org-mode macros inside code blocks and using babel - emacs

Inspired by this great post, I'm trying to use the combination of org-mode and babel for issuing queries to elasticsearch. For example, counting the number of entries in an index:
#+BEGIN_SRC sh
curl -XGET 'http://my.uri.example:8080/index/_count'
#+END_SRC
The above code can be evaluated using C-c C-c when the point is in the block.1
On the other hand, one can define macros in the org document. My question is: is it possible to define a macro
#+MACRO: live-db http://my.uri.example:8080
and rewrite the code block as follow:
#+BEGIN_SRC sh
curl -XGET '{{{live-db}}}/index/_count'
#+END_SRC
Out of the box, for me, it didn't work... It seems like babel is not expanding the macro before the evaluation of the block. Ideas?
Edit
Now, once I learned that I can use es-mode, I won't to fine tune my question. Consider the following two requests:
#+BEGIN_SRC es :url http://mu.uri.stage:8080
GET /users/_search?pretty
{
"query": {
"match_all":{}
}
}
#+END_SRC
and
#+BEGIN_SRC es :url http://mu.uri.live:8080
GET /users/_search?pretty
{
"query": {
"match_all":{}
}
}
#+END_SRC
They merely differ in the URL. I would like to define two macros:
#+MACRO staging http://my.uri.stage:8080
#+MACRO live http://my.uri.live:8080
and then use the macros as the variables of the blocks. Is it possible?
1 Make sure you enable the evaluation of sh. Add something like:
(org-babel-do-load-languages
'org-babel-load-languages
'((sh . t)))
to your .emacs.

macro expansion is not natively supported when executing code blocks, but the Noweb reference syntax which is supported is much more powerful.
However, I doubt that it will work using es-mode, since it passes the url in a header argument and not a variable.
This is a simple example for a sh code block:
#+name: staging
: http://my.uri.stage:8080
#+name: live
: http://my.uri.live:8080
#+name: test
#+begin_src sh :var url=staging
echo $url
#+end_src
#+call: test(live)
#+RESULTS:
: http://my.uri.live:8080
#+call: test(staging)
#+RESULTS:
: http://my.uri.stage:8080

Related

Literate programming, org mode, function declaration

I want to write a Literate Program with org-mode. Let's say I have the following function:
fn do_stuff()
{
// 200 lines of code go here.
}
I would write something like that in org-mode:
#+BEGIN_SRC rust :tangle /some/path
fn do_stuff()
{
#+END_SRC
// Many more blocks of `BEGIN_SRC` go here to exlpain 200 lines of code mentioned above.
// They will have formatting, and prose between them, and headings with different nesting levels.
// The whole shebang in short.
#+BEGIN_SRC rust :tangle /some/path
}
#+END_SRC
Now, here is the question. I hope I will explain it well, it's kinda hard to put in words.
What do I do with the first and last #+BEGIN_SRC blocks shown above?
How do I style the function declaration with org-mode and/or Literate Programming?
It seems kind of out of place with all the “formatting, prose, headings” of the 200 lines of code mentioned above.
I need ideas please :-)
Thanks in advance.
I would use noweb to tangle the full code without necessarily presenting it all in order. That is, I would do something like this:
The core code is
#+name: code1
#+begin_src rust :noweb yes :tangle no
...
#+end_src
More code etc. and then, at the end:
#+BEGIN_SRC rust :tangle /some/path
fn do_stuff()
{
<<code1>>
}
#+END_SRC
You may need :noweb yes on the full code block as well.

What can cause org-auto-tangle to result nil for all noweb code block evaluations

I'm using literate programming for some configuration files and would like to have some parts from elisp code block evaluations. I tried evaluating named code blocks with :noweb tangle but they always results nil and I do not see any errors in the *Messages*. Here's a simplified hello world example and the results I got.
Org file
#+title: Hello
#+PROPERTY: header-args :tangle hello.txt :cache no :exports none
#+auto_tangle: t
#+name: hello-world-output
#+begin_src emacs-lisp :tangle no :eval no-export :results output
(print "Hello world")
#+end_src
#+name: hello-world-value
#+begin_src emacs-lisp :tangle no :eval no-export :results value
"Hello world"
#+end_src
#+begin_src text :noweb tangle
<<hello-world-output>> -> <<hello-world-output()>>
<<hello-world-value>> -> <<hello-world-value()>>
#+end_src
Tangled results
(print "Hello world") -> nil
"Hello world" -> nil
I also checked that org-link-elisp-confirm-function and org-confirm-babel-evaluate both have nil value, so they should not be preventing evaluation.
EDIT: I forgot to mention that I used org-auto-tangle. Issue doesn't occur when calling org-bable-tangle directly.
I had the same issue and below solved it for me.
According to the code of org-auto-tangle the code will not be evaluated by default. In order to have the code auto evaluated you need to add your org file to the org-auto-tangle-babel-safelist. I've posted the definition of the variable and a link to the README with an example of how to set the variable.
(defvar org-auto-tangle-babel-safelist '()
"List of full path of files for which code blocks need to be evaluated.
By default, code blocks are not evaluated during the auto-tangle to avoid
possible code execution from unstrusted source. To enable code blocks evaluation
for a specific file, add its full path to this list.")
https://github.com/yilkalargaw/org-auto-tangle#babel-auto-tangle-safelist

Org-mode: overwriting globals #+PROPERTIES: in block headers?

