# Контекстно-свободная Parse::Yapp-грамматика шаблонизатора
#
# Для корректной работы нужен патченый LIME со следующими изменениями:
# (*) Подменой лексемы 'lit' на 'str' в метаграмматике.
#     Это нужно, чтобы можно было юзать строковые лексемы типа '<!--'.
# (*) Для корректной обработки ошибок нужно, чтобы метод eat() возвращал
#     false при ошибке и true при успехе. Т.к. подразумевается, что лексический
#     анализатор зависим от работы синтаксического, знает о его состоянии и
#     соответственно выдаёт либо лексемы "внутри" блоков кода, либо литералы
#     "вне" оных.
# Взять таковой можно здесь: https://github.com/vitalif/lime
#
# {{ двойные скобки }} нужно исключительно чтобы маркеры начала и конца подстановки
# были уникальны в грамматике. Вместо них обычно используются { одинарные }, а
# выбор корректной лексемы - скобки или маркера - делает лексический анализатор.
# Но зато вместо { фигурных скобок } можно выбрать себе любые другие маркеры!
#
# Все выражения представляются массивом из двух значений: [ код выражения, флаг экранирования ]
# Флаг экранирования == true, если это выражение HTML-безопасно. При включённом auto_escape
# небезопасные выражения прогоняются через экранирование.
#
# Кстати:
# * Олдстайл BEGIN .. END ликвидирован
# * Возможно, нужно добавить в каком-то виде foreach ... as key => value

%start template

%token literal
%token name

%token '..'
%token '||'
%token 'OR'
%token 'XOR'
%token 'AND'
%token '&&'
%token '&'
%token '=='
%token '!='
%token '<'
%token '>'
%token '<='
%token '>='
%token '+'
%token '-'
%token '*'
%token '/'
%token '%'
%token '('
%token ')'
%token '!'
%token 'NOT'
%token '{'
%token '}'
%token ','
%token '=>'
%token '['
%token ']'
%token '<!--'
%token '-->'
%token '{{'
%token '}}'

%left '..'
%left '||' 'OR' 'XOR'
%left '&&' 'AND'
%nonassoc '==' '!=' '<' '>' '<=' '>='
%left '+' '-'
%left '&'
%left '*' '/' '%'

