library mustache.scanner;

import 'token.dart';
import 'template_exception.dart';

class Scanner {
  
  Scanner(String source, this._templateName, String delimiters,
      {bool lenient: true})
   : _source = source,
     _lenient = lenient,
     _itr = source.runes.iterator {
    
    if (source == '') {
      _c = _EOF;
    } else {
      _itr.moveNext();
      _c = _itr.current;
    }
    
    if (delimiters == null) {
      _openDelimiter = _openDelimiterInner = _OPEN_MUSTACHE;
      _closeDelimiter = _closeDelimiterInner = _CLOSE_MUSTACHE;
    
    } else if (delimiters.length == 3) {
      _openDelimiter = delimiters.codeUnits[0];
      _closeDelimiter = delimiters.codeUnits[2];
    
    } else if (delimiters.length == 5) {
      _openDelimiter = delimiters.codeUnits[0];
      _openDelimiterInner = delimiters.codeUnits[1];
      _closeDelimiterInner = delimiters.codeUnits[3];
      _closeDelimiter = delimiters.codeUnits[4];
    
    } else {
      throw new TemplateException(
              'Invalid delimiter string $delimiters', null, null, null);   
    }
  }

  final String _templateName;
  final String _source;
  final bool _lenient;
  
  final Iterator<int> _itr;
  int _offset = 0;
  int _c = 0;
  
  final List<Token> _tokens = new List<Token>();

  // These can be changed by the change delimiter tag.
  int _openDelimiter;
  int _openDelimiterInner;
  int _closeDelimiterInner;
  int _closeDelimiter;
  
  List<Token> scan() {
    
    for (int c = _peek(); c != _EOF; c = _peek()) {
      
      // Scan text tokens.
      if (c != _openDelimiter) {
        _scanText();
        continue;
      }
        
      int start = _offset;
      
      // Read first open delimiter character.
      _read();
      
      // If only a single delimiter character then create a text token.
      if (_openDelimiterInner != null && _peek() != _openDelimiterInner) {
        var value = new String.fromCharCode(_openDelimiter);
        _append(TokenType.text, value, start, _offset);
        continue;
      }
        
      if (_openDelimiterInner != null) _expect(_openDelimiterInner);
      
      // Handle triple mustache.
      if (_openDelimiterInner == _OPEN_MUSTACHE &&
          _openDelimiter == _OPEN_MUSTACHE &&
          _peek() == _OPEN_MUSTACHE) {
        
        _read();
        _append(TokenType.openDelimiter, '{{{', start, _offset);
        _scanTagContent();
        _scanCloseTripleMustache();
                   
      } else {
      
        // Check to see if this is a change delimiter tag. {{= | | =}}
        // Need to skip whitespace and check for "=".
        int wsStart = _offset;
        var ws = _readWhile(_isWhitespace);
        
        if (_peek() == _EQUAL) {
          _parseChangeDelimiterTag(start);
          
        } else {
          // Scan standard mustache tag.
          var value = new String.fromCharCodes(_openDelimiterInner == null
              ? [_openDelimiter]
              : [_openDelimiter, _openDelimiterInner]);
          
          _append(TokenType.openDelimiter, value, start, wsStart);
          
          if (ws != '') _append(TokenType.whitespace, ws, wsStart, _offset);
          
          _scanTagContent();
          _scanCloseDelimiter();
        }
      }
    }
    return _tokens;
  }
  
  int _peek() => _c;
  
  int _read() {
    int c = _c;
    _offset++;
    _c = _itr.moveNext() ? _itr.current : _EOF;
    return c;
  }
  
  String _readWhile(bool test(int charCode)) {
    if (_c == _EOF) return '';      
    int start = _offset;
    while (_peek() != _EOF && test(_peek())) {
      _read();
    }
    int end = _peek() == _EOF ? _source.length : _offset;
    return _source.substring(start, end);
  }
  
  _expect(int expectedCharCode) {
    int c = _read();

    if (c == _EOF) {
      throw new TemplateException('Unexpected end of input',
          _templateName, _source, _offset - 1);

    } else if (c != expectedCharCode) {
      throw new TemplateException('Unexpected character, '
        'expected: ${new String.fromCharCode(expectedCharCode)}, '
        'was: ${new String.fromCharCode(c)}',
          _templateName, _source, _offset - 1);
    }
  }
  
  _append(TokenType type, String value, int start, int end) =>
      _tokens.add(new Token(type, value, start, end));
  
  bool _isWhitespace(int c)
    => const [_SPACE, _TAB , _NEWLINE, _RETURN].contains(c);
  
  // Scan text. This adds text tokens, line end tokens, and whitespace
  // tokens for whitespace at the begining of a line. This is because the
  // mustache spec requires special handing of whitespace.
  void _scanText() {
    int start = 0;
    TokenType token;
    String value;
  
    for (int c = _peek(); c != _EOF && c != _openDelimiter; c = _peek()) {
      start = _offset;
      
      switch (c) {
        case _SPACE:
        case _TAB:
          value = _readWhile((c) => c == _SPACE || c == _TAB);
          token = TokenType.whitespace;          
          break;
          
        case _NEWLINE:
          _read();
          token = TokenType.lineEnd;
          value = '\n';
          break;
        
        case _RETURN:
          _read();
          if (_peek() == _NEWLINE) {
            _read();
            token = TokenType.lineEnd;
            value = '\r\n';
          } else {
            token = TokenType.text;
            value = '\r';
          }
          break;
          
        default:
          value = _readWhile((c) => c != _openDelimiter && c != _NEWLINE);
          token = TokenType.text;
      }
      
