pbook.py: convert endlines to Unix format

[clinton/parenscript.git] / reference.lisp

;;;# ParenScript Language Reference

;;; This chapters describes the core constructs of ParenScript, as
;;; well as its compilation model. This chapter is aimed to be a
;;; comprehensive reference for ParenScript developers. Programmers
;;; looking for how to tweak the ParenScript compiler itself should
;;; turn to the ParenScript Internals chapter.

;;;# Statements and Expressions
;;;t \index{statement}
;;;t \index{expression}

;;; In contrast to Lisp, where everything is an expression, JavaScript
;;; makes the difference between an expression, which evaluates to a
;;; value, and a statement, which has no value. Examples for
;;; JavaScript statements are `for', `with' and `while'. Most
;;; ParenScript forms are expression, but certain special forms are
;;; not (the forms which are transformed to a JavaScript
;;; statement). All ParenScript expressions are statements
;;; though. Certain forms, like `IF' and `PROGN', generate different
;;; JavaScript constructs whether they are used in an expression
;;; context or a statement context. For example:

(+ i (if 1 2 3)) => i + (1 ? 2 : 3)

(if 1 2 3)
   => if (1) {
        2;
      } else {
        3;
      }

;;;# Symbol conversion
;;;t \index{symbol}
;;;t \index{symbol conversion}

;;; Lisp symbols are converted to JavaScript symbols by following a
;;; few simple rules. Special characters `!', `?', `#', `$', `@', `%',
;;; '/', `*' and `+' get replaced by their written-out equivalents
;;; "bang", "what", "hash", "dollar", "at", "percent", "slash",
;;; "start" and "plus" respectively. 

!?#$@% => bangwhathashdollaratpercent

;;; The `-' is an indication that the following character should be
;;; converted to uppercase. Thus, `-' separated symbols are converted
;;; to camelcase. The `_' character however is left untouched.

bla-foo-bar => blaFooBar

;;; If you want a JavaScript symbol beginning with an uppercase, you
;;; can either use a leading `-', which can be misleading in a
;;; mathematical context, or a leading `*'.

*array => Array

;;; The `.' character is left as is in symbols. This allows the
;;; ParenScript programmer to use a practical shortcut when accessing
;;; slots or methods of JavaScript objects. Instead of writing

(slot-value foobar 'slot)

;;; we can write

foobar.slot

;;; A symbol beggining and ending with `+' or `*' is converted to all
;;; uppercase, to signify that this is a constant or a global
;;; variable.

*global-array*        => GLOBALARRAY

*global-array*.length => GLOBALARRAY.length

;;;## Reserved Keywords
;;;t \index{keyword}
;;;t \index{reserved keywords}

;;; The following keywords and symbols are reserved in ParenScript,
;;; and should not be used as variable names.

! ~ ++ -- * / % + - << >> >>> < > <= >= == != ==== !== & ^ | && ||
*= /= %= += -= <<= >>= >>>= &= ^= |= 1- 1+
ABSTRACT AND AREF ARRAY BOOLEAN BREAK BYTE CASE CATCH CC-IF CHAR CLASS
COMMA CONST CONTINUE CREATE DEBUGGER DECF DEFAULT DEFUN DEFVAR DELETE
DO DOEACH DOLIST DOTIMES DOUBLE ELSE ENUM EQL EXPORT EXTENDS FALSE
FINAL FINALLY FLOAT FLOOR FOR FUNCTION GOTO IF IMPLEMENTS IMPORT IN INCF
INSTANCEOF INT INTERFACE JS LAMBDA LET LISP LIST LONG MAKE-ARRAY NATIVE NEW
NIL NOT OR PACKAGE PRIVATE PROGN PROTECTED PUBLIC RANDOM REGEX RETURN
SETF SHORT SLOT-VALUE STATIC SUPER SWITCH SYMBOL-MACROLET SYNCHRONIZED T
THIS THROW THROWS TRANSIENT TRY TYPEOF UNDEFINED UNLESS VAR VOID VOLATILE
WHEN WHILE WITH WITH-SLOTS

;;;# Literal values
;;;t \index{literal value}

;;;## Number literals
;;;t \index{number}
;;;t \index{number literal}

; number ::= a Lisp number

;;;
;;; ParenScript supports the standard JavaScript literal
;;; values. Numbers are compiled into JavaScript numbers. 

