Applying a Multi-Hunk Refactoring Patch

Multi-Hunk Patches

Real-world patches often contain multiple hunks — each targeting a different section of the file. The patch engine applies hunks from top to bottom, adjusting line offsets as previous hunks add or remove lines.

Example: Refactoring a Module

--- a/api.js
+++ b/api.js
@@ -2,7 +2,7 @@

-const API_URL = "http://localhost:3000";
+const API_URL = process.env.API_URL || "http://localhost:3000";

 async function fetchUsers() {
   const response = await fetch(API_URL + "/users");
@@ -15,6 +15,10 @@
   return data;
 }

+async function fetchUserById(id) {
+  const response = await fetch(API_URL + "/users/" + id);
+  return response.json();
+}
+
 module.exports = { fetchUsers };
@@ -22,3 +26,3 @@
-module.exports = { fetchUsers };
+module.exports = { fetchUsers, fetchUserById };

Hunk Application Order

1. Hunk 1 (line 2): Replaces hardcoded URL with environment variable — no line count change
2. Hunk 2 (line 15): Adds a new function — adds 4 lines, creating an offset of +4
3. Hunk 3 (line 22 + offset = 26): Updates exports — the offset from Hunk 2 shifts this hunk's target

Offset Tracking

After each hunk, the engine tracks a cumulative line offset:

• Hunk adds N lines: offset increases by N
• Hunk removes N lines: offset decreases by N
• Subsequent hunks have their target line number adjusted by the current offset

Selective Application

The Diff Patch Applier lets you accept or reject each hunk independently. For example, you might accept the URL change (Hunk 1) and the new function (Hunk 2) but reject the export change (Hunk 3) if you want to handle it differently.