Parse::BBCode(3) Module to parse BBCode and render it as HTML or text

SYNOPSIS

Parse::BBCode parses common bbcode like


[b]bold[/b] [size=10]big[/size]

short tags like

    [foo://test]

and custom bbcode tags.

For the documentation of short tags, see ``SHORT TAGS''.

To parse a bbcode string, set up a parser with the default HTML defintions of Parse::BBCode::HTML:

    # render bbcode to HTML
    use Parse::BBCode;
    my $p = Parse::BBCode->new();
    my $code = 'some [b]b code[/b]';
    my $rendered = $p->render($code);
    # parse bbcode, manipulate tree and render
    use Parse::BBCode;
    my $p = Parse::BBCode->new();
    my $code = 'some [b]b code[/b]';
    my $tree = $p->parse($code);
    # do something with $tree
    my $rendered = $p->render_tree($tree);

Or if you want to define your own tags:

    my $p = Parse::BBCode->new({
            tags => {
                # load the default tags
                Parse::BBCode::HTML->defaults,
                # add/override tags
                url => 'url:<a href="%{link}A">%{parse}s</a>',
                i   => '<i>%{parse}s</i>',
                b   => '<b>%{parse}s</b>',
                noparse => '<pre>%{html}s</pre>',
                code => sub {
                    my ($parser, $attr, $content, $attribute_fallback) = @_;
                    if ($attr eq 'perl') {
                        # use some syntax highlighter
                        $content = highlight_perl($content);
                    }
                    else {
                        $content = Parse::BBCode::escape_html($$content);
                    }
                    "<tt>$content</tt>"
                },
                test => 'this is klingon: %{klingon}s',
            },
            escapes => {
                klingon => sub {
                    my ($parser, $tag, $text) = @_;
                    return translate_into_klingon($text);
                },
            },
        }
    );
    my $code = 'some [b]b code[/b]';
    my $parsed = $p->render($code);

DESCRIPTION

If you set up the Parse::BBCode object without arguments, the default tags are loaded, and any text outside or inside of parseable tags will go through a default subroutine which escapes HTML and replaces newlines with <br> tags. If you need to change this you can set the options 'url_finder', 'text_processor' and 'linebreaks'.

METHODS

new
Constructor. Takes a hash reference with options as an argument.

    my $parser = Parse::BBCode->new({
        tags => {
            url => ...,
            i   => ...,
        },
        escapes => {
            link => ...,
        },
        close_open_tags   => 1, # default 0
        strict_attributes => 0, # default 1
        direct_attributes => 1, # default 1
        url_finder        => 1, # default 0
        smileys           => 0, # default 0
        linebreaks        => 1, # default 1
    );
tags
See ``TAG DEFINITIONS''
escapes
See ``ESCAPES''
url_finder
See ``URL FINDER''
smileys
If you want to replace smileys with an icon:

    my $parser = Parse::BBCode->new({
            smileys => {
                base_url => '/your/url/to/icons/',
                icons => { qw/ :-) smile.png :-( sad.png / },
                # sprintf format:
                # first argument url
                # second argument original text smiley (HTML escaped)
                format => '<img src="%s" alt="%s">',
                # if you need the url and text in a different order
                # see perldoc -f sprintf, e.g.
                # format => '<img alt="%2$s" src="%1$s">',
            },
        });

This subroutine will be applied during the url_finder (or first, if url_finder is 0), and the rest will get processed by the text procesor (default escaping html and replacing linebreaks).

Smileys are only replaced if surrounded by whitespace or start/end of line/text.

    [b]bold<hr> :-)[/b] :-(

In this example both smileys will be replaced. The first smiley is at the end of the text because the text inside [b][/b] is processed on its own.

Open to any suggestions here.

linebreaks
The default text processor replaces linebreaks with <br>\n. If you don't want this, set 'linebreaks' to 0.
text_processor
If you need to add any customized text processing (like smiley parsing, for example), you can pass a subroutine here. Note that this subroutine also needs to do HTML escaping itself!

