bug#8141: 23.2.94; python-fill-paragraph

bug#8141: 23.2.94; python-fill-paragraph

[Andreas Röhler]

[-- Attachment #1: Type: text/plain, Size: 302 bytes --]


Hi,

attached a test file, revealing a bug in python-fill-paragraph already 
reported at lp, which pertains to python.el and python-mode.el likewise.

Patch attached.

Andreas

--
https://code.launchpad.net/~a-roehler/python-mode/python-mode-components
https://code.launchpad.net/s-x-emacs-werkstatt/

[-- Attachment #2: fill-paragraph-problems-lp710373-test.el --]
[-- Type: text/x-emacs-lisp, Size: 2277 bytes --]

(defun py-fill-paragraph-problems-lp:710373-test (&optional arg load-branch-function)
  (interactive "p")
  (when load-branch-function (funcall load-branch-function))
  (py-fill-paragraph-problems-lp:710373-test-intern arg))

(defun py-fill-paragraph-problems-lp:710373-test-intern (&optional arg)
  "If no `load-branch-function' is specified, make sure the appropriate branch is loaded. Otherwise default python-mode will be checked. "
  (let ((tmp-dir "/tmp/") 
        (fpp-exec-buffer "fill-paragraph-problems-lp:710373")
        (diff-buffer "fpp-lp:710373-old")
        (fpp-teststring "
    \"\"\"
    triple-quoted string containing \"quotation\" marks.
    triple-quoted string containing \"quotation\" marks.
    triple-quoted string containing \"quotation\" marks.
    triple-quoted string containing \"quotation\" marks.
    triple-quoted string containing \"quotation\" marks.
    \"\"\"
")
        erg)
    (set-buffer (get-buffer-create diff-buffer))
    (erase-buffer)
    (insert fpp-teststring)
    (write-file (concat tmp-dir diff-buffer))
    (if arg
        (progn
          (set-buffer (get-buffer-create fpp-exec-buffer))
          (when arg (switch-to-buffer (current-buffer)))
          (erase-buffer)
          (py-fill-paragraph-problems-lp:710373-test-base))
      (with-temp-buffer (py-fill-paragraph-problems-lp:710373-test-base)))))

(defun py-fill-paragraph-problems-lp:710373-test-base () 
  (python-mode)
  (insert fpp-teststring)
  (set-buffer fpp-exec-buffer)
  (font-lock-mode 1)
  (font-lock-fontify-buffer)
  (goto-char (point-min))
  (forward-line 3)
  (message "I'm here: %s" (point))
  (set-buffer-modified-p nil)
  ;;     (describe-function 'py-fill-paragraph)
  (setq fill-column 55)
  (if (functionp 'py-fill-paragraph)
      (py-fill-paragraph)
    (python-fill-paragraph))
  (write-file (concat tmp-dir fpp-exec-buffer))
  (setq erg (diff (concat tmp-dir fpp-exec-buffer) (concat tmp-dir diff-buffer) "-u"))
  (if (featurep 'xemacs)
      (progn
        (set-buffer "*Diff Output*")
        (switch-to-buffer (current-buffer)))
    (set-buffer erg)
    (sit-for 1)
    (assert (numberp (progn (goto-char (point-min))(search-forward "no differences" nil t 1))) t)
    (message "%s" "py-fill-paragraph-problems-lp:710373 passed")))

[-- Attachment #3: python.el.diff --]
[-- Type: text/x-patch, Size: 597 bytes --]

--- python.el	2011/02/28 20:22:31	1.1
+++ python.el	2011/02/28 20:24:23
@@ -2022,7 +2022,9 @@
 		;; current line.
 		((eq t (nth 3 syntax))	; in fenced string
 		 (goto-char (nth 8 syntax)) ; string start
-		 (setq start (line-beginning-position))
+                 (if (looking-at "\\(?:'''\\|\"\"\"\\|'\\|\"\\)\\\\?\n?")
+                     (setq start (match-end 0))
+                   (setq start (line-beginning-position)))
 		 (setq end (condition-case () ; for unbalanced quotes
                                (progn (forward-sexp)
                                       (- (point) 3))

bug#8141: 23.2.94; python-fill-paragraph

[Christoph Scholtes]

Andreas Röhler <andreas.roehler@easy-emacs.de> writes:

> attached a test file, revealing a bug in python-fill-paragraph already
> reported at lp, which pertains to python.el and python-mode.el
> likewise.
>
> Patch attached.

Thanks for the bug report and the patch.

However, I would like to understand first what the problem is/was. The
initial bug report on Launchpad included a file (kn.py), whose
doc-string caused problems, right?

I can open this file and use python-fill-paragraph and fill-paragraph
with no problem in emacs 23.2 and the latest trunk. There is no error
and the doc string looks formatted correctly, also.

What is the problem with the code in python.el/python-mode.el?