# Директивы

%%

template: chunks {
    $_[0]->{template}->{st}->{functions}->{main}->{body} = "sub fn_main() {\nmy \$stack = [];\nmy \$t = '';\n".$_[1]."\nreturn \$t;\n}\n";
    '';
  }
;
chunks: {
    '';
  }
| chunks chunk {
    $_[1] . $_[2];
  }
;
chunk: literal {
    '$t .= ' . $_[1] . ";\n";
  }
| '<!--' code_chunk '-->' {
    $_[2];
  }
| '{{' exp '}}' {
    '$t .= ' . ($_[2][1] || !$_[0]->{template}->{options}->{auto_escape} ? $_[2][0] : $_[0]->{template}->compile_function($_[0]->{template}->{options}->{auto_escape}, [ $_[2] ])[0]) . ";\n";
  }
| error {
    '';
  }
;
code_chunk: c_if | c_set | c_fn | c_for | exp {
    '$t .= ' . ($_[1][1] || !$_[0]->{template}->{options}->{auto_escape} ? $_[1][0] : $_[0]->{template}->compile_function($_[0]->{template}->{options}->{auto_escape}, [ $_[1] ])[0]) . ";\n";
  }
;
c_if: 'IF' exp '-->' chunks '<!--' 'END' {
    "if (" . $_[2][0] . ") {\n" . $_[4] . "}\n";
  }
| 'IF' exp '-->' chunks '<!--' 'ELSE' '-->' chunks '<!--' 'END' {
    "if (" . $_[2][0] . ") {\n" . $_[4] . "} else {\n" . $_[8] . "}\n";
  }
| 'IF' exp '-->' chunks c_elseifs chunks '<!--' 'END' {
    "if (" . $_[2][0] . ") {\n" . $_[4] . $_[5] . $_[6] . "}\n";
  }
| 'IF' exp '-->' chunks c_elseifs chunks '<!--' 'ELSE' '-->' chunks '<!--' 'END' {
    "if (" . $_[2][0] . ") {\n" . $_[4] . $_[5] . $_[6] . "} else {\n" . $_[10] . "}\n";
  }
;
c_elseifs: '<!--' elseif exp '-->' {
    #{
    "} elsif (" . $_[3][0] . ") {\n";
    #}
  }
| c_elseifs chunks '<!--' elseif exp '-->' {
    #{
    $_[1] . $_[2] . "} elsif (" . $_[5][0] . ") {\n";
    #}
  }
;
c_set: 'SET' varref '=' exp {
    $_[2][0] . ' = ' . $_[4][0] . ";\n";
  }
| 'SET' varref '-->' chunks '<!--' 'END' {
    "push \@\$stack, \$t;\n\$t = '';\n" . $_[4] . $_[2][0] . " = \$t;\n\$t = pop(\@\$stack);\n";
  }
;
c_fn: fn name '(' arglist ')' '=' exp {
    $_[0]->{template}->{st}->{functions}->{$_[2]} = {
      'name' => $_[2],
      'args' => $_[4],
      'body' => 'sub fn_'.$_[2]." () {\nreturn ".$_[7].";\n}\n",
      //'line' => $line, Ой, я чо - аргументы не юзаю?
      //'pos' => $pos,
    };
    '';
  }
| fn name '(' arglist ')' '-->' chunks '<!--' 'END' {
    $_[0]->{template}->{st}->{functions}->{$_[2]} = {
      'name' => $_[2],
      'args' => $_[4],
      'body' => 'sub fn_'.$_[2]." () {\nmy \$stack = [];\nmy \$t = '';\n".$_[7]."\nreturn \$t;\n}\n",
      //'line' => $line,
      //'pos' => $pos,
    };
    '';
  }
;
c_for: for varref '=' exp '-->' chunks '<!--' 'END' {
    my @varref = @{$_[2]};
    my @exp = @_{$_[4]};
    my $cs = $_[6];
    #{
    my $varref_index = substr($varref[0], 0, -1) . ".'_index'}";
    "push \@\$stack, ".$varref[0].", ".$varref_index.", 0;
foreach my \$item (array1($exp[0])) {
".$varref[0]." = \$item;
".$varref_index." = \$stack[count(\$stack)-1]++;
".$cs."}
pop \@\$stack;
".$varref_index." = pop(\@\$stack);
".$varref[0]." = pop(\@\$stack);
";
  }
;
fn: 'FUNCTION' | 'BLOCK' | 'MACRO' ;
for: 'FOR' | 'FOREACH' ;
elseif: 'ELSE' 'IF' | 'ELSIF' | 'ELSEIF' ;