See ``TEXT PROCESSORS''

close_open_tags
Default: 0

If set to true (1), it will close open tags at the end or before block tags.

strict_attributes
Default: 1

If set to true (1), tags with invalid attributes are left unparsed. If set to false (0), the attribute for this tags will be empty.

An invalid attribute:

    [foo=bar far boo]...[/foo]

I might add an option to define your own attribute validation. Contact me if you'd like to have this.

direct_attributes
Default: 1

Normal tag syntax is:

  [tag=val1 attr2=val2 ...]

If set to 0, tag syntax is

  [tag attr2=val2 ...]
attribute_quote
You can change how the attribute values shuold be quoted. Default is a double quote (which is still optional):

  my $parser = Parse::BBCode->new(
      attribute_quote => '"',
      ...
  );
  [tag="foo" attr="bar" attr2=baz]...[/tag]

If you set it to single quote:

  my $parser = Parse::BBCode->new(
      attribute_quote => "'",
      ...
  );
  [tag='foo' attr=bar attr2='baz']...[/tag]

You can also set it to both: "'"". Then both quoting types are allowed:

  my $parser = Parse::BBCode->new(
      attribute_quote => q/'"/,
      ...
  );
  [tag='foo' attr="bar" attr2=baz]...[/tag]
attribute_parser
You can pass a subref that overrides the default attribute parsing. See ``ATTRIBUTE PARSING''
strip_linebreaks
Default: 1

Strips linebreaks at start/end of block tags

render
Input: The text to parse, optional hashref

Returns: the rendered text

    my $rendered = $parser->render($bbcode);

You can pass an optional hashref with information you need inside of your self-defined rendering subs. For example if you want to display code in a codebox with a link to download the code you need the id of the article (in a forum) and the number of the code tag.

    my $parsed = $parser->render($bbcode, { article_id => 23 });
    # in the rendering sub:
        my ($parser, $attr, $content, $attribute_fallback, $tag, $info) = @_;
        my $article_id = $parser->get_params->{article_id};
        my $code_id = $tag->get_num;
        # write downloadlink like
        # download.pl?article_id=$article_id;code_id=$code_id
        # in front of the displayed code

See examples/code_download.pl for a complete example of how to set up the rendering and how to extract the code from the tree. If run as a CGI skript it will give you a dialogue to save the code into a file, including a reasonable default filename.

parse
Input: The text to parse.

Returns: the parsed tree (a Parse::BBCode::Tag object)

    my $tree = $parser->parse($bbcode);
render_tree
Input: the parse tree

Returns: The rendered text

    my $parsed = $parser->render_tree($tree);

You can pass an optional hashref, for explanation see ``render''

forbid
    $parser->forbid(qw/ img url /);

Disables the given tags.

permit
    $parser->permit(qw/ img url /);

Enables the given tags if they are in the tag definitions.

escape_html
Utility to substitute

    <>&"'

with their HTML entities.

    my $escaped = Parse::BBCode::escape_html($text);
error
If the given bbcode is invalid (unbalanced or wrongly nested classes), currently Parse::BBCode::render() will either leave the invalid tags unparsed, or, if you set the option "close_open_tags", try to add closing tags. If this happened "error()" will return the invalid tag(s), otherwise false. To get the corrected bbcode (if you set "close_open_tags") you can get the tree and return the raw text from it:

    if ($parser->error) {
        my $tree = $parser->get_tree;
        my $corrected = $tree->raw_text;
    }
parse_attributes
You can inherit from Parse::BBCode and define your own attribute parsing. See ``ATTRIBUTE PARSING''.
new_tag
Returns a Parse::BBCode::Tag object. It just does:
    shift;
    Parse::BBCode::Tag->new(@_);

If you want your own tag class, inherit from Parse::BBCode and let it return Parse::BBCode::YourTag->new