Regards,
Christoph

bug#8141: 23.2.94; python-fill-paragraph

[Andreas Röhler]

Am 01.03.2011 03:59, schrieb Christoph Scholtes:
> Andreas Röhler<andreas.roehler@easy-emacs.de>  writes:
>
>> attached a test file, revealing a bug in python-fill-paragraph already
>> reported at lp, which pertains to python.el and python-mode.el
>> likewise.
>>
>> Patch attached.
>
> Thanks for the bug report and the patch.
>
> However, I would like to understand first what the problem is/was. The
> initial bug report on Launchpad included a file (kn.py), whose
> doc-string caused problems, right?
>
> I can open this file and use python-fill-paragraph and fill-paragraph
> with no problem in emacs 23.2 and the latest trunk. There is no error
> and the doc string looks formatted correctly, also.
>
> What is the problem with the code in python.el/python-mode.el?
>
> Regards,
> Christoph
>

Hi,

with M-q resp. fill-paragraph it's fine here too.

python-fill-paragraph displays the bug.

Did you try the test delivered?

AFAIS bug basically might reside in
fill-paragraph, where

    ;; 2. Try fill-paragraph-function.
    (and (not (eq fill-paragraph-function t))
         (or fill-paragraph-function
             (and (minibufferp (current-buffer))
                  (= 1 (point-min))))
         (let ((function (or fill-paragraph-function


should be written like

(cond (fill-paragraph-function
        DTRT)

       (t everything else



Unless I'm wrong...

Thanks

Andreas

bug#8141: 23.2.94; python-fill-paragraph

[Stefan Monnier]

> with M-q resp. fill-paragraph it's fine here too.
> python-fill-paragraph displays the bug.
[...]
> AFAIS bug basically might reside in
> fill-paragraph, where

If calling fill-paragraph works while python-fill-paragraph doesn't, it
would rather seem to indicate that the bug is in python-fill-paragraph
(maybe the bug is that python-fill-paragraph should never be called
directly as a command, for example).

>    ;; 2. Try fill-paragraph-function.
>    (and (not (eq fill-paragraph-function t))
>         (or fill-paragraph-function
>             (and (minibufferp (current-buffer))
>                  (= 1 (point-min))))
>         (let ((function (or fill-paragraph-function


> should be written like

> (cond (fill-paragraph-function
>        DTRT)

>       (t everything else

No, the current definition is on purpose.  As the docstring of
fill-paragraph-function indicates:

   ...
   If the function returns nil, then `fill-paragraph' does its normal work.
   ...


-- Stefan

bug#8141: 23.2.94; python-fill-paragraph

[Andreas Röhler]

Am 08.03.2011 03:08, schrieb Stefan Monnier:
>> with M-q resp. fill-paragraph it's fine here too.
>> python-fill-paragraph displays the bug.
> [...]
>> AFAIS bug basically might reside in
>> fill-paragraph, where
>
> If calling fill-paragraph works while python-fill-paragraph doesn't, it
> would rather seem to indicate that the bug is in python-fill-paragraph
> (maybe the bug is that python-fill-paragraph should never be called
> directly as a command, for example).
>
>>     ;; 2. Try fill-paragraph-function.
>>     (and (not (eq fill-paragraph-function t))
>>          (or fill-paragraph-function
>>              (and (minibufferp (current-buffer))
>>                   (= 1 (point-min))))
>>          (let ((function (or fill-paragraph-function
>
>
>> should be written like
>
>> (cond (fill-paragraph-function
>>         DTRT)
>
>>        (t everything else
>
> No, the current definition is on purpose.


Hi Stefan,

IMHO that introduces complexity, even if there are reasons for.
If modes introduce their own fill-function, it's better to leave them 
than. If the modes need some later stuff from fill.el, they may take it.


Andreas





   As the docstring of
> fill-paragraph-function indicates:
>
>     ...
>     If the function returns nil, then `fill-paragraph' does its normal work.
>     ...
>
>
> -- Stefan
>

bug#8141: 23.2.94; python-fill-paragraph

[Stefan Monnier]

>> No, the current definition is on purpose.

> IMHO that introduces complexity, even if there are reasons for.
> If modes introduce their own fill-function, it's better to leave them
> than. If the modes need some later stuff from fill.el, they may take it.

What complexity?  If the mode's fill function wants to do it all itself,
than it just needs to return a non-nil value.


        Stefan

bug#8141: 23.2.94; python-fill-paragraph

[Glenn Morris]

I have no idea what this report was supposed to be about.

It may have been about filling the following:

"""
triple-quoted string containing "quotation" marks.
triple-quoted string containing "quotation" marks.
triple-quoted string containing "quotation" marks.
triple-quoted string containing "quotation" marks.
triple-quoted string containing "quotation" marks.
"""

In 24.1, using M-x fill-paragraph and M-x python-fill-paragraph give
slightly different results.

In the current trunk, using M-x fill-paragraph and M-x
python-fill-paragraph-function (why is this interactive?) obviously give
the same results. However, the result is not the same as either form in
24.1. It looks like this, with an empty line at the end that presumably
should not be there?

"""triple-quoted string containing "quotation" marks.  triple-quoted
string containing "quotation" marks.  triple-quoted string containing
"quotation" marks.  triple-quoted string containing "quotation" marks.
triple-quoted string containing "quotation" marks.

"""

bug#8141: 23.2.94; python-fill-paragraph

[Fabián Ezequiel Gallina]

This works fine on current trunk, notice there's a new option that allow 
users to set the style for docstrings.