NAME
    Blatte - text macro/markup/template language

SYNOPSIS
      use Blatte;
      use Blatte::Builtins ':all';

      my $perl = &Blatte::Parse('...some Blatte program ...');
      if (defined($perl)) {
        my $result = eval $perl;
        if (defined($result)) {
          &Blatte::traverse($result, \&callback);
        } elsif ($@) {
          ...handle execution error...
        }
      } else {
        ...handle parsing error...
      }

DESCRIPTION
    Blatte is a very powerful text markup and transformation
    language with a very simple syntax. A Blatte document can be
    translated into a Perl program that, when executed, produces a
    transformed version of the input document.

    This module itself contains some utility functions for handling
    Blatte documents, described below in the FUNCTIONS section.
    However, writers of Blatte-based software will generally be more
    interested in other associated modules. See in particular the
    Blatte::Compiler manpage (for processing files full of Blatte
    code) and the Blatte::Builtins manpage (for a description of the
    Blatte language's intrinsic functions).

    Most casual end users will probably be interested in Blatte's
    ability to serve as a higher-level language for writing web
    pages. This requires the additional CPAN package Blatte::HTML.

    Most of the remainder of this document describes the syntax and
    semantics of the Blatte language.

THE BLATTE LANGUAGE
    Blatte has three metacharacters: \ { }. These are used to
    represent lists, variables, syntactic forms, function calls,
    string literals, comments, and a "forget-whitespace" operator
    explained below.

    Everything else in a Blatte document is either whitespace or is
    divided into "words," also explained below.

    To include a literal metacharacter, precede it with backslash:
    \\, \{, and \}.

    All Blatte expressions correspond to equivalent Perl
    expressions. When a Blatte document is parsed, its Blatte
    expressions are converted to Perl, which may then be evaluated.

    Here is a quick rundown of Blatte expression types.

    \VAR
        This is a variable reference. The identifier following the \
        must begin with a letter, and may be followed by letters,
        digits, or underscore.

        This corresponds to the Perl scalar variable $var. All
        values in Blatte are Perl scalars.

    {\define \VAR EXPR}
        This defines a new variable \VAR to contain the value of the
        Blatte expression EXPR.

        This corresponds to the Perl sequence

          use vars '$var';
          $var = EXPR;

        (where EXPR has been transformed to its Perl equivalent).

    {\set! \VAR EXPR}
        This sets the existing variable \VAR to the value of the
        Blatte expression EXPR.

        This corresponds to the Perl expression

          $VAR = EXPR

    {\if TEST THEN ELSE1 ELSE2 ...}
        This evaluates TEST. If the result is true, it evaluates and
        returns THEN; otherwise it evaluates the ELSEs and returns
        the value of the last one.

        All Blatte values are true except for 0 and the empty string
        (as in Perl) and the empty Blatte list, {} (which
        corresponds to the Perl array reference []).

        Perl equivalent:

          if (TEST) {
            THEN;
          } else {
            ELSE1;
            ELSE2;
            ...
          }

    {\and EXPR1 EXPR2 ...}
        Evaluates each EXPR in turn, stopping if one yields a false
        value. Returns the value of the last EXPR it evaluates.

        Perl equivalent:

          EXPR1 && EXPR2 && ...

    {\or EXPR1 EXPR2 ...}
        Evaluates each EXPR in turn, stopping if one yields a true
        value. Returns the true value if there is one, falsehood
        otherwise.

        Perl equivalent:

          EXPR1 || EXPR2 || ...

    {\cond {TEST1 THEN1a THEN1b ...} {TEST2 THEN2a THEN2b ...} ...}
        Evaluates each TEST in turn. If one yields a true value,
        evaluates the corresponding THENs in sequence and returns
        the value of the last one.

        Perl equivalent:

          if (TEST1) {
            THEN1a;
            THEN1b;
            ...
          } elsif (TEST2) {
            THEN2a;
            THEN2b;
            ...
          } ...

    {\while TEST EXPR1 EXPR2 ...}
        Evaluates TEST. If it's true, evaluates EXPR1 through EXPRn
        and starts again.

        Perl correspondence:

          while (TEST) {
            EXPR1;
            EXPR2;
            ...
          }

    {\lambda {PARAM1 PARAM2 ...} EXPR1 EXPR2 ...}
        Creates an anonymous subroutine. When the subroutine is
        invoked, its arguments will be assigned to the parameters
        PARAM1, PARAM2, etc., and the EXPRs evaluated in the
        resulting context. The value will be the value of the last
        EXPR.

        Each PARAM is one of the following:

    \VAR    An ordinary variable reference. This creates a *positional
            parameter*.

    \=VAR   This is a *named parameter*.

    \&VAR   This is a *rest parameter*. There may be at most one of
            these.

