// $Id: ParserNew.cs 1295 2007-05-11 16:52:51Z josd $
// PxButton | build | csc /o /doc:Euler.xml *.cs |

using System;
using System.Collections;
using System.Text;
using System.Xml;

/// <summary>N3 parser</summary>
/// <author>Jos De Roo</author>

public class Parser {
  internal String str;                      // N3 string
  internal int pos = 0;                     // tokenizer position
  internal ArrayList vt = null;             // quantified variables table
  internal Euler r = null;                  // root Euler object
  internal String u = null;                 // base URI of the RDF resource
  
  internal const String ST = "{}[]()<>\"';,.^! \t\r\n\\";
  internal const String LOGa = "a";
  internal const String LOGe = "=";
  internal const String LOGi = "=>";
  internal const String LOGimplies = "<http://www.w3.org/2000/10/swap/log#implies>";
  internal const String DPO = "<http://www.daml.org/2001/03/daml+oil";
  internal const String ONT = "<http://www.w3.org/2001/10/daml+oil";
  internal const String OWL = "<http://www.w3.org/2002/07/owl";
  internal const String OWLequivalentTo = "<http://www.w3.org/2002/07/owl#equivalentTo>";
  internal const String OWLsameIndividualAs = "<http://www.w3.org/2002/07/owl#sameIndividualAs>";
  internal const String OWLsameAs = "<http://www.w3.org/2002/07/owl#sameAs>";
  internal const String OWLsameClassAs = "<http://www.w3.org/2002/07/owl#sameClassAs>";
  internal const String OWLequivalentClass = "<http://www.w3.org/2002/07/owl#equivalentClass>";
  internal const String OWLsamePropertyAs = "<http://www.w3.org/2002/07/owl#samePropertyAs>";
  internal const String OWLequivalentProperty = "<http://www.w3.org/2002/07/owl#equivalentProperty>";
  internal const String OWLdifferentIndividualFrom = "<http://www.w3.org/2002/07/owl#differentIndividualFrom>";
  internal const String OWLdifferentFrom = "<http://www.w3.org/2002/07/owl#differentFrom>";
  internal const String OWLFunctionalProperty = "<http://www.w3.org/2002/07/owl#FunctionalProperty>";
  internal const String OWLInverseFunctionalProperty = "<http://www.w3.org/2002/07/owl#InverseFunctionalProperty>";
  internal const String OWLUnambiguousProperty = "<http://www.w3.org/2002/07/owl#UnambiguousProperty>";
  internal const String OWLUniqueProperty = "<http://www.w3.org/2002/07/owl#UniqueProperty>";
  internal const String XSDstring = "<http://www.w3.org/2001/XMLSchema#string>";
  internal const String RDFtype = "<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>";
  internal const String RDFfirst = "<http://www.w3.org/1999/02/22-rdf-syntax-ns#first>";
  internal const String RDFrest = "<http://www.w3.org/1999/02/22-rdf-syntax-ns#rest>";
  internal const String RDFnil = "<http://www.w3.org/1999/02/22-rdf-syntax-ns#nil>";
  
  /// <summary>constructs an N3 parser</summary>

  public Parser() {
  }
  
  /// <summary>constructs an N3 parser</summary>
  /// <param name="s">the N3 string</param>
  /// <param name="vartab">quantified variables table</param>
  /// <param name="root">Euler object</param>
  /// <param name="uri">URI of the RDF resource</param>

  public Parser(String s, ArrayList vartab, Euler root, String uri) {
    str = s;
    vt = vartab;
    r = root;
    u = uri;
  }
  
  /// <summary>N3 triple parse method</summary>
  /// <returns>Euler object</returns>

  public virtual Euler Parse() {
    String nt = tokenize();
    if (nt == null) return null;
    return parse(true, nt);
  }
  