1       => 1

123.123 => 123.123

;;; Note that the base is not conserved between Lisp and JavaScript.

#x10    => 16

;;;## String literals
;;;t \index{string}
;;;t \index{string literal}

; string ::= a Lisp string

;;; Lisp strings are converted into JavaScript literals.

"foobar"      => 'foobar'

"bratzel bub" => 'bratzel bub'

;;; Escapes in Lisp are not converted to JavaScript escapes. However,
;;; to avoid having to use double backslashes when constructing a
;;; string, you can use the CL-INTERPOL library by Edi Weitz.

;;;## Array literals
;;;t \index{array}
;;;t \index{ARRAY}
;;;t \index{MAKE-ARRAY}
;;;t \index{AREF}
;;;t \index{array literal}

; (ARRAY {values}*)
; (MAKE-ARRAY {values}*)
; (AREF array index)
; 
; values ::= a ParenScript expression
; array  ::= a ParenScript expression
; index  ::= a ParenScript expression

;;; Array literals can be created using the `ARRAY' form.

(array)       => [  ]

(array 1 2 3) => [ 1, 2, 3 ]

(array (array 2 3)
       (array "foobar" "bratzel bub"))
   => [ [ 2, 3 ], [ 'foobar', 'bratzel bub' ] ]

;;; Arrays can also be created with a call to the `Array' function
;;; using the `MAKE-ARRAY'. The two forms have the exact same semantic
;;; on the JavaScript side.

(make-array)       => new Array()

(make-array 1 2 3) => new Array(1, 2, 3)

(make-array
 (make-array 2 3)
 (make-array "foobar" "bratzel bub"))
  => new Array(new Array(2, 3), new Array('foobar', 'bratzel bub'))

;;; Indexing arrays in ParenScript is done using the form `AREF'. Note
;;; that JavaScript knows of no such thing as an array. Subscripting
;;; an array is in fact reading a property from an object. So in a
;;; semantic sense, there is no real difference between `AREF' and
;;; `SLOT-VALUE'.

;;;## Object literals
;;;t \index{CREATE}
;;;t \index{SLOT-VALUE}
;;;t \index{WITH-SLOTS}
;;;t \index{object literal}
;;;t \index{object}
;;;t \index{object property}
;;;t \index{property}

; (CREATE {name value}*)
; (SLOT-VALUE object slot-name)
; (WITH-SLOTS ({slot-name}*) object body)
;
; name      ::= a ParenScript symbol or a Lisp keyword
; value     ::= a ParenScript expression
; object    ::= a ParenScript object expression
; slot-name ::= a quoted Lisp symbol
; body      ::= a list of ParenScript statements

;;;
;;; Object literals can be create using the `CREATE' form. Arguments
;;; to the `CREATE' form is a list of property names and values. To be
;;; more "lispy", the property names can be keywords.

(create :foo "bar" :blorg 1)
   => { foo : 'bar', 
        blorg : 1 }

(create :foo "hihi"
        :blorg (array 1 2 3)
        :another-object (create :schtrunz 1))
   => { foo : 'hihi', 
        blorg : [ 1, 2, 3 ], 
        anotherObject : { schtrunz : 1 } }

;;; Object properties can be accessed using the `SLOT-VALUE' form,
;;; which takes an object and a slot-name.

(slot-value an-object 'foo) => anObject.foo

;;; A programmer can also use the "." symbol notation explained above.

an-object.foo => anObject.foo

;;; The form `WITH-SLOTS' can be used to bind the given slot-name
;;; symbols to a macro that will expand into a `SLOT-VALUE' form at
;;; expansion time.

(with-slots (a b c) this
  (+ a b c))
    => this.a + this.b + this.c;

;;;## Regular Expression literals
;;;t \index{REGEX}
;;;t \index{regular expression}
;;;t \index{CL-INTERPOL}

; (REGEX regex)
;
; regex ::= a Lisp string

;;; Regular expressions can be created by using the `REGEX' form. The
;;; regex form actually does nothing at all to its argument, and
;;; prints it as is.

(regex "/foobar/i") => /foobar/i

;;; Here CL-INTERPOL proves really useful.