        See below for how to invoke Blatte subroutines, and how
        argument parsing proceeds.

        Perl equivalent:

          sub {
            ...(argument parsing)...
            EXPR1;
            EXPR2;
            ...
          }

    {\define {\NAME PARAM1 PARAM2 ...} EXPR1 EXPR2 ...}
        Defines a subroutine and assigns it to the variable \NAME.
        This is shorthand for

          {\define \NAME {\lambda {PARAM1 PARAM2 ...} EXPR1 EXPR2 ...}}

    {\let {{\VAR1 VAL1} {\VAR2 VAL2} ...} EXPR1 EXPR2 ...}
        Evaluates the VALs, then assigns them to the vars, then
        evaluates the EXPRs in the resulting context, returning the
        value of the last one.

        Perl equivalence:

          {
            my($VAR1, $VAR2, ...) = (VAL1, VAL2, ...);

            EXPR1;
            EXPR2;
            ...
          }

    {\let* {{\VAR1 VAL1} {\VAR2 VAL2} ...} EXPR1 EXPR2 ...}
        Like \let, but each VAL is assigned to its \VAR as soon as
        it's computed, so later VALs can refer to earlier \VARs.

        Perl equivalence:

          {
            my $VAR1 = VAL1;
            my $VAR2 = VAL2;
            ...

            EXPR1;
            EXPR2;
            ...
          }

    {\letrec {{\VAR1 VAL1} {\VAR2 VAL2} ...} EXPR1 EXPR2 ...}
        Like \let and \let*, but evaluates all the VALs in a context
        where the VARs have been declared but not yet assigned. This
        allows them to refer to one another by reference.

        Perl equivalence:

          {
            my($VAR1, $VAR2, ...);

            ($VAR1, $VAR2, ...) = (VAL1, VAL2, ...);

            EXPR1;
            EXPR2;
            ...
          }

    {EXPR1 EXPR2 ... EXPRn}
        A sequence of one or more expressions enclosed in curly
        braces that isn't one of the expression types listed above
        is either a function call or a plain list.

        If the value of EXPR1 is a Blatte subroutine, it is a
        function call, and the values of EXPR2 through EXPRn are
        passed as arguments. In this case it corresponds to the
        following Perl:

          &{EXPR1}({ ...named parameter hash... }, ...other arguments...)

        See below for an explanation of Blatte subroutine argument
        parsing.

        If EXPR1 isn't a Blatte subroutine, then

          {EXPR1 EXPR2 ... EXPRn}

        is a plain list whose elements are the values of EXPR1
        through EXPRn. Blatte lists correspond to Perl array
        references:

          [EXPR1, EXPR2, ..., EXPRn]

    \"...\"
        This is a Blatte string. Strings may contain any characters,
        including whitespace. \ must be escaped like this: \\. Note,
        however, that " is *not* escaped.

    WORD
        Anything that contains no whitespace, isn't one of the above
        expression types, and escapes any instances of Blatte's
        three metacharacters, is a word.

    In addition to the expression types above, there are two
    additional pieces of Blatte syntax:

    \;  This introduces a comment, which continues to the end of the
        line.

    \/  This is the "forget-whitespace" operator. It cancels any
        whitespace immediately preceding it. See the section on
        Whitespace handling below.

  Argument parsing for function calls

    As mentioned above, Blatte subroutines (a.k.a. functions) have
    three kinds of parameters: positional, named, and rest.

    When calling a Blatte function, named parameters can be given
    values by writing

      \NAME=EXPR

    where \NAME is the parameter (which was given as \=NAME in the
    function definition).

    All remaining arguments in the function call are assigned to
    positional parameters in the order in which they were declared.
    If there aren't enough positional parameters, then all the
    remaining arguments are collected in a Blatte list and assigned
    to the rest parameter.

    When a Blatte function call is translated to Perl, the named
    parameter assignments are collected together in an anonymous
    HASH reference, which is passed as the first argument to the
    Perl subroutine corresponding to the Blatte function. (All
    Blatte functions are Perl subroutines that accept this anonymous
    HASH reference of named parameters as a first argument.) All
    remaining Blatte arguments are passed in sequence to the Perl
    subroutine.

    Inside the Perl subroutine, arguments are unpacked as follows:
    named parameters are extracted from the HASH reference and
    assigned to correspondingly named Perl variables; Perl variables
    for the positional parameters get the next N arguments from @_;
    and the remainder of @_ is turned into an ARRAY reference (a
    Blatte list) and assigned to the Perl variable denoting the rest
    parameter.

