Lisp expression replacements with query-regexp-replace

I rarely use query-replace-regexp's lisp expression replacement feature, but I came upon a good use for it today!

Recently, I've been testing out org-srs for practicing Mandarin vocab. So far, it works very well! When I asked to truncate the serialized data which gets written to the Org files to reduce diff noise and to make the state column fit on my small laptop screen, the package author kindly offered a workaround¹, which causes this:

# Without workaround:

#+NAME: srsitem:2d51efcd-8523-4432-b074-50d1ba4c45b3
| ! | timestamp            | rating |          stability |        difficulty | state     |
|---+----------------------+--------+--------------------+-------------------+-----------|
|   | 2024-12-11T18:37:19Z |        |                0.0 |               0.0 | :new      |
|   | 2024-12-11T18:38:23Z | :again |             0.4072 |            7.2102 | :learning |
|   | 2024-12-11T18:39:05Z | :easy  | 0.9375660842587888 | 6.078123501986166 | :review   |
|   | 2024-12-14T07:07:08Z | :easy  | 16.133939701409084 | 4.972537594025855 | :review   |
| * | 2024-12-30T07:07:08Z |        |                    |                   |           |

to become this:

# With workaround:

#+NAME: srsitem:2d51efcd-8523-4432-b074-50d1ba4c45b3
| ! | timestamp            | rating | stability | difficulty | state     |
|---+----------------------+--------+-----------+------------+-----------|
|   | 2024-12-11T18:37:19Z |        |       0.0 |        0.0 | :new      |
|   | 2024-12-11T18:38:23Z | :again |      0.41 |       7.21 | :learning |
|   | 2024-12-11T18:39:05Z | :easy  |      0.94 |       6.08 | :review   |
|   | 2024-12-14T07:07:08Z | :easy  |     16.13 |       4.97 | :review   |
| * | 2024-12-30T07:07:08Z |        |           |            |           |

However, I already had some existing data in my chinese.org file. To truncate that data, I used this lisp expression replacement with query-replace-regexp:

[0-9]+\.\([0-9]+\) → \,(if (length> \1 2) (format "%0.02f" \#0) \0)

The regexp matches the numbers in the stability and difficulty columns. My chinese.org file conveniently contains no other numbers in decimal-point notation. The lisp expression checks if the part after the decimal point contains more than two numbers, since I don't want 0.0 to become 0.00. If it does, then the format expression rounds it to two decimal places. Otherwise, the match is left as-is. Note that \#0 in the replacement text converts the match string to a number, which is required by the %f format type specifier.

After replacing all of the numbers in the buffer, I ran a quick keyboard macro to align all of the srsitem tables to fit the narrower contents, but I'll leave that as an exercise to the reader.

I hope this helps you. Feel free to contact me with suggestions, corrections, or questions!

(cl-defmethod org-srs-algorithm-repeat :around ((_ fsrs-scheduler) _)
  (let ((output (cl-call-next-method)))
    (setf (alist-get 'stability output) (format "%.02f" (alist-get 'stability output))
          (alist-get 'difficulty output) (format "%.02f" (alist-get 'difficulty output)))
    output))