Patchouli: an interactive patch file splitter

Have you ever committed a big batch of changes that need to be code reviewed?

Did it happen that the code got finetuned and bugfixed over time, but you want to split up the final change set in a way that makes it easier on the reviewers?

Does it so happen that the preferred review units don't line up with your commit units? Maybe not even with complete files?

If you've ever wanted to split a patch set into parts at the hunk level, and do so iteratively and interactively, patchouli is for you!

Installing

Patchouli is available via pip:

pip install patchouli

Usage

(I'll use an arbitrary Linux kernel patch file to demonstrate)

Run patchouli on your patch file(s):

$ patchouli example.patch

Patchouli will show you the first hunk and a prompt to start typing commands.

(Type 'create foo' then 'move foo' to start classifying hunks)
**************** linux/arch/i386/kernel/process.c.seg ****************
    * Save away %fs and %gs. No need to save %es and %ds, as
    * those are always kernel segments while inside the kernel.
    */
-	asm volatile("movl %%fs,%0":"=m" (*(int *)&prev->fs));
-	asm volatile("movl %%gs,%0":"=m" (*(int *)&prev->gs));
+	asm volatile("mov %%fs,%0":"=m" (prev->fs));
+	asm volatile("mov %%gs,%0":"=m" (prev->gs));

    /*
     * Restore %fs and %gs if needed.
unclassified (1/8)>

Patchouli calls a bunch of hunks a "change set". At the start, all of your hunks will be in the change set named "unclassified". To see all hunks, type ls:

> ls

You'll see the following:

->   1) linux/arch/i386/kernel/process.c.seg
     2) linux/arch/i386/kernel/vm86.c.seg
     3) linux/arch/x86_64/kernel/process.c.seg
     4) linux/arch/x86_64/kernel/process.c.seg
     5) linux/arch/x86_64/kernel/process.c.seg
     6) linux/arch/x86_64/kernel/process.c.seg
     7) linux/include/asm-i386/system.h.seg
     8) linux/include/asm-i386/system.h.seg
(Type 'hunk N' to go to a specific hunk, 'show' to show the current hunk)

Create a new change set to move changes into:

> create i386

Start moving the first two hunks into this new change set:

> move i386   (or simply 'm i386')
> m           (without a name repeats the last move)
> <enter>     (simply pressing enter repeats the last command)

Type set to see an overview of all change sets:

> set

   i386 (2 hunks)
-> unclassified (6 hunks)

Type write to write out the changes to individual .patch files:

> write

Wrote i386.patch
Wrote unclassified.patch

Helpful tips

• Type help for a list of all commands.
• undo undoes the last move.
• You can move through the hunk list by typing next and back.
• next, back and move can be shortened to n, b and m.
• The shell has tab completion, courtsey of Python's cmd library.