    Example:

      {\define {\fn \=n1 \=n2 \a \b \&r}
        ...do stuff...}

    becomes:

      $fn = sub {
        my($_named, $a, $b) = splice(@_, 0, 3);
        my $n1 = $_named->{n1};
        my $n2 = $_named->{n2};
        my $r = \@_;

        ...do stuff...
      };

    and the function call

      {\fn \n2=17 This is an example.}

    becomes:

      &$fn({n2 => 17}, 'This', 'is', 'an', 'example.')

    which means $n1 will be undef, $n2 will be 17, $a will be
    'This', $b will be 'is', and $r will be ['an', 'example'].

    (Almost. This example ignores Blatte's whitespace handling,
    which is explained below.)

  Whitespace handling

    Before recognizing a Blatte expression, the Blatte parser skips
    over a (possibly empty) sequence of whitespace. This whitespace
    is preserved; then, when the parser is finished, the result is
    wrapped in a Blatte::Ws object, or a "whitespace wrapper,"
    containing the preserved whitespace and the parsed expression.
    In this way, the whitespace preceding each expression is carried
    along with the expression itself. When Blatte expressions are
    parsed, evaluated, and rendered normally (see flatten() below),
    the output preserves the same whitespace found in the input.

FUNCTIONS
    Parse(INPUT)
        Parses the first Blatte expression in INPUT using the
        default parser. The result is converted to a string of Perl
        code and returned.

        (If you don't use the default parser, it's possible to
        change Blatte's syntax, to obtain the intermediate parse
        tree before conversion to Perl, and more. See the
        Blatte::Parser(3) manpage.)

        INPUT is either a string containing Blatte code, or a
        reference to such a string. If it's a reference, then after
        a successful parse, the matched expression will be removed
        from the beginning of the referenced string.

    traverse(OBJ, CALLBACK, [WS])
        Walks the data structure OBJ. If OBJ is an ARRAY reference,
        its elements are recursively traversed. If OBJ is a
        whitespace wrapper, its contents are recursively traversed.
        Otherwise, CALLBACK (a Perl CODE reference) is invoked on WS
        and OBJ. WS is a string of whitespace, possibly empty.

        CALLBACK should return truth if it uses its whitespace
        argument and falsehood if it doesn't (meaning that
        traverse() should reuse the whitespace value in subsequent
        calls, if necessary). "Uses" usually means that the
        whitespace is copied to some sort of output.

        This function supplies the logic that causes outer
        whitespace wrappers to take precedence over inner ones.

    flatten(OBJ, [WS])
        Renders OBJ as a string, with optional leading whitespace WS
        (which takes precedence over any outermost whitespace
        wrapper that might be present).

        This function is written in terms of traverse().

    wrapws(WS, OBJ)
        Creates a new whitespace wrapper whose whitespace string is
        WS and whose nested object is OBJ.

    unwrapws(OBJ)
        Unwraps as many layers of whitespace wrapper from OBJ as
        necessary to reach a non-wrapper object, then returns that.

    wsof(OBJ)
        This returns the whitespace string, possibly empty,
        associated with OBJ. Note that

          wsof(wrapws(X, Y))

        is always X, however

          unwrapws(wrapws(X, Y))

        is Y only when Y is not itself a whitespace wrapper.

    true(OBJ)
        Returns the Blatte truth value of OBJ (after whitespace-
        unwrapping, see above). As in Perl, 0 and the empty string
        are false, but so is an empty Blatte list (that is, a Perl
        ARRAY ref with zero elements); all other values are true.

    quote(STRING)
        Quote STRING using Blatte syntax so that it can be read back
        as a Blatte string literal.

PEDIGREE
    Blatte is a successor to an earlier language called Latte. The B
    is for "better" -- better because, whereas Latte's runtime
    facilities were fairly limited, Blatte's are effectively
    unlimited, since Blatte has the full power of Perl at its
    disposal. The implementation is also much faster and simpler
    than that of Latte. Latte users should beware that, despite
    being substantially similar, the Blatte language has
    siginificant differences from Latte.

    The design of the Blatte language was strongly influenced by
    Scheme, and was guided by these principles:

    *   The syntax should be simple, terse, and unobtrusive.

    *   There should be as few metacharacters as possible (contrast
        TeX), so as to have minimum impact on mostly textual Blatte
        documents.

    *   The language should be fully general, not a half-hearted macro
        system that makes many computations impossible.

    *   The correspondence between the source language (Blatte) and the
        target language (Perl) should be simple and direct.

DEDICATION
    Blatte is dedicated to the memory of Julie Epelboim, 1964-2001.
    Not that she ever would have used it. This was a woman who
    preferred to write *raw Postscript code* rather than use a page
    layout program. The tragedy of having lost her is dwarfed by the
    good fortune of having known her.

SEE ALSO
    the Blatte::Compiler(3) manpage, the Blatte::Builtins(3)
    manpage, the Blatte::Parser(3) manpage, the Blatte::Ws(3)
    manpage.