Merge pull request #8830 from uinstinct/ripgrep-regex

fix: regular expressions in ripgrep

Author: uinstinct

core/tools/definitions/grepSearch.ts

@@ -13,15 +13,15 @@ export const grepSearchTool: Tool = {
   function: {
     name: BuiltInToolNames.GrepSearch,
     description:
-      "Performs a regex search over the repository using ripgrep. Will not include results for many build, cache, secrets dirs/files. Output may be truncated, so use targeted queries",
+      "Performs a regular expression (regex) search over the repository using ripgrep. Will not include results for many build, cache, secrets dirs/files. Output may be truncated, so use targeted queries",
     parameters: {
       type: "object",
       required: ["query"],
       properties: {
         query: {
           type: "string",
           description:
-            "The search query to use. Must be the exact string to be searched or a valid ripgrep expression. Use regex with alternation (e.g., 'word1|word2|word3) or character classes to find multiple potential words in a single search.",
+            "The regex pattern to search for within file contents. Use regex with alternation (e.g., 'word1|word2|word3') or character classes to find multiple potential words in a single search.",
         },
       },
     },

core/util/regexValidator.ts

@@ -150,16 +150,6 @@ export function prepareQueryForRipgrep(query: string): {
   query: string;
   warning?: string;
 } {
-  // Check if it looks like a literal search that should be escaped
-  if (looksLikeLiteralSearch(query)) {
-    return {
-      query: escapeLiteralForRegex(query),
-      warning:
-        "Query contained special regex characters and was escaped for literal text search",
-    };
-  }
-
-  // Otherwise validate and sanitize as regex
   const validation = validateAndSanitizeRegex(query);
 
   return {

core/util/regexValidator.vitest.ts

@@ -111,8 +111,7 @@ describe("looksLikeLiteralSearch", () => {
 describe("prepareQueryForRipgrep", () => {
   it("should escape literal-looking queries", () => {
     const result = prepareQueryForRipgrep("hello.world");
-    expect(result.query).toBe("hello\\.world");
-    expect(result.warning).toContain("escaped for literal text search");
+    expect(result.query).toBe("hello.world");
   });
 
   it("should sanitize regex patterns", () => {
@@ -130,14 +129,12 @@ describe("prepareQueryForRipgrep", () => {
   describe("real-world examples", () => {
     it("should escape file patterns", () => {
       const result = prepareQueryForRipgrep("*.js");
-      expect(result.query).toBe("\\*\\.js");
-      expect(result.warning).toContain("escaped for literal text search");
+      expect(result.query).toBe("*.js");
     });
 
     it("should escape function calls", () => {
       const result = prepareQueryForRipgrep("console.log()");
-      expect(result.query).toBe("console\\.log\\(\\)");
-      expect(result.warning).toContain("escaped for literal text search");
+      expect(result.query).toBe("console.log()");
     });
 
     it("should not escape proper regex patterns", () => {
@@ -163,8 +160,7 @@ describe("problematic patterns that originally failed", () => {
 
   it("should handle dollar signs in shell patterns", () => {
     const result = prepareQueryForRipgrep("$(command)");
-    expect(result.query).toBe("\\$\\(command\\)");
-    expect(result.warning).toContain("escaped for literal text search");
+    expect(result.query).toBe("$(command)");
   });
 
   it("should handle escaped dollar signs", () => {
@@ -253,8 +249,7 @@ describe("patterns that should NOT be over-sanitized", () => {
   describe("edge cases that could trigger false positives", () => {
     it("should not treat mathematical expressions as problematic", () => {
       const result = prepareQueryForRipgrep("a + b * c");
-      expect(result.query).toBe("a \\+ b \\* c"); // Should be escaped as literal
-      expect(result.warning).toContain("escaped for literal text search");
+      expect(result.query).toBe("a + b * c"); // Should be escaped as literal
     });
 
     it("should not break regex that uses word boundaries", () => {
@@ -271,8 +266,7 @@ describe("patterns that should NOT be over-sanitized", () => {
 
     it("should not treat IPv4 addresses as problematic regex", () => {
       const result = prepareQueryForRipgrep("192.168.1.1");
-      expect(result.query).toBe("192\\.168\\.1\\.1"); // Should be escaped as literal
-      expect(result.warning).toContain("escaped for literal text search");
+      expect(result.query).toBe("192.168.1.1"); // Should be escaped as literal
     });
 
     it("should not break hex color codes", () => {
@@ -285,8 +279,7 @@ describe("patterns that should NOT be over-sanitized", () => {
   describe("patterns that should preserve user intent", () => {
     it("should not sanitize intentional regex alternation", () => {
       const result = prepareQueryForRipgrep("(foo|bar)");
-      expect(result.query).toBe("\\(foo\\|bar\\)"); // Should be escaped as literal since unescaped
-      expect(result.warning).toContain("escaped for literal text search");
+      expect(result.query).toBe("(foo|bar)"); // Should be escaped as literal since unescaped
     });
 
     it("should preserve escaped alternation in regex", () => {
@@ -300,11 +293,5 @@ describe("patterns that should NOT be over-sanitized", () => {
       expect(result.query).toBe("name LIKE '%john%'"); // No regex metacharacters here
       expect(result.warning).toBeUndefined();
     });
-
-    it("should not break CSS selectors when used as literal search", () => {
-      const result = prepareQueryForRipgrep(".class-name > div:nth-child(2)");
-      expect(result.query).toBe("\\.class-name > div:nth-child\\(2\\)");
-      expect(result.warning).toContain("escaped for literal text search");
-    });
   });
 });