feat(reportOnNewData): use hashing instead of dict

src/readers/queryService.ts

@@ -24,6 +24,7 @@ import { parseCypherQuery } from "../utils/cypher/queryParser";
 import { formatTimeDifference } from "ts-common/src/logger/logger";
 import { graphQueryBackend2graphQuery } from "../frontend/statistics";
 import { Query2BackendQuery } from "../utils/reactflow/query2backend";
+import { hashDictionary, hashIsEqual } from "../utils/hashing";
 
 export const queryService = async (db: DbConnection, query: string): Promise<GraphQueryResultMetaFromBackend> => {
   // TODO: only neo4j is supported for now
@@ -287,16 +288,13 @@ export const queryServiceReaderDiffCheck = async (type: QueryExecutionTypes) =>
       USER_MANAGEMENT_SERVICE_API
     );
 
-    // const ssInsight = await getUserSaveStateInsight(
-    //   headers.message.sessionData.userID,
-    //   headers.message.sessionData.saveStateID,
-    //   USER_MANAGEMENT_SERVICE_API
-    // );
-
-    const previousQueryResult = {
-      nodes: [],
-      edges: [],
-    };
+    const ssInsight = await getUserSaveStateInsight(
+      headers.message.sessionData.userID,
+      headers.message.sessionData.saveStateID,
+      USER_MANAGEMENT_SERVICE_API
+    );
+    // const previousQueryResult = ssInsight.previous_result_hash;
+    const previousQueryResult = "Get hash from db";
 
     log.debug("Received query request:", ss);
 
@@ -331,21 +329,14 @@ export const queryServiceReaderDiffCheck = async (type: QueryExecutionTypes) =>
           log.debug("Query result!");
           log.info(`Query executed in ${formatTimeDifference(Date.now() - startTime)}`);
 
-          const queryResult = {
+          const queryResult = hashDictionary({
             nodes: result.nodes.map((node) => node._id),
             edges: result.edges.map((edge) => edge._id),
-          };
-
-          // Compare query results
-          const nodesDifferent = queryResult.nodes.some((node, index) => node !== previousQueryResult.nodes[index]);
-          const edgesDifferent = queryResult.edges.some((edge, index) => edge !== previousQueryResult.edges[index]);
-
-          if (
-            queryResult.nodes.length !== previousQueryResult.nodes.length ||
-            queryResult.edges.length !== previousQueryResult.edges.length ||
-            nodesDifferent ||
-            edgesDifferent
-          ) {
+          });
+
+          log.info("Comparing hash values from current and previous query");
+
+          if (previousQueryResult && !hashIsEqual(queryResult, previousQueryResult)) {
             log.info("Different results, use Dennis code...");
           } else {
             log.info("No difference in result sets");
@@ -356,7 +347,7 @@ export const queryServiceReaderDiffCheck = async (type: QueryExecutionTypes) =>
             headers.message.sessionData.userID,
             headers.message.sessionData.saveStateID,
             USER_MANAGEMENT_SERVICE_API,
-            queryResult
+            { previous_result_hash: queryResult }
           );
 
           log.info("Updated node and edge ids in SaveState");

src/utils/hashing/index.ts

+const FNV_OFFSET_BASIS = 2166136261;
+
+export const hashDictionary = (dictionary: Record<"nodes" | "edges", string[]>): string => {
+  // Convert the dictionary into a consistent string format
+  const jsonString = JSON.stringify(dictionary, Object.keys(dictionary).sort());
+
+  // FNV-1a hash implementation
+  let hash = FNV_OFFSET_BASIS; // FNV offset basis
+  for (let i = 0; i < jsonString.length; i++) {
+    hash ^= jsonString.charCodeAt(i);
+    hash += (hash << 1) + (hash << 4) + (hash << 7) + (hash << 8) + (hash << 24); // FNV prime
+  }
+
+  // Convert hash to a hexadecimal string
+  return (hash >>> 0).toString(16); // Ensure unsigned integer representation
+};
+
+export const hashIsEqual = (hash1: string, hash2: string): boolean => {
+  return hash1 === hash2;
+};