emacs-orgmode
[Top][All Lists]
Advanced
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Prompts again WAS Re: [Orgmode] avoiding source block prompts

From: Charles C. Berry
Subject: Re: Prompts again WAS Re: [Orgmode] avoiding source block prompts
Date: Wed, 4 Aug 2010 20:09:30 -0700 
On Wed, 28 Jul 2010, Eric Schulte wrote:


Thanks Chuck,

I've just swapped in your new and improved regexp.

I share your concern both that there could be cases where prompts aren't
matched, or more seriously where the actual output looks like a prompt
(e.g. user returns a prompt-like string).

Before taking this explicit prompt removal approach I tried
automatically removing comment lines from the input before passing it to
the inferior R process, however that ran into similar problems of users
potentially inputting valid strings which look like comments -- which
would be hard to catch without implementing a fairly sophisticated R
parser as part of Babel.

This new regexp looks like the best option for now.

Thanks! -- Eric


Eric (and anyone else listening in),

I've put together a revised org-babel-R-evaluate.

With "R :session :results ouput" it  will gracefully handle commands like:

        cat("\n > > + > \n")
leaving the output in place, but trimming out the prompts and gratuitous lines. 

The code appears below along with a couple of test cases.

I did not touch anything in org-babel-R-evaluate above the second
'(output'; it looks like the ':session :results value' case will have trouble when a block ends with something like this 

        cat( "no newline" );invisible(1+1)

(say) and (inferior-ess-send-input) does not result in a newline, but
I am not sure how to engineer that. Maybe just
eval( parse( text=..ob.body ) ), but I don't know how to get the text in
body sent to R as ..ob.body.

If the elisp looks peculiar, it is because I am just picking it up.
Criticism and pointers welcome.

Chuck


#+begin_src emacs-lisp
  (setq org-babel-R-initiate-sink
       "..ob.tcon <- textConnection(\"..ob.tcon.txt\",\"w\")
       ..ob.eol <- \"\n\"
        sink( ..ob.tcon ) \n")

  (setq org-babel-R-finish-sink
       "\nsink()
        close( ..ob.tcon )
        cat( ..ob.eol )
        cat(..ob.tcon.txt, sep=..ob.eol)
        rm( ..ob.tcon.txt )")

(defmacro sinkout (bd)
  (list 'concat org-babel-R-initiate-sink bd org-babel-R-finish-sink))

(defun drop-two-lines (linestring)
  "Search for second '\\n' and delete all characters up to an including it."
  (substring  linestring
              (progn (string-match "\n.*\n" linestring)
                     (match-end 0))))

(defun org-babel-R-evaluate
  (session body result-type column-names-p row-names-p)
  "Pass BODY to the R process in SESSION.
  If RESULT-TYPE equals 'output then return a list of the outputs
  of the statements in BODY, if RESULT-TYPE equals 'value then
  return the value of the last statement in BODY, as elisp."
  (if (not session)
      ;; external process evaluation
      (case result-type
        (output (org-babel-eval org-babel-R-command body))
        (value
         (let ((tmp-file (make-temp-file "org-babel-R-results-")))
           (org-babel-eval org-babel-R-command
                           (format org-babel-R-wrapper-method
                                   body tmp-file
                                   (if row-names-p "TRUE" "FALSE")
                                   (if column-names-p
                                       (if row-names-p "NA" "TRUE")
                                     "FALSE")))
           (org-babel-R-process-value-result
            (org-babel-import-elisp-from-file
             (org-babel-maybe-remote-file tmp-file)) column-names-p))))
    ;; comint session evaluation
    (case result-type
      (value
       (let ((tmp-file (make-temp-file "org-babel-R"))
             broke)
         (org-babel-comint-with-output (session org-babel-R-eoe-output)
           (insert (mapconcat
                    #'org-babel-chomp
                    (list
                     body
                     (format org-babel-R-wrapper-lastvar
                             tmp-file
                             (if row-names-p "TRUE" "FALSE")
                             (if column-names-p
                                 (if row-names-p "NA" "TRUE")
                               "FALSE"))
                     org-babel-R-eoe-indicator) "\n"))
           (inferior-ess-send-input))
         (org-babel-R-process-value-result
          (org-babel-import-elisp-from-file
           (org-babel-maybe-remote-file tmp-file))  column-names-p)))
      (output
       (drop-two-lines
        (mapconcat
         #'org-babel-chomp
         (butlast
          (delq nil
                (mapcar
                 #'identity
                 (org-babel-comint-with-output (session org-babel-R-eoe-output)
                   (insert (mapconcat #'org-babel-chomp
                                      (list (sinkout body) 
org-babel-R-eoe-indicator)
                                      "\n"))
                   (inferior-ess-send-input)))) 2) "\n"))))))
#+end_src

#+results:
: org-babel-R-evaluate

#+begin_src R :session :results output
cat(" >\n")
1+1
#+end_src

#+results:
:  >
: [1] 2

#+begin_src R :session :results output
  cat(" > start of block < \n" )
  x <-
      rnorm(3)
  x
  y <- as.data.frame(diag(2))
  y+x
  cat(" >end of block< ") # no newline
#+end_src

#+results:
: > start of block < : [1] -1.4569661 0.1888919 -0.3685594 
:           V1         V2
: 1 -0.4569661 -0.3685594
: 2  0.1888919 -0.4569661
:  >end of block<

Add an extra newline for :session :results value ?

