format=flowed incorrect line wrapping

[Nicolas Bock]

On Fri, Dec 21 2018, David Edmondson wrote: 

> On Friday, 2018-12-21 at 11:54:46 GMT, David Edmondson wrote: 
> 
>> On Thursday, 2018-12-20 at 07:00:08 -07, Nicolas Bock wrote: 
>> 
>>> sorry I didn't explain my problem very well. What I am looking 
>>> for  is to reflow (to borrow a term from NeoMutt [1]) an email 
>>> message  that I am _reading_ to the buffer width.  Currently 
>>> format=flowed  emails are simply shown as they are in the raw 
>>> email body. Editing  and sending emails in format=flowed works 
>>> fine. 
>> 
>> Currently emacs doesn't know that the text/plain part has the 
>> format=flowed attribute, so it can't do anything about it. 
>> 
>> Two options occur to me: - ensure that emacs knows the 
>> attributes of the part so that it could be 
>>   extended to do something useful, 
> 
> Here's a hacky patch that takes this approach. Does it do what 
> you would expect? 

Hi David,

thanks for the patch! I applied it but it's not doing anything. I 
am not sure though that I am actually using the patch, so the fact 
that there is no change shouldn't be taken too seriously :) Are 
you on IRC? Maybe moving our discussion there would make it 
easier.

Thanks!

Nick

> A complication is the indentation of parts, which makes it hard to set a
> good value for `fill-flowed-display-column' (because you don't know the
> indentation depth).
>
> A wash function (which would be the cleaner approach, presuming that we
> passed the part to all wash functions) *does* know the indentation
> depth, and so could perhaps adjust the value of
> `fill-flowed-encode-column' appropriately.
>
> diff --git a/emacs/notmuch-show.el b/emacs/notmuch-show.el
> index 78f1af47..18cbc690 100644
> --- a/emacs/notmuch-show.el
> +++ b/emacs/notmuch-show.el
> @@ -744,6 +744,11 @@ will return nil if the CID is unknown or cannot be retrieved."
>      (save-excursion
>        (save-restriction
>  	(narrow-to-region start (point-max))
> +	;; XXX dme: It would be much nicer to do this using a wash
> +	;; function attached to the hook, but that doesn't currently
> +	;; get access to the part and so cannot check the format.
> +	(when (string= (plist-get part :format) "flowed")
> +	  (fill-flowed))
>  	(run-hook-with-args 'notmuch-show-insert-text/plain-hook msg depth))))
>    t)
>  
> diff --git a/notmuch-show.c b/notmuch-show.c
> index 07e9a5db..92d18bb8 100644
> --- a/notmuch-show.c
> +++ b/notmuch-show.c
> @@ -708,6 +708,19 @@ format_part_sprinter (const void *ctx, sprinter_t *sp, mime_node_t *node,
>      }
>  
>      if (GMIME_IS_PART (node->part)) {
> +	/* Include the format parameter of text/plain parts if it
> +	 * exists, allowing the UI to adapt the display of the part as
> +	 * appropriate.
> +	 */
> +	if (g_mime_content_type_is_type (content_type, "text", "plain")) {
> +	    const char *format =
> +		g_mime_object_get_content_type_parameter (node->part, "format");
> +	    if (format) {
> +		sp->map_key (sp, "format");
> +		sp->string (sp, format);
> +	    }
> +	}
> +
>  	/* For non-HTML text parts, we include the content in the
>  	 * JSON. Since JSON must be Unicode, we handle charset
>  	 * decoding here and do not report a charset to the caller.
>
> dme.
> -- 
> When I grow up I'll be stable.