Re: Exporting blocks without removing indentation

[Dan Davison <davison@stats.ox.ac.uk>]

Carsten Dominik <carsten.dominik@gmail.com> writes:

> Hi Dan,
>
> this looks really good.  I have applied the patch and again
> made the necessary entries in lisp/ChangeLog -  it would be nice if
> you could make them next time.  

Thanks, I think I will remember next time!

<...>
> I recommend to do this in the very
> end,
> to minimize merge conflicts - unfortunately ChamgeLog almost always
> creates
> a simple conflict because changes are always written to the top of the
> file.
> Usually I create a git diff in Emacs (if you use magit, press "D" and
> then
> for example "origin/master" as base.  In the dif buffer, you can press
> RET
> at each change to get to the source buffer, and then C-x 4 a to create
> the
> ChangeLog entry.

Thanks, that's really useful. I didn 't know any of that.

Dan

>
> I know this is annoying because git of course does record its changes so
> nicely - but such are the rules for code in Emacs.
>
> Also I have changed the indentation in the docstring that it complied to
> Emacs rules: First line indented like TAB does it, but the second and
> following lines not indented so that, when the docstring is displayed,
> you do not get those lines with initial whitespace.
>
> Thanks!
>
> - Carsten
>
>
> Changes at master~1
> 	Modified lisp/ChangeLog
> diff --git a/lisp/ChangeLog b/lisp/ChangeLog
> index 3bea93f..41f3734 100755
> --- a/lisp/ChangeLog
> +++ b/lisp/ChangeLog
> @@ -1,3 +1,16 @@
> +2009-11-01  Carsten Dominik  <carsten.dominik@gmail.com>
> +
> +	* org-src.el (org-src-preserve-indentation): Document that this
> +	variable is also used during export.
> +
> +	* org-exp.el (org-export-format-source-code-or-example): Preserve
> +	indentation if a block has a -i option, or if
> +	`org-src-preserve-indentation' is set.
> +
> +	* org-exp-blocks.el (org-export-blocks-preprocess): Preserve
> +	indentation if a block has a -i option, or if
> +	`org-src-preserve-indentation' is set.
> +
>  2009-10-31  Carsten Dominik  <carsten.dominik@gmail.com>
>
>  	* org.el (org-mode-map): Define the new archiving keys.
> 	Modified lisp/org-src.el
> diff --git a/lisp/org-src.el b/lisp/org-src.el
> index de8f8a2..eba8ef3 100644
> --- a/lisp/org-src.el
> +++ b/lisp/org-src.el
> @@ -87,13 +87,12 @@ These are the regions where each line starts with
> a colon."
>
>  (defcustom org-src-preserve-indentation nil
>    "If non-nil, leading whitespace characters in source code
> -  blocks are preserved on export, and when switching between the
> -  org buffer and the language mode edit buffer. If this variable
> -  is nil then, after editing with \\[org-edit-src-code], the
> -  minimum (across-lines) number of leading whitespace characters
> -  are removed from all lines, and the code block is uniformly
> -  indented according to the value of
> -  `org-edit-src-content-indentation'."
> +blocks are preserved on export, and when switching between the
> +org buffer and the language mode edit buffer. If this variable
> +is nil then, after editing with \\[org-edit-src-code], the
> +minimum (across-lines) number of leading whitespace characters
> +are removed from all lines, and the code block is uniformly
> +indented according to the value of `org-edit-src-content-indentation'."
>    :group 'org-edit-structure
>    :type 'boolean)
>
>
>
> On Nov 1, 2009, at 2:36 AM, Dan Davison wrote:
>
>> Carsten Dominik <carsten.dominik@gmail.com> writes:
>>
>>> Hi Dan,
>>>
>>> maybe the easiest way to implement this would be a -i (or whatever)
>>> switch a the src block.
>>
>> Thanks, I have gone ahead with your suggestion. As you suggest, I've
>> added a -i switch that prevents the indentation of a block from being
>> removed during export. If the variable org-src-preserve-indentation[1]
>> is non-nil, then that is equivalent to having a -i switch on every
>> source block.
>>
>> Then the issue remained that an org-exp-block plugin is unable to
>> return
>> a block with the original indentation if the indentation is removed
>> before it receives it. So I have made that indentation removal
>> contingent on the same criteria (non-nil variable or -i switch).
>>
>> Those 2 commits are in branch ded-indent at
>> git://repo.or.cz/org-mode/babel.git, and also pasted below.
>>
>> Dan
>>
>> Footnotes:
>>