(regex #?r"/([^\s]+)foobar/i") => /([^\s]+)foobar/i

;;;## Literal symbols
;;;t \index{T}
;;;t \index{FALSE}
;;;t \index{NIL}
;;;t \index{UNDEFINED}
;;;t \index{THIS}
;;;t \index{literal symbols}
;;;t \index{null}
;;;t \index{true}

; T, FALSE, NIL, UNDEFINED, THIS

;;; The Lisp symbols `T' and `FALSE' are converted to their JavaScript
;;; boolean equivalents `true' and `false'.  

T     => true

FALSE => false

;;; The Lisp symbol `NIL' is converted to the JavaScript keyword
;;; `null'.

NIL => null

;;; The Lisp symbol `UNDEFINED' is converted to the JavaScript keyword
;;; `undefined'.

UNDEFINED => undefined

;;; The Lisp symbol `THIS' is converted to the JavaScript keyword
;;; `this'.

THIS => this

;;;# Variables
;;;t \index{variable}
;;;t \index{symbol}

; variable ::= a Lisp symbol

;;; All the other literal Lisp values that are not recognized as
;;; special forms or symbol macros are converted to JavaScript
;;; variables. This extreme freedom is actually quite useful, as it
;;; allows the ParenScript programmer to be flexible, as flexible as
;;; JavaScript itself.

variable    => variable

a-variable  => aVariable

*math       => Math

*math.floor => Math.floor

;;;# Function calls and method calls
;;;t \index{function}
;;;t \index{function call}
;;;t \index{method}
;;;t \index{method call}

; (function {argument}*)
; (method   object {argument}*)
;
; function ::= a ParenScript expression or a Lisp symbol
; method   ::= a Lisp symbol beginning with .
; object   ::= a ParenScript expression
; argument ::= a ParenScript expression

;;; Any list passed to the JavaScript that is not recognized as a
;;; macro or a special form (see "Macro Expansion" below) is
;;; interpreted as a function call. The function call is converted to
;;; the normal JavaScript function call representation, with the
;;; arguments given in paren after the function name.

(blorg 1 2) => blorg(1, 2)

(foobar (blorg 1 2) (blabla 3 4) (array 2 3 4))
   => foobar(blorg(1, 2), blabla(3, 4), [ 2, 3, 4 ])

((aref foo i) 1 2) => foo[i](1, 2)

;;; A method call is a function call where the function name is a
;;; symbol and begins with a "." . In a method call, the name of the
;;; function is append to its first argument, thus reflecting the
;;; method call syntax of JavaScript. Please note that most method
;;; calls can be abbreviated using the "." trick in symbol names (see
;;; "Symbol Conversion" above).

(.blorg this 1 2) => this.blorg(1, 2)

(this.blorg 1 2) => this.blorg(1, 2)

(.blorg (aref foobar 1) NIL T)
   => foobar[1].blorg(null, true)

;;;# Operator Expressions
;;;t \index{operator}
;;;t \index{operator expression}
;;;t \index{assignment operator}
;;;t \index{EQL}
;;;t \index{NOT}
;;;t \index{AND}
;;;t \index{OR}

; (operator {argument}*)
; (single-operator argument)
;
; operator ::= one of *, /, %, +, -, <<, >>, >>>, < >, EQL,
;              ==, !=, =, ===, !==, &, ^, |, &&, AND, ||, OR.
; single-operator ::= one of INCF, DECF, ++, --, NOT, !
; argument ::= a ParenScript expression

;;; Operator forms are similar to function call forms, but have an
;;; operator as function name. 
;;;
;;; Please note that `=' is converted to `==' in JavaScript. The `='
;;; ParenScript operator is not the assignment operator. Unlike
;;; JavaScript, ParenScript supports multiple arguments to the
;;; operators.

(* 1 2)   => 1 * 2

(= 1 2)   => 1 == 2

(eql 1 2) => 1 == 2

;;; Note that the resulting expression is correctly parenthized,
;;; according to the JavaScript operator precedence that can be found
;;; in table form at:

    http://www.codehouse.com/javascript/precedence/

(* 1 (+ 2 3 4) 4 (/ 6 7))
  => 1 * (2 + 3 + 4) * 4 * (6 / 7)

;;; The pre/post increment and decrement operators are also
;;; available. `INCF' and `DECF' are the pre-incrementing and
;;; pre-decrementing operators, and `++' and `--' are the
;;; post-decrementing version of the operators. These operators can
;;; take only one argument.

