This is latte.info, produced by makeinfo version 4.8 from latte.texi. INFO-DIR-SECTION Latte - the Language for Transforming Text START-INFO-DIR-ENTRY * latte: (latte). Writing Latte documents. END-INFO-DIR-ENTRY This is the user guide for Latte, the Language for Transforming Text. Copyright (C) 1998 Zanshin Inc. The contents of this file are subject to the Zanshin Public License Version 1.0 (the "License"); you may not use this file except in compliance with the License. You should have received a copy of the License with Latte; see the file COPYING. You may also obtain a copy of the License at . Documents distributed under the License are distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License for the specific language governing rights and limitations under the License. The Original Code is Latte. The Initial Developer of the Original Code is Zanshin, Inc. This product includes software developed by the University of California, Berkeley and its contributors. File: latte.info, Node: Top, Next: Introduction, Prev: (dir), Up: (dir) This is the user guide for Latte, the Language for Transforming Text. * Menu: * Introduction:: * Latte examples:: * Latte syntax:: * Latte builtins:: * Writing HTML with Latte:: * The latte-text translator:: * Future directions:: * Further information:: * Pedigree:: * Function and Variable Index:: File: latte.info, Node: Introduction, Next: Latte examples, Prev: Top, Up: Top 1 Introduction ************** Latte, the Language for Transforming Text, is a very simple and powerful language for including "markup" in text documents. Markup refers to information in the text that isn't strictly part of the text but describes it--for example, to specify that a certain word should appear in boldface, or that a certain phrase is to be rendered as a chapter heading. Many languages for text markup already exist. The best-known one is HTML (Hypertext Markup Language), the language of World-Wide Web documents. Other well-known markup languages are SGML, TeX, and troff. Why create a new markup language? Because the existing languages lack generality. HTML can't easily be used for anything except web documents. TeX can't easily be used for anything except printed books, articles, business letters, and the like. Latte, on the other hand, defines a very simple syntax that is suitable for every task requiring text markup. Latte also provides a software library that makes it easy for programmers to create "translators" from Latte into other languages. One such translator, `latte-html', comes with Latte and can convert Latte documents into HTML. Another, `latte-text', produces a plain text version of the very same source document. Many other Latte translators are planned. HTML authors generally find it easier to write in Latte than in HTML (*note Latte vs. HTML::). Latte documents can therefore be used for everything HTML is used for _and_ everything TeX is used for _and_ everything troff is used for, etc. In fact, a single Latte document can be run through different translators to produce HTML, TeX, and plain text versions. Latte is dedicated to my sister, the amazing Suzanne Glickstein. File: latte.info, Node: Latte examples, Next: Latte syntax, Prev: Introduction, Up: Top 2 Latte examples **************** Here are some examples illustrating the use of Latte to write World Wide Web documents. {\head {\title Our document}} {\body {\h1 Our document} Here is our first WWW document produced with {\a \href=http://www.latte.org/ Latte}. We can't wait to get started on our second document!} This short Latte document produces the following HTML when run through `latte-html -f': Our document

Our document

Here is our first WWW document produced with Latte.

We can't wait to get started on our second document! Observe that Latte uses curly braces (`{' and `}') to group text, and a backslash (`\') to introduce markup keywords. This example only uses Latte keywords that are exact analogs of HTML tags: `head', `title', `body', `h1', `a', and the attribute `href'. It's possible to write Latte documents that contain nothing but HTML analogs, but the real power of Latte is revealed when its own features are combined with those of the target language. For instance, note that the document title of the previous example, `Our document', appears twice in the Latte and HTML versions--once inside the `title' tag and once inside the `h1' tag. It shouldn't be necessary to write the document's title twice, though. It should suffice to write it once and by some mechanism have it appear in the two places it's needed. Also, writing it twice is an invitation to error: if the document title ever changes, it's possible to update it in one place and forget to update it in the other. Here's a version of the first example that solves that problem: {\def \doctitle {Our document}} {\head {\title \doctitle}} {\body {\h1 \doctitle} Here is our first WWW document produced with {\a \href=http://www.latte.org/ Latte}. We can't wait to get started on our second document!} This version begins by using the Latte procedure `\def' to define a new "variable" named `\doctitle'. The value of `\doctitle' is `Our document'. This value is substituted in the two places that `\doctitle' is used when this file is processed with `latte-html'. Now suppose the enthusiastic authors of this example decide that every WWW document they write will have the same text in the `title' section as in an `h1' section at the beginning of the document body. Rather than structure every Latte file along these lines: {\def \doctitle ...} {\head {\title \doctitle}} {\body {\h1 \doctitle} ...} they can define a "function" to do that work for them: {\def {\document \doctitle \&content} {{\head {\title \doctitle}} {\body {\h1 \doctitle}} \content}} and can now write each document in abbreviated fashion: {\document {Our second document} This was as exciting to write as our first!} File: latte.info, Node: Latte syntax, Next: Latte builtins, Prev: Latte examples, Up: Top 3 Latte syntax ************** When a Latte translator program such as `latte-html' reads a Latte file, it considers the contents as a sequence of Latte "expressions". An expression is one of the following things: * A "word"--that is, any sequence of characters containing no whitespace, no backslash, and no curly braces; * A "quoted string"--that is, any sequence of characters enclosed in a pair of `\" ... \"' delimiters. * A "variable reference"--that is, a backslash followed by the name of a variable. Variable names in Latte must begin with a letter or an underscore, and may be followed by zero or more letters, underscores, or digits, plus the characters `?', `!', `+', and `-'. * A "group", which is a pair of curly braces surrounding zero or more "subexpressions". * A "quoted expression"--that is, any Latte expression preceded by one of Latte's quote symbols: `\'', `\`', `\,', and `\,@'. These are all covered in more detail in the sections that follow. Other constructs that may appear in a Latte file are: * Whitespace, naturally. Latte handles whitespace in a special way; see *Note Whitespace::. * Comments. These begin with `\;' and continue to the end of the line. Latte discards comments when reading Latte files. * The "forget-whitespace" symbol, written `\/'. Wherever this appears, it has the effect of canceling all the whitespace Latte has read since the last expression. * Menu: * Words and strings:: * Variables:: * Groups:: * Quoted expressions:: * Whitespace:: * Latte vs. HTML:: File: latte.info, Node: Words and strings, Next: Variables, Prev: Latte syntax, Up: Latte syntax 3.1 Words and strings ===================== Most Latte files consist mainly of "words." Simply put, words are those sequences of characters that aren't categorized as any other kind of Latte construct. To be more precise, a Latte word is a sequence of characters that does not include whitespace, backslashes, or curly braces. However, any of those characters _can_ be included in a word by "escaping" them with a backslash. Here's an example: In a Latte "word," you must use \\ to escape \\, \{ and \}. This example contains these fourteen Latte words: 1. In 2. a 3. Latte 4. "word," 5. you 6. must 7. use 8. \ 9. to 10. escape 11. \, 12. { 13. and 14. }. In Latte, whitespace is not normally part of a word (it's handled specially; *note Whitespace::). Sometimes, however, it's desirable to force Latte to consider a single word to include some whitespace, especially when you wish to affix whitespace to the beginning or end of a word. This is best accomplished by using a quoted string, which Latte treats exactly like a word. A quoted string begins with `\"' and ends with `\"'. Inside a quoted string, only backslash must be escaped (with a backslash); every other character, including `"', `{', and `}', can be written normally. Quoted strings are therefore also useful for text that contains a lot of curly braces, such as fragments of C or Perl programs. Note that the only way to represent an empty word in Latte is with quoting: `\"\"'. File: latte.info, Node: Variables, Next: Groups, Prev: Words and strings, Up: Latte syntax 3.2 Variables ============= A Latte variable can hold a single Latte expression as its value. That value is substituted wherever the variable is used. The expression {\def \VARNAME EXPRESSION} creates a new variable named VARNAME whose value is EXPRESSION. The expression {\set! \VARNAME EXPRESSION} assigns the new value EXPRESSION to VARNAME, which must previously have been defined with `\def'. The expression {\let {{\VARNAME1 EXPRESSION1} {\VARNAME2 EXPRESSION2} ...