TAG DEFINITIONS

Here is an example of all the current definition possibilities:

    my $p = Parse::BBCode->new({
            tags => {
                i   => '<i>%s</i>',
                b   => '<b>%{parse}s</b>',
                size => '<font size="%a">%{parse}s</font>',
                url => 'url:<a href="%{link}A">%{parse}s</a>',
                wikipedia => 'url:<a href="http://wikipedia.../?search=%{uri}A">%{parse}s</a>',
                noparse => '<pre>%{html}s</pre>',
                quote => 'block:<blockquote>%s</blockquote>',
                code => {
                    code => sub {
                        my ($parser, $attr, $content, $attribute_fallback) = @_;
                        if ($attr eq 'perl') {
                            # use some syntax highlighter
                            $content = highlight_perl($$content);
                        }
                        else {
                            $content = Parse::BBCode::escape_html($$content);
                        }
                        "<tt>$content</tt>"
                    },
                    parse => 0,
                    class => 'block',
                },
                hr => {
                    class => 'block',
                    output => '<hr>',
                    single => 1,
                },
            },
        }
    );

The following list explains the above tag definitions:

%s
    i => '<i>%s</i>'
    [i] italic <html> [/i]
    turns out as
    <i> italic &lt;html&gt; </i>

So %s stands for the tag content. By default, it is parsed itself, so that you can nest tags.

"%{parse}s"
    b   => '<b>%{parse}s</b>'
    [b] bold <html> [/b]
    turns out as
    <b> bold &lt;html&gt; </b>

"%{parse}s" is the same as %s because 'parse' is the default.

%a
    size => '<font size="%a">%{parse}s</font>'
    [size=7] some big text [/size]
    turns out as
    <font size="7"> some big text </font>

So %a stands for the tag attribute. By default it will be HTML escaped.

url tag, %A, "%{link}A"
    url => 'url:<a href="%{link}a">%{parse}s</a>'

the first thing you can see is the "url:" at the beginning - this defines the url tag as a tag with the class 'url', and urls must not be nested. So this class definition is mainly there to prevent generating wrong HTML. if you nest url tags only the outer one will be parsed.

another thing you can see is how to apply a special escape. The attribute defined with "%{link}a" is checked for a valid URL. "javascript:" will be filtered.

    [url=/foo.html]a link[/url]
    turns out as
    <a href="/foo.html">a link</a>

Note that a tag like

    [url]http://some.link.example[/url]

will turn out as

    <a href="">http://some.link.example</a>

In the cases where the attribute should be the same as the content you should use %A instead of %a which takes the content as the attribute as a fallback. You probably need this in all url-like tags:

    url => 'url:<a href="%{link}A">%{parse}s</a>',
"%{uri}A"
You might want to define your own urls, e.g. for wikipedia references:

    wikipedia => 'url:<a href="http://wikipedia/?search=%{uri}A">%{parse}s</a>',

"%{uri}A" will uri-encode the searched term:

    [wikipedia]Harold & Maude[/wikipedia]
    [wikipedia="Harold & Maude"]a movie[/wikipedia]
    turns out as
    <a href="http://wikipedia/?search=Harold+%26+Maude">Harold &amp; Maude</a>
    <a href="http://wikipedia/?search=Harold+%26+Maude">a movie</a>
Don't parse tag content
Sometimes you need to display verbatim bbcode. The simplest form would be a noparse tag:

    noparse => '<pre>%{html}s</pre>'
    [noparse] [some]unbalanced[/foo] [/noparse]

With this definition the output would be

    <pre> [some]unbalanced[/foo] </pre>

So inside a noparse tag you can write (almost) any invalid bbcode. The only exception is the noparse tag itself:

    [noparse] [some]unbalanced[/foo] [/noparse] [b]really bold[/b] [/noparse]

Output:

    [some]unbalanced[/foo] <b>really bold</b> [/noparse]

Because the noparse tag ends at the first closing tag, even if you have an additional opening noparse tag inside.

The "%{html}s" defines that the content should be HTML escaped. If you don't want any escaping you can't say %s because the default is 'parse'. In this case you have to write "%{noescape}".