(++ i)   => i++

(-- i)   => i--

(incf i) => ++i

(decf i) => --i

;;; The `1+' and `1-' operators are shortforms for adding and
;;; substracting 1.

(1- i) => i - 1

(1+ i) => i + 1

;;; The `not' operator actually optimizes the code a bit. If `not' is
;;; used on another boolean-returning operator, the operator is
;;; reversed.

(not (< i 2))   => i >= 2

(not (eql i 2)) => i != 2

;;;# Body forms
;;;t \index{body form}
;;;t \index{PROGN}
;;;t \index{body statement}

; (PROGN {statement}*) in statement context
; (PROGN {expression}*) in expression context
;
; statement  ::= a ParenScript statement
; expression ::= a ParenScript expression

;;; The `PROGN' special form defines a sequence of statements when
;;; used in a statement context, or sequence of expression when used
;;; in an expression context. The `PROGN' special form is added
;;; implicitly around the branches of conditional executions forms,
;;; function declarations and iteration constructs.

;;; For example, in a statement context:

(progn (blorg i) (blafoo i))
   => blorg(i);
      blafoo(i);

;;; In an expression context:

(+ i (progn (blorg i) (blafoo i)))
   => i + (blorg(i), blafoo(i))

;;; A `PROGN' form doesn't lead to additional indentation or
;;; additional braces around it's body.

;;;# Function Definition
;;;t \index{function}
;;;t \index{method}
;;;t \index{function definition}
;;;t \index{DEFUN}
;;;t \index{LAMBDA}
;;;t \index{closure}
;;;t \index{anonymous function}

; (DEFUN name ({argument}*) body)
; (LAMBDA ({argument}*) body)
;
; name     ::= a Lisp Symbol
; argument ::= a Lisp symbol
; body     ::= a list of ParenScript statements

;;; As in Lisp, functions are defined using the `DEFUN' form, which
;;; takes a name, a list of arguments, and a function body. An
;;; implicit `PROGN' is added around the body statements.

(defun a-function (a b)
  (return (+ a b)))
  => function aFunction(a, b) {
       return a + b;
     }

;;; Anonymous functions can be created using the `LAMBDA' form, which
;;; is the same as `DEFUN', but without function name. In fact,
;;; `LAMBDA' creates a `DEFUN' with an empty function name.

(lambda (a b) (return (+ a b)))
  => function (a, b) {
       return a + b;
     }

;;;# Assignment
;;;t \index{assignment}
;;;t \index{SETF}
;;;t \index{assignment operator}

; (SETF {lhs rhs}*)
;
; lhs ::= a ParenScript left hand side expression
; rhs ::= a ParenScript expression

;;; Assignment is done using the `SETF' form, which is transformed
;;; into a series of assignments using the JavaScript `=' operator.

(setf a 1) => a = 1

(setf a 2 b 3 c 4 x (+ a b c))
   => a = 2;
      b = 3;
      c = 4;
      x = a + b + c;

;;; The `SETF' form can transform assignments of a variable with an
;;; operator expression using this variable into a more "efficient"
;;; assignment operator form. For example:

(setf a (1+ a))          => a++

(setf a (* 2 3 4 a 4 a)) => a *= 2 * 3 * 4 * 4 * a

(setf a (- 1 a))         => a = 1 - a

;;;# Single argument statements
;;;t \index{single-argument statement}
;;;t \index{RETURN}
;;;t \index{THROW}
;;;t \index{THROW}
;;;t \index{function}

; (RETURN {value}?)
; (THROW  {value}?)
;
; value ::= a ParenScript expression

;;; The single argument statements `return' and `throw' are generated
;;; by the form `RETURN' and `THROW'. `THROW' has to be used inside a
;;; `TRY' form. `RETURN' is used to return a value from a function
;;; call.

