NAME

"Parse::Token" - Definition of tokens used by "Parse::Lex"

SYNOPSIS

        require 5.005;

        use Parse::Lex;
        @token = qw(
            ADDOP    [-+]
            INTEGER  [1-9][0-9]*
           );

        $lexer = Parse::Lex->new(@token);
        $lexer->from(\*DATA);

        $content = $INTEGER->next;
        if ($INTEGER->status) {
          print "$content\n";
        }
        $content = $ADDOP->next;
        if ($ADDOP->status) {
          print "$content\n";
        }
        if ($INTEGER->isnext(\$content)) {
          print "$content\n";
        }
        __END__
        1+2

DESCRIPTION

The "Parse::Token" class and its derived classes permit defining the tokens used by "Parse::Lex" or "Parse::LexEvent".

The creation of tokens can be done by means of the "new()" or "factory()" methods. The "Lex::new()" method of the "Parse::Lex" package indirectly creates instances of the tokens to be recognized.

The "next()" or "isnext()" methods of the "Parse::Token" package permit interfacing the lexical analyzer with a syntactic analyzer of recursive descent type. For interfacing with "byacc", see the "Parse::YYLex" package.

"Parse::Token" is included indirectly by means of "use Parse::Lex" or "use Parse::LexEvent".

Methods

action

Returns the anonymous subroutine defined within the "Parse::Token" object.

factory LIST

factory ARRAY_REF

The "factory(LIST)" method creates a list of tokens from a list of specifications, which include for each token: a name, a regular expression, and possibly an anonymous subroutine. The list can also include objects of class "Parse::Token" or of a class derived from it.

The "factory(ARRAY_REF)" method permits creating tokens from specifications of type attribute-value:

        Parse::Token->factory([Type => 'Simple', 
                               Name => 'EXAMPLE', 
                               Regex => '.+']);

"Type" indicates the type of each token to be created (the package prefix is not indicated).

"factory()" creates a series of tokens but does not import these tokens into the calling package.

You could for example write:

        %keywords = 
          qw (
              PROC  undef
              FUNC  undef
              RETURN undef
              IF    undef
              ELSE  undef
              WHILE undef
              PRINT undef
              READ  undef
             );
        @tokens = Parse::Token->factory(%keywords);

and install these tokens in a symbol table in the following manner:

        foreach $name (keys %keywords) {
          ${$name} = pop @tokens;
          $symbol{"\L$name"} = [${$name}, ''];
        }

"${$name}" is the token instance.

During the lexical analysis phase, you can use the tokens in the following manner:

        qw(IDENT [a-zA-Z][a-zA-Z0-9_]*),  sub {
           $symbol{$_[1]} = [] unless defined $symbol{$_[1]};
           my $type = $symbol{$_[1]}[0];
           $lexer->setToken((not defined $type) ? $VAR : $type);
           $_[1];  # THE TOKEN TEXT
         }

This permits indicating that any symbol of unknown type is a variable.

In this example we have used $_[1] which corresponds to the text recognized by the regular expression. This text associated with the token must be returned by the anonymous subroutine.

get EXPR

"get" obtains the value of the attribute named by the result of evaluating EXPR. You can also use the name of the attribute as a method name.

getText

Returns the character string that was recognized by means of this "Parse::Token" object.

Same as the text() method.

isnext EXPR

isnext

Returns the status of the token. The consumed string is put into EXPR if it is a reference to a scalar.

name

Returns the name of the token.

next

Activate searching for the lexeme defined by the regular expression contained in the object. If this lexeme is recognized on the character stream to analyze, "next" returns the string found and sets the status of the object to true.

new SYMBOL_NAME, REGEXP, SUB

new SYMBOL_NAME, REGEXP

Creates an object of type "Parse::Token::Simple" or "Parse::Token::Segmented". The arguments of the "new()" method are, respectively: a symbolic name, a regular expression, and possibly an anonymous subroutine. The subclasses of "Parse::Token" permit specifying tokens by means of a list of attribute-values.