# Выражения

exp: exp '..' exp {
    [ '(' . $_[1][0] . ' . ' . $_[3][0] . ')', $_[1][1] && $_[3][1] ];
  }
| exp '||' exp {
    [ '(' . $_[1][0] . ' || ' . $_[3][0] . ')', $_[1][1] && $_[3][1] ];
  }
| exp 'OR' exp {
    [ '(' . $_[1][0] . ' || ' . $_[3][0] . ')', $_[1][1] && $_[3][1] ];
  }
| exp 'XOR' exp {
    [ '(' . $_[1][0] . ' XOR ' . $_[3][0] . ')', 1 ];
  }
| exp '&&' exp {
    [ '(' . $_[1][0] . ' && ' . $_[3][0] . ')', 1 ];
  }
| exp 'AND' exp {
    [ '(' . $_[1][0] . ' && ' . $_[3][0] . ')', 1 ];
  }
| exp '==' exp {
    [ '$self->eq(' . $_[1][0] . ', ' . $_[3][0] . ')', 1 ];
  }
| exp '!=' exp {
    [ '!$self->eq(' . $_[1][0] . ', ' . $_[3][0] . ')', 1 ];
  }
| exp '<' exp {
    [ '$self->lt(' . $_[1][0] . ', ' . $_[3][0] . ')', 1 ];
  }
| exp '>' exp {
    [ '$self->gt(' . $_[1][0] . ', ' . $_[3][0] . ')', 1 ];
  }
| exp '<=' exp {
    [ '!$self->gt(' . $_[1][0] . ', ' . $_[3][0] . ')', 1 ];
  }
| exp '>=' exp {
    [ '!$self->lt(' . $_[1][0] . ', ' . $_[3][0] . ')', 1 ];
  }
| exp '+' exp {
    [ '(' . $_[1][0] . ' + ' . $_[3][0] . ')', 1 ];
  }
| exp '-' exp {
    [ '(' . $_[1][0] . ' - ' . $_[3][0] . ')', 1 ];
  }
| exp '&' exp {
    [ '(' . $_[1][0] . ' & ' . $_[3][0] . ')', 1 ];
  }
| exp '*' exp {
    [ '(' . $_[1][0] . ' * ' . $_[3][0] . ')', 1 ];
  }
| exp '/' exp {
    [ '(' . $_[1][0] . ' / ' . $_[3][0] . ')', 1 ];
  }
| exp '%' exp {
    [ '(' . $_[1][0] . ' % ' . $_[3][0] . ')', 1 ];
  }
| p10
;
p10: p11
| '-' p11 {
    [ '(-'.$_[2][0].')', 1 ];
  }
;
p11: nonbrace
| '(' exp ')' varpath {
    [ '('.$_[2][0].')'.$_[4], 0 ];
  }
| '!' p11 {
    [ '(!'.$_[2][0].')', 1 ];
  }
| 'NOT' p11 {
    [ '(!'.$_[2][0].')', 1 ];
  }
;
nonbrace: '{' hash '}' {
    [ "{ " . $_[2] . " }", 1 ];
  }
| literal {
    [ $1, 1 ];
  }
| varref
| name '(' ')' {
    $_[0]->{template}->compile_function($_[1], []);
  }
| name '(' list ')' {
    $_[0]->{template}->compile_function($_[1], $_[3]);
  }
| name '(' gthash ')' {
    [ "\$self->{parent}->call_block('".addcslashes($_[1], "'\\")."', { ".$_[3]." }, '".addcslashes($this->template->lexer->errorinfo(), "'\\")."')", 1 ];
  }
| name nonbrace {
    $_[0]->{template}->compile_function($_[1], [ $_[3] ]);
  }
| method '(' ')' {
    [ $_[1].'()', 1 ];
  }
| method '(' list ')' {
    my $argv = [];
    foreach my $a (@{$_[3]}) {
      push @$argv, $a->[0];
    }
    [ $_[1].'('.join(', ', @$argv).')', 1 ];
  }
;
method: varref '.' name {
    $_[1][0].'->'.$_[3];
  }
;
list: exp {
    [ $_[1] ];
  }
| exp ',' list {
    [ $_[1], @{$_[3]} ];
  }
;
arglist: name {
    [ $_[1] ];
  }
| name ',' arglist {
    [ $_[1], @{$_[3]} ];
  }
| {
    [];
  }
;
hash: pair
| pair ',' hash {
    $_[1] . ', ' . $_[3];
  }
| {
    '';
  }
;
gthash: gtpair
| gtpair ',' gthash {
    $_[1] . ', ' . $_[3];
  }
;
pair: exp ',' exp {
    $_[1][0] . ' => ' . $_[3][0];
  }
| gtpair
;
gtpair: exp '=>' exp {
    $_[1][0] . ' => ' . $_[3][0];
  }
;
varref: name {
    [ "\$self->{tpldata}{'".addcslashes($_[1], "\\\'")."'}", 0 ];
  }
| varref varpart {
    [ $_[1][0] . $_[2], 0 ];
  }
;
varpart: '.' name {
    "{'".addcslashes($_[1], "\\\'")."'}";
  }
| '[' exp ']' {
    ($_[2][1] eq 'i' ? '['.$_[2][0].']' : "{".$_[2][0]."}");
  }
;
varpath: {
    '';
  }
| varpath varpart {
    $_[1] . $_[2];
  }
;

%%

# Possible tokens consisting of special characters
my $chartokens = '+ - = * / % ! , . < > ( ) { } [ ] & .. || && == != <= >= =>';

# Reserved keywords
my $keywords_str = 'OR XOR AND NOT IF ELSE ELSIF ELSEIF END SET FOR FOREACH FUNCTION BLOCK MACRO';

sub _Lexer
{
    my ($parser) = shift;

    if ($parser->YYEndOfInput)
    {
        my $input = <STDIN>;
        return('', undef) unless $input;
        $parser->input($input);
        my $lex = $parser->{__lexer} = {
            options => {} ???,

            # Current position in code
            codelen => strlen($input),
            pos => 0,
            lineno => 0,

            # Preprocessed keyword tokens
            nchar => {},
            lens => [],
            keywords => { map { $_ => 1 } split / /, $keywords_str },

