Skip to content

Commit

Permalink
Create 'table' for post processing of Babel result
Browse files Browse the repository at this point in the history
  • Loading branch information
@howardabrams
howardabrams committed Aug 14, 2015
1 parent a195d1b commit ba7ff6e
Showing 1 changed file with 71 additions and 107 deletions.
178 changes: 71 additions & 107 deletions babel/sh-formatting.org
View file
Original file line number Diff line number Diff line change
Expand Up @@ -7,116 +7,80 @@
This is part of my Library of Babel collection. Remember to hit the
=C-c C-v i= to add these sections.

* Table Separator

Instead of relying on the org-mode to separate the output into a
table based on spaces, let’s allow me to do a :post call and
break up the cells based on a regular expression.

To begin, let’s have some sample data to play with:

#+NAME: example-table
#+BEGIN_EXAMPLE
a b c
d ei ei o f
g h i
#+END_EXAMPLE

And

#+NAME: example-table2
#+BEGIN_EXAMPLE
a - b c
doh - ei ei o f
g - h i
#+END_EXAMPLE

Since the results from an =sh= block come in as a single string, we
use the above example code to pass into this block:

#+NAME: table_sep
#+BEGIN_SRC elisp :var data=example-table regex=" +" :results code
(defun split-line-by-colpos (str positions)
"Split a string, STR, based on a list of column POSITIONS."
(let ((end-col (car positions)))
(if (and end-col str (not (string-empty-p str)))
(cons
(string-trim (substring str 0 end-col))
(split-line-by-colpos (substring str end-col) (cdr positions)))
(list (string-trim str)))))

;; (split-line-by-colpos "foo bar bazzies bo" '(4 4 8)) => ("foo" "bar" "bazzies" "bo")

(defun analyze-line-for-colpos (str &optional regex)
"Calculates columns of first line, STR, based on a regular expression, REGEX.
Defaults for columns with multiple spaces (not a single space)."
(unless regex (setq regex " *"))
(if (and str (not (string-empty-p str)))
(let ((pos (string-match regex str))
(epos (match-end 0)))
(when pos
(cons epos
(analyze-line-for-colpos (substring str epos) regex))))))

;; (analyze-line-for-colpos "a b c") ;; => (6 11)

;; (let* ((s "abc defghi jkl")
;; (pos (analyze-line-for-colpos s)))
;; (split-line-by-colpos s pos)) ;; => ("abc" "defghi" "jkl")

;; (let* ((s "abc - def ghi jkl")
;; (pos (analyze-line-for-colpos s " - ")))
;; (split-line-by-colpos s pos)) ;; => ("abc -" "def ghi jkl")

(defun string-to-table-with-separator (str regex)
(prin1-to-string
(let* ((lines (split-string str "\n"))
(pos (analyze-line-for-colpos (car lines) regex)))
;; (message "First: '%s'" (car lines))
;; (princ pos)
(mapcar (lambda (line) (split-line-by-colpos line pos)) lines))))

;; Function to apply these functions
(if (stringp data)
(string-to-table-with-separator data regex)
(message "Huh ... data is something else")
data)
#+END_SRC

#+RESULTS: table_sep
#+BEGIN_SRC elisp
(org-babel-script-escape "((\"a\" \"b\" \"c\") (\"d\" \"ei ei o\" \"f\") (\"g\" \"h\" \"i\"))")
#+END_SRC

#+RESULTS:
| a | b | c |
| d | ei ei o | f |
| g | h | i |

So, we can call =docker= to get some tabular data:

#+BEGIN_SRC sh :results output
* Table by Separator

The following is a simple approach looking for a regular expression
that works on *every* row. We also run every line we separate
through a filter to remove any blank entries... maybe that should be
configurable option.

#+NAME: table
#+BEGIN_SRC elisp :results value table :var data=example-table regex=" +"
(flet ((split-the-line (line)
(filter (lambda (cell) (not (string-empty-p cell)))
(split-string (string-trim line) regex))))
(if (stringp data)
(mapcar 'split-the-line (split-string data "[\n\r]+" t))
data)) ;; Not a string? Just return the value...
#+END_SRC

#+RESULTS: table
| a | b | c |
| d | ei ei o | f |
| g | h | i |

Let’s try with a different type of table, formatted more like a CSV:

#+HEADER: :post table(data=*this*, regex=",")
#+BEGIN_SRC sh :results output
echo "a, b,c"
echo "doh,ei ei o,f"
echo "g,h, i"
#+END_SRC

#+RESULTS:
| a | b | c |
| doh | ei ei o | f |
| g | h | i |

Once more, but this time with output that is already in some sort of
strange table output (can you say nova?):

#+HEADER: :post table(data=*this*, regex="[\|\+]")
#+BEGIN_SRC sh :results output
echo "| a | b | c |"
echo "| doh | ei ei o | f |"
echo "| g | h | i |"
#+END_SRC

#+RESULTS:
| a | b | c |
| doh | ei ei o | f |
| g | h | i |

Need a real world example, so call =docker= to get some tabular data:

#+HEADER: :prologue $(/usr/local/bin/boot2docker shellinit)
#+BEGIN_SRC sh :results output
docker images
#+END_SRC
#+END_SRC

#+RESULTS:
: REPOSITORY TAG IMAGE ID CREATED VIRTUAL SIZE
: <none> <none> 511136ea3c5a 2 years ago 0 B

The end result is something like:

#+BEGIN_SRC sh :post table_sep(data=*this*) :results output
docker images
#+END_SRC

#+RESULTS:
: "'((\"REPOSITORY\" \"TAG\" \"IMAGE ID\" \"CREATED\" \"VIRTUAL SIZE\") (\"<none>\" \"<none>\" \"511136ea3c5a\" \"2 years ago\" \"0 B\") (\"\"))"
#+RESULTS:
: REPOSITORY TAG IMAGE ID CREATED VIRTUAL SIZE
: tomcat 6.0 8c4e1a4ca737 4 weeks ago 345.8 MB
: rand/docker-clojurescript latest f6e0e40b4408 7 months ago 1.091 GB

Now, let’s see if we really have something good here:

#+HEADER: :prologue $(/usr/local/bin/boot2docker shellinit)
#+BEGIN_SRC sh :post table(data=*this*) :results output
docker images
#+END_SRC

#+RESULTS:
| REPOSITORY | TAG | IMAGE ID | CREATED | VIRTUAL SIZE |
| tomcat | 6.0 | 8c4e1a4ca737 | 4 weeks ago | 345.8 MB |
| rand/docker-clojurescript | latest | f6e0e40b4408 | 7 months ago | 1.091 GB |

:RESULTS:
(("REPOSITORY" "TAG" "IMAGE ID" "CREATED" "VIRTUAL SIZE")
("<none>" "<none>" "511136ea3c5a" "2 years ago" "0 B")
(""))
:END:
Now, if we can just teach org-mode that the first line is a header,
we’d be golden.

0 comments on commit ba7ff6e

Please sign in to comment.