CGI::FormMagick::TagMaker is a Perl module created to generate HTML tags.
my $html = CGI::FormMagick::TagMaker->new();
$html->input( type => 'submit' ),
This Perl 5 object class can be used to generate any HTML tags in a format that is consistent with the W3C HTML 4.0 standard. There are no restrictions on what tags are named, however; you can ask for any new or unsupported tag that comes along from Netscape or Microsoft, and it will be made. Additionally, you can generate lists of said tags with one method call, or just parts of said tags (but not both at once).
In this implementation, "standard format" means that tags are made as pairs () by default, unless they are known to be "no pair" tags. Tags that I know to be "no pair" are [basefont, img, area, param, br, hr, input, option, tbody, frame, comment, isindex, base, link, meta]. However, you can force any tag to be "pair" or "start only" or "end only" by appropriately modifying your call to the tag making method.
Also, "standard format" means that tag modifiers are formatted as "key=value" by default, unless they are known to be "no value" modifiers. Modifiers that I know to be "no value" are [ismap, noshade, compact, checked, multiple, selected, nowrap, noresize, param]. These are formatted simply as "key" because their very presence indicates positive assertion, while their absense means otherwise. For modifiers with values, the values will always become bounded by quotes, which ensures they work with both string and numerical quantities (eg: key="value").
Note that this class is a subclass of Class::ParamParser, and inherits all of its methods, "params_to_hash()" and "params_to_array()".
Through the magic of autoloading, this class can make any html tag by calling a class method with the same name as the tag you want. For examples, use "hr()" to make a "< HR >" tag, or "p('text')" to make "< P >text< /P >". This also means that if you mis-spell any method name, it will still make a new tag with the mis-spelled name. For autoloaded methods only, the method names are case-insensitive.
If you call a class method whose name ends in either of ['_start', '_end', '_pair'], this will be interpreted as an instruction to make just part of one tag whose name are the part of the method name preceeding that suffix. For example, calling "p_start( 'text' )" results in "< P >text" rather than "< P >text< /P >". Similarly, calling "p_end()" will generate a "< /P >" only. Using the '_pair' suffix will force tags to be made as a pair, whether or not they would do so naturally. For example, calling "br_pair" would produce a "< BR >< /BR >" rather than the normal "< BR >". When using either of ['_start','_pair'], the arguments you pass the method are exactly the same as the unmodified method would use, and there are no other symantec differences. However, when using the '_end' suffix, any arguments are ignored, as the latter member of a tag pair never carries any attributes anyway.
If you call a class method whose name ends in "_group", this will be interpreted as an instruction to make a list of tags whose name are the part of the method name preceeding the "_group". For example, calling "td_group( ['here','we','are'] )" results in "< TD >here< /TD >< TD >we< /TD >< TD >are< /TD >" being generated. The arguments that you call this method are exactly the same as for calling a method to make a single tag of the same name, except that the extra optional parameter "list" can be used to force an ARRAY ref of the new tags to be returned instead of a scalar. The symantec difference is that any arguments whose values are ARRAY refs are interpreted as a list of values where each one is used in a separate tag; for a single tag, the literal ARRAY ref itself would be used. The number of tags produced is equal to the length of the longest ARRAY ref passed as an argument. For any other arguments who have fewer than this count, their last value is replicated and appended enough times as necessary to make them the same length. The value of a scalar argument is used for all the tags. For example, calling "input_group( type => checkbox, name => 'letters', value => ['a','b','c'] )" produces '< INPUT TYPE="checkbox" NAME="letters" VALUE="a" >< INPUT TYPE="checkbox" NAME="letters" VALUE="b" >< INPUT TYPE="checkbox" NAME="letters" VALUE="c" >'.
All autoloaded methods require their parameters to be in named format. These names and values correspond to attribute names and values for the new tags. Since "no value" attributes are essentially booleans, they can have any true or false value associated with them in the parameter list, which won't be printed. If an autoloaded method is passed exactly one parameter, it will be interpreted as the "text" that goes between the tag pair (< TAG >text< /TAG >) or after "start tags" (< TAG >text). The same result can be had explicitely by passing the named parameter "text". Most static (non-autoloaded) methods require positional parameters, except for start_html(), which can take either format. The names of any named parameters can optionally start with a "-".