Merge pull request #2848 from OriginTrail/v6/fix/blazegraph-handling-…

…utf-8 Handle Blazegraph corruption of special characters
OriginTrail · Dec 25, 2023 · 49d9a74 · 49d9a74
2 parents 7ce252e + c83c41b
commit 49d9a74
Show file tree

Hide file tree

Showing 2 changed files with 46 additions and 3 deletions.
diff --git a/src/modules/triple-store/implementation/ot-blazegraph/ot-blazegraph.js b/src/modules/triple-store/implementation/ot-blazegraph/ot-blazegraph.js
@@ -4,6 +4,8 @@ import OtTripleStore from '../ot-triple-store.js';
 class OtBlazegraph extends OtTripleStore {
     async initialize(config, logger) {
         await super.initialize(config, logger);
+        // this regex will match \Uxxxxxxxx but will exclude cases where there is a double slash before U (\\U)
+        this.unicodeRegex = /(?<!\\)\\U([a-fA-F0-9]{8})/g;
 
         await Promise.all(
             Object.keys(this.repositories).map(async (repository) => {
@@ -47,6 +49,40 @@ class OtBlazegraph extends OtTripleStore {
         }
     }
 
+    hasUnicodeCodePoints(input) {
+        return this.unicodeRegex.test(input);
+    }
+
+    decodeUnicodeCodePoints(input) {
+        const decodedString = input.replace(this.unicodeRegex, (match, hex) => {
+            const codePoint = parseInt(hex, 16);
+            return String.fromCodePoint(codePoint);
+        });
+
+        return decodedString;
+    }
+
+    async _executeQuery(repository, query, mediaType) {
+        const result = await this.queryEngine.query(
+            query,
+            this.repositories[repository].queryContext,
+        );
+        const { data } = await this.queryEngine.resultToString(result, mediaType);
+
+        let response = '';
+
+        for await (const chunk of data) {
+            response += chunk;
+        }
+
+        // Handle Blazegraph special characters corruption
+        if (this.hasUnicodeCodePoints(response)) {
+            response = this.decodeUnicodeCodePoints(response);
+        }
+
+        return response;
+    }
+
     async healthCheck(repository) {
         try {
             const response = await axios.get(

diff --git a/src/service/triple-store-service.js b/src/service/triple-store-service.js
@@ -1,6 +1,6 @@
 import { formatAssertion } from 'assertion-tools';
 
-import { SCHEMA_CONTEXT, TRIPLE_STORE_REPOSITORIES } from '../constants/constants.js';
+import { SCHEMA_CONTEXT, TRIPLE_STORE_REPOSITORIES, MEDIA_TYPES } from '../constants/constants.js';
 
 class TripleStoreService {
     constructor(ctx) {
@@ -87,7 +87,14 @@ class TripleStoreService {
         tokenId,
         keyword,
     ) {
-        const assertion = await this.getAssertion(fromRepository, assertionId);
+        let assertion;
+        // Try-catch to prevent infinite processing loop when unexpected error is thrown while getting KA
+        try {
+            assertion = await this.getAssertion(fromRepository, assertionId);
+        } catch (e) {
+            this.logger.error(`Error while getting assertion for moving asset: ${e.message}`);
+            return;
+        }
 
         // copy metadata and assertion
         await this.localStoreAsset(
@@ -246,7 +253,7 @@ class TripleStoreService {
             repository,
             assertionId,
         );
-        nquads = await this.dataService.toNQuads(nquads, 'application/n-quads');
+        nquads = await this.dataService.toNQuads(nquads, MEDIA_TYPES.N_QUADS);
 
         this.logger.debug(
             `Assertion: ${assertionId} ${