Re: [RFC PATCH] gnus: Add piem-gnus-mid-to-thread

[Kyle Meyer <kyle@kyleam.com>]

Xinglu Chen writes:

> Inserts a string of the whole thread in mbox format.

Great, thanks for looking into this.

> ---
> The function works as expected when running
> 'M-: (funcall (piem-gnus-mid-thread t))', but when running
> piem-b4-am-from-mid interactively, it just inserts an empty string.
>
> Any ideas of why this happens?  Sorry, my elisp skills are not that great.

Yes, I think you just need to move the lambda deeper so that the
messages are collected outside of the returned function.

> diff --git a/piem-gnus.el b/piem-gnus.el
> index 5f10be8..f91f91c 100644
> --- a/piem-gnus.el
> +++ b/piem-gnus.el
> @@ -58,6 +58,30 @@
>  ;; If there is an easy way to generate an mbox for a thread in Gnus, a
>  ;; function for `piem-mid-to-thread-functions' should be defined.
>  
> +(defun piem-gnus-mid-to-thread (_mid)

Hmm, yeah, I don't think we can convince Gnus to do much useful with the
message ID here.  However, perhaps there should at least be a check that
the given message ID matches the one for the article at point before
returning a function (i.e. claiming piem-gnus-mid-to-thread can generate
a thread for the given MID).

Also, I think this would need to be guarded by

  (derived-mode-p 'gnus-summary-mode)

or perhaps

  (derived-mode-p 'gnus-article-mode 'gnus-summary-mode)

to make sure we don't start doing this collection in buffers where it
won't work.

> +  (lambda ()

This is the lambda I was referring to in the beginning of the message.
It should be moved to deeper so that it contains only (insert ...).

> +    (save-excursion
> +      ;; Cursor has to be at the root of the thread
> +      (gnus-summary-refer-parent-article most-positive-fixnum)
> +      (let ((articles (gnus-summary-articles-in-thread))
> +	    message
> +	    ;; Just show raw message
> +	    (gnus-have-all-headers t)
> +	    gnus-article-prepare-hook
> +	    gnus-article-decode-hook
> +	    gnus-display-mime-function
> +	    gnus-break-pages)

Is the above set of variables (or this approach in general) based off of
an example from the Gnus source code?

If so, I should probably read through it to orient myself.  I am mostly
clueless when it comes to writing Gnus-related elisp and pretty much
stumble around its source code looking for something that seems close to
what I want.

> +	(mapc (lambda (article)
> +		(gnus-summary-display-article article)
> +		(push (format "From mboxrd@z Thu Jan  1 00:00:00 1970\n%s\n"
> +			      (with-current-buffer gnus-article-buffer
> +				(buffer-substring-no-properties
> +				 (point-min)
> +				 (point-max))))
> +			      message))
> +		articles)

For mboxrd, I suppose there needs to be some From-munging.

> +	      (insert (apply #'concat (nreverse message)))))))