} BODY ...} assigns EXPRESSION1 to VARNAME1, EXPRESSION2 to VARNAME2, and so on, then executes BODY with those variables in effect. Outside of the `\let' expression, those variables do not exist. * Menu: * Scope:: * Function parameters:: File: latte.info, Node: Scope, Next: Function parameters, Prev: Variables, Up: Variables 3.2.1 Scope ----------- It is possible to have a variable outside a `\let' with the same name as a variable inside a `\let'. Suppose the variable is `\foo'. Then the definition inside the `\let' "hides" or "shadows" the definition outside the `\let', like so: {\def \foo 7} {\let {{\foo 12}} The value of foo is \foo} Now the value of foo is \foo This example yields the following text: The value of foo is 12 Now the value of foo is 7 The region in which a variable is visible is called its "scope". Different computer languages have different "scoping rules". Latte uses "lexical scope", also sometimes called "static scope". This means that you can always tell by looking at the Latte file exactly which variable is meant by each variable reference--it's always the one defined in the textually innermost enclosing scope.(1) ---------- Footnotes ---------- (1) Under another popular set of scoping rules called "dynamic scope", the variable meant by a particular variable reference can actually change from moment to moment, depending on differences in the order in which functions are executing. File: latte.info, Node: Function parameters, Prev: Scope, Up: Variables 3.2.2 Function parameters ------------------------- The parameters of Latte functions (*note Groups::) are variables whose scope is the body of the function for which they're defined. There are three kinds of function parameter in Latte: "positional parameter" This is the simplest kind of parameter. In the definition of the function, a positional parameter is written exactly like an ordinary variable. When the function is invoked, the first actual argument is assigned to the first positional parameter, the second actual argument is assigned to the second positional parameter, and so on.(1) Here's an example of a function with two positional parameters: {\def {\function \param1 \param2} {My arguments are \param1 and \param2.}} {\function red green} => My arguments are red and green. If fewer arguments are given than there are positional parameters, then the excess parameters get the default value `{}' (i.e., the empty group). If too many arguments are given and there is no "rest parameter" (see below), the excess arguments are silently ignored. "named parameter" When a function is invoked, a named parameter only gets a value when the caller explicitly assigns it a value by name. In the parameter list of the function definition, a named parameter is indicated by `\=PARAM'. When the function is called, its named parameters are given values by writing `\PARAM=VALUE'. (There must be no whitespace between `\PARAM' and the `=' that follows it.) In the body of a function, named parameters are referenced like ordinary variables. Example: {\def {\function \=param} {\if \param {My param is \param}}} {\function \param={full of eels}} => My param is full of eels Any named parameter not given a value gets the value `{}' by default. Note that one difference between positional parameters and named parameters is that the caller of a function need not know the names of positional parameters in order to use them, but must know the names of named parameters in order to use those. When a function is invoked, any named parameter assignments do not affect matching up arguments to _positional_ parameters. Named parameter assignments may be given in any order, anywhere in the argument list, without changing the meaning of the function call. "rest parameter" Every function may have one rest parameter, which is denoted in the parameter list by `\&PARAM'. The value of this parameter is a Latte group containing excess arguments that could not be assigned to positional parameters. Example: {\def {\function \param1 \param2 \&rest} {The first two parameters are \param1 and \param2 The remaining parameters are \rest}} {\function a b c d e} => The first two parameters are a and b The remaining parameters are c d e Latte makes extensive use of rest parameters in functions that are HTML analogs, allowing Latte to represent an arbitrary amount of enclosed text easily (*note Balanced tag functions::). For example, here is a simplified definition of `\a' (with some essential details removed for clarity), which emits an HTML anchor: {\def {\a \=href \&rest} { \rest }} Now when `\a' is called as follows: {\a \href=http://www.latte.org/ The Language for Transforming Text} the five excess arguments--`The', `Language', `for', `Transforming', and `Text'--are grouped together in `\rest' and emitted between the `' and `' tags, yielding The Language for Transforming Text (One of the details removed for clarity is the precise whitespace arrangement in the output of this example, but we return to this example with greater precision in *Note Whitespace::.) ---------- Footnotes ---------- (1) What's the difference between a "parameter" and an "argument"? The terms are usually used interchangeably, but technically speaking, "parameter" refers to the variable in the function definition, while "argument" is the value that gets passed in when the function is called. The value of the argument is assigned to the parameter. File: latte.info, Node: Groups, Next: Quoted expressions, Prev: Variables, Up: Latte syntax 3.3 Groups ========== A group in Latte is an expression composed of a sequence of zero or more subexpressions surrounded by `{' and `}'. Groups serve two purposes. The first is to gather multiple expressions into a single expression, for use in contexts where only a single expression is allowed. For example, only a single expression may be passed as the value when defining a variable with `\def'. If the value happens to be a single word, grouping isn't needed: {\def \var Hello} but for more complex values, grouping is needed: {\def \var {Hello there}} The other use of groups in Latte is for performing "function calls". When the first subexpression of a group is a function (that is, when it is a variable reference whose _value_ is a function, or when it is some other Latte expression yielding a function), then that function is called, passing the remaining subexpressions to it as arguments. (The rules for how arguments are interpreted by functions are described in *Note Function parameters::.) The entire group is then replaced with the value produced by the function call. Of special interest is the variant of `\def' that looks like this: {\def {\var ...} BODY ...} When the first argument to `\def' is a group rather than a variable reference, then it is interpreted as a function definition. For details on using `\def', see *Note Latte builtins::. File: latte.info, Node: Quoted expressions, Next: Whitespace, Prev: Groups, Up: Latte syntax 3.4 Quoted expressions ====================== A quoted expression is a Latte expression preceded by a quote symbol, of which there are four kinds described below. Quoting expressions is for preventing them from being evaluated, or exerting control over _when_ they're evaluated. In Latte this is generally necessary only when writing macros (*note Functions on functions::). Here are the four kinds of quoted expressions: `\'EXPR' This is an "ordinary quote" that, when evaluated, yields EXPR exactly as written. Example: \'{\add 3 4} => {\add 3 4} `\`EXPR' This is called "quasiquote" and works just like the ordinary quote, except that individual subexpressions of EXPR may be "unquoted". The result of evaluating a quasiquoted expression is that expression itself, except with any unquoted subexpressions replaced by their values. (See below for example.) `\,EXPR' This is an "unquote". When it appears inside a quasiquoted expression, it causes EXPR to be evaluated. Example: \`{\add \,{\multiply 3 4} 5} => {\add 12 5} `\,@EXPR' This is called "unquote-splicing", or "splicing unquote". It works just like unquote, except that if the result of evaluating EXPR is a group, then the result of `\,@EXPR' is _the individual elements of that group_ rather than _the group containing the elements_. This is used for "splicing" the elements of a list into the containing list at the same nesting level. Example: \`{a b \,@{\group c d e} f} => {a b c d e f} See the documentation for `\macro' for an example using the different forms of quote. File: latte.info, Node: Whitespace, Next: Latte vs. HTML, Prev: Quoted expressions, Up: Latte syntax 3.5 Whitespace ============== As mentioned earlier, a Latte translator divides a Latte file into expressions. Whitespace is not considered an expression. Instead, each expression gets associated with _the whitespace that precedes it in the Latte file_. For example, when given the following input: {\document {Our second document} This was as exciting to write as our first!