Block tags
    quote => 'block:<blockquote>%s</blockquote>',

To force valid html you can add classes to tags. The default class is 'inline'. To declare it as a block add "'block:"" to the start of the string. Block tags inside of inline tags will either close the outer tag(s) or leave the outer tag(s) unparsed, depending on the option "close_open_tags".

Define subroutine for tag
All these definitions might not be enough if you want to define your own code, for example to add a syntax highlighter.

Here's an example:

    code => {
        code => sub {
            my ($parser, $attr, $content, $attribute_fallback, $tag, $info) = @_;
            if ($attr eq 'perl') {
                # use some syntax highlighter
                $content = highlight_perl($$content);
            }
            else {
                $content = Parse::BBCode::escape_html($$content);
            }
            "<tt>$content</tt>"
        },
        parse => 0,
        class => 'block',
    },

So instead of a string you define a hash reference with a 'code' key and a sub reference. The other key is "parse" which is 0 by default. If it is 0 the content in the tag won't be parsed, just as in the noparse tag above. If it is set to 1 you will get the rendered content as an argument to the subroutine.

The first argument to the subroutine is the Parse::BBCode object itself. The second argument is the attribute, the third is the already rendered tag content as a scalar reference and the fourth argument is the attribute fallback which is set to the content if the attribute is empty. The fourth argument is just for convenience. The fifth argument is the tag object (Parse::BBCode::Tag) itself, so if necessary you can get the original tag content by using:

    my $original = $tag->raw_text;

The sixth argument is an info hash. It contains:

    my $info = {
        tags => $tags,
        stack => $stack,
        classes => $classes,
    };

The variable $tags is a hashref which contains all tag names which are outside the current tag, with a count. This is convenient if you have to check if the current processed tag is inside a certain tag and you want to behave differently, like

    if ($info->{tags}->{special}) {
        # we are somewhere inside [special]...[/special]
    }

The variable $stack contains an array ref with all outer tag names. So while processing the tag 'i' in

    [quote][quote][b]bold [i]italic[/i][/b][/quote][/quote]

it contains
    [qw/ quote quote b i /]

The variable $classes contains a hashref with all tag classes and their counts outside of the current processed tag. For example if you want to process URIs with URI::Find, and you are already in a tag with the class 'url' then you don't want to use URI::Find here.

    unless ($info->{classes}->{url}) {
        # not inside of a url class tag ([url], [wikipedia, etc.)
        # parse text for urls with URI::Find
    }
Single-Tags
Sometimes you might want single tags like for a horizontal line:

    hr => {
        class => 'block',
        output => '<hr>',
        single => 1,
    },

The hr-Tag is a block tag (should not be inside inline tags), and it has no closing tag (option "single")

    [hr]
    Output:
    <hr>

ESCAPES

    my $p = Parse::BBCode->new({
        ...
        escapes => {
            link => sub {
            },
        },
    });

You can define or override escapes. Default escapes are html, uri, link, email, htmlcolor, num. An escape functions as a validator and filter. For example, the 'link' escape looks if it got a valid URI (starting with "/" or "\w+://") and html-escapes it. It returns the empty string if the input is invalid.

See ``default_escapes'' in Parse::BBCode::HTML for the detailed list of escapes.

URL FINDER

Usually one wants to also create hyperlinks from any url found in the bbcode, not only in url tags. The following code will use URI::Find to search for all types of urls (unless inside of a url tag itself), create a link in the given format and html-escape the rest. If the url is longer than 50 chars, it will cut the link title and append three dots. If you set max_length to 0, the title won't be cut.

    my $p = Parse::BBCode->new({
            url_finder => {
                max_length  => 50,
                # sprintf format:
                format      => '<a href="%s" rel="nofollow">%s</a>',
            },
            tags => ...
        });

Note: If you use the special tag '' in the tag definitions you will overwrite the url finder and have to do that yourself.

Alternative:

    my $p = Parse::BBCode->new({
            url_finder => 1,
            ...