It is often the case that I want to have :results silent for all the code blocks in an Org-mode document. Do simplify my block headers, I define this in my Org-mode document:
#+PROPERTY: header-args :results silent
That works properly: all the codeblocks uses the :results silent option in the blocks' header.
However, if I specify :results output in one of the code block, it will still be silent. I would have expected that it would overwrite the global setting, but it doesn't seem so.
Am I right by saying that or is there something I am missing to get this behavior?
Here is an example of what I would like to do:
#+PROPERTY: header-args :results silent
...
#+BEGIN_SRC clojure
;; this one is silent
(def foo "bar)
#+END_SRC
...
#+BEGIN_SRC clojure :results output
;; this one is being outputted
(def foo "bar)
#+END_SRC
This seems to be a bug, present even in the most up-to-date version of org-mode. I've reported it on the org-mode mailing list.
EDIT: Charles Berry pointed out on the ML that this is not a bug. The opposite of "silent" is "replace", so the second source block should read:
#+BEGIN_SRC elisp :results output replace
;; this one is being outputted
(princ "foo")
#+END_SRC
See http://thread.gmane.org/gmane.emacs.orgmode/108001/focus=108008
and the :results section in the manual http://orgmode.org/org.html#results
Note that you get to pick one value for every section (collection, type, format, handling) - if you don't pick a value for a section, a default value is picked. In the above, there was no explicit value for "handling", so the default value from the property still controlled.

org-babel: want all noweb block references to appear verbatim on export

Consider the following MVE in org-mode -- it contains my full question in detail. But, in summary, with some code blocks, some noweb references to other code blocks are substituted inline when I export the document, and, with other code blocks, the noweb references, in double broket quotes, are copied verbatim into the exported PDF. I do not know what causes this difference in behavior and I don't know how to control it, but I'd like to. I'd like to be able to specify that some blocks have behavior 1 (references substituted) and other blocks have behavior 2 (references verbatim).
The PDF that results from org-export is at this link
#+BEGIN_COMMENT
The emacs lisp block must export results, even though the results are none,
otherwise the block will not be eval'ed on export, and we will get
unacceptable confirmation requests for all the subsequent python blocks.
#+END_COMMENT
#+BEGIN_SRC emacs-lisp :exports results :results none
(setq org-confirm-babel-evaluate nil)
#+END_SRC
** PyTests
Define the test and cases. This code must be tangled out to an external file
so =py.test= can see it.
When I /export/ this to PDF, the noweb references, namely =<<imports>>= and
=<<definitions>>=, are substituted inline, so the typeset version of this
block in the PDF shows ALL the code. This is not what I want.
#+NAME: test-block
#+BEGIN_SRC python :noweb yes :tangle test_foo.py
<<imports>>
<<definitions>>
def test_smoke ():
np.testing.assert_approx_equal (foo_func (), foo_constant)
#+END_SRC
#+RESULTS: test-block
: None
The following blocks import prerequisites and do a quick smoke test:
** Do Some Imports
#+NAME: imports
#+BEGIN_SRC python
import numpy as np
#+END_SRC
#+RESULTS: imports
: None
** Define Some Variables
However, in the typeset PDF, the noweb reference =<<foo-func>>= in the block
below is /not/ substituted in-line, but rather appears verbatim. I want /all/
noweb references to appear verbatim in the exported, typeset, PDF document,
just like this one.
#+NAME: definitions
#+BEGIN_SRC python
foo_constant = 42.0
<<foo-func>>
#+END_SRC
#+RESULTS: definitions
** Define Some Functions
*** Foo Function is Really Interesting
#+NAME: foo-func
#+BEGIN_SRC python
def foo_func () :
return 42.000
#+END_SRC
#+RESULTS: foo-func
: None
We want results from pytest whether it succeeds or fails, hence the /OR/ with
=true= in the shell
#+BEGIN_SRC sh :results output replace :exports both
py.test || true
#+END_SRC
#+RESULTS:
: ============================= test session starts ==============================
: platform darwin -- Python 2.7.10, pytest-2.8.0, py-1.4.30, pluggy-0.3.1
: rootdir: /Users/bbeckman/foo, inifile:
: collected 1 items
:
: test_foo.py .
:
: =========================== 1 passed in 0.06 seconds ===========================
Found the appropriate references here
Here is a corrected PDF exported from the following .org file.
And here is the corrected MVE (it, itself, explains the correction):
#+BEGIN_COMMENT
The emacs lisp block must export results, even though the exports are none,
otherwise the block will not be eval'ed on export, and we will get unacceptable
confirmation requests for all the subsequent python blocks.
#+END_COMMENT
#+BEGIN_SRC emacs-lisp :exports results :results none
(setq org-confirm-babel-evaluate nil)
#+END_SRC
** PyTests
Define the test and cases. This code must be tangled out to an external file
so =py.test= can see it.
When I /export/ this to PDF, the noweb references, namely =<<imports>>= and
=<<definitions>>=, are *NOT* substituted inline, but typeset verbatim. This
is what I want. You get this behavior by saying =:noweb no-export= in the
header.
#+NAME: test-block
#+BEGIN_SRC python :tangle test_foo.py :noweb no-export :exports code :results none
dummy_for_org_mode = True
<<imports>>
<<definitions>>
def test_smoke ():
np.testing.assert_approx_equal (foo_func (), foo_constant)
#+END_SRC
The following blocks import prerequisites and do a quick smoke test:
** Do Some Imports
#+NAME: imports
#+BEGIN_SRC python :exports code :results none
import numpy as np
#+END_SRC
** Define Some Variables and Functions
In this block, I want the noweb reference =<<foo-func>>= in the block to be
substituted in-line and not to appear verbatim. Do that by saying
=:noweb yes= in the header.
#+NAME: definitions
#+BEGIN_SRC python :noweb yes :exports code :results none
foo_constant = 42.0
<<foo-func>>
#+END_SRC
** Define Some Functions
*** Foo Function is Really Interesting
Here, I want to talk about the implementation of foo function in detail, but I
don't want its code to be exported again, just to appear in the original
=.org= file as I reminder or note to me.
#+NAME: foo-func
#+BEGIN_SRC python :exports none :results none
def foo_func () :
return 42.000
#+END_SRC
** Run the Tests
We want results from pytest whether it succeeds or fails, hence the /OR/ with
=true= in the shell
#+BEGIN_SRC sh :results output replace :exports both
py.test || true
#+END_SRC
#+RESULTS:
: ============================= test session starts ==============================
: platform darwin -- Python 2.7.10, pytest-2.8.0, py-1.4.30, pluggy-0.3.1
: rootdir: /Users/bbeckman/foo, inifile:
: collected 1 items
:
: test_foo.py .
:
: =========================== 1 passed in 0.08 seconds ===========================