>> [1] org-src-preserve-indentation also acts in org-src-mode, where it
>> prevents indentation from being removed when switching between org and
>> language mode buffers.
>>
>> From d9447ad7b08332cac005bcb602ac1e55cf0588cf Mon Sep 17 00:00:00 2001
>> From: Dan Davison <davison@stats.ox.ac.uk>
>> Date: Sat, 31 Oct 2009 15:43:42 -0400
>> Subject: [PATCH 1/2] Allow code indentation to be preserved on export
>>
>> This is controlled on a per-block basis with a new src block flag
>> -i. It can be controlled globally using the variable
>> org-src-preserve-indentation.
>> ---
>> lisp/org-exp.el |    4 +++-
>> lisp/org-src.el |   12 +++++++-----
>> 2 files changed, 10 insertions(+), 6 deletions(-)
>>
>> diff --git a/lisp/org-exp.el b/lisp/org-exp.el
>> index f08ce41..c6ffe75 100644
>> --- a/lisp/org-exp.el
>> +++ b/lisp/org-exp.el
>> @@ -2273,6 +2273,8 @@ INDENT was the original indentation of the
>> block."
>> 	    rpllbl (string-match "-r\\>" opts)
>> 	    keepp (string-match "-k\\>" opts)
>> 	    textareap (string-match "-t\\>" opts)
>> +	    preserve-indentp (or org-src-preserve-indentation
>> +				 (string-match "-i\\>" opts))
>> 	    cols (if (string-match "-w[ \t]+\\([0-9]+\\)" opts)
>> 		     (string-to-number (match-string 1 opts))
>> 		   80)
>> @@ -2285,7 +2287,7 @@ INDENT was the original indentation of the
>> block."
>> 	;; we cannot use numbering or highlighting.
>> 	(setq num nil cont nil lang nil))
>>       (if keepp (setq rpllbl 'keep))
>> -      (setq rtn (org-remove-indentation code))
>> +      (setq rtn (if preserve-indentp code (org-remove-indentation
>> code)))
>>       (when (string-match "^," rtn)
>> 	(setq rtn (with-temp-buffer
>> 		    (insert rtn)
>> diff --git a/lisp/org-src.el b/lisp/org-src.el
>> index 463bfa8..de8f8a2 100644
>> --- a/lisp/org-src.el
>> +++ b/lisp/org-src.el
>> @@ -87,11 +87,13 @@ These are the regions where each line starts
>> with a colon."
>>
>> (defcustom org-src-preserve-indentation nil
>>   "If non-nil, leading whitespace characters in source code
>> -  blocks are preserved. Otherwise, after editing with
>> -  \\[org-edit-src-code], the minimum (across-lines) number of
>> -  leading whitespace characters are removed from all lines, and
>> -  the code block is then uniformly indented according to the
>> -  value of `org-edit-src-content-indentation'."
>> +  blocks are preserved on export, and when switching between the
>> +  org buffer and the language mode edit buffer. If this variable
>> +  is nil then, after editing with \\[org-edit-src-code], the
>> +  minimum (across-lines) number of leading whitespace characters
>> +  are removed from all lines, and the code block is uniformly
>> +  indented according to the value of
>> +  `org-edit-src-content-indentation'."
>>   :group 'org-edit-structure
>>   :type 'boolean)
>>
>> --
>> 1.6.3.3
>>
>> From 60a5a63e6c6d66d1925f8cfc1191502bc34a7269 Mon Sep 17 00:00:00 2001
>> From: Dan Davison <davison@stats.ox.ac.uk>
>> Date: Sat, 31 Oct 2009 20:39:57 -0400
>> Subject: [PATCH 2/2] Allow org-exp-blocks plugins to return block
>> with indentation unaltered.
>>
>> With these changes, if org-src-preserve-indentation is non-nil, or if
>> the block has a -i switch, then the behaviour of org-exp-blocks is
>> altered as follows:
>>
>> 1. Indentation is not removed before passing the block contents to the
>>   block-transforming plugin.
>>
>> 2. The result returned by the plugin is not re-indented.
>> ---
>> lisp/org-exp-blocks.el |   17 ++++++++++-------
>> 1 files changed, 10 insertions(+), 7 deletions(-)
>>
>> diff --git a/lisp/org-exp-blocks.el b/lisp/org-exp-blocks.el
>> index 646688d..c92a9fa 100644
>> --- a/lisp/org-exp-blocks.el
>> +++ b/lisp/org-exp-blocks.el
>> @@ -166,7 +166,7 @@ specified in BLOCKS which default to the value of
>>   (save-window-excursion
>>     (let ((case-fold-search t)
>> 	  (types '())
>> -	  indentation type func start)
>> +	  indentation type func start body headers preserve-indent)
>>       (flet ((interblock (start end)
>> 			 (mapcar (lambda (pair) (funcall (second pair) start end))
>> 				 org-export-interblocks)))
>> @@ -176,17 +176,20 @@ specified in BLOCKS which default to the value
>> of
>> 		"^\\([ \t]*\\)#\\+begin_\\(\\S-+\\)[
>> \t]*\\(.*\\)?[\r\n]\\([^
>> \000]*?\\)[\r\n][ \t]*#\\+end_\\S-+.*" nil t)
>>           (setq indentation (length (match-string 1)))
>> 	  (setq type (intern (match-string 2)))