            # Last directive start position, directive and substitution start/end counters
            last_start => 0,
            last_start_line => 0,
            in_code => 0,
            in_subst => 0,
            force_literal => 0,
        };
        foreach (split(/ /, $chartokens))
        {
            $lex->{nchar}{strlen($_)}{$_} = 1;
        }
        # Add code fragment finishing tokens
        $lex->{nchar}{strlen($lex->{options}->{end_code})}{$lex->{options}->{end_code}} = 1;
        if ($this->options->end_subst)
        {
            $lex->{nchar}{strlen($lex->{options}->{end_subst})}{$lex->{options}->{end_subst}} = 1;
        }
        # Reverse-sort lengths
        $lex->{lens} = [ sort { $b <=> $a } keys %{$lex->{nchar}} ];
    }

    my $lex = $parser->{__lexer};

    for (${$parser->YYInput})
    {
        m/\G[ \t]*/gc;
        m/\G([0-9]+(?:\.[0-9]+)?)/gc and return('NUM',$1);
        m/\G([A-Za-z][A-Za-z0-9_]*)/gc and return('VAR',$1);
        m/\G(.)/gcs and return($1,$1);
        return('', undef);
    }
}

__PACKAGE__->lexer(\&_Lexer);

class VMXTemplateLexer
{
    function feed($parser)
    {
        try
        {
            $parser->reset();
            $in = false;
            while ($t = $this->read_token())
            {
                $success = $parser->eat($t[0], $t[1]);
                if (!$success)
                {
                    // Pass $in from last step so we skip to the beginning
                    // of directive even if it just ended and $this->in_* == 0
                    $this->skip_error(end($parser->parser->errors), $in);
                }
                $in = $this->in_code || $this->in_subst;
            }
            $parser->eat_eof();
        }
        catch (parse_error $e)
        {
            $this->options->error($e->getMessage());
        }
    }

    function set_code($code)
    {
        $this->code = $code;
        $this->codelen = strlen($this->code);
        $this->pos = $this->lineno = 0;
    }

    function errorinfo()
    {
        $linestart = strrpos($this->code, "\n", $this->pos-$this->codelen-1) ?: -1;
        $lineend = strpos($this->code, "\n", $this->pos) ?: $this->codelen;
        $line = substr($this->code, $linestart+1, $this->pos-$linestart-1);
        $line .= '^^^';
        $line .= substr($this->code, $this->pos, $lineend-$this->pos);
        return " in {$this->options->input_filename}, line ".($this->lineno+1).", byte {$this->pos}, marked by ^^^ in $line";
    }

    function warn($text)
    {
        $this->options->error($text.$this->errorinfo());
    }

    /**
     * Skip a directive
     */
    function skip_error($e, $force = false)
    {
        if (substr($e, 0, 18) !== 'error not expected')
        {
            $this->warn($e);
            if ($this->in_code || $this->in_subst || $force)
            {
                $this->in_code = $this->in_subst = 0;
                $this->pos = $this->last_start;
                $this->lineno = $this->last_start_line;
                $this->force_literal = 1;
            }
        }
    }

