<?php
/* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4: */

require_once('parser.php');
require_once('haskellTerminal.php');

class haskellTokenizer
{
    public $debug = false;
    public $source = null;
    public $tokens = array();

    public function &getTerminal()
    {
        $terminal = new haskellTerminal();
        $terminal->setToken($this->getToken());
        return $terminal;
    }

    public function &getToken()
    {
        $token =& $this->getTokenByPhpTokenizer();
        if (is_string($token)) {
            switch ($token) {
            case '*':
            case '+':
                $token = array(VAROP, $token, 0);
                break;
            case '<':
                $next_token =& $this->getTokenByPhpTokenizer();
                if (is_string($next_token) && $next_token == '-') {
                    $token = array(VAROP, '<-', 0);
                } else {
                    $this->ungetTokenByPhpTokenizer($next_token);
                }
            }
            if ($this->debug) {
                if (is_string($token)) {
                    $this->log("STRING: $token");
                } else {
                    $this->log(sprintf("%s: %s", $token[0], $token[1]));
                }
            }
            return $token;
        } else if (!is_array($token)) {
            return $token;
        }

        $token_name = token_name($token[0]);

        switch ($token[0]) {
        case T_LNUMBER:
            $token[0] = NUMLIT;
            break;
        case T_PRINT:
        case T_STRING:
            $token[0] = VARID;
            break;
        case T_CONSTANT_ENCAPSED_STRING:
            $token[0] = STRINGLIT;
            break;
        default:
            $token[0] = 0;
            break;
        }

        switch ($token[1]) {
        case "as":
            $token[0] = ASMOD;
            break;
        case "case":
            $token[0] = CASEXP; 
            break;
        case "class":
            $token[0] = TCLASS; 
            break;
        case "data":
            $token[0] = DATA;
            break;
        case "default":
            $token[0] = DEFAULT_T;
            break;
        case "deriving":
            $token[0] = DERIVING;
            break;
        case "do":
            $token[0] = DO_T; 
            break;
        case "else":
            $token[0] = ELSE_T;
            break;
        case "hiding":
            $token[0] = HIDING;
            break;
        case "if":
            $token[0] = IF_T;
            break;
        case "import":
            $token[0] = IMPORT; 
            break;
        case "in":
            $token[0] = IN;
            break;
        case "infix":
            $token[0] = INFIXN; 
            break;
        case "infixl":
            $token[0] = INFIXL; 
            break;
        case "infixr":
            $token[0] = INFIXR; 
            break;
        case "instance":
            $token[0] = TINSTANCE;
            break;
        case "let":
            $token[0] = LET;
            break;
        case "module":
            $token[0] = TMODULE;
            break;
        case "newtype":
            $token[0] = TNEWTYPE;
            break;
        case "of":
            $token[0] = OF;
            break;
        case "primitive":
            $token[0] = PRIMITIVE;
            break;
        case "qualified":
            $token[0] = QUALIFIED;
            break;
        case "then":
            $token[0] = THEN;
            break;
        case "type":
            $token[0] = TYPE;
            break;
        case "where":
            $token[0] = WHERE;
            break;
        }

        if ($this->debug) {
            if ($token[0]) {
                $this->log(sprintf("%s: %s", $token[0], $token[1]));
            } else {
                $this->log(sprintf("%s: %s", $token_name, $token[1]));
            }
        }
        return $token;
    }

    public function &getTokens($force = false)
    {
        if (!$this->tokens || $force) {
            $this->tokens = token_get_all($this->source);
        }
        return $this->tokens;
    }

    protected function log($str)
    {
        echo "{$str}\n";
    }

    public function &setSource(&$source)
    {
        $this->source =& $source;
        $this->getTokens(true);
        return $source;
    }

    protected function &getTokenByPhpTokenizer()
    {
        $token = array_shift($this->tokens);

        if (is_string($token) || !is_array($token)) {
            return $token;
        }

        switch ($token[0]) {
        case T_CLOSE_TAG:
        case T_COMMENT:
        case T_DOC_COMMENT:
        case T_INLINE_HTML:
        case T_OPEN_TAG:
        case T_WHITESPACE:
            return $this->getTokenByPhpTokenizer();
        }

        return $token;
    }

    protected function ungetTokenByPhpTokenizer(&$token)
    {
        return array_unshift($this->tokens, $token);
    }
}