This will use the default like shown above (max length 50 chars).

Default is 0.

ATTRIBUTES

There are two types of tags. The default (option direct_attributes=1):

    [foo=bar a=b c=d]
    [foo="text with space" a=b c=d]

The parsed attribute structure will look like:

    [ ['bar'], ['a' => 'b'], ['c' => 'd'] ]

Another bbcode variant doesn't use direct attributes:

    [foo a=b c=d]

The resulting attribute structure will have an empty first element:

    [ [''], ['a' => 'b'], ['c' => 'd'] ]

ATTRIBUTE PARSING

If you have bbcode attributes that don't fit into the two standard syntaxes you can inherit from Parse::BBCode and overwrite the parse_attributes method, or you can pass an option attribute_parser contaning a subref.

Example:

    [size=10]big[/size] [foo|bar|boo]footext[/foo] end

The size tag should be parsed normally, the foo tag needs different parsing.

    sub parse_attributes {
        my ($self, %args) = @_;
        # $$text contains '|bar|boo]footext[/foo] end
        my $text = $args{text};
        my $tagname = $args{tag}; # 'foo'
        if ($tagname eq 'foo') {
            # work on $$text
            # result should be something like:
            # $$text should contain 'footext[/foo] end'
            my $valid = 1;
            my @attr = ( [''], [1 => 'bar'], [2 => 'boo'] );
            my $attr_string = '|bar|boo';
            return ($valid, [@attr], $attr_string, ']');
        }
        else {
            return shift->SUPER::parse_attributes(@_);
        }
    }
    my $parser = Parse::BBCode->new({
        ...
        attribute_parser => \&parse_attributes,
    });

If the attributes are not valid, return

    0, [ [''] ], '|bar|boo', ']'

If you don't find a closing square bracket, return:

    0, [ [''] ], '|bar|boo', ''

TEXT PROCESSORS

If you set url_finder and linebreaks to 1, the default text processor will work like this:

    my $post_processor = \&sub_for_escaping_HTML;
    $text = code_to_replace_urls($text, $post_processor);
    $text =~ s/\r?\n|\r/<br>\n/g;
    return $text;

It will be applied to text outside of bbcode and inside of parseable bbcode tags (and not to code tags or other tags with unparsed content).

If you need an additional post processor this usually cannot be done after the HTML escaping and url finding. So if you write a text processor it must do the HTML escaping itself. For example if you want to replace smileys with image tags you cannot simply do:

    $text =~ s/ :-\) /<img src=...>/g;

because then the image tag would be HTML escaped after that. On the other hand it's usually not possible to do something like that *after* the HTML escaping since that might introduce text sequences that look like a smiley (or whatever you want to replace).

So a simple example for a customized text processor would be:

    ...
    url_finder     => 1,
    linebreaks     => 1,
    text_processor => sub {
        # for $info hash description see render() method
        my ($text, $info) = @_;
        my $out = '';
        while ($text =~ s/(.*)( |^)(:\))(?= |$)//mgs) {
            # match a smiley and anything before
            my ($pre, $sp, $smiley) = ($1, $2, $3);
            # escape text and add smiley image tag
            $out .= Parse::BBCode::escape_html($pre) . $sp . '<img src=...>';
        }
        # leftover text
        $out .= Parse::BBCode::escape_html($text);
        return $out;
    },

This will result in: Replacing urls, applying your text_processor to the rest of the text and after that replace linebreaks with <br> tags.

If you want to completely define the plain text processor yourself (ignoring the 'linebreak', 'url_finder', 'smileys' and 'text_processor' options) you define the special tag with the empty string:

    my $p = Parse::BBCode->new({
        tags => {
            '' => sub {
                my ($parser, $attr, $content, $info) = @_;
                return frobnicate($content);
                # remember to escape HTML!
            },
            ...