  /// <summary>N3 node parse method</summary>
  /// <param name="nt">next token</param>
  /// <returns>Euler object</returns>

  public virtual Euler Parse(String nt) {
    return parse(false, nt);
  }
  
  internal Euler parse(bool b, String nt) {
    Euler e = new Euler();
    try {
      if (b) {
        if (nt == null) e.subj = null;
        else if (nt.Equals("{")) {
          nt = tokenize();
          if (nt.Equals(".")) e.subj = parse(false, "{}");
          else e.subj = parse(true, nt);
          Euler el = e.subj;
          nt = tokenize();
          while (el.near != null) el = el.near;
          while (!nt.Equals("}")) {
            el.near = parse(true, nt);
            while (el.near != null) el = el.near;
            nt = tokenize();
          }
        }
        else if (nt.Equals("[")) {
          e.subj = parse(true, null);
          if (e.subj.verb == null) r.verb = ";]";
          nt = tokenize();
          if (!nt.Equals("]")) Console.Error.WriteLine("** (p1) expecting ] at " + e + " but got " + nt);
        }
        else if (nt.Equals("(")) {
          list(e);
          e.subj = e.obj;
          e.obj = null;
        }
        else {
          e.subj = parse(false, nt);
          if (nt.Equals("this")) e.subj.verb = '<' + u + "#frag" + e.GetHashCode() + '>';
        }
        path(e.subj);
        nt = tokenize();
      }
      if (nt == null || nt.Equals(";")) {
        e.bound = true;
        return e;
      }
      else if (e.subj != null && nt.StartsWith("_:")) {
        e.cverb = nt;
        if (!vt.Contains(nt)) vt.Add(nt);
      }
      else if (nt.Equals("[")) {
        String a = "_:" + e.GetHashCode();
        if (!vt.Contains(a)) vt.Add(a);
        Euler ap = parse(true, a);
        e.cverb = a;
        if (ap.verb != null) r.verb = ";]";
        nt = tokenize();
        if (!nt.Equals("]")) Console.Error.WriteLine("** (p2) expecting ] at " + e + " but got " + nt);
        e.near = ap;
      }
      else if (nt.Equals("(")) {
        list(e);
        return e.obj;
      }
      else e.cverb = nt;
      e.verb = e.cverb;
      if (e.verb.EndsWith("@@")) e.verb = e.verb.Substring(0, e.verb.Length - 2);
      String s1 = e.verb;
      String s2 = "";
      if (e.verb.IndexOf('"') != -1) {
        s1 = e.verb.Substring(0, e.verb.IndexOf('"'));
        s2 = e.verb.Substring(e.verb.IndexOf('"'));
      }
      if (r != null && s1.Length > 0 && !s1.Equals(LOGa) && !s1.Equals(LOGe) &&
          s1[0] != '(' && s1[0] != '<' && s1[0] != '[' &&
          s1[0] != '_' && s1[0] != '"' && s1[0] != '\'' && s1.IndexOf(':') != - 1) {
        String pf = s1.Substring(0, s1.IndexOf(':') + 1);
        String pg = (String) r.nsp[pf];
        if (pg != null) pf = pg;
        String qf = (String) r.hshtNs[pf];
        if (qf == null) {
          Console.Error.WriteLine("** no @prefix " + pf + " found, taking <" + u + "#>");
          qf = "<" + u + "#>";
          r.hshtNs[pf] = qf;
        }
        StringBuilder sb = new StringBuilder(pf);
        sb.Append(e.cverb.Substring(e.cverb.IndexOf(':') + 1));
        e.cverb = sb.ToString();
        sb = new StringBuilder(qf);
        sb.Insert(sb.Length - 1, s1.Substring(s1.IndexOf(':') + 1));
        e.verb = sb.ToString() + s2;
      }
      else if (u != null && s1.Length > 0 && s1[0] == '<' && s1.IndexOf(':') == - 1)
        e.verb = e.cverb = '<' + toURI(s1.Substring(1, s1.Length - 2)) + '>' + s2;
      if (e.verb.StartsWith("'")) e.verb = "'" + Double.Parse(e.verb.Substring(1, e.verb.Length - 2)) + "'";
      if (e.verb.Equals(LOGa)) e.verb = RDFtype;
      if (e.verb.Equals(LOGe)) e.verb = OWLsameAs;
      if (e.verb.Equals(LOGi)) e.verb = LOGimplies;
      if (e.verb.StartsWith(DPO)) e.verb = OWL + e.verb.Substring(DPO.Length);
      if (e.verb.StartsWith(ONT)) e.verb = OWL + e.verb.Substring(ONT.Length);
      if (e.verb.Equals(OWLequivalentTo)) e.verb = OWLsameAs;
      if (e.verb.Equals(OWLsameIndividualAs)) e.verb = OWLsameAs;
      if (e.verb.Equals(OWLsameClassAs)) e.verb = OWLequivalentClass;
