Add/remove column spreadsheet features in Emacs?

Question

Given delimited data in the following format, how can I insert and delete columns?

abc|efg|123|xyz123abc|yes
xxx|bbb|cc|ddd|no

Say, for example, that I wanted to add a 3rd column with a default value of 1 and remove the 4th column so the data would look like this:

abc|efg|1|123|yes
xxx|bbb|1|cc|no

I was looking into org-mode tables as a starting point, however it puts extra spaces around the data.

| abc | efg | 123 | xyz123abc |
| xxx | bbb | cc  | ddd       |
|     |     |     |           |

Answer 1

Well I just found out how to import data into SES (there are no "ses-import" functions, but it turns out you can simply yank tab-separated data), so this answer is even simpler than using table-capture :) (unless your data contains tabs, in which case you'll need to work around that).

SES is an actual spreadsheet application within Emacs. See its info node for details:
C-hidm ses RET, or M-: (info "ses") RET

To process the data:

1. Convert your original data to TSV with, say, M-%|RETTABRET!
2. Kill that text (e.g. select region, then C-w)
3. Create a temporary buffer and change to ses-mode: C-xb *ses* RET
   M-x ses-mode RET
   (and of course, C-hm for a description of ses-mode, and its keybindings)
4. Yank the data into the spreadsheet: C-y
5. Move point to the column to delete, and kill it with M-k
6. Copy the result back to the kill ring in TSV format: C-xhxt
   (i.e. mark-whole-buffer + ses-export-tsv)
7. Kill the *ses* buffer with C-xkRET
8. Yank the modified TSV data back into the original buffer with C-y
9. Convert the tabs back to |s with M-%TABRET|RET!

Caveat: I'm actually having issues yanking into SES under my standard config, but this process works fine under emacs -q. YMMV.

Answer 2

As well as org-mode's table facilities, Emacs has some other table editing features, which will get you most of the way there.

abc|efg|123|xyz123abc|yes
xxx|bbb|cc|ddd|no

Selecting the region followed by
M-x table-capture RET | RET $ RET RET 1 RET

gives us:

+---+---+---+---------+---+
|abc|efg|123|xyz123abc|yes|
+---+---+---+---------+---+
|xxx|bbb|cc |ddd      |no |
+---+---+---+---------+---+

Now move point inside a cell of the column you wish to delete, and use
M-x table-delete-column RET

+---+---+---+---+
|abc|efg|123|yes|
+---+---+---+---+
|xxx|bbb|cc |no |
+---+---+---+---+

'Unrecognising' the table will switch off the table editing mode, so that you can safely mess with the formatting.

M-x table-unrecognize-table RET

Then it's just a matter of search-and-replace to revert back to the original format.

You can delete all the row separators by killing the first row with C-kC-k, and the remainder with M-%C-yRETRET!)

|abc|efg|123|yes|
|xxx|bbb|cc |no |

Then you just need to clean up the row prefix and suffix, and the padding within cells:

M-C-% +| RET | RET ! (n.b. there's a space before the +)
M-C-% ^|\||$ RET RET !

abc|efg|123|yes
xxx|bbb|cc|no

You could record the clean-up steps as a keyboard macro (or write a function) in order to execute them in a single step.

Answer 3

I can think of two ways to update it manually (though that might not be what you're after):

Adding or deleting characters using rectangles:

cua-mode
<C-return> ;; cua-set-rectangle-mark on first row
;; create a rectangle one character wide from first to last rows
;; type '1|' to create your new column.

Record a macro to add/delete a column entry and repeat for the entire file.