSON values.
+ */
+export type SampleValue =
+  | string // String, Symbol, ObjectId, Binary, RegExp, Code, etc. (converted to string)
+  | number // Number, Int32, Long, Double, Decimal128, Timestamp (converted via valueOf())
+  | boolean
+  | Date
+  | null
+  | undefined;
+
+/**
+ * Schema field information (for LLM processing)
+ */
+export interface FieldInfo {
+  type: MongoDBFieldType; // MongoDB primitive type
+  sample_values?: SampleValue[]; // Primitive sample values (limited to 10)
+  probability?: number; // 0.0 - 1.0 field frequency
+}
+
 export type SchemaAnalysisCompletedState = {
   status: typeof SCHEMA_ANALYSIS_STATE_COMPLETE;
-  schema: Schema;
+  processedSchema: Record<string, FieldInfo>;
   sampleDocument: Document;
   schemaMetadata: {
     maxNestingDepth: number;