      if (e.verb.Equals(OWLsamePropertyAs)) e.verb = OWLequivalentProperty;
      if (e.verb.Equals(OWLdifferentIndividualFrom)) e.verb = OWLdifferentFrom;
      if (e.verb.Equals(OWLUnambiguousProperty)) e.verb = OWLInverseFunctionalProperty;
      if (e.verb.Equals(OWLUniqueProperty)) e.verb = OWLFunctionalProperty;
      if (e.verb.StartsWith("?") && !vt.Contains(e.verb)) vt.Add(e.verb);
      if (vt != null && vt.Count > 0) e.varid = vt.IndexOf(e.verb);
      if (e.varid == - 1) e.bound = true;
      if (nt.Equals(".") && (str[pos] == ' ' || str[pos] == '\n')) {
        e.obj = parse(false, "");
        e.verb = "";
        e.cverb = "";
        e.bound = true;
        return e;
      }
      if (str != null && pos < str.Length && str[pos] == '^' && str[pos + 1] == '^') {
        pos = pos + 2;
        Euler et = parse(false, tokenize());
        if (!et.verb.Equals(XSDstring)) {
          e.subj = e.copy();
          e.verb = e.cverb = "^^";
          e.bound = true;
          e.obj = et;
          String.Intern(e.obj.verb);
          if (e.subj.bound) Datatype.Compare(e.obj.verb, r.getLit(e.subj), r.getLit(e.subj));
        }
      }
      if (b) {
        nt = tokenize();
        if (nt.Equals("{")) {
          nt = tokenize();
          if (nt.Equals(".")) e.obj = parse(false, "{}");
          else e.obj = parse(true, nt);
          Euler el = e.obj;
          nt = tokenize();
          while (nt != null && !nt.Equals("}")) {
            el.near = parse(true, nt);
            el = el.near;
            nt = tokenize();
          }
        }
        else if (nt.Equals("[")) {
          e.obj = parse(true, null);
          if (e.obj.verb != null) r.verb = ";]";
          nt = tokenize();
          if (!nt.Equals("]")) Console.Error.WriteLine("** (p3) expecting ] at " + e + " but got " + nt);
        }
        else if (nt.Equals("(")) list(e);
        else e.obj = parse(false, nt);
        path(e.obj);
        nt = tokenize();
        if (nt.EndsWith("\"\"\"")) {
          e.obj.cverb = e.obj.cverb + nt;
          e.obj.verb = e.obj.cverb;
          nt = tokenize();
        }
        Euler el2 = e;
        while (nt.Equals(";") || nt.Equals(",")) {
          while (el2.near != null) el2 = el2.near;
          if (nt.Equals(";")) {
            nt = tokenize();
            if (nt.Equals("]")) {
              r.verb = ";]";
              nt = ";";
              break;
            }
            else if (nt.Equals(".")) break;
          }
          else nt = el2.cverb;
          el2.near = parse(false, nt);
          el2.near.subj = e.subj;
          nt = tokenize();
          if (nt.Equals("{")) {
            nt = tokenize();
            if (nt.Equals(".")) el2.near.obj = parse(false, "{}");
            else el2.near.obj = parse(true, nt);
            Euler ef = el2.near.obj;
            nt = tokenize();
            while (nt != null && !nt.Equals("}")) {
              ef.near = parse(true, nt);
              ef = ef.near;
              nt = tokenize();
            }
          }
          else if (nt.Equals("[")) {
            el2.near.obj = parse(true, null);
            if (el2.near.obj.verb != null) r.verb = ";]";
            nt = tokenize();
            if (!nt.Equals("]")) Console.Error.WriteLine("** (p4) expecting ] at " + e + " but got " + nt);
          }
          else if (nt.Equals("(")) list(el2.near);
          else el2.near.obj = parse(false, nt);
          nt = tokenize();
          if (nt.EndsWith("\"\"\"")) {
            el2.near.obj.cverb = el2.near.obj.cverb + nt;
            el2.near.obj.verb = el2.near.obj.cverb;
            nt = tokenize();
          }
          swap(el2.near);
        }
        if (!nt.Equals(".") && !nt.Equals(";"))
          Console.Error.WriteLine("** (p5) expecting . or ; at " + e + " but got " + nt);
      }
      swap(e);
      e.far = r;
      return e;
    }
    catch (NullReferenceException exc) {
      Console.Error.WriteLine(exc.StackTrace);
      return e;
    }
  }
  