>> +	  (setq headers (save-match-data (split-string (match-string 3))))
>> +	  (setq body (match-string 4))
>> +	  (setq preserve-indent (or org-src-preserve-indentation
>> (member "-
>> i" headers)))
>> +	  (unless preserve-indent
>> +	    (setq body (save-match-data (org-remove-indentation body))))
>> 	  (unless (memq type types) (setq types (cons type types)))
>> 	  (save-match-data (interblock start (match-beginning 0)))
>> 	  (if (setq func (cadr (assoc type org-export-blocks)))
>> 	      (progn
>>                 (replace-match (save-match-data
>> -                                 (if (memq type org-export-blocks-
>> witheld)
>> -                                     ""
>> -                                   (apply func (save-match-data
>> (org-remove-indentation (match-string 4)))
>> -                                          (split-string (match-
>> string 3) " ")))) t t)
>> -                ;; indent block
>> -                (indent-code-rigidly (match-beginning 0) (match-end
>> 0) indentation)))
>> +                                 (if (memq type org-export-blocks-
>> witheld) ""
>> +                                   (apply func body headers))) t t)
>> +                (unless preserve-indent
>> +		  (indent-code-rigidly (match-beginning 0) (match-end
>> 0) indentation))))
>> 	  (setq start (match-end 0)))
>> 	(interblock start (point-max))))))
>>
>> --
>> 1.6.3.3
>>
>>
>>
>>>
>>> Switches are being processed already, so it will be simple to add
>>> one,
>>> I think.
>>>
>>> - Carsten
>>>
>>> On Oct 28, 2009, at 2:29 PM, Dan Davison wrote:
>>>
>>>> Fairly late in the export process, org-exp removes the common
>>>> indentation from lines inside blocks (line 2288 of org-exp.el):
>>>>
>>>> (setq rtn (org-remove-indentation code))
>>>>
>>>> However, what do we do if
>>>>
>>>> 1. A user has a block that she wishes to export with the indentation
>>>> intact?
>>>>
>>>> 2. An org-exp-blocks plugin has a block that it wishes to export
>>>> with
>>>>  the indentation intact?
>>>>
>>>> I propose that we make (1) and (2) possible. An example where (1)
>>>> arises
>>>> is if, when documenting a moderately complex program, different
>>>> parts of
>>>> a class or function are contained in separate blocks.
>>>>
>>>> Q1. Is that aim approved of?
>>>>
>>>> (2) is a bit trickier than (1). To solve (1), we could introduce a
>>>> fix
>>>> in org-exp.el like
>>>>
>>>> -      (setq rtn (org-remove-indentation code))
>>>> +      (setq rtn (if org-src-preserve-indentation code
>>>> +                 (org-remove-indentation code)))
>>>>
>>>> However in solving (2) the plugin cannot assume that the variable
>>>> org-src-preserve-indentation is in effect (in fact it very likely
>>>> isn't). So we need to come up with a way in which the block
>>>> resulting
>>>> from an org-exp-blocks plugin can have its indentation survive
>>>> through
>>>> to the export product, even if org-src-preserve-indentation is
>>>> globally
>>>> nil. Two possibilities are
>>>>
>>>> 1. We adopt a convention whereby if all lines start with some
>>>> character
>>>>  (say ,) then that character is removed (but the remaining
>>>> indentation
>>>>  is left intact.)
>>>>
>>>> 2. We transplant the indentation-removing work from org-exp.el into
>>>>  org-exp-blocks.el, and arrange that it is only performed if the
>>>> block
>>>>  is *not* handled by an org-exp-blocks plugin.
>>>>
>>>> Q2. Has this made sense? What solution do you suggest?
>>>>
>>>> Q3. Should we use org-src-preserve-indentation in this way to affect
>>>> export? (It currently has effect when switching between the org
>>>> buffer
>>>> and the code edit buffer.)
>>>>
>>>> Dan
>>>>
>>>>
>>>> _______________________________________________
>>>> Emacs-orgmode mailing list
>>>> Remember: use `Reply All' to send replies to the list.
>>>> Emacs-orgmode@gnu.org
>>>> http://lists.gnu.org/mailman/listinfo/emacs-orgmode
>>>
>>> - Carsten
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Emacs-orgmode mailing list
>>> Remember: use `Reply All' to send replies to the list.
>>> Emacs-orgmode@gnu.org
>>> http://lists.gnu.org/mailman/listinfo/emacs-orgmode
>>
>
> - Carsten
>
>
>
>
>
> _______________________________________________
> Emacs-orgmode mailing list
> Remember: use `Reply All' to send replies to the list.
> Emacs-orgmode@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-orgmode