What is a unified diff? How to read the ---, +++ and @@ lines

Here's the patch for the example, a small Python file where greet gained a second argument and main gained a second call:

--- a/greet.py
+++ b/greet.py
@@ -1,9 +1,10 @@
-def greet(name):
-    print("Hello, " + name)
+def greet(name, greeting="Hello"):
+    print(f"{greeting}, {name}")
 
 
 def main():
     greet("world")
+    greet("team", greeting="Hi")
 
 
 if __name__ == "__main__":

The first two lines name the files. --- a/greet.py is the old version and +++ b/greet.py is the new one. The a/ and b/ prefixes are a convention from git. They let git apply and patch -p1 strip one folder level and find the real path.

Next comes a hunk, one section of the change. Its header is @@ -1,9 +1,10 @@. The part after the minus sign is about the old file: the hunk starts at line 1 and covers 9 lines. The part after the plus sign says the same for the new file, line 1 and 10 lines. When a count is 1, it's left out, so @@ -4 +4 @@ means one line on each side.

Inside the hunk, every line starts with one character:

• A space marks context, a line that's the same in both files.
• A minus sign marks a line that's only in the old file.
• A plus sign marks a line that's only in the new file.

Context lines are there so a tool can find the right spot even if the file shifted a little since the patch was made. This tool keeps up to three lines of context, like diff -u and git diff. When two changes sit within six lines of each other, their context would overlap, so they share one hunk. That's why the example has 2 changes but a single hunk.

One special line can show up at the end: \ No newline at end of file. It means the line above it is the last line of a file that doesn't end with a line break. Some editors add that final line break and some don't, so it's a common cause of small, confusing diffs.