The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.


Hardware::Vhdl::Lexer - Split VHDL code into lexical tokens


    use Hardware::Vhdl::Lexer;
    # Open the file to get the VHDL code from
    my $fh;
    open $fh, '<', 'device_behav.vhd' || die $!
    # Create the Lexer object
    my $lexer = Hardware::Vhdl::Lexer->new({ linesource => $fh });
    # Dump all the tokens
    my ($token, $type);
    while( (($token, $type) = $lexer->get_next_token) && defined $token) {
        print "# type = '$type' token='$token'\n";


Hardware::Vhdl::Lexer splits VHDL code into lexical tokens. To use it, you need to first create a lexer object, passing in something which will supply chunks of VHDL code to the lexer. Repeated calls to the get_next_token method of the lexer will then return VHDL tokens (in scalar context) or a token type code and the token (in list context). get_next_token returns undef when there are no more tokens to be read.

NB: in this documentation I refer to "lines" of VHDL code and "line" sources etc., but in fact the chunks of code don't have to be broken up at line-ends - they can be broken anywhere that isn't in the middle of a token. New-line characters just happen to be a simple and safe way to split up a file. You don't even have to split up the VHDL at all, you can pass in the whole thing as the first and only "line".


        new({ linesource => <source> [, nhistory => N] })

Note that from version 1.0 of this module the arguments must now be given as a hash reference rather than a hash, so the curly brackets above are required.

The linesource argument is required: it defines where the VHDL source code will be taken from (see below).

The optional nhistory argument sets how many "code" tokens (see the get_next_token method) will be remembered for access by the history method.

new({ linesource => $filehandle_reference [, nhistory => N] })

To read from a file, pass in the filehandle reference like this:

    use Hardware::Vhdl::Lexer;
    my $fh;
    open $fh, '<', $filename || die $!;
    my $lexer = Hardware::Vhdl::Lexer->new({ linesource => $fh });
new({ linesource => \@array_of_lines [, nhistory => N] })
new({ linesource => \$scalar_containing_vhdl [, nhistory => N] })

To read VHDL source that is already in program memory, the linesource argument can be a reference to either an array of lines or a single string which can have embedded newlines.

new({ linesource => $object_with_get_next_line_method [, nhistory => N] })

The linesource argument can be an object with a get_next_line method. This method must return undef when there are no more lines to read.

new({ linesource => \&subroutine_that_returns_lines [, nhistory => N] })

If none of the above input methods suits your needs, you can give a subroutine reference and wrap whatever code you need to get the VHDL source. When called, this subroutine must return each line of source code in turn, and then return undef when there are no more lines.



Returns the linesource argument passed into the constructor. Before version 1.0 of this module, this method was called linesource().


In scalar context, returns the next VHDL token.

In list context, returns a token type code and the token

Nothing is removed from the source code: if you concatenate all the tokens returned by get_next_token(), you will get the same result as if you concatenate all the strings returned by the linesource object.

The token type codes are 1 or 2-character strings. When the codes are 2 characters, the first character gives the general class of the token and the second indicates its type more specifically. The first character will be 'w' for whitespace, 'r' for comments (remarks) or 'c' for code. It should be possible to remove all comment tokens, and change whitespace tokens for different whitespace, and always end up with functionally equivalent code.

The token type codes are:


Whitespace:Newline. This could be any of \012, \015, \015\012 or \012\015.


Whitespace:Spaces. A group of whitespace characters which match the /s regexp pattern but which do not include any carriage-return or linefeed characters.


Remark. The token will start with two dashes and include the remainder of the source code line, not including any newline characters. The next token will either be a newline or undef.


Code:String literal. The lexer accepts multi-line strings, even though the VHDL specification does not allow them.


Code:Character literal.


Code:Bit_vector literal. For example, B"001_1010" or O"7720" or H"A7_DEAD".


Code:Numeric literal. This could be a specified-base literal like 8#7720# or a simple integer or floating-point value.


Code:Identifier or keyword. For example, package or my_signal_23 or /extended identifier$%!/..


Code:Punctuation. A 1 or 2-character group of punctuation symbols that is part of VHDL syntax. For example, '<=' is returned as a single 'cp' token, as is '&', but '#' would be returned as an unexpected character (see below).


Unexpected character. Any character in the source that does not match any of the above definitions, and cannot be part of valid VHDL code. Note that prior to version 1.0 of this module, these would be returned with the 'cp' token type code.


Returns previous code tokens. N must not be larger than the nhistory argument passed to the constructor. history(0) will return the text of the last token returned by get_next_token whose type started with a 'c', history(1) will return the code token before that, and so on.


Michael Attenborough, <michael.attenborough at>


This module requires the following modules to be available:

    Carp: any version Class::Std: any version Readonly: version 1.03 or later


"Argument to Hardware::Vhdl::Lexer->new() must be hash reference"

Have you remembered to put curly brackets around the argument list? Pre-1.0 versions of this module used to take the arguments to new() as a direct hash, but version 1.0 onwards need a hash reference. This means that the curly brackets need to be added when migrating from pre-1.0 to 1.0 or later.

    # Old style (argument list is hash) - doesn't work any more
    my $lexer = Hardware::Vhdl::Lexer->new( linesource => $fh );

    # New style (argument is a hash ref) - do it this way now
    my $lexer = Hardware::Vhdl::Lexer->new({ linesource => $fh });
"Hardware::Vhdl::Lexer constructor requires a linesource to be specified"

The 'linesource' argument to Hardware::Vhdl::Lexer->new() is required, and it is a fatal error not to provide one.

"Hardware::Vhdl::Lexer->new 'linesource' parameter is not of a valid type (it is not a reference)"

The linesource parameter needs to be a reference to something. If your VHDL code to be passed is in a scalar string, you need to pass in a reference to the string, not the string itself.

"Hardware::Vhdl::Lexer->new 'linesource' parameter is not of a valid type (type is '<type>')"

The linesource parameter that you have passed to new() does not appear to be a reference to a scalar, a list, a filehandle, a subroutine or an object with a get_next_line method. You have passed a reference to something (otherwise you would see the previous message) and the error message will tell you what it appears to be a reference to.

"Internal error (token failed to match anything)"

This is a "this should never happen" type of error, and is a sign that I have included a bug. If you ever see this error, I would appreciate a bug report describing how to reproduce the error.


Please report any bugs or feature requests to bug-hardware-vhdl-lexer at, or through the web interface at I will be notified, and then you'll automatically be notified of progress on your bug as I make changes.


You can find documentation for this module with the perldoc command.

    perldoc Hardware::Vhdl::Lexer

You can also look for information at:


Copyright 2006 Michael Attenborough, all rights reserved.

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.