NAME Template::Parser::CET - CGI::Ex::Template based parser for the TT2 engine SYNOPSIS use Template; use Template::Parser::CET; my $t = Template->new( PARSER => Template::Parser->new ); # you can override all instances of TT # by any of the following methods use Template::Parser::CET activate => 1; # OR use Template::Parser::CET; Template::Parser::CET->activate; # OR use Template::Config; $Template::Config::PARSER = 'Template::Parser::CET'; my $t = Template->new; DESCRIPTION Template::Parser::CET provides much or most of the TT3 syntax and runs on the current TT2 engine. CGI::Ex::Template (CET) provides a fast implementation of TT2 and TT3. There are some cases where Template::Toolkit is faster. There are also some cases where shops have custom providers, or custom stashes that require the use of the current TT2 engine. In these cases, Template::Parser::CET provides the best of both worlds - offering TT2 AND TT3 syntax and running on the existing platform making use of all of your current work (In many cases CET should be able to do this anyway). This module may eventually be made obsolete when the final real Template::Toolkit 3 engine by Andy Wardley is released. But that would only be a good thing. If the TT3 engine doesn't provide full backward compatibility this module will. CET has provided TT3 features since Spring of 2006 but there has been little reported uptake. The TT3 features/extended syntax are very compelling. For various reasons people chose not to use CET. Now people can use TT2 and get the features of TT3 (through CET) today. Hopefully Template::Parser::CET and CGI::Ex::Template can be used in the same spirit as Pugs is used for Perl 6. All of the code from CET and Template::Parser::CET are free for use in TT3. SPEED All speed is relative and varies tremendously depending upon the size and content of your template. Template::Parser::CET generally compiles documents at the same speed as Template::Parser and Template::Grammar. CGI::Ex::Template compiles documents to its AST (abastract syntax tree) very quickly, but Template::Paser::CET then has to emit a TT2 style compiled Template::Document perl document. So even though CGI::Ex::Template has a speed advantage, the advantage is lost in Template::Parser::CET. If you use compiled in memory templates - they will execute as quickly as the normal TT2 documents. In all other cases Template::Parser::CET will prepare the documents at about the same speed. FEATURES So what exactly are the features and syntax that Template::Parser::CET provides? The following is a list of most of the features that will be in TT3 and are in Template::Parser::CET. All of the listed features are in addition to those provided natively by Template::Toolkit. Grammar CGI::Ex::Template provides Template::Parser::CET with a recursive grammar. This provides a range of benefits including speed, better error reporting, more consistent syntax, more possibilities for extending the grammar. Numerical hash keys work [% a = {1 => 2} %] All hash key parsing is a little more sane. Not entirely more since CET needs to be backwards compatible. Quoted hash key interpolation is fine [% a = {"$foo" => 1} %] Multiple ranges in same array constructor [% a = [1..10, 21..30] %] Constructor types can call virtual methods. (TT3) [% a = [1..10].reverse %] [% "$foo".length %] [% 123.length %] # = 3 [% 123.4.length %] # = 5 [% -123.4.length %] # = -5 ("." binds more tightly than "-") [% (a ~ b).length %] [% "hi".repeat(3) %] # = hihihi [% {a => b}.size %] # = 1 The "${" and "}" variable interpolators can contain expressions, not just variables. [% [0..10].${ 1 + 2 } %] # = 4 [% {ab => 'AB'}.${ 'a' ~ 'b' } %] # = AB [% color = qw/Red Blue/; FOR [1..4] ; color.${ loop.index % color.size } ; END %] # = RedBlueRedBlue You can use regular expression quoting. [% "foo".match( /(F\w+)/i ).0 %] # = foo [% a = /a b c . e/xs %] Tags can be nested. [% f = "[% (1 + 2) %]" %][% f | eval %] # = 3 Reserved names are less reserved. [% GET GET %] # gets the variable named "GET" [% GET $GET %] # gets the variable who's name is stored in "GET" Pipe "|" can be used anywhere dot "." can be and means to call the virtual method. [% a = {size => "foo"} %][% a.size %] # = foo [% a = {size => "foo"} %][% a|size %] # = 1 (size of hash) Added V2PIPE configuration item Restores the behavior of the pipe operator to be compatible with TT2. With V2PIPE = 1 [% PROCESS a | repeat(2) %] # = value of block or file a repeated twice With V2PIPE = 0 (default) [% PROCESS a | repeat(2) %] # = process block or file named a ~ a Added "fmt" scalar, list, and hash virtual methods which work similar to the Perl 6 methods. [% text.fmt("%s") %] [% list.fmt("%s", ", ") %] [% hash.fmt("%s => %s", "\n") %] Added "pick" list virtual method which picks a random value. [% ["a".."z"].pick(8).join %] Added "rand" text virtual method which gives a random number between 0 and the item. [% 20.rand %] Added "0" text virtual method which returns the item itself. This blurs the line between list and text items. [% a = "20" %][% a.0 IF a.size %] Added "int" text virtual method which returns the integer portion of a value. [% "2.3343".int %] Whitespace is less meaningful. [% 2-1 %] # = 1 (fails in TT2) Added pow operator. [% 2 ** 3 %] [% 2 pow 3 %] # = 8 8 Added self modifiers (+=, -=, *=, /=, %=, **=, ~=). [% a = 2; a *= 3 ; a %] # = 6 [% a = 2; (a *= 3) ; a %] # = 66 Added pre and post increment and decrement (++ --). [% ++a ; ++a %] # = 12 [% a-- ; a-- %] # = 0-1 Added qw// contructor. [% a = qw(a b c); a.1 %] # = b [% qw/a b c/.2 %] # = c Added regex contructor. [% "FOO".match(/(foo)/i).0 %] # = FOO [% a = /(foo)/i; "FOO".match(a).0 %] # = FOO Allow for scientific notation. (TT3) [% a = 1.2e-20 %] [% 123.fmt('%.3e') %] # = 1.230e+02 Allow for hexidecimal input. [% a = 0xff0000 %][% a %] # = 16711680 [% a = 0xff2 / 0xd; a.fmt('%x') %] # = 13a Post operative directives can be nested. Andy Wardley calls this side-by-side effect notation. [% one IF two IF three %] same as [% IF three %][% IF two %][% one %][% END %][% END %] [% a = [[1..3], [5..7]] %][% i FOREACH i = j FOREACH j = a %] # = 123567 Semi-colons on directives in the same tag are optional. [% SET a = 1 GET a %] [% FOREACH i = [1 .. 