REGEXP is either a simple regular expression, or a reference to an array containing from one to three regular expressions. In the first case, the instance belongs to the "Parse::Token::Simple" class. In the second case, the instance belongs to the "Parse::Token::Segmented" class. The tokens of this type permit recognizing structures of type character string delimited by quotation marks, comments in a C program, etc. The regular expressions are used to recognize:

1. The beginning of the lexeme,

2. The "body" of the lexeme; if this second expression is missing, "Parse::Lex" uses "(?:.*?)",

3. the end of the lexeme; if this last expression is missing then the first one is used. (Note! The end of the lexeme cannot span several lines).

Example:

          qw(STRING), [qw(" (?:[^"\\\\]+|\\\\(?:.|\n))* ")],

These regular expressions can recognize multi-line strings delimited by quotation marks, where the backslash is used to quote the quotation marks appearing within the string. Notice the quadrupling of the backslash.

Here is a variation of the previous example which uses the "s" option to include newline in the characters recognized by ""."":

          qw(STRING), [qw(" (?s:[^"\\\\]+|\\\\.)* ")],

(Note: it is possible to write regular expressions which are more efficient in terms of execution time, but this is not our objective with this example. See Mastering Regular Expressions.)

The anonymous subroutine is called when the lexeme is recognized by the lexical analyzer. This subroutine takes two arguments: $_[0] contains the token instance, and $_[1] contains the string recognized by the regular expression. The scalar returned by the anonymous subroutine defines the character string memorized in the token instance.

In the anonymous subroutine you can use the positional variables $1, $2, etc. which correspond to the groups of parentheses in the regular expression.

regexp

Returns the regular expression of the "Token" object.

set LIST

Allows marking a token with a list of attribute-value pairs.

An attribute name can be used as a method name.

setText EXPR

The value of "EXPR" defines the character string associated with the lexeme.

Same as the "text(EXPR)" method.

status EXPR

status

Indicates if the last search of the lexeme succeeded or failed. "status EXPR" overrides the existing value and sets it to the value of EXPR.

text EXPR

text

"text()" returns the character string recognized by means of the token. The value of "EXPR" sets the character string associated with the lexeme.

trace OUTPUT

trace

Class method which activates/deactivates a trace of the lexical analysis.

"OUTPUT" can be a file name or a reference to a filehandle to which the trace will be directed.

Subclasses of Parse::Token

Subclasses of the "Parse::Token" class are being defined. They permit recognizing specific structures such as, for example, strings within double-quotes, C comments, etc. Here are the subclasses which I am working on:

"Parse::Token::Simple" : tokens of this class are defined by means of a single regular expression.

"Parse::Token::Segmented" : tokens of this class are defined by means of three regular expressions. Reading of new data is done automatically.

"Parse::Token::Delimited" : permits recognizing, for example, C language comments.

"Parse::Token::Quoted" : permits recognizing, for example, character strings within quotation marks.

"Parse::Token::Nested" : permits recognizing nested structures such as parenthesized expressions. NOT DEFINED.

These classes are recently created and no doubt contain some bugs.

Parse::Token::Action

Tokens of the "Parse::Token::Action" class permit inserting arbitrary Perl expressions within a lexical analyzer. An expression can be used for instance to print out internal variables of the analyzer:

$LEX_BUFFER : contents of the buffer to be analyzed
$LEX_LENGTH : length of the character string being analyzed
$LEX_RECORD : number of the record being analyzed
$LEX_OFFSET : number of characters already consumed since the start of the analysis.
$LEX_POS : position reached by the analysis as a number of characters since the start of the buffer.

The class constructor accepts the following attributes:

"Name" : the name of the token
"Expr" : a Perl expression

Example :

        $ACTION = new Parse::Token::Action(
                                      Name => 'ACTION',
                                      Expr => q!print "LEX_POS: $LEX_POS\n" .
                                      "LEX_BUFFER: $LEX_BUFFER\n" .
                                      "LEX_LENGTH: $LEX_LENGTH\n" .
                                      "LEX_RECORD: $LEX_RECORD\n" .
                                      "LEX_OFFSET: $LEX_OFFSET\n" 
                                      ;!,
                                     );

Parse::Token::Simple

The class constructor accepts the following attributes:

"Handler" : the value indicates the name of a function to call during an analysis performed by an analyzer of class "Parse::LexEvent".
"Name" : the associated value is the name of the token.
"Regex" : the associated value is a regular expression corresponding to the pattern to be recognized.
"ReadMore" : if the associated value is 1, the recognition of the token continues after reading a new record. The strings recognized are concatenated. This attribute only has effect during analysis of a character stream.
"Sub" : the associated value must be an anonymous subroutine to be executed after the token is recognized. This function is only used with analyzers of class "Parse::Lex" or "Parse::CLex".

