CONTENTS

NAME

POE::Filter::Line - serialize and parse terminated records (lines)

SYNOPSIS

#!perl

use POE qw(Wheel::FollowTail Filter::Line);

POE::Session->create(
  inline_states => {
    _start => sub {
      $_[HEAP]{tailor} = POE::Wheel::FollowTail->new(
        Filename => "/var/log/system.log",
        InputEvent => "got_log_line",
        Filter => POE::Filter::Line->new(),
      );
    },
    got_log_line => sub {
      print "Log: $_[ARG0]\n";
    }
  }
);

POE::Kernel->run();
exit;

DESCRIPTION

POE::Filter::Line parses stream data into terminated records. The default parser interprets newlines as the record terminator, and the default serializer appends network newlines (CR/LF, or "\x0D\x0A") to outbound records.

Record terminators are removed from the data POE::Filter::Line returns.

POE::Filter::Line supports a number of other ways to parse lines. Constructor parameters may specify literal newlines, regular expressions, or that the filter should detect newlines on its own.

PUBLIC FILTER METHODS

POE::Filter::Line's new() method has some interesting parameters.

new

new() accepts a list of named parameters.

In all cases, the data interpreted as the record terminator is stripped from the data POE::Filter::Line returns.

InputLiteral may be used to parse records that are terminated by some literal string. For example, POE::Filter::Line may be used to parse and emit C-style lines, which are terminated with an ASCII NUL:

my $c_line_filter = POE::Filter::Line->new(
  InputLiteral => chr(0),
  OutputLiteral => chr(0),
);

OutputLiteral allows a filter to put() records with a different record terminator than it parses. This can be useful in applications that must translate record terminators.

Literal is a shorthand for the common case where the input and output literals are identical. The previous example may be written as:

my $c_line_filter = POE::Filter::Line->new(
  Literal => chr(0),
);

An application can also allow POE::Filter::Line to figure out which newline to use. This is done by specifying InputLiteral to be undef:

my $whichever_line_filter = POE::Filter::Line->new(
  InputLiteral => undef,
  OutputLiteral => "\n",
);

InputRegexp may be used in place of InputLiteral to recognize line terminators based on a regular expression. In this example, input is terminated by two or more consecutive newlines. On output, the paragraph separator is "---" on a line by itself.

my $paragraph_filter = POE::Filter::Line->new(
  InputRegexp => "([\x0D\x0A]{2,})",
  OutputLiteral => "\n---\n",
);

MaxBuffer sets the maximum amount of data that the filter will hold onto while trying to find a line ending. Defaults to 512 MB.

MaxLength sets the maximum length of a line. Defaults to 64 MB.

If either the MaxLength or MaxBuffer constraint is exceeded, POE::Filter::Line will throw an exception.

PUBLIC FILTER METHODS

POE::Filter::Line has no additional public methods.

SUBCLASSING

POE::Filter::Line exports the FIRST_UNUSED constant. This points to the first unused element in the $self array reference. Subclasses should store their own data beginning here, and they should export their own FIRST_UNUSED constants to help future subclassers.

SEE ALSO

Please see POE::Filter for documentation regarding the base interface.

The SEE ALSO section in POE contains a table of contents covering the entire POE distribution.

BUGS

The default input newline parser is a regexp that has an unfortunate race condition. First the regular expression:

/(\x0D\x0A?|\x0A\x0D?)/

While it quickly recognizes most forms of newline, it can sometimes detect an extra blank line. This happens when a two-byte newline character is broken between two reads. Consider this situation:

some stream dataCR
LFother stream data

The regular expression will see the first CR without its corresponding LF. The filter will properly return "some stream data" as a line. When the next packet arrives, the leading "LF" will be treated as the terminator for a 0-byte line. The filter will faithfully return this empty line.

It is advised to specify literal newlines or use the autodetect feature in applications where blank lines are significant.

AUTHORS & COPYRIGHTS

Please see POE for more information about authors and contributors.