What is unified diff format?

Unified diff format (also called unidiff) shows changes between two files in a compact, human-readable way. It was introduced by GNU diff and is now the default output format for git diff. Each diff starts with file headers (--- and +++ lines), followed by one or more hunks that show the actual changes with surrounding context lines.

// Unified diff format anatomy

--- a/src/app.ts              // original file
+++ b/src/app.ts              // modified file
@@ -10,7 +10,8 @@            // hunk header: line numbers
 // context line (unchanged)   // starts with space
-const port = 3000;           // removed line (starts with -)
+const port = parseInt(       // added line (starts with +)
+  process.env.PORT || "3000" // added line
+);
 // more context               // unchanged

// Hunk header: @@ -oldStart,oldCount +newStart,newCount @@
// -10,7 means: starting at line 10, showing 7 lines from original
// +10,8 means: starting at line 10, showing 8 lines in modified

Anatomy of a unified diff

A unified diff has several parts: (1) The diff header line starting with 'diff --git a/file b/file'. (2) File metadata lines like 'index abc1234..def5678'. (3) The old file name prefixed with '--- a/'. (4) The new file name prefixed with '+++ b/'. (5) Hunk headers like '@@ -10,7 +10,8 @@' indicating line ranges. (6) Context lines starting with a space. (7) Removed lines starting with '-'. (8) Added lines starting with '+'.

Understanding hunk headers

The hunk header @@ -10,7 +10,8 @@ means: in the old file, this hunk starts at line 10 and spans 7 lines; in the new file, it starts at line 10 and spans 8 lines. The numbers after the comma indicate the number of lines shown (context + changes). Some hunks include a function name after the closing @@ for additional context.