SHORT TAGS

It can be very convenient to have short tags like [foo://id]. This is not really a part of BBCode, but I consider it as quite similar, so I added it to this module. For example to link to threads, cpan modules or wikipedia articles:

    [thread://123]
    [thread://123|custom title]
    # can be implemented so that it links to thread 123 in the forum
    # and additionally fetch the thread title.
    [cpan://Module::Foo|some useful module]
    [wikipedia://Harold & Maude]

You can define a short tag by adding the option "short". The tag will work as a classic tag and short tag. If you only want to support the short version, set the option "classic" to 0.

    my $p = Parse::BBCode->new({
            tags => {
                Parse::BBCode::HTML->defaults,
                wikipedia => {
                    short   => 1,
                    output  => '<a href="http://wikipedia/?search=%{uri}A">%{parse}s</a>',
                    class   => 'url',
                    classic => 0, # don't support classic [wikipedia]...[/wikipedia]
                },
                thread => {
                    code => sub {
                        my ($parser, $attr, $content, $attribute_fallback) = @_;
                        my $id = $attribute_fallback;
                        if ($id =~ tr/0-9//c) {
                            return '[thread]' . encode_entities($id) . '[/thread]';
                        }
                        my $name;
                        if ($attr) {
                            # custom title will be in $attr
                            # [thread=123]custom title[/thread]
                            # [thread://123|custom title]
                            # already escaped
                            $name = $$content;
                        }
                        return qq{<a href="/thread/$id">$name</a>};
                    },
                    short   => 1,
                    classic => 1, # default is 1
                },
            },
        }
    );

WHY ANOTHER BBCODE PARSER

I wrote this module because HTML::BBCode is not extendable (or I didn't see how) and BBCode::Parser seemed good at the first glance but has some issues, for example it says that the following bbode

    [code] foo [b] [/code]

is invalid, while I think you should be able to write unbalanced code in code tags. Also BBCode::Parser dies if you have invalid code or not-permitted tags, but in a forum you'd rather show a partly parsed text then an error message.

What I also wanted is an easy syntax to define own tags, ideally - for simple tags - as plain text, so you can put it in a configuration file. This allows forum admins to add tags easily. Some forums might want a tag for linking to perlmonks.org, other forums need other tags.

Another goal was to always output a result and don't die. I might add an option which lets the parser die with unbalanced code.

WHY BBCODE?

Some forums and blogs prefer a kind of pseudo HTML for user comments. The arguments against bbcode is usually: ``Why should people learn an additional markup language if they can just use HTML?'' The problem is that many people don't know HTML.

BBCode is often a bit shorter, for example if you have a code tag with an attribute that tells the parser what language the content is in.

    [code=perl]...[/code]
    <code language="perl">...</code>

Also, forum HTML is usually not real HTML. It is usually a subset and sometimes with additional tags. So in the backend you need to parse it anyway to turn it into real HTML.

BBCode is widely known and used. Unfortunately though, there is no specification; some forums only allow attributes in double quotes, some forums implement only one attribute that can be separated by spaces, which makes it difficult to parse if you want to support more than one attribute.

I tried to support the most common syntax (attributes without quotes, in single or double quotes) and tags. If you need additional tags it's relatively easy to implement them. For example in my forum I implemented a [more] tag that hides long text or code in thread view. Without Javascript you will see the expanded content when clicking on the single article, or with Javascript the content will be added inline via Ajax.

TODO

BBCode to Textile|Markdown
There is a Parse::BBCode::Markdown module which is only roughly tested.
API
The main syntax is likely to stay, only the API for callbacks might change. At the moment it is not possible to add callbacks to the parsing process, only for the rendering phase.

REQUIREMENTS

perl >= 5.8.0, Class::Accessor::Fast, URI::Escape

AUTHOR

Tina Mueller

CREDITS

Thanks to Moritz Lenz for his suggestions about the implementation and the test cases.

Viacheslav Tikhanovskii

Sascha Kiefer

COPYRIGHT AND LICENSE

Copyright (C) 2014 by Tina Mueller

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.6.1 or, at your option, any later version of Perl 5 you may have available.