  internal String tokenize() {
    String nt = token();
    while (pos < str.Length && nt == null) nt = token();
    return nt;
  }
  
  internal String next() {
    int start = pos;
    while (pos < str.Length && ST.IndexOf(str[pos]) < 0) pos++;
    if (start == pos && ST.IndexOf(str[pos]) >= 0) pos++;
    return str.Substring(start, (pos) - (start));
  }
  
  internal String token() {
    if (r.verb != null && r.verb.Equals(".}")) return r.verb = "}";
    if (r.verb != null && r.verb.Equals(";]")) return r.verb = "]";
    if (pos >= str.Length) return null;
    int start = pos;
    String t = null;
    String nt = next();
    if (nt.IndexOf('#') != - 1) {
      t = nt.Substring(0, nt.IndexOf('#'));
      String v = r.verb;
      while (nt != null && !nt.Equals("\n") && pos < str.Length) nt = next();
      r.verb = v;
      if (t.Equals("")) return null;
    }
    else if (nt.Equals("@prefix") || nt.Equals("bind")) {
      String nsc = tokenize();
      if (nsc.Equals("default")) nsc = ":";
      String nsd = nsc;
      String nsu = tokenize();
      if (u != null && nsu.Length > 0 && nsu[0] == '<' && nsu.IndexOf(':') == - 1)
        nsu = '<' + toURI(nsu.Substring(1, nsu.Length - 2)) + '>';
      String nsv = (String) r.hshtNs[nsd];
      while (nsv != null && !nsu.Equals(nsv)) {
        nsd = "ns" + nsd;
        nsv = (String) r.hshtNs[nsd];
      }
      r.nsp[nsd] = nsd;
      r.nsp[nsc] = nsd;
      r.hshtNs[nsd] = nsu;
      tokenize();
      return null;
    }
    else if (nt.Equals("\"")) {
      StringBuilder sb = new StringBuilder("\"");
      if (str[pos] == '"' && str[pos + 1] == '"') {
        sb.Append(next());
        sb.Append(next());
        while (true) {
          sb.Append(next());
          if (sb.ToString().EndsWith("\"\"\"") && str[pos] != '"' && str[pos + 1] != '"') break;
        }
        t = sb.ToString();
      }
      else {
        nt = next();
        while (!nt.Equals("\"")) {
          if (nt.Equals("\\")) {
            sb.Append(nt);
            nt = next();
          }
          sb.Append(nt);
          if (pos >= str.Length) {
            Console.Error.WriteLine("** (t1) expecting \" at " + sb);
            break;
          }
          nt = next();
        }
        sb.Append("\"");
        String lang = "";
        if (str[pos] == '@') {
          lang = token();
          int i = lang.IndexOf('-');
          if (i != -1) sb.Append(lang.Substring(0, i) + lang.Substring(i).ToUpper());
          else sb.Append(lang);
        }
        t = sb.ToString();
      }
    }
    else if (nt.Equals("'")) {
      StringBuilder sb = new StringBuilder("'");
      nt = next();
      while (!nt.Equals("'")) {
        sb.Append(nt);
        if (pos >= str.Length) {
          Console.Error.WriteLine("** (t2) expecting ' at " + sb);
          break;
        }
        nt = next();
      }
      t = sb.Append("'").ToString();
    }
    else if (nt.Equals("<")) {
      StringBuilder sb = new StringBuilder("<");
      nt = next();
      while (!nt.Equals(">")) {
        sb.Append(nt);
        if (pos >= str.Length) {
          Console.Error.WriteLine("** (t2) expecting > at " + sb);
          break;
        }
        nt = next();
      }
      t = sb.Append(">").ToString();
    }
    else if (nt.Equals("=")) {
      if (str[pos] == '>') t = nt + next();
      else t = nt;
    }
    else if (nt.Equals("is")) {
      t = tokenize();
      nt = tokenize();
      if (!nt.Equals("of")) Console.Error.WriteLine("** (t4) expecting \"of\" but got " + nt);
      t = t + "@@";
    }
    else if (nt.Equals("has")) {
      t = tokenize();
      nt = tokenize();
      if (!nt.Equals("of")) Console.Error.WriteLine("** (t5) expecting \"of\" but got " + nt);
    }
    else if (nt.Equals(" ") || nt.Equals("\t") || nt.Equals("\r") || nt.Equals("\n") ||
             nt.Equals("-") || nt.Equals(">"))
      return null;
    else if (nt.StartsWith("_:") && !nt.EndsWith("_" + Euler.doc)) t = nt + "_" + Euler.doc;
    else t = nt;
    if (pos < str.Length && ST.IndexOf(t) == -1 && t[0] != '"' && str[pos] == '"') t = t + token();
    if (pos < str.Length && str[pos] == '.' &&
        str[pos + 1] != ' ' && str[pos + 1] != '\n' &&
        str[pos + 1] != ']' && str[pos + 1] != '}' && str[pos + 1] != '#') {
      int opos = pos;
      try {
        t = XmlConvert.ToDouble(t + next() + next()).ToString();
      }
      catch (Exception) {
        pos = opos;
      }
    }
    if (r.verb != null && t != null && !r.verb.Equals(".") && t.Equals("}")) {
      r.verb = ".}";
      return ".";
    }
    if (r.verb != null && t != null && !r.verb.Equals(";") && t.Equals("]")) {
      r.verb = ";]";
      return ";";
    }
    return r.verb = t;
  }
  