Org-mode fails to export when passing arguments to code blocks that exports results

When passing arguments to code block that exports results Org-mode fails to export with the error "Wrong type argument: listp". How can I fix this?
Here is an example. When it is exported it gives the error 'Wrong type argument: listp, "bar"'.
#+TITLE: Example
#+SOURCE: example-one
#+BEGIN_SRC emacs-lisp :exports results
(setq foo "bar")
#+END_SRC
#+SOURCE: example-two
#+BEGIN_SRC emacs-lisp :exports results :var x=example-one
(setq foo (concat x x))
#+END_SRC
I am running Org-mode 7.6 in Emacs 23.3.1.
This issue might simply be with the older copy of Org that you're running. Tested it today with a recent git pull and get the results below. As pmr suggested, you might have better luck asking on the mailing list ( emacs-orgmode#gnu.org ) since there might be someone there who would know what caused this issue and what might have been changed to resolve it in later versions.
The features and examples discussed in the manual are based on the current release version (7.8.03 in this case) so they will not always be compatible with older versions. Does the info-node in your version indicate that it should work?
These 2 commands will show you the associated info-nodes for that section of the Org Manual
; The node itself
(Info-goto-node "(org) var")
; Parent node, in case the first node isn't present
(Info-goto-node "(org) Working With Source Code")
Test Results
Org
* Test variable passing
Headlines are created to split the code blocks apart. When trying to eval on export I'm getting a syntax read error which was reported here: http://lists.gnu.org/archive/html/emacs-orgmode/2012-01/msg00993.html
** Ex 1
#+name: example-one
#+BEGIN_SRC emacs-lisp :exports results
(setq foo "bar")
#+END_SRC
** Ex 2
#+name: example-two
#+BEGIN_SRC emacs-lisp :exports results :var x=example-one
(setq foo (concat x x))
#+END_SRC
Latex
\vspace*{1cm}
Headlines are created to split the code blocks apart. When trying to eval on export I'm getting a syntax read error which was reported here: \href{http://lists.gnu.org/archive/html/emacs-orgmode/2012-01/msg00993.html}{http://lists.gnu.org/archive/html/emacs-orgmode/2012-01/msg00993.html}
\section{Ex 1}
\label{sec-1}
\begin{verbatim}
bar
\end{verbatim}
\section{Ex 2}
\label{sec-2}
\begin{verbatim}
barbar
\end{verbatim}
in your .emacs file - make sure you have the following line:
(setq org-babel-load-languages
(quote ((emacs-lisp . t))))