Babel C-mode corrupts double-quoted strings in output

["Martin Jerabek" <om@mailservice.ms>]

Hi!

I recently started to use Org Babel for C++ programs. One of the programs outputs several lines with double-quoted strings, similar to this:

#+NAME: doublequotes_cpp
#+begin_src cpp :includes <iostream> :results output verbatim raw
std::cout << "\"line 1\"\n";
std::cout << "\"line 2\"\n";
std::cout << "\"line 3\"\n";
#+end_src

#+RESULTS: doublequotes_cpp
line 1

As you can see, only the first line is copied to the RESULTS block, and it is stripped of the double quotes.

I tracked down the problem to org-babel-read (in ob-core.el). org-babel-C-execute (in ob-C.el) calls this function with the output of the C++ program. The problem is the following line:

((eq (string-to-char cell) ?\") (read cell))

i.e. if the output of the program starts with a double quote, it is passed to read which reads only the first string and also removes the double quotes, resulting in the observed output.

The original version of this piece of code was added with the following commit:

commit 60a8ba556d682849eafb0f84e689967cd2965549
Author: Eric Schulte <schulte.eric@gmail.com>
Date:   Wed Mar 2 07:55:39 2011 -0700

    ob: read string variable values wrapped in double quotes, removing the quotes
    
    * lisp/ob.el (org-babel-read): Read string variable values wrapped in
      double quotes, removing the quotes.

AFAICT this modification was done in response to the email thread "[Orgmode] org-babel-read should have option NOT to interpret as elisp" started on 2011-02-27, more specifically the email on "Wed, 02 Mar 2011 07:56:45 -0700" from Eric Schulte. This was obviously done for parsing variables in the header line, not for the program output, but the Babel C mode uses org-babel-read also for the output.

I assumed that ":results output verbatim raw" would prevent any postprocessing of the output but this is not the case for C mode.

I am not sure how to fix this without breaking backward compatibility. I assume it should be fixed directly in org-babel-C-execute, not in a central function like org-babel-read to minimize the impact. Surprisingly (for me) the equivalent shell script works as expected:

#+NAME: doublequotes
#+begin_src shell :results output verbatim raw
echo '"line 1"'
echo '"line 2"'
echo '"line 3"'
#+end_src

#+RESULTS: doublequotes
"line 1"
"line 2"
"line 3"

because org-babel-execute:shell does not process the output with org-babel-read. I do not know if languages other than the C family (C, C++, D) are affected.

At the very least, the documentation of org-babel-read should be expanded to document the fact that if the CELL parameter starts with a double quote, it is processed by the read function.

Best regards
Martin Jerabek