Reshape a list into rows

Question

I am looking for an easy way to turn a list such as this:

one,
two,
three,
four,
five,
...

into something more like this:

one,      two,      three,   four,
five,     six,      seven,   eight,
nine,     ten,      eleven,  twelve,
thirteen, fourteen, fifteen, sixteen,

It seems relatively straight-forward to do "column-major" reshaping by using rectangles, but more often than not it is the "row-major" reshaping that I want, since many programming languages accept a list in this format.

I am a big fan of the multiple-cursors package, so I would gladly accept an answer which presents a clever usage of mc features. That is not a requirement though. I find that multiple-cursors can get a bit laggy after >100 cursors, so there may be better ways to reshape a list when the list is long anyway.

Answer 1

Solution

This solution is derived from the below discussion. The only difference is to not manually insert spaces and use an alternative regex for alignment with does not need spaces and also works with text that has spaces in it.

1. Re-ordering: M-< F3 C-u M-^ C-x z z C-n F4 M-0 F4
2. Select whole buffer: C-x h
3. Align
   • C-u M-x align-regexp
   • Replace the content in the minibuffer with ,\(\) RET
   • RET
   • RET
   • y

Result

one,       two,       three,    four,    
five,      six,       seven,    eight,   
nine,      ten,       eleven,   twelve,  
thirteen,  fourteen,  fifteen,  sixteen,  

Discussion

1. M-x beginning-of-buffer
2. M-x start-kbd-macro -- Start recording macro
3. Do below 3 times
   • C-u M-x join-line (I have this bound to C-j for my convenience)
   • Hit SPACE -- Enter a space character which will later help with alignment. This is required only in emacs-elisp-mode. But even if this space is added regardless, this solution will still work.
4. M-x next-line (C-n)
5. M-x kmacro-end-or-call-macro -- Stop recording macro
6. M-0 C-x e -- Run the macro till the end of buffer is reached. (ref)
7. M-x mark-whole-buffer (C-x h)
8. C-u M-x align-regexp -> SPACE RET, RET, RET, y -- Do the regex align by Space character for all columns. Note that if you use , instead of SPACE, you can align content having the space characters too.

Explanation about when it is required to add that extra space in Step 3 above

The function fixup-whitespace called in delete-indentation function (which join-line aliases to) is responsible for adding that space automatically for major-modes not equal to emacs-lisp-mode.

The reason is the following line in fixup-whitespace function (in simple.el):

(looking-at "$\\|\\s(\\|\\s'"))`

If this expression is true, space is NOT inserted after finishing the join operation.

That expression is true when any one of these is true:

• if the point is at the end of line
• if the current character is an open parenthesis char as defined in the major-mode's syntax table
• if the current character is an expression prefix as defined in the major-mode's syntax table

And ... ', , and # are counted as expression prefixes in emacs-lisp-mode. At the end of joining operation, the point is on the , character at the end of the line (e.g. one, or two, or three,) and so the above looking-at expression becomes true when the major mode is emacs-lisp-mode; and thus the need to add that extra space manually in Step 3 for this mode.

Of course you don't have to worry about the space not being auto inserted if alignment is done at comma instead of space character in Step 8.

Answer 2

I too love multiple cursors, but for this one I'd use keyboard macros.

F3 C-3 C-n M-^ M-^ M-^ C-n C-0 F4

Which means:

1. Start kbd macro.
2. Move down 3 lines.
3. Join 3 lines up.
4. Move to next line.
5. Finish and repeat kbd macro indefinitely.

---

You can then proceed to use align-regexp like the other answers explain.

Answer 3

1. M-x set-fill-column as desired.
2. M-q: fill-paragraph.
3. M-x align-cols.

The code for align-cols is here

Answer 4

• Mark the region
• C-uM-| column -x RET

or column -x -c N for lines of (up to) N characters in length.

column is a BSD-derived shell command, which you quite likely have if you're on a Unix-like OS.

The -x argument gives the "row-major" output that you wanted.

For example, column -x -c 70 gives you the exact number of rows and columns that you showed in the question:

one,        two,        three,      four,
five,       six,        seven,      eight,
nine,       ten,        eleven,     twelve,
thirteen,   fourteen,   fifteen,    sixteen,

I don't see a way to specify the minimum spacing between columns, but the main problem with this is that the tab width used by column might not match your tab-width in Emacs, which might give you some wonky-looking output (even though it's formatted nicely for some other value).

However, once you have the basic arrangement of rows and columns, you can use align-regexp to tidy up if necessary:

C-uM-x align-regexp

The pattern would be: ,\(\s-*\) (i.e. prefix a comma to the default pattern). The next two options are default, and then say "Yes" to "Repeat throughout the line".

n.b. I believe you're going to run into trouble with some of the other answers as soon as your data contains spaces, which is why I think using a tool (column) already designed for this task is sensible, and why I recommend using the commas for alignment if you use align-regexp, rather than spaces.

Note that when indent-tabs-mode is non-nil, align-regexp will insert tabs for alignment. You can either untabify to convert them to spaces, or use https://stackoverflow.com/a/8129994/324105 to prevent it from happening.