} Latte divides it up as follows: * A group preceded by no whitespace, consisting of: * The variable reference `\document' preceded by no whitespace; * A group preceded by a single space, consisting of: * The word `Our' preceded by no whitespace; * The word `second' preceded by a single space; * The word `document' preceded by a single space * The word `This' preceded by one newline and two spaces; * The word `was' preceded by one space; * The word `as' preceded by one space; * The word `exciting' preceded by one space; * The word `to' preceded by one space; * The word `write' preceded by one newline and two spaces; * The word `as' preceded by one space; * The word `our' preceded by one space; * The word `first!' preceded by one space. The whitespace attached to an expression follows the expression wherever it goes (e.g., whether the expression is assigned to variables, passed to functions, etc.). However, that may not be the whitespace used when and if the expression is emitted as output. In general, when emitting output, expressions inherit the preceding-whitespace of the context in which they're used, not the context in which they're defined. Here are some examples: {\def \foo Hello} "\foo" => "Hello" In this example, even though `Hello' is preceded by a single space where it's assigned to `\foo', the _use_ of `\foo' has zero spaces preceding it, so the result contains no space between `"' and `Hello'. {\def {\a \=href \&rest} { \rest }} {\a \href=foo bar} => bar Why, in the output, is there just one space between `a' and `href'? In the definition of `\a', the subexpression `{href="\href"}' is preceded by a newline and several spaces. However, it's used inside an `\if' expression, which itself is preceded by only one space. When the `\if' expression is replaced by the `{href="\href"}' subexpression, the subexpression inherits its whitespace. A word about the "forget-whitespace" symbol, `\/'. When this appears in a Latte file, all the whitespace that precedes it is canceled. This can be useful when you want the Latte source to include whitespace for readability, but you don't want that whitespace to appear in the output. File: latte.info, Node: Latte vs. HTML, Prev: Whitespace, Up: Latte syntax 3.6 Latte vs. HTML ================== HTML was never supposed to be something that you would see... it staggers me that people have actually put up with having to write HTML by hand. --Tim Berners-Lee, inventor of HTML Here are some shortcomings of HTML that are remedied by Latte. Matching tags In HTML, most markup tags come in beginning-ending pairs that must match and nest properly. For instance, `text' is how you'd put `text' in a bold-italic typeface. But if you write `text', it's an error. Plus, all those paired tags create a lot of visual clutter, making HTML hard for humans to read. And most text editors can't offer much help in identifying and navigating matching sets of tags. On the other hand, Latte uses curly braces to group text, which most text editors are already able to handle well. And Latte only requires beginning tags, not ending tags, reducing visual clutter and eliminating the potential for mismatches. Paragraph tags HTML requires `

' to appear at the beginning of each paragraph. Since most of us are accustomed to denoting paragraph breaks with just a blank line, it's easy to forget the `

'. When the `

' is remembered, it adds more clutter to the text. When `latte-html' outputs one or more blank lines, it automatically includes a `

'. *Note Automatic HTML processing::. Metacharacters HTML defines numerous "metacharacters"--characters that trigger special processing. These include `<', `>', `&', and `"'. Wherever you wish to include these characters in a document, you must use a named "character entity" instead: `<', `>', `&', and `"'. Not only are these characters bad choices for metacharacters in documents that are mostly textual (and may therefore be expected to contain several occurrences of `"' and `&' at least), but their character entity replacements are verbose and are not visually related to the characters they replace. In Latte, the only metacharacters are `\', `{', and `}', and to use these characters literally, it is only necessary to escape them with a backslash. When `latte-html' encounters HTML metacharacters such as `&', it automatically emits the corresponding HTML character entity. Repeated text and constructs HTML has no macro facility; no way to store text that is used repeatedly in a document, and no way to define shorthand for oft-used idioms. Consequently, HTML documents may contain redundant stretches of text (as we saw in the first example; *note Latte examples::) or complex, repeated, and difficult-to-diagnose layout directives. Latte is a full-fledged programming language. Not only can repeated constructs be encapsulated in variables and functions, but these variables and functions can have descriptive names that indicate their purpose. Instead of dozens of nested `', `', and `
' tags, for instance, each with dozens of attribute settings, one can lay out a complex page with something simple like: {\layout \top-margin={...} \left-margin={...} ...main body...} File: latte.info, Node: Latte builtins, Next: Writing HTML with Latte, Prev: Latte syntax, Up: Top 4 Latte builtins **************** This chapter is a reference guide to the built-in functions of Latte. It does not describe the HTML-related functions of `latte-html'. For those, see *Note HTML functions::. Most of these functions are intrinsic to Latte. However, some are loaded at runtime from the library file `standard.latte'. * Menu: * Control structures:: * Functions on variables:: * Functions on functions:: * Group functions:: * Boolean functions:: * Text functions:: * Arithmetical functions:: * File functions:: * Other functions:: * Predefined variables:: File: latte.info, Node: Control structures, Next: Functions on variables, Prev: Latte builtins, Up: Latte builtins 4.1 Control structures ====================== -- Function: if test then-expr else-expr Evaluates TEST. If the result is true (*note Boolean functions::), evaluates THEN-EXPR, otherwise evaluates ELSE-EXPR. The result of the `\if' is the result of whichever of the THEN-EXPR or ELSE-EXPR is invoked. If ELSE-EXPR is omitted, it is taken to be `{}' by default. -- Function: while test body ... Evaluates TEST. If the result is true, evaluates the BODY subexpressions, then repeats. The result is the _concatenation_ of the results of each invocation of the BODY expressions. For example: {\def \i 0} {\while {\less? \i 3} i is \i {\set! \i {\add \i 1}}} => i is 0 i is 1 i is 2 -- Function: foreach \var group body ... Evaluates GROUP. Then, for each element in the result, sets \VAR to that value and evaluates the BODY subexpressions. File: latte.info, Node: Functions on variables, Next: Functions on functions, Prev: Control structures, Up: Latte builtins 4.2 Functions on variables ========================== -- Function: def \var value Creates a new variable \VAR in the innermost scope (*note Scope::), evaluates VALUE, and assigns the result to the new variable. -- Function: def {\var params ...} body ... Creates a new function whose parameters are given by PARAMS and whose body consists of BODY; assigns the function to the new variable \VAR in the innermost scope. The parameter list PARAMS consists of zero or more positional parameters and named parameters, plus one optional rest parameter. *Note Function parameters::. When the function is invoked, the BODY expressions are evaluated one at a time in sequence. The return value of the function is the value of the last BODY expression. *This behavior differs from that of Latte versions prior to 2.0!* In those versions, user-defined functions constructed an implicit group containing the values of all BODY expressions. That behavior introduced an ambiguity in the case of functions that sometimes returned a group and sometimes returned some other kind of value. The new behavior more closely matches that of other related programming languages (*note Pedigree::). To ensure that a function written for the old Latte works under the current Latte, it should generally suffice to convert this: {\def {\func ARGS ...} BODY1 BODY2 ...} into this: {\def {\func ARGS ...} {BODY1 BODY2 ...}} To aid in this transition, a new kind of error is detected by Latte. Invoking a user-defined function that contains subexpressions that have no side effects and don't participate in the return value signals a "Useless subexpression" error. Note that `{\def {\var PARAMS ...} BODY ...}' is exactly equivalent to {\def \VAR {\lambda {PARAMS ...} BODY ...}} The `\lambda' function is described in *Note Functions on functions::. -- Function: defmacro {\var params ...} body ... Like `{\def {\VAR PARAMS ...} BODY ...', but creates a macro instead of a function. See `\macro' (*note Functions on functions::). -- Function: let {{\var value} ...} body ... Assigns to each \\VAR the result of evaluating each VALUE, then evaluates BODY in the scope of those variables. Note that every VALUE is evaluated before any of the \VARs is assigned, so the VALUEs cannot refer to the \VARs. In other words, this is an error: {\let {{\list {x y z}} {\first {\front \list}}} ...} To get the desired effect, you can do this: {\let {{\list {x y z}}} {\let {{\first {\front \list}}} ...}} or this: {\let {{\list {x y z}} {\first {}}} {\set! \first {\front \list}} ...