import sys
import string
import re
import util
import ntriples
import rdfstore

S, P, O, C = 0, 1, 2, 3
V, T = 0, 1
URI = 'tag:infomesh.net,2001-08-07:URI'
LIT = 'tag:infomesh.net,2001-08-07:Literal'
ANON = 'tag:infomesh.net,2001-08-07:Anon'
VAR = 'tag:infomesh.net,2001-08-07:Var'
RC = 'tag:infomesh.net,2001-08-07:RootContext'
CONT = 'tag:infomesh.net,2001-08-07:Context' # For extensibility

# NTriples = ntriples.NTriples

class SWIPTQL(ntriples.NTriples): 
   """A simple query language: NTriples with ?x vars."""

   def __init__(self):
      ntriples.NTriples.__init__(self)
      self.q = r'(<[^>]+>|_:[^\s]+|\?[^\s]+|\"(?:\\\"|[^"])*\")'
      self.qt = r'[ \t]*'+self.q+r'[ \t]+'+self.q+r'[ \t]+'+self.q+self.eol
      self.regq = re.compile(self.qt, re.S)
      self.uvs = []

   def appendq(self, line):
      """Parses and validates a line, and then adds it to self.quads"""
      # print str(line)
      if self.regq.match(line):
         terms = self.regq.findall(line)[0]
         for term in terms: 
            if term[0] == '<' and term[-1] == '>': # Term is a URI-view
               self.CurrentQuad.append((term[1:-1], URI))
            elif term[:2] == '_:': # Term is an unlabelled node
               bnode = term[2:]
               # This filters out incorrect bNodes! CWM produced these...
               if re.compile(r'[A-Za-z][A-Za-z0-9]*', re.S).match(bnode): 
                  # Check the bnode isn't already used in a different context
                  bnode = self.correctbnode(bnode)
                  self.bnodes.append((bnode, self.CurrentContext))
                  self.CurrentQuad.append((bnode, ANON))
               else: raise 'bnode: "'+str(term[2:])+'" is not a valid bnode'
            elif term[0] == '"' and term[-1] == '"': 
               self.CurrentQuad.append((util.unescape(term[1:-1]), LIT))
            elif term[0] == '?': # Term is an unlabelled node
               bnode = term[1:]
               # This filters out incorrect bNodes! CWM produced these...
               if re.compile(r'[A-Za-z][A-Za-z0-9]*', re.S).match(bnode): 
                  # Check the bnode isn't already used in a different context
                  bnode = self.correctbnode(bnode)
                  self.uvs.append((bnode, self.CurrentContext))
                  self.CurrentQuad.append((term[1:], VAR))
               else: raise 'bnode: "'+str(term[1:])+'" is not a valid bnode'
            else: raise 'Term '+str(term)+' is not a valid NTriples term.'
         self.CurrentQuad.append((self.CurrentContext, RC))
         self.add(self.CurrentQuad)
         self.CurrentQuad = [] # Reset the current quad!
      elif self.regc.match(line): pass # Line is just whitespace, or is a comment
      else: raise 'line: "'+line+'" isn\'t fine' # Validity error!

def run(): 
   x = SWIPTQL()
   x.parsen(sys.argv[1])
   x.parsen(sys.argv[1])
   x.parsen(sys.argv[1])
   x.printquads()

# Main program

if __name__ == "__main__":
    run()

# Phew