Add filtering for low-quality install scripts and deprioritize specified patterns

Author: Explosion-Scratch

lib/sources.js

@@ -545,6 +545,12 @@ const SCRIPT_DETECTION_CONFIG = {
     "brew update",
     "brew upgrade",
     "sed",
+    "copy and paste",
+    "```",
+    "sudo rm",
+    "http://",
+    "open powershell",
+    "open terminal",
   ],
   SCORE_WEIGHTS: {
     HIGH: 10,
@@ -693,17 +699,22 @@ const findInstallScripts = async (owner, repo, body) => {
   const uniqueSnippets = [];
   const seenCodes = new Set();
 
-  snippets
-    .sort((a, b) => b.score - a.score)
-    .forEach((snippet) => {
-      const normalizedCode = normalizeCode(snippet.code);
-      if (!seenCodes.has(normalizedCode)) {
-        seenCodes.add(normalizedCode);
-        uniqueSnippets.push(snippet);
-      }
-    });
+  const sortedSnippets = snippets
+    .sort((a, b) => b.score - a.score);
+
+  sortedSnippets.forEach((snippet) => {
+    const normalizedCode = normalizeCode(snippet.code);
+    if (!seenCodes.has(normalizedCode)) {
+      seenCodes.add(normalizedCode);
+      uniqueSnippets.push(snippet);
+    }
+  });
+
+  // Filter out the lowest 50% scoring items
+  const keepCount = Math.max(1, Math.ceil(uniqueSnippets.length / 2));
+  const filteredSnippets = uniqueSnippets.slice(0, keepCount);
 
-  return uniqueSnippets;
+  return filteredSnippets;
 };
 
 const hasHighPriorityInstallScript = (installScripts) => {