(return 1)       => return 1

(throw "foobar") => throw 'foobar'

;;;# Single argument expression
;;;t \index{single-argument expression}
;;;t \index{object creation}
;;;t \index{object deletion}
;;;t \index{DELETE}
;;;t \index{VOID}
;;;t \index{TYPEOF}
;;;t \index{INSTANCEOF}
;;;t \index{NEW}
;;;t \index{new}

; (DELETE     {value})
; (VOID       {value})
; (TYPEOF     {value})
; (INSTANCEOF {value})
; (NEW        {value})
;
; value ::= a ParenScript expression

;;; The single argument expressions `delete', `void', `typeof',
;;; `instanceof' and `new' are generated by the forms `DELETE',
;;; `VOID', `TYPEOF', `INSTANCEOF' and `NEW'. They all take a
;;; ParenScript expression.

(delete (new (*foobar 2 3 4))) => delete new Foobar(2, 3, 4)

(if (= (typeof blorg) *string)
    (alert (+ "blorg is a string: " blorg))
    (alert "blorg is not a string"))
  => if (typeof blorg == String) {
       alert('blorg is a string: ' + blorg);
     } else {
       alert('blorg is not a string');
     }

;;;# Conditional Statements
;;;t \index{conditional statements}
;;;t \index{IF}
;;;t \index{WHEN}
;;;t \index{UNLESS}
;;;t \index{conditionals}

; (IF conditional then {else})
; (WHEN condition then)
; (UNLESS condition then)
;
; condition ::= a ParenScript expression
; then      ::= a ParenScript statement in statement context, a
;                 ParenScript expression in expression context
; else      ::= a ParenScript statement in statement context, a
;                 ParenScript expression in expression context

;;; The `IF' form compiles to the `if' javascript construct. An
;;; explicit `PROGN' around the then branch and the else branch is
;;; needed if they consist of more than one statement. When the `IF'
;;; form is used in an expression context, a JavaScript `?', `:'
;;; operator form is generated.

(if (blorg.is-correct)
    (progn (carry-on) (return i))
    (alert "blorg is not correct!"))
   => if (blorg.isCorrect()) {
        carryOn();
        return i;
      } else {
        alert('blorg is not correct!');
      }

(+ i (if (blorg.add-one) 1 2))
   => i + (blorg.addOne() ? 1 : 2)

;;; The `WHEN' and `UNLESS' forms can be used as shortcuts for the
;;; `IF' form.

(when (blorg.is-correct)
  (carry-on)
  (return i))
    => if (blorg.isCorrect()) {
         carryOn();
         return i;
       }

(unless (blorg.is-correct)
  (alert "blorg is not correct!"))
    => if (!blorg.isCorrect()) {
         alert('blorg is not correct!');
       }

;;;# Variable declaration
;;;t \index{variable}
;;;t \index{variable declaration}
;;;t \index{binding}
;;;t \index{scoping}
;;;t \index{DEFVAR}
;;;t \index{LET}

; (DEFVAR var {value}?)
; (LET ({var | (var value)) body)
;
; var   ::= a Lisp symbol
; value ::= a ParenScript expression
; body  ::= a list of ParenScript statements

;;; Variables (either local or global) can be declared using the
;;; `DEFVAR' form, which is similar to its equivalent form in
;;; Lisp. The `DEFVAR' is converted to "var ... = ..." form in
;;; JavaScript.

(defvar *a* (array 1 2 3)) => var A = [ 1, 2, 3 ];

(if (= i 1)
    (progn (defvar blorg "hallo")
           (alert blorg))
    (progn (defvar blorg "blitzel")
           (alert blorg)))
   => if (i == 1) {
        var blorg = 'hallo';
        alert(blorg);
      } else {
        var blorg = 'blitzel';
        alert(blorg);
      }

;;; A more lispy way to declare local variable is to use the `LET'
;;; form, which is similar to its Lisp form.

(if (= i 1)
    (let ((blorg "hallo"))
      (alert blorg))
    (let ((blorg "blitzel"))
      (alert blorg)))
   => if (i == 1) {
        var blorg = 'hallo';
        alert(blorg);
      } else {
        var blorg = 'blitzel';
        alert(blorg);
      }

;;; However, beware that scoping in Lisp and JavaScript are quite
;;; different. For example, don't rely on closures capturing local
;;; variables in the way you'd think they would.