} The value of the `\let' expression is the result of evaluating the last BODY expression. *This behavior differs from that of Latte versions prior to 2.0!* In those versions, `\let' expressions constructed an implicit group containing the values of all BODY expressions. See `\def' for a discussion of this change. To ensure that a `\let' expression written for the old Latte works under the current Latte, it should generally suffice to convert this: {\let {{\VAR VAL} ...} BODY1 BODY2 ...} into this: {\let {{\VAR VAL} ...} {BODY1 BODY2 ...}} To aid in this transition, a new kind of error is detected by Latte. Invoking a `\let' expression that contains subexpressions that have no side effects and don't participate in the return value signals a "Useless subexpression" error. Note that {\let {{\VAR VAL} ...} BODY ...} is equivalent to {{\lambda {\VAR ...} BODY ...} VAL ...} -- Function: set! \var value Assigns to existing variable \VAR the result of evaluating VALUE. File: latte.info, Node: Functions on functions, Next: Group functions, Prev: Functions on variables, Up: Latte builtins 4.3 Functions on functions ========================== -- Function: funcall function args ... Evaluates FUNCTION; the result must be a Latte function. Then evaluates all of the ARGS. The function yielded by FUNCTION is then invoked, with ARGS passed in as its arguments. {\funcall \add 2 3 4} => 9 -- Function: apply function args ... last-arg Exactly like `\funcall', except that if the value of LAST-ARG is a group, then _its elements_, not the group, are passed individually to FUNCTION. {\apply \add 2 3 4} => 9 {\let {{\numbers {2 3 4}}} {\apply \add 6 \numbers}} => 15 -- Function: lambda {params ...} body ... Creates a Latte function with parameter list given by PARAMS and body given by BODY. A full discussion of user-defined functions in Latte appears in the description for `\def' (*note Functions on variables::). -- Function: macro {params ...} body ... Like `\lambda', this produces a new Latte function with the given parameter list and body; but the function produced is a special kind called a "macro". When the macro is evaluated by passing it some argument: 1. the arguments are _not evaluated_; 2. the macro body is evaluated using the non-evaluated argument values; 3. the result of _that_ is then evaluated. For example, here's how to use a macro to define a function called `\unless'. {\def \unless {\macro {\test \&body} \`{\if {\not \,\test} \,\body}}} The idea is for `{\unless TEST BODY ...}' to evaluate BODY only when TEST is false. Here's how the macro definition works when `\unless' is called like this: {\unless {\zero? \x} {\set! \x {\subtract \x 1}}} 1. The argument value `{\zero? \x}' is assigned to the macro parameter `\test', and the remaining arguments (in this example, the sole body expression) are gathered into a group and assigned to `\body' (so `\body' equals `{{\set! \x {\subtract \x 1}}}}'). 2. The body of the macro is evaluated. This is a quasiquoted expression (*note Quoted expressions::), so first, nested unquoted expressions are evaluated: 1. `\,\test' => `{\zero? \x}' 2. `\,\body' => `{{\set! \x {\subtract \x 1}}}}' and then the quasiquoted expression is returned with the unquoted subexpressions replaced by their values: {\if {\not {\zero? \x}} {{\set! \x {\subtract \x 1}}}}} 3. The resulting expression is evaluated normally. -- Function: compose f1 f2 Produces a new function of one argument that yields `{F1 {F2 X}}' (where X is the argument). Naturally, F1 and F2 must both be functions of one argument. This is used in the definition of `\cadr' et al. (*note Group functions::): {\def \cadr {\compose \car \cdr}} -- Function: lmap function group Evaluates FUNCTION, which must yield a function, and GROUP, which must yield a group. Then applies FUNCTION to each element of GROUP in turn, yielding a group containing the result of each function call. Example: {\def {\add1 \x} {\add \x 1}} {\lmap \add1 {3 4 5}} => 4 5 6 File: latte.info, Node: Group functions, Next: Boolean functions, Prev: Functions on functions, Up: Latte builtins 4.4 Group functions =================== -- Function: append expr ... Each EXPR is evaluated, and a group of expressions is constructed as follows: if EXPR's value is a group, its elements are individually added to the result; otherwise EXPR's value is added as a single element. -- Function: back group Yields the last element of the given GROUP. -- Function: car group Yields the first element from the given GROUP. Synonym for `\front'. -- Function: cdr group Yields the elements of GROUP minus the first element. Example: {\cdr {a b c}} => b c -- Function: caar expr -- Function: cadr expr -- Function: cdar expr -- Function: cddr expr Produces {\car {\car EXPR}}, {\car {\cdr EXPR}}, {\cdr {\car EXPR}}, {\cdr {\cdr EXPR}}, respectively. These functions are defined in `standard.latte' using `\compose' (*note Functions on functions::). -- Function: cons expr group Constructs a new group consisting of EXPR followed by the elements of GROUP. Synonym for `\push-front'. -- Function: empty? group Yields a true value if GROUP is empty, false otherwise. *Note Boolean functions::. Equivalent to {\equal? 0 {\length GROUP}} -- Function: front group Yields the first element from the given GROUP. Synonym for `\car'. -- Function: group expr ... Constructs a group consisting of all the given EXPRs. -- Function: length group Returns the number of elements in GROUP. Can also be used on text strings; see *Note Text functions::. -- Function: member? expr group Returns a true value if EXPR is `\equal?' to any member of GROUP, false otherwise. -- Function: nth n group Returns the Nth element of GROUP, counting from 0. Can also be used on text strings. -- Function: push-back expr group Constructs a new group consisting of the elements of GROUP followed by EXPR. Synonym for `\snoc'. -- Function: push-front expr group Constructs a new group consisting of EXPR followed by the elements of GROUP. Synonym for `\cons'. -- Function: rdc group Yields the elements of GROUP minus the last element. Example: {\rdc {a b c}} => a b -- Function: reverse group Reverses the elements of GROUP. Does not reverse the elements of nested groups. Examples: {\reverse {a b c}} => c b a {\reverse {a {b c} d}} => d b c a -- Function: snoc expr group Constructs a new group consisting of the elements of GROUP followed by EXPR. Synonym for `\push-back'. -- Function: subseq group from to Constructs a new group consisting of the elements of GROUP beginning at FROM and ending before TO. Both FROM and TO count from 0. If either is negative, positions are counted backward from the end of GROUP rather than forward from the front. If TO is omitted, the new group contains the elements from FROM through the end of GROUP. Examples: {\subseq {a b c d e} 1 3} => b c {\subseq {a b c d e} -2} => d e File: latte.info, Node: Boolean functions, Next: Text functions, Prev: Group functions, Up: Latte builtins 4.5 Boolean functions ===================== "Boolean" values are truth and falsehood. In Latte, every value is considered "true" (for purposes of the test clauses in `\if' and `\while') except for the empty group, `{}', which is false. Note that the value `0' (zero), which is false in languages such as C and Perl, is true in Latte. -- Function: and expr ... Evaluates each EXPR in turn until one yields falsehood. If no expression yields falsehood, returns the value of the last one. Otherwise returns `{}'. -- Function: not expr Negates the truth value of EXPR. If EXPR is true, yields `{}'. If EXPR is false, yields the Latte "truth object." (The "truth object" is a boolean value with no displayable representation that is only used for this purpose.) -- Function: or expr ... Evaluates each EXPR in turn until one yields truth, then returns that value. If no expression is true, returns `{}'. File: latte.info, Node: Text functions, Next: Arithmetical functions, Prev: Boolean functions, Up: Latte builtins 4.6 Text functions ================== -- Function: concat string ... Constructs a new string by concatenating all the STRING arguments. Synonym for `\string-append'. -- Function: downcase string ... Constructs a group containing all the STRING values converted to lower case. -- Function: explode string ... Constructs a group whose elements are the individual characters of the given STRING values. -- Function: length string Returns the number of characters in STRING. Can also be used on groups; see *Note Group functions::. -- Function: nth n string Returns the Nth character of STRING, counting from 0. Can also be used on groups. -- Function: string-append string ... Constructs a new string by concatenating all the STRING arguments. Synonym for `\concat'. -- Function: string-ge? string ... Returns truth if each STRING is greater than or equal to the ones following it ("monotonically non-increasing"), else returns falsehood. Synonym for `\string-greater-equal?'. Example: {\string-ge? three three ten ten seven one} is true. -- Function: string-greater-equal? string ... Returns truth if each STRING is greater than or equal to the ones following it ("monotonically non-increasing"), else returns falsehood. Synonym for `\string-ge?'. -- Function: string-greater? string ... Returns truth if each STRING is strictly greater than the ones following it ("monotonically decreasing"), else returns falsehood. Synonym for `\string-gt?'