xinclude/table.py

   1 #!/usr/bin/python
   2 # -*- coding: utf-8 -*-
   3
   4 import sys
   5 import re
   6
   7 from urlparse import urlparse
   8 from lxml import etree
   9 from lxml.builder import ElementMaker
  10 from treecutter import constants as const
  11
  12 def linkify(text):
  13     db = ElementMaker(namespace=const.DB_NS, nsmap=const.NSMAP)
  14     r = re.search(r"(?P<url>https?://[^ ]+)\|(?P<title>[\w\-\.]+)", text)
  15     if r:
  16         rep = r.groups(r.group(1))
  17         text = db.link(rep[1],**{const.XLINK+"href": rep[0]})
  18     return text
  19
  20 class Table(object):
  21     def __init__(self, tablefile, title):
  22         self.tablefile  = tablefile
  23         self.title = title
  24         self.cols = []
  25
  26     def parse(self):
  27         f = file(self.tablefile, 'r')
  28         for line in f:
  29             c = line.split()
  30             self.cols.append(c)
  31
  32     def db_xml(self):
  33         db = ElementMaker(namespace=const.DB_NS, nsmap=const.NSMAP)
  34         cols = self.cols
  35         nrcol = str(len(cols[0]))
  36         if cols[0][0][0] == '*':
  37             cols[0][0] = cols[0][0][1:]
  38             h = cols.pop(0)
  39             row = db.row()
  40             for e in h:
  41                 row.append(db.entry(linkify(e), align="center"))
  42             head = db.thead(row)
  43         body = db.tbody()
  44         for r in cols:
  45             row = db.row()
  46             body.append(row)
  47             for e in r:
  48                 row.append(db.entry(linkify(e)))
  49         tab = db.table(db.title(self.title),
  50                        db.tgroup(head,body,cols=nrcol,
  51                                  colsep='1',rowsep='1',align='left'),
  52                        frame='all')
  53         return tab
  54
  55 if __name__ == "__main__":
  56     for arg in sys.argv[1:]:
  57         al = arg.split("=")
  58         if al[0] == "lang":
  59             lang = al[1]
  60         if al[0] == "xptr":
  61             argument = al[1].decode('utf-8')
  62
  63     (tablefile, title) = argument.split('|')
  64     tab = Table(tablefile,title)
  65     tab.parse()
  66     txml = tab.db_xml()
  67
  68     sys.stdout.write(etree.tostring(txml,encoding='UTF-8',pretty_print=False))