Fix regex mapping

Author: jg-rp

src/path/functions/match.ts

@@ -1,5 +1,6 @@
 import { LRUCache } from "../lru_cache";
 import { FilterFunction, FunctionExpressionType } from "./function";
+import { mapRegexp } from "./pattern";
 
 export type MatchFilterFunctionOptions = {
   /**
@@ -58,56 +59,11 @@ export class Match implements FilterFunction {
 
   protected fullMatch(pattern: string): string {
     const parts: string[] = [];
-    let nonCaptureGroup = false;
-
-    if (!pattern.startsWith("^") && !pattern.startsWith("^(")) {
-      nonCaptureGroup = true;
-      parts.push("^(?:");
-    }
-    parts.push(this.mapRegexp(pattern));
-
-    if (nonCaptureGroup && !pattern.endsWith("$") && !pattern.endsWith(")$")) {
-      parts.push(")$");
-    }
-
-    return parts.join("");
-  }
-
-  // See https://datatracker.ietf.org/doc/html/rfc9485#name-ecmascript-regexps
-  protected mapRegexp(pattern: string): string {
-    let escaped = false;
-    let charClass = false;
-    const parts: string[] = [];
-    for (const ch of pattern) {
-      switch (ch) {
-        case ".":
-          if (!escaped && !charClass) {
-            parts.push("(?:(?![\r\n])\\P{Cs}|\\p{Cs}\\p{Cs})");
-          } else {
-            parts.push(ch);
-            escaped = false;
-          }
-          break;
-        case "\\":
-          escaped = true;
-          parts.push(ch);
-          break;
-        case "[":
-          charClass = true;
-          escaped = false;
-          parts.push(ch);
-          break;
-        case "]":
-          charClass = false;
-          escaped = false;
-          parts.push(ch);
-          break;
-        default:
-          escaped = false;
-          parts.push(ch);
-          break;
-      }
-    }
+    const explicitCaret = pattern.startsWith("^");
+    const explicitDollar = pattern.endsWith("$");
+    if (!explicitCaret && !explicitDollar) parts.push("^(?:");
+    parts.push(mapRegexp(pattern));
+    if (!explicitCaret && !explicitDollar) parts.push(")$");
     return parts.join("");
   }
 }

src/path/functions/pattern.ts

@@ -0,0 +1,39 @@
+// See https://datatracker.ietf.org/doc/html/rfc9485#name-ecmascript-regexps
+export function mapRegexp(pattern: string): string {
+  let escaped = false;
+  let charClass = false;
+  const parts: string[] = [];
+  for (const ch of pattern) {
+    if (escaped) {
+      parts.push(ch);
+      escaped = false;
+      continue;
+    }
+
+    switch (ch) {
+      case ".":
+        if (!charClass) {
+          parts.push("(?:(?![\r\n])\\P{Cs}|\\p{Cs}\\p{Cs})");
+        } else {
+          parts.push(ch);
+        }
+        break;
+      case "\\":
+        escaped = true;
+        parts.push(ch);
+        break;
+      case "[":
+        charClass = true;
+        parts.push(ch);
+        break;
+      case "]":
+        charClass = false;
+        parts.push(ch);
+        break;
+      default:
+        parts.push(ch);
+        break;
+    }
+  }
+  return parts.join("");
+}

src/path/functions/search.ts

@@ -1,5 +1,6 @@
 import { LRUCache } from "../lru_cache";
 import { FilterFunction, FunctionExpressionType } from "./function";
+import { mapRegexp } from "./pattern";
 
 export type SearchFilterFunctionOptions = {
   /**
@@ -48,50 +49,12 @@ export class Search implements FilterFunction {
     }
 
     try {
-      const re = new RegExp(this.mapRegexp(pattern), "u");
+      const re = new RegExp(mapRegexp(pattern), "u");
       if (this.cacheSize > 0) this.#cache.set(pattern, re);
       return !!s.match(re);
     } catch (error) {
       if (this.throwErrors) throw error;
       return false;
     }
   }
-
-  // See https://datatracker.ietf.org/doc/html/rfc9485#name-ecmascript-regexps
-  protected mapRegexp(pattern: string): string {
-    let escaped = false;
-    let charClass = false;
-    const parts: string[] = [];
-    for (const ch of pattern) {
-      switch (ch) {
-        case ".":
-          if (!escaped && !charClass) {
-            parts.push("(?:(?![\r\n])\\P{Cs}|\\p{Cs}\\p{Cs})");
-          } else {
-            parts.push(ch);
-            escaped = false;
-          }
-          break;
-        case "\\":
-          escaped = true;
-          parts.push(ch);
-          break;
-        case "[":
-          charClass = true;
-          escaped = false;
-          parts.push(ch);
-          break;
-        case "]":
-          charClass = false;
-          escaped = false;
-          parts.push(ch);
-          break;
-        default:
-          escaped = false;
-          parts.push(ch);
-          break;
-      }
-    }
-    return parts.join("");
-  }
 }

tests/path/regex_filters.test.ts

@@ -26,6 +26,41 @@ describe("match filter", () => {
       SyntaxError,
     );
   });
+  test("don't replace dot in character group", () => {
+    const env = new JSONPathEnvironment();
+    const query = "$[?match(@, 'ab[.c]d')]";
+    const data = ["abcd", "ab.d", "abxd"];
+    const rv = env.query(query, data);
+    expect(rv.values()).toStrictEqual(["abcd", "ab.d"]);
+  });
+  test("don't replace escaped dots", () => {
+    const env = new JSONPathEnvironment();
+    const query = "$[?match(@, 'ab\\\\.d')]";
+    const data = ["abcd", "ab.d", "abxd"];
+    const rv = env.query(query, data);
+    expect(rv.values()).toStrictEqual(["ab.d"]);
+  });
+  test("handle escaped right square bracket in character group", () => {
+    const env = new JSONPathEnvironment();
+    const query = "$[?match(@, 'ab[\\\\].c]d')]";
+    const data = ["abcd", "ab.d", "abxd"];
+    const rv = env.query(query, data);
+    expect(rv.values()).toStrictEqual(["abcd", "ab.d"]);
+  });
+  test("explicit start caret", () => {
+    const env = new JSONPathEnvironment();
+    const query = "$[?match(@, '^ab.*')]";
+    const data = ["abcd", "ab.d", "axc"];
+    const rv = env.query(query, data);
+    expect(rv.values()).toStrictEqual(["abcd", "ab.d"]);
+  });
+  test("explicit end dollar", () => {
+    const env = new JSONPathEnvironment();
+    const query = "$[?match(@, '.*?bc$')]";
+    const data = ["abcd", "abc", "axc"];
+    const rv = env.query(query, data);
+    expect(rv.values()).toStrictEqual(["abc"]);
+  });
 });
 
 describe("search filter", () => {