. -- Function: string-gt? string ... Returns truth if each STRING is strictly greater than the ones following it ("monotonically decreasing"), else returns falsehood. Synonym for `\string-greater?'. Example: {\string-gt? three ten seven one} is true. -- Function: string-le? string ... Returns truth if each STRING is less than or equal to the ones following it ("monotonically non-decreasing"), else returns falsehood. Synonym for `\string-less-equal?'. Example: {\string-le? one seven ten ten three three} is true. -- Function: string-less-equal? string ... Returns truth if each STRING is less than or equal to the ones following it ("monotonically non-decreasing"), else returns falsehood. Synonym for `\string-le?'. -- Function: string-less? string ... Returns truth if each STRING is strictly less than the ones following it ("monotonically increasing"), else returns falsehood. Synonym for `\string-lt?'. -- Function: string-lt? string ... Returns truth if each STRING is strictly less than the ones following it ("monotonically increasing"), else returns falsehood. Example: {\string-lt? one seven ten three} is true. -- Function: substr string from to Constructs a new string consisting of the characters of STRING beginning at FROM and ending before TO. Both FROM and TO count from 0. If either is negative, positions are counted backward from the end of STRING rather than forward from the front. If TO is omitted, the new string contains the characters from FROM through the end of STRING. Examples: {\substr abcde 1 3} => bc {\substr abcde -2} => de -- Function: upcase string ... Constructs a group containing all the STRING values converted to upper case. File: latte.info, Node: Arithmetical functions, Next: File functions, Prev: Text functions, Up: Latte builtins 4.7 Arithmetical functions ========================== When Latte is installed, it can be configured for integer arithmetic only or for integer and floating-point arithmetic. If integer arithmetic only, then a number is any Latte string consisting of one or more digits, with an optional leading plus or minus sign. If integer and floating-point, then a number is any integer (as described) optionally followed by a decimal point and zero or more digits. -- Function: add number ... Adds all the given NUMBERs. -- Function: ceil number Rounds NUMBER up to the next integer. If Latte is configured for integer-only arithmetic, NUMBER is returned unchanged. -- Function: divide number ... Divides the first NUMBER by each successive argument. An attempt to divide by zero triggers an error. -- Function: floor number Rounds NUMBER down to the next integer. If Latte is configured for integer-only arithmetic, NUMBER is returned unchanged. -- Function: ge? number ... Returns truth if each NUMBER is greater than or equal to the ones following it ("monotonically non-increasing"), else returns falsehood. Synonym for `\greater-equal?'. Example: {\ge? 10 10 7 3 3 1} is true. -- Function: greater-equal? number ... Returns truth if each NUMBER is greater than or equal to the ones following it ("monotonically non-increasing"), else returns falsehood. Synonym for `\ge?'. -- Function: greater? number ... Returns truth if each NUMBER is strictly greater than the ones following it ("monotonically decreasing"), else returns falsehood. Synonym for `\gt?'. -- Function: gt? number ... Returns truth if each NUMBER is strictly greater than the ones following it ("monotonically decreasing"), else returns falsehood. Synonym for `\greater?'. Example: {\gt? 10 7 3 1} is true. -- Function: le? number ... Returns truth if each NUMBER is less than or equal to the ones following it ("monotonically non-decreasing"), else returns falsehood. Synonym for `\less-equal?'. Example: {\le? 1 3 3 7 10 10} is true. -- Function: less-equal? number ... Returns truth if each NUMBER is less than or equal to the ones following it ("monotonically non-decreasing"), else returns falsehood. Synonym for `\le?'. -- Function: less? number ... Returns truth if each NUMBER is strictly less than the ones following it ("monotonically increasing"), else returns falsehood. Synonym for `\lt?'. -- Function: lt? number ... Returns truth if each NUMBER is strictly less than the ones following it ("monotonically increasing"), else returns falsehood. Synonym for `\less?'. Example: {\lt? 1 3 7 10} is true. -- Function: modulo a b Returns A modulo B. -- Function: multiply number ... Multiplies all the given NUMBERs. -- Function: random n Returns a random integer greater than or equal to 0 and less than N. -- Function: subtract number ... Subtracts from the first NUMBER each successive argument. If only one NUMBER is given, negates the number. -- Function: zero? number Returns truth if NUMBER is 0, falsehood otherwise. Equivalent to: {\equal? 0 NUMBER} File: latte.info, Node: File functions, Next: Other functions, Prev: Arithmetical functions, Up: Latte builtins 4.8 File functions ================== -- Function: file-contents filename Returns, as a Latte string, the contents of the file named by FILENAME. -- Function: process-output program args ... Runs PROGRAM with arguments ARGS, returning the output of the program as a Latte string. -- Function: load-file filename Evaluates the contents of the Latte file FILENAME. Textual results are discarded, but variable and function definitions and other side effects are retained. -- Function: load-library filename Like `\load-file', but searches for FILENAME using Latte's library-search algorithm. For each directory in the Latte search path, tries to load `FILENAME' from that directory, or if that fails, `FILENAME.latte'. The default Latte search path is set at installation time, usually to the directories `/usr/local/share/latte' and `.'. It can be changed by setting the environment variable `LATTE_PATH' to a colon-separated list of directories. -- Function: include filename This works like `\load-file' but the text of FILENAME is not discarded. This can therefore be used to assemble an aggregate Latte document out of smaller pieces. File: latte.info, Node: Other functions, Next: Predefined variables, Prev: File functions, Up: Latte builtins 4.9 Other functions =================== -- Function: equal? expr ... Returns truth if all the EXPRs are equal, falsehood otherwise. Equality is defined recursively on groups: two groups are `\equal?' if and only if each of their corresponding subexpressions are `\equal?'. -- Function: getenv name Returns the value of the environment variable named NAME. If no such environment variable exists, returns falsehood. -- Function: error text ... Exits from Latte with an error message given by TEXT plus the file location where `\error' was called. -- Function: group? expr Returns truth if the given EXPR is a group, falsehood otherwise. -- Function: operator? expr Returns truth if the given EXPR is an operator (i.e., function), falsehood otherwise. -- Function: string? expr Returns truth if the given EXPR is a string, falsehood otherwise. -- Function: warn text ... Emits a warning message given by TEXT plus the file location where `\warn' was called. File: latte.info, Node: Predefined variables, Prev: Other functions, Up: Latte builtins 4.10 Predefined variables ========================= -- Variable: __latte-version__ This variable is predefined to contain the version number of `latte-html'. File: latte.info, Node: Writing HTML with Latte, Next: The latte-text translator, Prev: Latte builtins, Up: Top 5 Writing HTML with Latte ************************* Latte was originally created to provide a saner alternative to HTML. Although it has expanded in purpose, simplifying the production of HTML documents remains its primary use, at least for now. This chapter discusses how. An understanding of HTML is assumed. * Menu: * Automatic HTML processing:: * HTML functions:: * Nonstandard HTML:: * Predefined variables in latte-html:: * Invoking latte-html:: * Makefile rules:: * Latte mode for Emacs:: File: latte.info, Node: Automatic HTML processing, Next: HTML functions, Prev: Writing HTML with Latte, Up: Writing HTML with Latte 5.1 Automatic HTML processing ============================= As `latte-html' evaluates top-level Latte expressions, it performs two kinds of automatic HTML processing on the result before sending it to the output: Automatic `
'-tag generation Wherever one or more blank lines appear, any subsequent text is preceded with a `
' tag. Automatic character-entity translation Wherever an HTML metacharacter is encountered, its character-entity code is substituted; e.g., `<' and `>' for `<' and `>'. These processing steps can be controlled using the `\_pre' and `\html' functions; see *Note HTML functions::. File: latte.info, Node: HTML functions, Next: Nonstandard HTML, Prev: Automatic HTML processing, Up: Writing HTML with Latte 5.2 HTML functions ================== This section describes the HTML functions available in `latte-html'. Most are defined in `html.latte', but some are intrinsic to `latte-html'. -- Function: _pre text ... This is a subroutine of the HTML tag function `\pre' (which is used for including preformatted text). Any text enclosed in `{\_pre ...}' is not subject to automatic `
'-tag generation. So whereas `latte-html' would normally turn this: First paragraph. Second paragraph. into this: First paragraph.