Example. new Parse::Token::Simple(Name => 'remainder', Regex => '[^/\'\"]+', ReadMore => 1);

Parse::Token::Segmented

The definition of these tokens includes three regular expressions. During analysis of a data stream, new data is read as long as the end of the token has not been reached.

The class constructor accepts the following attributes:

"Handler" : the value indicates the name of a function to call during analysis performed by an analyzer of class "Parse::LexEvent".
"Name" : the associated value is the name of the token.
"Regex" : the associated value must be a reference to an array that contains three regular expressions.
"Sub" : the associated value must be an anonymous subroutine to be executed after the token is recognized. This function is only used with analyzers of class "Parse::Lex" or "Parse::CLex".

Parse::Token::Quoted

"Parse::Token::Quoted" is a subclass of "Parse::Token::Segmented". It permits recognizing character strings within double quotes or single quotes.

Examples.

      ---------------------------------------------------------
       Start    End            Escaping
      ---------------------------------------------------------
        '        '              ''
        "        "              ""
        "        "              \
      ---------------------------------------------------------

The class constructor accepts the following attributes:

"End" : The associated value is a regular expression permitting recognizing the end of the token.
"Escape" : The associated value indicates the character used to escape the delimiter. By default, a double occurrence of the terminating character escapes that character.
"Handler" : the value indicates the name of a function to be called during an analysis performed by an analyzer of class "Parse::LexEvent".
"Name" : the associated value is the name of the token.
"Start" : the associated value is a regular expression permitting recognizing the start of the token.
"Sub" : the associated value must be an anonymous subroutine to be executed after the token is recognized. This function is only used with analyzers of class "Parse::Lex" or "Parse::CLex".

Example. new Parse::Token::Quoted(Name => 'squotes', Handler => 'string', Escape => '\\', Quote => qq!\'!, );

Parse::Token::Delimited

"Parse::Token::Delimited" is a subclass of "Parse::Token::Segmented". It permits, for example, recognizing C language comments.

Examples.

      ---------------------------------------------------------
        Start   End     Constraint
                        on the contents
      ---------------------------------------------------------
        /*       */                         C Comment
        <!--     -->      No '--'           XML Comment
        <!--     -->                        SGML Comment
        <?       ?>                         Processing instruction
                                            in SGML/XML
      ---------------------------------------------------------

The class constructor accepts the following attributes:

"End" : The associated value is a regular expression permitting recognizing the end of the token.
"Handler" : the value indicates the name of a function to be called during an analysis performed by an analyzer of class "Parse::LexEvent".
"Name" : the associated value is the name of the token.
"Start" : the associated value is a regular expression permitting recognizing the start of the token.
"Sub" : the associated value must be an anonymous subroutine to be executed after the token is recognized. This function is only used with analyzers of class "Parse::Lex" or "Parse::CLex".

Example. new Parse::Token::Delimited(Name => 'comment', Start => '/[*]', End => '[*]/' );

Parse::Token::Nested - Not defined

Examples.

      ----------------------------------------------------------
        Start   End
      ----------------------------------------------------------
        (        )                      Symbolic Expressions
        {        }                      Rich Text Format Groups
      ----------------------------------------------------------

BUGS

The implementation of subclasses of tokens is not complete for analyzers of the "Parse::CLex" class. I am not too keen to do it, since an implementation for classes "Parse::Lex" and "Parse::LexEvent" seems quite sufficient.

AUTHOR

Philippe Verdret. Documentation translated to English by Vladimir Alexiev and Ocrat.

ACKNOWLEDGMENTS

Version 2.0 owes much to suggestions made by Vladimir Alexiev. Ocrat has significantly contributed to improving this documentation. Thanks also to the numerous persons who have made comments or sometimes sent bug fixes.

REFERENCES

Friedl, J.E.F. Mastering Regular Expressions. O'Reilly & Associates 1996.

Mason, T. & Brown, D. - Lex & Yacc. O'Reilly & Associates, Inc. 1990.