add non-game content filter

lib/nonGameTerms.json

@@ -18,6 +18,7 @@
     "editor",
     "emulator",
     "expansion",
+    "figurine",
     "firmware",
     "guide",
     "hack",
@@ -26,6 +27,7 @@
     "installer",
     "intro",
     "json",
+    "jpg",
     "manual",
     "mod",
     "movie",
@@ -44,6 +46,7 @@
     "sdk",
     "setup",
     "soundtrack",
+    "sqlite",
     "terms",
     "tool",
     "trainer",
@@ -51,7 +54,8 @@
     "update",
     "utility",
     "video",
-    "Virtual Console",
-    "wallpaper"
+    "virtual console",
+    "wallpaper",
+    "xml"
   ]
 }

lib/services/elasticsearch.js

@@ -1,6 +1,9 @@
 import { Client } from '@elastic/elasticsearch';
 import debugPrint from '../debugprint.js';
 import { File } from '../models/index.js';
+import { readFileSync } from 'fs';
+import { fileURLToPath } from 'url';
+import { dirname, resolve } from 'path';
 
 const client = new Client({
   node: process.env.ELASTICSEARCH_URL || 'http://localhost:9200'
@@ -8,6 +11,23 @@ const client = new Client({
 
 const INDEX_NAME = 'myrient_files';
 
+// Cache for nonGameTerms
+let nonGameTermsCache = null;
+
+function getNonGameTerms() {
+  if (nonGameTermsCache) {
+    return nonGameTermsCache;
+  }
+
+  const __filename = fileURLToPath(import.meta.url);
+  const __dirname = dirname(__filename);
+
+  const nonGameTermsPath = resolve(__dirname, '../../lib/nonGameTerms.json');
+  nonGameTermsCache = JSON.parse(readFileSync(nonGameTermsPath, 'utf8'));
+
+  return nonGameTermsCache;
+}
+
 export async function initElasticsearch() {
   try {
     const indexExists = await client.indices.exists({ index: INDEX_NAME });
@@ -111,7 +131,7 @@ export async function search(query, options) {
   const searchQuery = {
     index: INDEX_NAME,
     body: {
-      size: 1000,
+      size: 1500,
       query: {
         bool: {
           must: buildMustClauses(query, options),
@@ -132,7 +152,6 @@ export async function search(query, options) {
   try {
     const startTime = process.hrtime();
     const response = await client.search(searchQuery);
-    const elapsed = parseHrtimeToSeconds(process.hrtime(startTime));
 
     // Fetch full records from PostgreSQL for the search results
     const ids = response.hits.hits.map(hit => hit._id);
@@ -146,13 +165,28 @@ export async function search(query, options) {
       return map;
     }, {});
 
-    // Combine Elasticsearch results with full PostgreSQL records
+    // Build results with full PostgreSQL records
+    let results = response.hits.hits.map(hit => ({
+      ...recordMap[hit._id].dataValues,
+      score: hit._score,
+      highlights: hit.highlight
+    }));
+
+    // Apply non-game content filtering in JavaScript if the option is enabled
+    if (options.hideNonGame) {
+      const nonGameTerms = getNonGameTerms();
+      const termPatterns = nonGameTerms.terms.map(term => new RegExp(term, 'i'));
+
+      // Filter results in JavaScript (much faster than complex Elasticsearch queries)
+      results = results.filter(item => {
+        // Check if filename contains any of the non-game terms
+        return !termPatterns.some(pattern => pattern.test(item.filename));
+      });
+    }
+
+    const elapsed = parseHrtimeToSeconds(process.hrtime(startTime));
     return {
-      items: response.hits.hits.map(hit => ({
-        ...recordMap[hit._id].dataValues,
-        score: hit._score,
-        highlights: hit.highlight
-      })),
+      items: results,
       elapsed
     };
   } catch (error) {