Second paragraph. it turns this: {\_pre First paragraph. Second paragraph.} into this: First paragraph. Second paragraph. Note that `\_pre' does _not_ turn off automatic character-entity translation (see `\html' below), though it is possible to use `\_pre' in combination with `\html' to achieve that effect. -- Function: html text ... Text enclosed in `{\html ...}' is not subject to automatic HTML character-entity translation. In other words, where `latte-html' would normally translate this: "Penn & Teller" into this: "Penn & Teller" it turns this: {\html "Penn & Teller"} into this: "Penn & Teller" This is of course essential to implementing all of the HTML-tag Latte functions, since they must be able to emit HTML metacharacters (particularly `<' and `>') literally. Note that `\html' does _not_ turn off automatic `
'-tag generation, though it is possible to use `\html' in combination with `\_pre' to achieve that effect. All remaining HTML functions fall into three categories: character-entity functions; balanced tag functions; and non-balanced tag functions. * Menu: * Character entity functions:: * Balanced tag functions:: * Non-balanced tag functions:: File: latte.info, Node: Character entity functions, Next: Balanced tag functions, Prev: HTML functions, Up: HTML functions 5.2.1 Character entity functions -------------------------------- For each named HTML character entity `&FOO;', `html.latte' defines a function of zero arguments named `\c-FOO'. So, for example, to emit a copyright symbol, write `{\c-copy}' (which produces `©'). The `\c-FOO' functions are implemented in terms of `\ch'. Write `{\ch NAME}' to emit `&NAME;'. There is also `\chx', which takes a two-digit hexadecimal number as an argument. It produces `&#NUM;'. File: latte.info, Node: Balanced tag functions, Next: Non-balanced tag functions, Prev: Character entity functions, Up: HTML functions 5.2.2 Balanced tag functions ---------------------------- "Balanced tag functions" are Latte functions that produce a balanced pair of HTML tags surrounding some text, with optional HTML attributes in the opening tag. For example, `\b' is a balanced tag function because {\b some text} produces some text There is a balanced tag function corresponding to every HTML tag defined in the HTML 4.0 "Transitional" standard (which includes deprecated forms). The Latte names of these functions are always all-lowercase. Every tag function has optional named parameters (*note Function parameters::) corresponding to the attributes permitted for that tag. For instance, `\a' has named parameters `\href' and `\name', among others. These attribute names are always all-lowercase too. When you specify a value for an HTML attribute, the value is automatically surrounded with double quotes. For example, {\a \href=http://www.latte.org/ The Latte language} produces this: The Latte language Some tags have boolean attributes that do not take values. For instance, in Some text `readonly' is a boolean attribute. The Latte way to include this attribute is to write `\readonly=1'. The `1' is ignored; any true value will do. A false value will cause the attribute not to appear in the tag. Every balanced tag function in `latte-html' accepts an arbitrary amount of text (as its rest parameter; *note Function parameters::) to be enclosed by the paired beginning-ending HTML tags. File: latte.info, Node: Non-balanced tag functions, Prev: Balanced tag functions, Up: HTML functions 5.2.3 Non-balanced tag functions -------------------------------- Non-balanced tags are those that are not paired with a closing tag and do not enclose text; for example, picture of foo (There is no `' tag.) Non-balanced tag functions in `latte-html' work exactly like balanced tag functions do, except that they only accept various named parameters as arguments (corresponding to their HTML attributes) and do not accept an arbitrary amount of text to enclose. File: latte.info, Node: Nonstandard HTML, Next: Predefined variables in latte-html, Prev: HTML functions, Up: Writing HTML with Latte 5.3 Nonstandard HTML ==================== It is possible to write nonstandard HTML tags and attributes even though `html.latte' doesn't predefine tag functions or named parameters for them. Nonstandard tags ---------------- To write a nonstandard balanced tag, use `\_bal-tag'. To write a nonstandard non-balanced tag, use `\_tag'. These functions are used to define all the HTML tag functions in `html.latte'. -- Function: _bal-tag name attrs bools deprs nonstandard \=depr \&rest Emits a balanced pair of HTML tags named NAME. The tag's attributes and their values are given by ATTRS, which is a group of the form {ATTRIBUTE VALUE ATTRIBUTE VALUE ...} An attribute in ATTRS with a false VALUE does not appear in the output. The tag's boolean attributes and their (boolean) values are given by BOOLS, which has the same form as ATTRS. The tag's deprecated attributes and their values are given by DEPRS, which also has the same form as ATTRS. Any attributes in DEPRS with true values will cause a warning to be emitted if `\strict-html4' is true (*note Predefined variables::). Any nonstandard attributes and their values are listed in NONSTANDARD and will also cause warnings to be emitted if `\strict-html4' is true. The named parameter `\depr', if set to true, indicates that this tag is deprecated and should generate a warning if used and `\strict-html4' is true. The rest parameter accumulates text to appear between the opening and closing HTML tags. Example: {\_bal-tag whee {foo bar right wrong} {x 1 y {}} {} {} The text in the tags} yields The text in the tags Of greater utility would be defining a tag function named `\whee' along these lines: {\def {\whee \=foo \=right \=x \=y \&rest} {\_bal-tag whee {foo \foo right \right} {x \x y \y} {} {} \rest}} so that you can then write {\whee \foo=bar \right=wrong \x=1 The text in the tags} to get the same result as above. -- Function: _tag name params bools deprs nonstandard \=depr This works exactly like `\_bal-tag' (see above) except that it does not have a rest parameter, does not enclose text, and does not emit an "ending tag." Example: {\_tag zoom {whoosh zing} {} {} {}} => Nonstandard attributes ---------------------- Including nonstandard attributes in a standard HTML tag is quite a bit easier. Every tag function in `html.latte' includes a named parameter called `\nonstandard' whose value is a list of attribute-value pairs to include in the tag. So, for example, {\b \nonstandard={boldness high} Very bold!} => Very bold! even though `\b' does not have a `boldness' parameter. Use of `\nonstandard' will trigger warnings if `\strict-html4' is true. File: latte.info, Node: Predefined variables in latte-html, Next: Invoking latte-html, Prev: Nonstandard HTML, Up: Writing HTML with Latte 5.4 Predefined variables in latte-html ====================================== In addition to the variables predefined by all Latte translators (*note Predefined variables::), `latte-html' predefines the following additional variables: -- Variable: __FILE__ The name of the file being processed. -- Variable: strict-html4 This variable contains a true value if `latte-html' was invoked with the `--strict' option (*note Invoking latte-html::), otherwise it contains falsehood. File: latte.info, Node: Invoking latte-html, Next: Makefile rules, Prev: Predefined variables in latte-html, Up: Writing HTML with Latte 5.5 Invoking latte-html ======================= Usage: `latte-html [OPTIONS] [FILE]' If FILE is not given or is `-', the standard input is processed. Options are: `-v' `--version' Print the version number of `latte-html' and exit. `-h' `--help' Print a short help summary and exit. `-l LIBRARY' `--load=LIBRARY' Load the Latte library file LIBRARY before processing FILE. Any number of library files may be preloaded this way. Loading works as in `\load-library' (*note File functions::). Libraries requested with this option are loaded after the default libraries (see `-n'). `-n' `--no-default' Do not load the default libraries. Normally, `latte-html' implicitly loads `standard.latte' (*note Latte builtins::) and `html.latte' (*note HTML functions::) on startup. If you wish to suppress loading of just one default library file, use `-n' and also use `-l' to explicitly load the other library file, like so: latte-html -n -l standard ... `-o FILE' `--output=FILE' Place output in FILE. Without this option, `latte-html' places its output on the standard output. `-f' `--fragment' Produce an HTML fragment rather than a complete HTML document. This causes `latte-html' to suppress the `' and `' tags with which it normally surrounds its output. This can be useful when multiple Latte documents must be processed separately then combined to create a single HTML document. `-s' `--strict' Causes `latte-html' to emit warnings whenever HTML tags or attributes are used that have been deprecated in the HTML 4.0 standard. The Latte variable `--strict' is set to true if this option is given, otherwise