#+begin_src R :sesison :results value
## use C-g to return after C-c C-c
## a user might do this by mistake:
cat("no newline")
invisible(1+1)
#+end_src

#+results:
: 2



"Charles C. Berry" <address@hidden> writes:


Eric,

In ob-R.el, changing this

          (if (string-match "^\\([ >]+\\)\\[[0-9]+\\]" line)

to this

          (if (string-match "^\\([ ]*[>+][ ]?\\)+\\([[0-9]+\\|[ ]\\)" line)

seems to fix things (note I added plus sign which is the continuation
prompt in R).

I tried it on a bunch of varied source blocks and it seems to work OK,
but I worry that there may be case where it will fail.

Chuck


On Tue, 27 Jul 2010, Charles C. Berry wrote:



Eric,

I spoke too soon.

Have a look at the following. You'll see that the prompts show up there. :-(

Chuck

#+begin_src R :session :results output
### create x
x <- data.frame(a=rnorm(2),b=rnorm(2))
### now print the result
x
#+end_src

#+results:
:
: > >            a          b
:  1 0.2702748 -0.2938296
:  2 1.1095136  0.1769425



On Tue, 27 Jul 2010, Charles C. Berry wrote:


 On Tue, 27 Jul 2010, Eric Schulte wrote:


  Hi Charles,

  I just pushed up a fix for the extra prompts in your output.

You will

  still have a blank line (the output from "x <- rnorm(1)"), however you
  can suppress that line by placing the "x <- rnorm(1)" command in a
  previous code block in the same session.


 Eric,

 Thanks. That works for me.

 Chuck


  Best -- Eric
  "Charles C. Berry" <address@hidden> writes:

  I think this is a new feature or perhaps a bug, which I

noticed when

  upgrading to version 7.01.

  The newline and the two prompts '> >' in the results

block below did

  not show up in the earlier versions I used.

  If this is a 'feature', is there a clean way to change

this behavior?

  If this is a bug, is there a simple patch to fix it?
  #+begin_src R :session :results output

  ### create x
  x <- rnorm(1)
  ### now print the result
  x
  #+end_src

  #+results:

  :
: > >   [1] 2.186783

  What I would have liked is to have something that

looks like this:

  #+results:

:   [1] 2.186783

  Thanks,
  Chuck

  Charles C. Berry                            (858)

534-2098

                                              Dept of

Family/Preventive > >   Medicine

  E mailto:address@hidden                   UC San Diego
  http://famprevmed.ucsd.edu/faculty/cberry/  La Jolla, San

Diego > >   92093-0901

  _______________________________________________

  Emacs-orgmode mailing list
  Please use `Reply All' to send replies to the list.
  address@hidden
  http://lists.gnu.org/mailman/listinfo/emacs-orgmode




 Charles C. Berry                            (858) 534-2098
                                            Dept of Family/Preventive
 Medicine
 E mailto:address@hidden                    UC San Diego
 http://famprevmed.ucsd.edu/faculty/cberry/  La Jolla, San Diego 92093-0901



 _______________________________________________
 Emacs-orgmode mailing list
 Please use `Reply All' to send replies to the list.
 address@hidden
 http://lists.gnu.org/mailman/listinfo/emacs-orgmode




Charles C. Berry                            (858) 534-2098
                                           Dept of Family/Preventive
Medicine
E mailto:address@hidden             UC San Diego
http://famprevmed.ucsd.edu/faculty/cberry/  La Jolla, San Diego 92093-0901



_______________________________________________
Emacs-orgmode mailing list
Please use `Reply All' to send replies to the list.
address@hidden
http://lists.gnu.org/mailman/listinfo/emacs-orgmode



Charles C. Berry                            (858) 534-2098
                                            Dept of Family/Preventive Medicine
E mailto:address@hidden             UC San Diego
http://famprevmed.ucsd.edu/faculty/cberry/  La Jolla, San Diego 92093-0901




Charles C. Berry                            (858) 534-2098
                                            Dept of Family/Preventive Medicine
E mailto:address@hidden             UC San Diego
http://famprevmed.ucsd.edu/faculty/cberry/  La Jolla, San Diego 92093-0901
reply via email to
[Prev in Thread] Current Thread [Next in Thread]