  internal void swap(Euler e) {
    if (e != null && e.subj == null && e.cverb.EndsWith("@@")) {
      e.subj = e.obj;
      e.obj = parse(false, e.cverb.Substring(0, e.cverb.Length - 2));
      e.verb = e.cverb = ".";
    }
    else if (e.subj != null && e.obj != null && e.cverb.EndsWith("@@")) {
      Euler el = e.subj;
      e.subj = e.obj;
      e.obj = el;
      e.cverb = e.cverb.Substring(0, e.cverb.Length - 2);
    }
  }
  
  internal void list(Euler e) {
    Euler el = e;
    while (true) {
      String nt = tokenize();
      if (nt.StartsWith("@")) {
        el.obj = parse(false, '?' + nt.Substring(1));
        nt = tokenize();
        if (!nt.Equals(")")) Console.Error.WriteLine("** (p6) expecting ) at " + el + " but got " + nt);
        break;
      }
      else if (nt.Equals(")")) {
        el.obj = parse(false, RDFnil);
        break;
      }
      else el.obj = parse(false, RDFfirst);
      if (nt.Equals("[")) {
        el.obj.obj = parse(true, null);
        tokenize();
      }
      else el.obj.obj = parse(false, nt);
      path(el.obj.obj);
      el.obj.near = parse(false, RDFrest);
      el = el.obj.near;
    }
    path(e.obj);
  }

  internal void path(Euler e) {
    if (str != null && pos < str.Length - 1 && str[pos] == '.' &&
        str[pos + 1] != ' ' && str[pos + 1] != '\n' &&
        str[pos + 1] != ']' && str[pos + 1] != '}' && str[pos + 1] != '#') {
      pos = pos + 1;
      e.subj = e.copy();
      e.verb = e.cverb = ".";
      e.bound = true;
      e.obj = parse(false, tokenize());
      String.Intern(e.obj.verb);
      path(e);
    }
    if (str != null && pos < str.Length - 1 && str[pos] == '^' &&
        str[pos + 1] != '^' && str[pos + 1] != ' ' && str[pos + 1] != '\n' &&
        str[pos + 1] != ']' && str[pos + 1] != '}' && str[pos + 1] != '#') {
      pos = pos + 1;
      e.subj = e.copy();
      e.verb = e.cverb = "^";
      e.bound = true;
      e.obj = parse(false, tokenize());
      String.Intern(e.obj.verb);
      path(e);
    }
  }

  internal String toURI(String s) {
    if (s.Equals("") || s.Equals("#")) s = u;
    if (!u.StartsWith("file:")) {
      try {
        if (s.EndsWith("#")) s = new Uri(new Uri(u.Substring(0, u.LastIndexOf('/') + 1)), s.Substring(0, s.Length -1)).ToString() + '#';
        else s = new Uri(new Uri(u), s).ToString();
      }
      catch (Exception e) {
        Console.Error.WriteLine(u + " " + s + " " + e);
      }
    }
    if (s.EndsWith(".n3#")) s = s.Substring(0, s.Length - 4) + "#";
    return s;
  }
}