it's set to false. `-L LANG' `--lang=LANG' Set the language code of the resulting HTML document to LANG. This causes the `' tag that is automatically emitted to include the `lang=LANG' attribute. `-d FLAGS' `--debug=FLAGS' Turn on Latte debugging; FLAGS is a comma-separated list of flags chosen from the set `eval' and `mem'. When the `eval' debugging flag is turned on, each Latte evaluation and result is displayed, with its nesting level indicated with indentation. When the `mem' debugging flag is turned on, a summary of extant and total objects allocated is periodically displayed. The output format of `--debug=eval' is designed to be used with the "selective display" feature of Emacs (q.v.), which is able to hide lines with greater than a given amount of indentation. File: latte.info, Node: Makefile rules, Next: Latte mode for Emacs, Prev: Invoking latte-html, Up: Writing HTML with Latte 5.6 Makefile rules ================== It is convenient to delegate the work of invoking `latte-html' to `make', especially when your project involves multiple Latte files that must be converted into multiple HTML files. Here's a suffix rule you can add to your `Makefile': .latte.html: -rm -f $@ latte-html -o $@ $< You may also find it convenient to list your HTML targets like so: HTMLFILES = foo.html bar.html ... all: $(HTMLFILES) This, combined with the suffix rule above, will cause `make all' to create `foo.html' from `foo.latte', `bar.html' from `bar.latte', and so on. Autoloading shared Latte definitions ------------------------------------ It is typical to place common Latte function definitions into a separate file that is shared by all the other Latte files in a project. If you have such definitions in `defns.latte' (say), you may wish to change the suffix rule to .latte.html: -rm -f $@ latte-html --load=defns -o $@ $< so that `defns.latte' is automatically loaded as each Latte file is processed. (The alternative is to begin each Latte file with a call to `{\load-library defns}'.) Managing multi-author documents ------------------------------- Latte can aid in producing HTML documents from multiple sources, such as documents written by multiple authors, each of whom "owns" a section of the document. Each section can be its own Latte file, allowing all authors to edit their portions concurrently. The final document can be a Latte file that uses `\include' to include the subparts (*note HTML functions::), or it can be an HTML template that, when glued together with HTML fragments generated from the individual Latte files, forms a complete HTML file. Here's an example of the first approach. Suppose Hank and Lorna are each responsible for one piece of `article.html', which is produced from the Latte file `article.latte'. Hank edits his portion in `hank.latte', and Lorna edits hers in `lorna.html'. The file `article.latte' may look like this: {\head {\title Our article}} {\body {\h1 Thesis} {\include hank.latte} {\h1 Antithesis} {\include lorna.latte}} and the final version can be produced simply by running `latte-html article.latte'. In this example, the files `hank.latte' and `lorna.latte' need not (indeed, should not) contain any calls to `\head' or `\body'. The second approach, where the output of individual Latte files is pasted together into an HTML template, is slightly trickier, but not much. Imagine that `article.html' is to be produced from `head.html' and `foot.html', which presumably contain HTML boilerplate for beginning and ending files, with the middle of the document supplied by `hank.latte' and `lorna.latte'. The `Makefile' rules for creating this document must generate HTML files from the Latte pieces, then concatenate all the HTML parts with the Unix command `cat'. Here's how it can do that: hank.html: hank.latte -rm -f hank.html latte-html --fragment -o hank.html hank.latte lorna.html: lorna.latte -rm -f lorna.html latte-html --fragment -o lorna.html lorna.latte article.html: head.html foot.html hank.html lorna.html -rm -f article.html cat head.html hank.html lorna.html foot.html > article.html Note the use of the `--fragment' option to `latte-html' to suppress the automatic generation of an HTML preamble (containing a `DOCTYPE' declaration and commentary), an `' tag, and a closing `' tag. File: latte.info, Node: Latte mode for Emacs, Prev: Makefile rules, Up: Writing HTML with Latte 5.7 Latte mode for Emacs ======================== The Latte package comes with `latte.el', which is an implementation of an Emacs editing mode for Latte files. It's based on Emacs's text mode and provides some consistent indentation rules plus syntax-based coloring of Latte language elements. When Latte is installed, `latte.el' (and a "byte-compiled" version of it, `latte.elc') is installed the proper place for Emacs extensions. To use Latte mode while editing a Latte file, first ensure that the library is loaded by typing `M-x load-library latte ' then invoke `latte-mode' with `M-x latte-mode ' Since this is somewhat cumbersome, you may wish to have Emacs invoke Latte mode automatically when you edit files whose names end in `.latte'. To arrange this, add the following lines to your `.emacs' file: (autoload 'latte-mode "latte" "Latte mode" t) (setq auto-mode-alist (cons '("\\.latte$" . latte-mode) auto-mode-alist)) File: latte.info, Node: The latte-text translator, Next: Future directions, Prev: Writing HTML with Latte, Up: Top 6 The latte-text translator *************************** Latte includes a program called `latte-text' that works very much like `latte-html' except that it produces plain text as output. The implementation is very rudimentary at present, but it's suitable for experimenting with. In future releases, `latte-text' will become more sophisticated. In principle, a file that is written to be processed with `latte-html' can also be processed with `latte-text' to obtain a reasonable plain-text facsimile of the HTML version. The library file `text.latte' defines most of the same markup functions as does `html.latte' but gives them non-HTML definitions. For example, in `html.latte', the function `\b' ("boldface") surrounds its arguments with `' and `'. In `text.latte', the same function surrounds its arguments with *asterisks*. File: latte.info, Node: Future directions, Next: Further information, Prev: The latte-text translator, Up: Top 7 Future directions ******************* Here's what's planned for Latte beyond the current release. Other translators and tools Translators to languages other than HTML would increase the usefulness of Latte. One vision is for marked-up text to always be writable in Latte format, then translatable to whatever language is required for its final processing step (HTML for documents destined for the web, TeX for documents destined for the printer, ASCII for documents destined for plain-text displays, etc.). Whether that's actually a desirable goal is an open question, but it's certainly an interesting idea. At any rate, additional translators can only be a good thing. There is a niche for tools that process Latte files without necessarily translating them. For instance, a Latte dependency-tracker can determine what files a particular one depends on via calls to `\file-contents', `\include', and so on. This would aid in writing accurate dependencies in Makefiles. Higher-level libraries The functions now available in `latte-html' provide a bare minimum of basic functionality plus HTML equivalence. They don't include any useful higher-level functions to assist with such things as page layout--you still have to write HTML-analogous code. Of course, Latte allows you to encapsulate that code as reusable functions, and indeed many such functions have been written by Latte users for specific purposes. Some of those functions should be cleaned up, documented, made more general, and assembled into a useful library. Another set of high-level functions could abstract away the details of the target language. It should be possible to write a single Latte document that can produce both HTML and TeX, for example. But a Latte file containing calls to `\img' and `\h3' clearly has an HTML bias, whereas a Latte file containing calls to (hypothetical functions) `\setlength' and `\verbatim' clearly has a TeX bias. A set of common markup functions that translates well into all target languages could be just what the doctor ordered. Richer string handling Latte needs some Perl-like facilities for pattern matching, composition, and decomposition of strings. Improved system interface Latte should be able to perform file operations such as linking and unlinking files, testing access permissions, reading directories, and so on, as well as other system operations such as querying the current time, getting and setting environment variables, and so on. Character set awareness Presently, `latte-html' presumes the character set of the text in its input is ASCII or a superset thereof (such as ISO Latin-1). It should become possible to advise `latte-html' that