;;;# Iteration constructs
;;;t \index{iteration}
;;;t \index{iteration construct}
;;;t \index{loop}
;;;t \index{array traversal}
;;;t \index{property}
;;;t \index{object property}
;;;t \index{DO}
;;;t \index{DOTIMES}
;;;t \index{DOLIST}
;;;t \index{DOEACH}
;;;t \index{WHILE}

; (DO ({var | (var {init}? {step}?)}*) (end-test) body)
; (DOTIMES (var numeric-form) body)
; (DOLIST (var list-form) body)
; (DOEACH (var object) body)
; (WHILE end-test body)
;
; var          ::= a Lisp symbol
; numeric-form ::= a ParenScript expression resulting in a number
; list-form    ::= a ParenScript expression resulting in an array
; object       ::= a ParenScript expression resulting in an object
; init         ::= a ParenScript expression
; step         ::= a ParenScript expression
; end-test     ::= a ParenScript expression
; body         ::= a list of ParenScript statements

;;; The `DO' form, which is similar to its Lisp form, is transformed
;;; into a JavaScript `for' statement. Note that the ParenScript `DO'
;;; form does not have a return value, that is because `for' is a
;;; statement and not an expression in JavaScript.

(do ((i 0 (1+ i))
     (l (aref blorg i) (aref blorg i)))
    ((or (= i blorg.length)
         (eql l "Fumitastic")))
  (document.write (+ "L is " l)))
   => for (var i = 0, l = blorg[i]; 
           !(i == blorg.length || l == 'Fumitastic');
           i = i + 1, l = blorg[i]) {
        document.write('L is ' + l);
      }

;;; The `DOTIMES' form, which lets a variable iterate from 0 upto an
;;; end value, is a shortcut for `DO'.

(dotimes (i blorg.length)
  (document.write (+ "L is " (aref blorg i))))
   => for (var i = 0; i < blorg.length; i = i + 1) {
        document.write('L is ' + blorg[i]);
      }

;;; The `DOLIST' form is a shortcut for iterating over an array. Note
;;; that this form creates temporary variables using a function called
;;; `JS-GENSYM', which is similar to its Lisp counterpart `GENSYM'.

(dolist (l blorg)
  (document.write (+ "L is " l)))
   => {
        var tmpArr1 = blorg;
        for (var tmpI2 = 0; tmpI2 < tmpArr1.length;
          tmpI2 = tmpI2 + 1) {
          var l = tmpArr1[tmpI2];
          document.write('L is ' + l);
        }
      }


;;; The `DOEACH' form is converted to a `for (var .. in ..)' form in
;;; JavaScript. It is used to iterate over the enumerable properties
;;; of an object.

(doeach (i object)
   (document.write (+ i " is " (aref object i))))
   => for (var i in object) {
        document.write(i + ' is ' + object[i]);
      }

;;; The `WHILE' form is transformed to the JavaScript form `while',
;;; and loops until a termination test evaluates to false.

(while (film.is-not-finished)
  (this.eat (new *popcorn)))
   => while (film.isNotFinished()) {
        this.eat(new Popcorn);
      }
      
;;;# The `CASE' statement
;;;t \index{CASE}
;;;t \index{switch}

; (CASE case-value clause*)
;
; clause     ::= (value body)
; case-value ::= a ParenScript expression
; value      ::= a ParenScript expression
; body       ::= a list of ParenScript statements

;;; The Lisp `CASE' form is transformed to a `switch' statement in
;;; JavaScript. Note that `CASE' is not an expression in
;;; ParenScript. The default case is not named `T' in ParenScript, but
;;; `DEFAULT' instead.

(case (aref blorg i)
  (1 (alert "one"))
  (2 (alert "two"))
  (default (alert "default clause")))
    => switch (blorg[i]) {
         case 1:   alert('one');
         case 2:   alert('two');
         default:   alert('default clause');
       }

;;;# The `WITH' statement
;;;t \index{WITH}
;;;t \index{dynamic scope}
;;;t \index{binding}
;;;t \index{scoping}
;;;t \index{closure}

; (WITH (object) body)
;
; object ::= a ParenScript expression evaluating to an object
; body   ::= a list of ParenScript statements

;;; The `WITH' form is compiled to a JavaScript `with' statements, and
;;; adds the object `object' as an intermediary scope objects when
;;; executing the body.