10] i END %] Note: a semi-colon is still required in front of any block directive that can be used as a post-operative directive. [% 1 IF 0 2 %] # prints 2 [% 1; IF 0 2 END %] # prints 1 Added a DUMP directive. Used for Data::Dumpering the passed variable or expression. [% DUMP a.a %] # dumps contents of a.a [% DUMP %] # dumps entire stash The Dumping is configurable via a DUMP configuration item. Added CONFIG directive. [% CONFIG ANYCASE => 1 PRE_CHOMP => '-' %] There is better line information When debug dirs is on, directives on different lines separated by colons show the line they are on rather than a general line range. Parse errors actually know what line and character they occured at and tell you about it. USING Template::Parser::CET There are several ways to get TT to use Template::Parser::CET. Pass in object during configuration. use Template; use Template::Parser::CET; my $t = Template->new( PARSER => Template::Parser::CET->new(\%config), ); Override the current program (option 1). use Template::Parser::CET activate => 1; Override the current program (option 2). use Template::Parser::CET; Template::Parser::CET->activate; You can then deactivate if youy want to use the normal parser by using: Template::Parser::CET->deactivate; Override the current program (option 3). use Template::Parser::CET; use Template::Config; local $Template::Config::PARSER = 'Template:Parser::CET'; Override all default instances. Modify the $PARSER value in Template/Config.pm to be 'Template::Parser::CET' rather than 'Template::Parser'. DOCUMENTATION Template::Toolkit and CGI::Ex::Template already cover everything that would be covered here. If you are running Template::Parser::CET then you already have both Template::Toolkit and CGI::Ex::Template installed. Please refer to their documentation for complete configuration and syntax examples. For any of the items in the FEATURES section you will need to refer to the CGI::Ex::Template documentation. BUGS / TODO * Template::Parser::CET is as non-invasive as it can be. It does no modification to the existing TT2 install. In order to provide features such as inline filters, self modifying operators, pre and post decrement and increment, and CONFIG and DUMP directive support, the abstraction to Template::Directive was broken. This means that projects such as Jemplate can't use these extended features directly (but projects such as Jemplate could write faster smaller templates if they used CGI::Ex::Template's compiled AST directly). * Cleanup compiled document output. * Add more line numbers to the compiled output. * Actually add the VObjects to the compile phase to get the compile time speed benefit. * Override filter generation code to allow for fall back to the SCALAR_OPS methods if a filter can't be found by the passed name. TT2 SYNTAX THAT WILL BREAK Pipe (FILTER alias) operators in ambiguous places. Under TT2 the following line: [% BLOCK a %]b is [% b %][% END %][% PROCESS a b => 234 | repeat(2) %] Would print: b is 234b is 234 Under CET and TT3 that line will print b is 234234 This is because the "|" has been used to allow for filter operations to be used inline on variables and also to call vmethods. The configuration option V2PIPE can be used to restore the old behavior. When V2PIPE is set to true (default is false), then CET will parse the block the same as TT2. When false it will parse the same as CET or TT3. You can use the CONFIG directive to set the option around some chunks of code that use the old syntax. [% CONFIG V2PIPE 1 -%] [% BLOCK a %]b is [% b %][% END %][% PROCESS a b => 234 | repeat(2) %] [%- CONFIG V2PIPE 0 %] Would print b is 234b is 234 Inline comments that end with the tag and not a newline. Because of the way the TT2 engine matches tags, the following works in TT2: [% a # GET THE value of a %] Because CET is recursive in nature, the closing tag has not been matched by the time the comment is removed. You will get a parse error saying not sure how to handle the tag. Simply change the previous example to the following: [% a # GET THE value of a %] All other commenting constructs parse just fine. The qw variable parse error If your template had a variable named qw - there will most likely be a parse error. In TT2 there was no qw() construct but there is in CET and TT3. [% a = qw %] Works fine in TT2 but is a parse error in TT3 [% a = qw(Foo Bar) %] Works fine in TT3 but is a parse error in TT2 TT2 TESTS THAT FAIL The following is a list of tests that will fail as of the TT2.19 test suite. All of the failed tests are caused by behavior that will be obsoleted by TT3. t/compile3.t - Fails 1 test Both CET and TT2 return the same error - but the error isn't formatted the same. t/debug.t - Fails 1 test CET debugs INTERPOLATED GETS - TT2 doesn't. There is an INTERPOLATED value that TT2 doesn't debug. t/fileline.t - Fails 4 tests CET is warn clean - even when performing numeric operations on non-numeric data - TT2 isn't and is testing for warnings. t/filter.t - Fails 1 test CET parses { 1 2 3 } as a hashref just fine - TT2 doesn't and expects an error. t/vars.t - Fails 8 tests (4 really, but parsing is failing) TT2 is allowing inline comments with closing tag on the same line. CET is recursive, the closing tag isn't matched before the closing tag - changing the closing tag to be on a separate line fixes the issue. AUTHOR Paul Seamons LICENSE This module may be distributed under the same terms as Perl itself.