    /**
     * Read next token from the stream
     * Returns array($token, $value) or false for EOF
     */
    sub _Lexer
    {
        if ($this->pos >= $this->codelen)
        {
            // End of code
            return false;
        }
        if ($this->in_code <= 0 && $this->in_subst <= 0)
        {
            $code_pos = strpos($this->code, $this->options->begin_code, $this->pos+$this->force_literal);
            $subst_pos = strpos($this->code, $this->options->begin_subst, $this->pos+$this->force_literal);
            $this->force_literal = 0;
            if ($code_pos === false && $subst_pos === false)
            {
                $r = array('literal', "'".addcslashes(substr($this->code, $this->pos), "'\\")."'");
                $this->lineno += substr_count($r[1], "\n");
                $this->pos = $this->codelen;
            }
            elseif ($subst_pos === false || $code_pos !== false && $subst_pos > $code_pos)
            {
                // Code starts closer
                if ($code_pos > $this->pos)
                {
                    // We didn't yet reach the code beginning
                    $str = substr($this->code, $this->pos, $code_pos-$this->pos);
                    if ($this->options->eat_code_line)
                    {
                        $str = preg_replace('/\n[ \t]*$/s', "\n", $str);
                    }
                    $r = array('literal', "'".addcslashes($str, "'\\")."'");
                    $this->lineno += substr_count($r[1], "\n");
                    $this->pos = $code_pos;
                }
                elseif ($code_pos !== false)
                {
                    // We are at the code beginning ($this->pos == $code_pos)
                    $i = $this->pos+strlen($this->options->begin_code);
                    while ($i < $this->codelen && (($c = $this->code{$i}) == ' ' || $c == "\t"))
                    {
                        $i++;
                    }
                    if ($i < $this->codelen && $this->code{$i} == '#')
                    {
                        // Strip comment
                        $i = strpos($this->code, $this->options->end_code, $i);
                        $this->pos = $i ? $i+strlen($this->options->end_code) : $this->codelen;
                        return $this->read_token();
                    }
                    $r = array('<!--', $this->options->begin_code);
                    $this->last_start = $this->pos;
                    $this->last_start_line = $this->lineno;
                    $this->pos += strlen($this->options->begin_code);
                    $this->in_code = 1;
                }
            }
            else
            {
                // Substitution is closer
                if ($subst_pos > $this->pos)
                {
                    $r = array('literal', "'".addcslashes(substr($this->code, $this->pos, $subst_pos-$this->pos), "'\\")."'");
                    $this->lineno += substr_count($r[1], "\n");
                    $this->pos = $subst_pos;
                }
                else
                {
                    $r = array('{{', $this->options->begin_subst);
                    $this->last_start = $this->pos;
                    $this->last_start_line = $this->lineno;
                    $this->pos++;
                    $this->in_subst = 1;
                }
            }
            return $r;
        }
        while ($this->pos < $this->codelen)
        {
            // Skip whitespace
            $t = $this->code{$this->pos};
            if ($t == "\n")
                $this->lineno++;
            elseif ($t != "\t" && $t != ' ')
                break;
            $this->pos++;
        }
        if ($this->pos >= $this->codelen)
        {
            // End of code
            return false;
        }
        if (preg_match('#[a-z_][a-z0-9_]*#Ais', $this->code, $m, 0, $this->pos))
        {
            $this->pos += strlen($m[0]);
            if (isset($this->keywords[$l = strtoupper($m[0])]))
            {
                // Keyword
                return array($l, $m[0]);
            }
            // Identifier
            return array('name', $m[0]);
        }
        elseif (preg_match(
            '/((\")(?:[^\"\\\\]+|\\\\.)*\"|\'(?:[^\'\\\\]+|\\\\.)*\''.
            '|0\d+|\d+(\.\d+)?|0x\d+)/Ais', $this->code, $m, 0, $this->pos))
        {
            // String or numeric non-negative literal
            $t = $m[1];
            if (isset($m[2]))
            {
                $t = str_replace('$', '\\$', $t);
            }
            $this->pos += strlen($m[0]);
            return array('literal', $t);
        }
        else
        {
            // Special characters
            foreach ($this->lens as $l)
            {
                $a = $this->nchar[$l];
                $t = substr($this->code, $this->pos, $l);
                if (isset($a[$t]))
                {
                    $this->pos += $l;
                    if ($this->in_code)
                    {
                        $this->in_code += ($t === $this->options->begin_code);
                        $this->in_code -= ($t === $this->options->end_code);
                        if (!$this->in_code)
                        {
                            if ($this->options->eat_code_line)
                            {
                                $p = $this->pos;
                                while ($p < $this->codelen && (($c = $this->code{$p}) == ' ' || $c == "\t" || $c == "\r"))
                                {
                                    $p++;
                                }
                                if ($p < $this->codelen && $this->code{$p} == "\n")
                                {
                                    $p++;
                                    if ($p < $this->codelen && $this->code{$p} == "\r")
                                    {
                                        $p++;
                                    }
                                    $this->pos = $p;
                                }
                            }
                            return array('-->', $t);
                        }
                    }
                    elseif ($this->in_subst)
                    {
                        $this->in_subst += ($t === $this->options->begin_subst);
                        $this->in_subst -= ($t === $this->options->end_subst);
                        if (!$this->in_subst)
                        {
                            return array('}}', $t);
                        }
                    }
                    return array($t, false);
                }
            }
            // Unknown character
            $this->skip_error(
                "Unexpected character '".$this->code{$this->pos}."'"
            );
            return array('error', false);
        }
    }
}