Re: Make doc-view's pdf->png conversion customizable

[Jambunathan K]

Hello Elias

(Sorry for the top post)

This more like a braindump and the intention here is not to comment on
your code but to share my thought process.

I had more or less similar requirement wrt viewing of OpenDocument/MS
files [1]. The patch has been pending for a while.

After some thought, I found out that it would be most convenient to
implement this in terms of a unified format-spec.

Search for the defcustom and format-spec in the attached code snippet
[2] (to get my drift).

A quick look at the patch suggests that you can define a format-spec for
each of the following params:

1. resolution 
2. pdf 
3. png 
4. page

and define the converter function as a string with "%" specifiers.

Do you see any issues with this approach?

I also think it might be a good to have some alist like the one below 

((PNG->PDF . PNG->PDF-SPECIFIC-FORMAT-SPEC)
 (DOC->PDF . DOC->PDF-SPECIFIC-FORMAT-SPEC))

in the doc-view so that users can plug in their own converters in an
easy way.

[1] http://lists.gnu.org/archive/html/emacs-devel/2011-05/msg00239.html

[2] Code snippet from my local area

--8<---------------cut here---------------start------------->8---
(defcustom org-export-convert-process
  '("soffice" "-norestore" "-invisible" "-headless" 
"\"macro:///BasicODConverter.Main.Convert(%I,%f,%O)\"")
  "Command to covert a Org exported format to other formats.
The variable is an list of the form (PROCESS ARG1 ARG2 ARG3
...).  Format specifiers used in the ARGs are replaced as below.
%i input file name in full
%I input file name as a URL
%f format of the output file
%o output file name in full
%O output file name as a URL
%d output dir in full
%D output dir as a URL"
  :group 'org-export)

(defun org-export-convert (&optional in-file fmt)
  "Convert file from one format to another using a converter.
IN-FILE is the file to be converted.  If unspecified, it defaults
to variable `buffer-file-name'.  FMT is the desired output format.  If the
backend has registered a CONVERT-METHOD via it's get function
then that converter is used.  Otherwise
`org-export-conver-process' is used."
  (interactive
   (let* ((input (if (featurep 'ido) 'ido-completing-read 'completing-read))
          (in-file (read-file-name "File to be converted: "
                                   nil buffer-file-name t))
          (fmt (funcall input "Output format:  "
                        (or (ignore-errors
                              (org-lparse-get-other-backends
                               (file-name-extension in-file)))
                            (org-lparse-all-backends))
                        nil nil nil)))
     (list in-file fmt)))
  (require 'browse-url)
  (let* ((in-file (expand-file-name (or in-file buffer-file-name)))
         (fmt (or fmt "doc") )
         (out-file (concat (file-name-sans-extension in-file) "." fmt))
         (out-dir (file-name-directory in-file))
         (backend (when (boundp 'org-lparse-backend) org-lparse-backend))
         (convert-process
          (or (ignore-errors (org-lparse-backend-get backend 'CONVERT-METHOD))
              org-export-convert-process))
         program arglist)

    (setq program (and convert-process (consp convert-process)
                       (car convert-process)))
    (unless (executable-find program)
      (error "Unable to locate the converter %s"  program))

    (setq arglist
          (mapcar (lambda (arg)
                    (format-spec arg `((?i . ,in-file)
                                       (?I . ,(browse-url-file-url in-file))
                                       (?f . ,fmt)
                                       (?o . ,out-file)
                                       (?O . ,(browse-url-file-url out-file))
                                       (?d . ,out-dir)
                                       (?D . ,(browse-url-file-url out-dir)))))
                  (cdr convert-process)))
    (ignore-errors (delete-file out-file))

    (message "Executing %s %s" program (mapconcat 'identity arglist " "))
    (apply 'call-process program nil nil nil arglist)

    ;; blah blah
    ))
--8<---------------cut here---------------end--------------->8---



Elias Pipping <address@hidden> writes:

> On Sat, Jun 25, 2011 at 4:35 PM, Elias Pipping
> <address@hidden> wrote:
>> Hello,
>>
>> mupdf[1] provides the program "pdfdraw" that can convert pdf files to
>> png images. I've attached a patch to doc-view.el (against the emacs-23
>> branch, can be applied to 23.3 without any changes as well) that (most
>> prominently) adds a new (customizable) variable
>> doc-view-pdf->png-converter-invocation, which can be set to one of the
>> pre-defined functions
>>
>>  doc-view-pdf->png-converter-invocation-ghostscript
>>  doc-view-pdf->png-converter-invocation-mupdf
>>
>> or a user-specified function. doc-view-pdf/ps->png and
>> doc-view-pdf/ps->png-1 were made to use them.
>>
>>
>> Best regards,
>>
>> Elias Pipping
>>
>> [1] http://mupdf.com/
>
> I attached an old version of the patch by mistake.
>
> I've attached the new and correct version to this email.
>
> From 870bacaa0e0db0f93d47c333349560fbae64f2fb Mon Sep 17 00:00:00 2001
> From: Elias Pipping <address@hidden>
> Date: Sat, 25 Jun 2011 13:50:11 +0200
> Subject: [PATCH] Make doc-view-pdf->png-1 customizable
>
> ---
>  lisp/doc-view.el |   71 ++++++++++++++++++++++++++++++++++++++++-------------
>  1 files changed, 53 insertions(+), 18 deletions(-)
>
> diff --git a/lisp/doc-view.el b/lisp/doc-view.el
> index af6e4f3..8a921a7 100644
> --- a/lisp/doc-view.el
> +++ b/lisp/doc-view.el
> @@ -153,6 +153,26 @@
>    :type 'file
>    :group 'doc-view)
>  
> +(defcustom doc-view-pdfdraw-program (executable-find "pdfdraw")
> +  "Program to convert PDF files to PNG."
> +  :type 'file
> +  :group 'doc-view)
> +
> +(defcustom doc-view-pdf->png-converter-invocation
> +  'doc-view-pdf->png-converter-invocation-ghostscript
> +  "Called to convert a PDF file into a PNG file"
> +  :type '(radio (function-item 
> doc-view-pdf->png-converter-invocation-ghostscript :doc "Use ghostscript")
> +             (function-item doc-view-pdf->png-converter-invocation-mupdf 
> :doc "Use mupdf")
> +                function)
> +  :group 'doc-view)
> +
> +(defcustom doc-view-ps->png-converter-invocation
> +  'doc-view-ps->png-converter-invocation-ghostscript
> +  "Called to convert a PS file into a PNG file"
> +  :type '(radio (function-item 
> doc-view-ps->png-converter-invocation-ghostscript :doc "Use ghostscript")
> +                function)
> +  :group 'doc-view)
> +
>  (defcustom doc-view-ghostscript-options
>    '("-dSAFER" ;; Avoid security problems when rendering files from untrusted
>             ;; sources.
> @@ -692,15 +712,35 @@ Should be invoked when the cached images aren't 
> up-to-date."
>                           (list "-o" pdf dvi)
>                           callback)))
>  
> +(defun doc-view-pdf->png-converter-invocation-ghostscript (resolution pdf 
> png &optional page)
> +  `((command . ,doc-view-ghostscript-program)
> +    (arguments . (,@doc-view-ghostscript-options
> +               ,(format "-r%d" resolution)
> +               ,@(if page `(,(format "-dFirstPage=%d" page)))
> +               ,@(if page `(,(format "-dLastPage=%d" page)))
> +               ,(concat "-sOutputFile=" png)
> +               ,pdf))))
> +
> +(defalias 'doc-view-ps->png-converter-invocation-ghostscript
> +  'doc-view-pdf->png-converter-invocation-ghostscript)
> +
> +(defun doc-view-pdf->png-converter-invocation-mupdf (resolution pdf png 
> &optional page)
> +  `((command . ,doc-view-pdfdraw-program)
> +    (arguments . (,(concat "-o" png)
> +               ,(format "-r%d" resolution)
> +               ,pdf
> +               ,@(if page `(,(format "%d" page)))))))
>  
>  (defun doc-view-pdf/ps->png (pdf-ps png)
>    "Convert PDF-PS to PNG asynchronously."
> -  (doc-view-start-process
> -   "pdf/ps->png" doc-view-ghostscript-program
> -   (append doc-view-ghostscript-options
> -           (list (format "-r%d" (round doc-view-resolution))
> -                 (concat "-sOutputFile=" png)
> -                 pdf-ps))
> +  (let ((invocation (case doc-view-doc-type
> +                       (pdf (funcall doc-view-pdf->png-converter-invocation
> +                                     (round doc-view-resolution) pdf-ps png))
> +                       (t   (funcall doc-view-ps->png-converter-invocation
> +                                     (round doc-view-resolution) pdf-ps 
> png)))))
> +    (doc-view-start-process
> +     "pdf/ps->png" (cdr (assoc 'command invocation))
> +     (cdr (assoc 'arguments invocation))
>     (lexical-let ((resolution doc-view-resolution))
>       (lambda ()
>         ;; Only create the resolution file when it's all done, so it also
> @@ -712,7 +752,7 @@ Should be invoked when the cached images aren't 
> up-to-date."
>         (when doc-view-current-timer
>           (cancel-timer doc-view-current-timer)
>           (setq doc-view-current-timer nil))
> -       (doc-view-display (current-buffer) 'force))))
> +       (doc-view-display (current-buffer) 'force)))))
>    ;; Update the displayed pages as soon as they're done generating.
>    (when doc-view-conversion-refresh-interval
>      (setq doc-view-current-timer
> @@ -723,17 +763,12 @@ Should be invoked when the cached images aren't 
> up-to-date."
>  (defun doc-view-pdf->png-1 (pdf png page callback)
>    "Convert a PAGE of a PDF file to PNG asynchronously.
>  Call CALLBACK with no arguments when done."
> -  (doc-view-start-process
> -   "pdf->png-1" doc-view-ghostscript-program
> -   (append doc-view-ghostscript-options
> -           (list (format "-r%d" (round doc-view-resolution))
> -                 ;; Sadly, `gs' only supports the page-range
> -                 ;; for PDF files.
> -                 (format "-dFirstPage=%d" page)
> -                 (format "-dLastPage=%d" page)
> -                 (concat "-sOutputFile=" png)
> -                 pdf))
> -   callback))
> +  (let ((invocation (funcall doc-view-pdf->png-converter-invocation
> +                          (round doc-view-resolution) pdf png page)))
> +    (doc-view-start-process
> +     "pdf/ps->png" (cdr (assoc 'command invocation))
> +     (cdr (assoc 'arguments invocation))
> +     callback)))
>  
>  (declare-function clear-image-cache "image.c" (&optional filter))

--