the text is in some other character set. This would produce character set information in the generated HTML file, and would also affect which characters undergo automatic character-entity translation (*note Automatic HTML processing::), and what entities they're translated to. Better debugging The debugging output produced by `latte-html -d eval' is voluminous and useful only to the very dauntless. It should be possible to usefully restrict what output is seen, and that output should become more representative of the actions of the Latte engine. Apache module It should be possible to write a module for the Apache HTTP server that would allow it to serve Latte documents without their needing to be translated to HTML first; the translation would occur on the fly at document-serving time. File: latte.info, Node: Further information, Next: Pedigree, Prev: Future directions, Up: Top 8 Further information ********************* Further information about Latte can be found on the Latte home page at `http://www.latte.org/' To participate in Latte-related discussion, you may subscribe to the Latte mailing list by sending a request to To reach the developers, send mail to . Use this address for reporting bugs. Please make sure your bug reports contain as much relevant information as possible, including: * The version number of Latte (find out by running `latte-html --version'); * Your operating system type and version number; * The arguments you used to invoke `latte-html'; * A concise description of the problem, including what you expected should happen and what actually happened; * The smallest data sample you can devise that reliably demonstrates the problem, and the erroneous output, if any, that it produces. Before reporting a bug, please be sure that the problem isn't a Latte usage error on your part. The Latte "Frequently Asked Questions" page can help you make this determination. It's at `http://www.latte.org/faq.html' Latte is "open source" software, licensed under the terms of the Zanshin Public License. A copy of the license comes with Latte in the file `COPYING'. The license can also be found at `http://www.zanshin.com/ZPL.html' Latte is a product of Zanshin, Inc. More about Zanshin can be found on the Zanshin home page at `http://www.zanshin.com/' File: latte.info, Node: Pedigree, Next: Function and Variable Index, Prev: Further information, Up: Top Appendix A Pedigree ******************* The language designer's task is consolidation, not innovation. --C.A.R. Hoare (paraphrased) Some users may recognize the influence of other languages in the design of Latte. Latte borrows the choice of metacharacters (`\', `{', and `}') from TeX. One of TeX's drawbacks is that there actually are numerous other metacharacters, some of which are active in some contexts and not in others. This makes it tricky to write syntactically correct TeX, so Latte specifically rejects making any other characters "meta," even though some of TeX's metacharacters are frequently convenient (such as `~' for a non-breaking space). Many of Latte's other features come from Scheme (a dialect of Lisp), including: prefix notation, in which function calls are written with the operator preceding the operands; lexical scope, described in section *Note Scope::, where the binding of a variable is fixed at lexical-analysis time (i.e., when the program is parsed rather than when it runs); manifest types, which means that "type" is a property of a value rather than of a variable; quasiquoting, which aids in writing macros; and first-class functions, meaning that function objects (created with `\lambda' and `\macro'; *note Functions on functions::) can be assigned to variables or passed to and returned from other functions. File: latte.info, Node: Function and Variable Index, Prev: Pedigree, Up: Top Function and Variable Index *************************** [index] * Menu: * __FILE__: Predefined variables in latte-html. (line 11) * __latte-version__: Predefined variables. (line 7) * _bal-tag: Nonstandard HTML. (line 18) * _pre: HTML functions. (line 11) * _tag: Nonstandard HTML. (line 78) * add: Arithmetical functions. (line 14) * and: Boolean functions. (line 13) * append: Group functions. (line 7) * apply: Functions on functions. (line 15) * back: Group functions. (line 14) * caar: Group functions. (line 31) * cadr: Group functions. (line 32) * car: Group functions. (line 18) * cdar: Group functions. (line 33) * cddr: Group functions. (line 34) * cdr: Group functions. (line 23) * ceil: Arithmetical functions. (line 18) * compose: Functions on functions. (line 90) * concat: Text functions. (line 7) * cons: Group functions. (line 42) * def: Functions on variables. (line 7) * defmacro: Functions on variables. (line 59) * divide: Arithmetical functions. (line 23) * downcase: Text functions. (line 12) * empty?: Group functions. (line 47) * equal?: Other functions. (line 7) * error: Other functions. (line 20) * explode: Text functions. (line 17) * file-contents: File functions. (line 7) * floor: Arithmetical functions. (line 28) * foreach: Control structures. (line 32) * front: Group functions. (line 54) * funcall: Functions on functions. (line 7) * ge?: Arithmetical functions. (line 33) * getenv: Other functions. (line 15) * greater-equal?: Arithmetical functions. (line 45) * greater?: Arithmetical functions. (line 51) * group: Group functions. (line 58) * group?: Other functions. (line 25) * gt?: Arithmetical functions. (line 57) * html: HTML functions. (line 44) * if: Control structures. (line 7) * include: File functions. (line 35) * lambda: Functions on functions. (line 26) * le?: Arithmetical functions. (line 69) * length <1>: Group functions. (line 62) * length: Text functions. (line 22) * less-equal?: Arithmetical functions. (line 81) * less?: Arithmetical functions. (line 87) * let: Functions on variables. (line 65) * lmap: Functions on functions. (line 102) * load-file: File functions. (line 17) * load-library: File functions. (line 23) * lt?: Arithmetical functions. (line 93) * macro: Functions on functions. (line 34) * member?: Group functions. (line 67) * modulo: Arithmetical functions. (line 105) * multiply: Arithmetical functions. (line 109) * not: Boolean functions. (line 19) * nth <1>: Text functions. (line 27) * nth: Group functions. (line 72) * operator?: Other functions. (line 29) * or: Boolean functions. (line 26) * ordinary quote: Quoted expressions. (line 16) * process-output: File functions. (line 12) * push-back: Group functions. (line 77) * push-front: Group functions. (line 82) * quasiquote: Quoted expressions. (line 23) * quote: Quoted expressions. (line 16) * random: Arithmetical functions. (line 113) * rdc: Group functions. (line 87) * reverse: Group functions. (line 95) * set!: Functions on variables. (line 121) * snoc: Group functions. (line 106) * splicing unquote: Quoted expressions. (line 38) * strict-html4: Predefined variables in latte-html. (line 15) * string-append: Text functions. (line 32) * string-ge?: Text functions. (line 37) * string-greater-equal?: Text functions. (line 49) * string-greater?: Text functions. (line 55) * string-gt?: Text functions. (line 61) * string-le?: Text functions. (line 73) * string-less-equal?: Text functions. (line 85) * string-less?: Text functions. (line 91) * string-lt?: Text functions. (line 97) * string?: Other functions. (line 34) * subseq: Group functions. (line 111) * substr: Text functions. (line 108) * subtract: Arithmetical functions. (line 118) * unquote: Quoted expressions. (line 31) * unquote-splicing: Quoted expressions. (line 38) * upcase: Text functions. (line 123) * useless: Functions on variables. (line 42) * warn: Other functions. (line 38) * while: Control structures. (line 16) * zero?: Arithmetical functions. (line 123) Tag Table: Node: Top1121 Node: Introduction1505 Node: Latte examples3353 Node: Latte syntax6459 Node: Words and strings8168 Node: Variables9786 Node: Scope10663 Ref: Scope-Footnote-111671 Node: Function parameters11915 Ref: Function parameters-Footnote-116246 Node: Groups16586 Node: Quoted expressions18099 Node: Whitespace19919 Node: Latte vs. HTML22906 Node: Latte builtins26360 Node: Control structures27045 Node: Functions on variables28161 Node: Functions on functions32591 Node: Group functions36134 Node: Boolean functions39402 Node: Text functions40477 Node: Arithmetical functions44103 Node: File functions47605 Node: Other functions48972 Node: Predefined variables50139 Node: Writing HTML with Latte50405 Node: Automatic HTML processing51026 Node: HTML functions51813 Node: Character entity functions53911 Node: Balanced tag functions54520 Node: Non-balanced tag functions56266 Node: Nonstandard HTML56880 Node: Predefined variables in latte-html60311 Node: Invoking latte-html60961 Node: Makefile rules63739 Node: Latte mode for Emacs67508 Node: The latte-text translator68621 Node: Future directions69589 Node: Further information73469 Node: Pedigree75112 Node: Function and Variable Index76637 End Tag Table