(with ((create :foo "foo" :i "i"))
  (alert (+ "i is now intermediary scoped: " i)))
   => with ({ foo : 'foo', 
              i : 'i' }) {
        alert('i is now intermediary scoped: ' + i);
      }

;;;# The `TRY' statement
;;;t \index{TRY}
;;;t \index{CATCH}
;;;t \index{FINALLY}
;;;t \index{exception}
;;;t \index{error handling}

; (TRY body {(:CATCH (var) body)}? {(:FINALLY body)}?)
;
; body ::= a list of ParenScript statements
; var  ::= a Lisp symbol

;;; The `TRY' form is converted to a JavaScript `try' statement, and
;;; can be used to catch expressions thrown by the `THROW'
;;; form. The body of the catch clause is invoked when an exception
;;; is catched, and the body of the finally is always invoked when
;;; leaving the body of the `TRY' form.

(try (throw "i") 
 (:catch (error)
   (alert (+ "an error happened: " error)))
 (:finally
   (alert "Leaving the try form")))
   => try {
        throw 'i';
      } catch (error) {
        alert('an error happened: ' + error);
      } finally {
        alert('Leaving the try form');
      }

;;;# The HTML Generator
;;;t \index{HTML}
;;;t \index{HTML generation}
;;;t \index{CSS}
;;;t \index{CSS generation}


; (HTML html-expression)

;;; The HTML generator of ParenScript is very similar to the HTML
;;; generator included in AllegroServe. It accepts the same input
;;; forms as the AllegroServer HTML generator. However, non-HTML
;;; construct are compiled to JavaScript by the ParenScript
;;; compiler. The resulting expression is a JavaScript expression.

(html ((:a :href "foobar") "blorg"))
  => '<a href=\"foobar\">blorg</a>'

(html ((:a :href (generate-a-link)) "blorg"))
  => '<a href=\"' + generateALink() + '\">blorg</a>'

;;; We can recursively call the JS compiler in a HTML expression.

(document.write
  (html ((:a :href "#"
            :onclick (js-inline (transport))) "link")))
  => document.write
     ('<a href=\"#\" onclick=\"' + 'javascript:transport();' + '\">link</a>')

; (CSS-INLINE css-expression)

;;; Stylesheets can also be created in ParenScript.

(css-inline :color "red"
            :font-size "x-small")
  => 'color:red;font-size:x-small'

(defun make-color-div(color-name)
    (return (html ((:div :style (css-inline :color color-name))
                   color-name " looks like this."))))
  => function makeColorDiv(colorName) {
       return '<div style=\"' + ('color:' + colorName) + '\">' + colorName
         + ' looks like this.</div>';
     }

;;;# Macrology
;;;t \index{macro}
;;;t \index{macrology}
;;;t \index{DEFJSMACRO}
;;;t \index{MACROLET}
;;;t \index{SYMBOL-MACROLET}
;;;t \index{JS-GENSYM}
;;;t \index{compiler}

; (DEFJSMACRO name lambda-list macro-body)
; (MACROLET ({name lambda-list macro-body}*) body)
; (SYMBOL-MACROLET ({name macro-body}*) body)
; (JS-GENSYM {string}?)
;
; name        ::= a Lisp symbol
; lambda-list ::= a lambda list
; macro-body  ::= a Lisp body evaluating to ParenScript code
; body        ::= a list of ParenScript statements
; string      ::= a string

;;; ParenScript can be extended using macros, just like Lisp can be
;;; extended using Lisp macros. Using the special Lisp form
;;; `DEFJSMACRO', the ParenScript language can be
;;; extended. `DEFJSMACRO' adds the new macro to the toplevel macro
;;; environment, which is always accessible during ParenScript
;;; compilation. For example, the `1+' and `1-' operators are
;;; implemented using macros.

(defjsmacro 1- (form)
  `(- ,form 1))

(defjsmacro 1+ (form)
  `(+ ,form 1))

;;; A more complicated ParenScript macro example is the implementation
;;; of the `DOLIST' form (note how `JS-GENSYM', the ParenScript of
;;; `GENSYM', is used to generate new ParenScript variable names):

(defjsmacro dolist (i-array &rest body)
  (let ((var (first i-array))
        (array (second i-array))
        (arrvar (js-gensym "arr"))
        (idx (js-gensym "i")))
    `(let ((,arrvar ,array))
      (do ((,idx 0 (++ ,idx)))
          ((>= ,idx (slot-value ,arrvar 'length)))
        (let ((,var (aref ,arrvar ,idx)))
          ,@body)))))

;;; Macros can be added dynamically to the macro environment by using
;;; the ParenScript `MACROLET' form (note that while `DEFJSMACRO' is a
;;; Lisp form, `MACROLET' and `SYMBOL-MACROLET' are ParenScript forms).