      _append(token, value, start, _offset);
    }
  }
  
  // Scan contents of a tag and the end delimiter token.
  void _scanTagContent() {
    
    int start;
    TokenType token;
    String value;
    
    bool isCloseDelimiter(int c) => 
          (_closeDelimiterInner == null && c == _closeDelimiter)
          || (_closeDelimiterInner != null && c == _closeDelimiterInner);      
    
    for (int c = _peek(); c != _EOF && !isCloseDelimiter(c); c = _peek()) {
      
      start = _offset;
      
      switch (c) {
        case _HASH:
        case _CARET:
        case _FORWARD_SLASH:
        case _GT:
        case _AMP:
        case _EXCLAIM:        
          _read();
          token = TokenType.sigil;
          value = new String.fromCharCode(c);
          break;
          
        case _SPACE:
        case _TAB:
        case _NEWLINE:
        case _RETURN:
          token = TokenType.whitespace;
          value = _readWhile(_isWhitespace);
          break;
          
        case _PERIOD:
          _read();
          token = TokenType.dot;
          value = '.';  
          break;
          
        default:          
          // Identifier can be any other character in lenient mode.
          token = TokenType.identifier;
          value = _readWhile((c) => !(const [ _HASH, _CARET, _FORWARD_SLASH,
            _GT, _AMP, _EXCLAIM, _SPACE, _TAB, _NEWLINE, _RETURN,
            _PERIOD].contains(c)) &&
            c != _closeDelimiterInner &&
            c != _closeDelimiter);
      }
      _append(token, value, start, _offset);    
    }    
  }

  // Scan close delimiter token.
  void _scanCloseDelimiter() {
    
    if (_peek() != _EOF) {
      int start = _offset;
      
      if (_closeDelimiterInner != null) _expect(_closeDelimiterInner);
      _expect(_closeDelimiter);
      
      String value = new String.fromCharCodes(_closeDelimiterInner == null
          ? [_closeDelimiter]
          : [_closeDelimiterInner, _closeDelimiter]);
      
      _append(TokenType.closeDelimiter, value, start, _offset);
    }    
  }

  // Scan close triple mustache delimiter token.
  void _scanCloseTripleMustache() {
    if (_peek() != _EOF) {
      int start = _offset;
      
      _expect(_CLOSE_MUSTACHE);
      _expect(_CLOSE_MUSTACHE);      
      _expect(_CLOSE_MUSTACHE);
      
      _append(TokenType.closeDelimiter, '}}}', start, _offset);
    }    
  }  

  // Open delimiter characters have already been read.
  void _parseChangeDelimiterTag(int start) {
    
    _expect(_EQUAL);
    
    var delimiterInner = _closeDelimiterInner;
    var delimiter = _closeDelimiter;
    
    _readWhile(_isWhitespace);
    
    int c;
    c = _read();
    
    if (c == _EQUAL) throw _error('Incorrect change delimiter tag.');
    _openDelimiter = c;
    
    c = _read();
    if (_isWhitespace(c)) {
      _openDelimiterInner = null;
    } else {
      _openDelimiterInner = c;
    }
    
    _readWhile(_isWhitespace);
    
    c = _read();
    
    if (_isWhitespace(c) || c == _EQUAL)
      throw _error('Incorrect change delimiter tag.');
    
    if (_isWhitespace(_peek()) || _peek() == _EQUAL) {
      _closeDelimiterInner = null;
      _closeDelimiter = c;
    } else {
      _closeDelimiterInner = c;
      _closeDelimiter = _read();
    }
    
    _readWhile(_isWhitespace);
    
    _expect(_EQUAL);
    
    _readWhile(_isWhitespace);     
     
    if (delimiterInner != null) _expect(delimiterInner);
     _expect(delimiter);
    
     // Create delimiter string.
     var buffer = new StringBuffer();
     buffer.writeCharCode(_openDelimiter);
     if (_openDelimiterInner != null) buffer.writeCharCode(_openDelimiterInner);
     buffer.write(' ');
     if (_closeDelimiterInner != null) {
       buffer.writeCharCode(_closeDelimiterInner);
     }
     buffer.writeCharCode(_closeDelimiter);
     var value = buffer.toString();
          
     _append(TokenType.changeDelimiter, value, start, _offset);
  }
  
  TemplateException _error(String message) {
    return new TemplateException(message, _templateName, _source, _offset);
  }

}

const int _EOF = -1;
const int _TAB = 9;
const int _NEWLINE = 10;
const int _RETURN = 13;
const int _SPACE = 32;
const int _EXCLAIM = 33;
const int _QUOTE = 34;
const int _APOS = 39;
const int _HASH = 35;
const int _AMP = 38;
const int _PERIOD = 46;
const int _FORWARD_SLASH = 47;
const int _LT = 60;
const int _EQUAL = 61;
const int _GT = 62;
const int _CARET = 94;

const int _OPEN_MUSTACHE = 123;
const int _CLOSE_MUSTACHE = 125;

const int _A = 65;
const int _Z = 90;
const int _a = 97;
const int _z = 122;
const int _0 = 48;
const int _9 = 57;

const int _UNDERSCORE = 95;
const int _MINUS = 45;