# -*- coding: UTF-8 -*-

import re
import sys

def make_re (s) :
  # So there is no need to import re in the calling module. 
  return re.compile(s, re.U)

### Some useful regular expressions

re_whitespaces_star     = make_re(u"\s*")
re_whitespaces_plus     = make_re(u"\s+")
re_non_whitespaces_star = make_re(u"\S*")
re_non_whitespaces_plus = make_re(u"\S+")
re_words_plus           = make_re(u"[a-zA-Z]+")
re_alphanumerics_plus   = make_re(u"\w+")

re_open_bracket         = make_re(u"\(")
re_close_bracket        = make_re(u"\)")
re_open_square          = make_re(u"\[")
re_close_square         = make_re(u"\]")
re_open_angle           = make_re(u"<")
re_close_angle          = make_re(u">")

re_single_quote         = make_re(u"'")
re_double_quote         = make_re(u'\"')
re_comma                = make_re(u",")
re_dot                  = make_re(u"\.")
re_ampersand            = make_re(u"&")
re_percent              = make_re(u"%")
re_hash                 = make_re(u"#")
re_equal_sign           = make_re(u"=")
re_forward_slash        = make_re(u"/")
# re_backward_slash???

re_any_but_double_quote = make_re(u'[^\"]*')

re_integer              = make_re(u"[0-9]+")
re_float                = make_re(u"([+-]*[0-9]*\.)?[0-9]+(e[+-][0-9]+)?")


### Exceptions

class DigestException (Exception) :
  pass

### Functions

def digest (regexp, inp, pos, exception_on_fail=False) :
  # Returns a tuple (unicode, pos)
  # Typical use: s, pos = digest (regexp, inp, pos) 
  
  match = regexp.match(inp, pos)
  
  if match == None :
    if exception_on_fail :
      before = unicode(inp[max(0, pos-20):pos])
      after  = unicode(inp[pos:min(len(inp), pos+20)])
      raise DigestException(before, after)
    else : 
      return (u"", pos)
  else : 
    pos2 = match.end()
    return (inp[pos:pos2], pos2)

def discard (regexp, inp, pos, exception_on_fail=False) :
  # Returns an integer
  # Typical use: pos = discard (regexp, inp, pos) 
  
  match = regexp.match(inp, pos)
  
  if match == None :
    if exception_on_fail : 
      before = unicode(inp[max(0, pos-20):pos])
      after  = unicode(inp[pos:min(len(inp), pos+20)])
      raise DigestException(before, after)
    else : 
      print "No match."
      return pos
  else :
    return match.end()