;;; ParenScript also supports symbol macros, which can be introduced
;;; using the ParenScript form `SYMBOL-MACROLET'. A new macro
;;; environment is created and added to the current macro environment
;;; list while compiling the body of the `SYMBOL-MACROLET' form. For
;;; example, the ParenScript `WITH-SLOTS' is implemented using symbol
;;; macros.

(defjsmacro with-slots (slots object &rest body)
  `(symbol-macrolet ,(mapcar #'(lambda (slot)
                                 `(,slot '(slot-value ,object ',slot)))
                             slots)
    ,@body))

;;;# The ParenScript Compiler
;;;t \index{compiler}
;;;t \index{ParenScript compiler}
;;;t \index{JS-COMPILE}
;;;t \index{JS-TO-STRINGS}
;;;t \index{JS-TO-STATEMENT-STRINGS}
;;;t \index{JS-TO-STRING}
;;;t \index{JS-TO-LINE}
;;;t \index{JS}
;;;t \index{JS-INLINE}
;;;t \index{JS-FILE}
;;;t \index{JS-SCRIPT}
;;;t \index{nested compilation}

; (JS-COMPILE expr)
; (JS-TO-STRINGS compiled-expr position)
; (JS-TO-STATEMENT-STRINGS compiled-expr position)
;
; compiled-expr ::= a compiled ParenScript expression
; position      ::= a column number
;
; (JS-TO-STRING expression)
; (JS-TO-LINE expression)
;
; expression ::= a Lisp list of ParenScript code
;
; (JS body)
; (JS-INLINE body)
; (JS-FILE body)
; (JS-SCRIPT body)
;
; body ::= a list of ParenScript statements

;;; The ParenScript compiler can be invoked from withing Lisp and from
;;; within ParenScript itself. The primary API function is
;;; `JS-COMPILE', which takes a list of ParenScript, and returns an
;;; internal object representing the compiled ParenScript. 

(js-compile '(foobar 1 2))
  => #<JS::FUNCTION-CALL {584AA5DD}>

;;; This internal object can be transformed to a string using the
;;; methods `JS-TO-STRINGS' and `JS-TO-STATEMENT-STRINGS', which
;;; interpret the ParenScript in expression and in statement context
;;; respectively. They take an additional parameter indicating the
;;; start-position on a line (please note that the indentation code is
;;; not perfect, and this string interface will likely be
;;; changed). They return a list of strings, where each string
;;; represents a new line of JavaScript code. They can be joined
;;; together to form a single string.

(js-to-strings (js-compile '(foobar 1 2)) 0)
  => ("foobar(1, 2)")

;;; As a shortcut, ParenScript provides the functions `JS-TO-STRING'
;;; and `JS-TO-LINE', which return the JavaScript string of the
;;; compiled expression passed as an argument.

(js-to-string '(foobar 1 2))
  => "foobar(1, 2)"

;;; For static ParenScript code, the macros `JS', `JS-INLINE',
;;; `JS-FILE' and `JS-SCRIPT' avoid the need to quote the ParenScript
;;; expression. All these forms add an implicit `PROGN' form around
;;; the body. `JS' returns a string of the compiled body, where the
;;; other expression return an expression that can be embedded in a
;;; HTML generation construct using the AllegroServe HTML
;;; generator. `JS-SCRIPT' generates a "SCRIPT" node, `JS-INLINE'
;;; generates a string to be used in node attributs, and `JS-FILE'
;;; prints the compiled ParenScript code to the HTML stream.

;;; These macros are also available inside ParenScript itself, and
;;; generate strings that can be used inside ParenScript code. Note
;;; that `JS-INLINE' in ParenScript is not the same `JS-INLINE' form
;;; as in Lisp, for example. The same goes for the other compilation
;;; macros.