org2json.py

   1 # org2json
   2 #  parse a org-mode file and convert it to JSON
   3
   4 __author__ = "Sridhar Ratnakumar <http://nearfar.org/>"
   5
   6 import re
   7 from itertools import count
   8
   9 from simplejson import dumps
  10
  11 def rindexed(seq):
  12     """
  13     >>> l = [5,7,9]
  14     >>> print list(rindexed(l))
  15     [(2, 9), (1, 7), (0, 5)]
  16     """
  17     return zip(
  18         range(len(seq))[::-1],
  19         reversed(seq))
  20
  21 def reverse(iter):
  22     l = list(iter)
  23     l.reverse()
  24     return l
  25
  26
  27 def org2py(orgtext):
  28     """Parse the given org file text and return the Python data structure
  29
  30     >>> j = org2py(open('sample.org').read())
  31     >>> j=list(j)
  32     >>> j[-1][0]
  33     '* Projects'
  34     >>> j[-2][0]
  35     '* Whims'
  36     >>> j[-3][0]
  37     '* Online Stuff'
  38     >>> j[-4][0]
  39     '* Stuff'
  40     >>> j[-5][0]
  41     '* Travels'
  42     """
  43     lines = orgtext.splitlines()
  44     def e(i1, i2):
  45         return '\n'.join(lines[i1:i2])
  46
  47     def by_star():
  48         last_index = len(lines)
  49         for index, line in rindexed(lines):
  50             if line.startswith("*"):
  51                 yield [index, e(index, last_index)]
  52                 last_index = index
  53
  54     def hier(items):
  55         STARS_PAT = re.compile(r"^(\**) ")
  56         def nstars(line):
  57             """
  58             >>> nstars("*** Foo Bar")
  59             3
  60             """
  61             return len(STARS_PAT.match(line).group(1))
  62
  63         def node(text, children):
  64             return (text, children)
  65
  66         items = list(items) # exhaust the iterator, if any.
  67
  68         # Create [[], [], [], ...] to hold stack values
  69         # for level=k at index k.
  70         # hence, we need a stack of length 1+max(k)
  71         max_level = max([nstars(t) for i,t in items])
  72         istack = [[] for x in range(1+max_level)]
  73
  74         pn = nstars(items[0][1]) # last item
  75         for index, text in items:
  76             n = nstars(text)
  77             assert n>0
  78             if n == pn:
  79                 # previous sibling
  80                 istack[pn].append(node(text, []))
  81             elif n < pn:
  82                 # up to the parent
  83                 istack[n].append(node(text, reverse(istack[pn])))
  84                 istack[pn] = []
  85             elif n > pn:
  86                 # children of one of the top nodes
  87                 istack[n].append(node(text, []))
  88             pn = n
  89
  90         return reverse(istack[1])
  91
  92     return hier(by_star())
  93
  94 def org2json(orgtext):
  95     """
  96     >>> org2py(open('sample.org').read())
  97     """
  98     return dumps(org2py(orgtext))
  99
 100
 101 if __name__ == '__main__':
 102     from doctest import testmod
 103     testmod()