▼Scribble: Racket Documentation Tool
►1 Getting Started
On this page:
1.1 A First Example
1.2 Multiple Sections
1.3 Splitting the Document Source
1.4 Document Styles
1.5 More Functions
1.6 Text Mode vs. Racket Mode for Arguments
1.7 @ Syntax Basics
1.8 Decoding Sequences
1.9 Next Steps
  top← prev  up  next → 

1 Getting Started

No matter what you want to do with Scribble, it’s best to start by generating a few simple HTML and/or PDF documents. This chapter steps you through the basics, and it ends in Next Steps with goal-specific advice on how to continue.

1.1 A First Example

Create a file "mouse.scrbl" with this content:

#lang scribble/base
 
@title{On the Cookie-Eating Habits of Mice}
 
If you give a mouse a cookie, he's going to ask for a
glass of milk.

The first line’s #lang scribble/base indicates that the file implements a Scribble document. The document starts in “text mode,” and the @ character escapes to operators like title, where the curly braces return to text mode for the arguments to the operator. The rest is document content.

Now run the scribble command-line program, specifying a mode for the kind of document that you want as output:

1.2 Multiple Sections

Add more text to "mouse.scrbl" so that it looks like this:

#lang scribble/base
 
@title{On the Cookie-Eating Habits of Mice}
 
If you give a mouse a cookie, he's going to ask for a
glass of milk.
 
@section{The Consequences of Milk}
 
That ``squeak'' was the mouse asking for milk. Let's
suppose that you give him some in a big glass.
 
He's a small mouse. The glass is too big---way too
big. So, he'll probably ask you for a straw. You might as
well give it to him.
 
@section{Not the Last Straw}
 
For now, to handle the milk moustache, it's enough to give
him a napkin. But it doesn't end there... oh, no.

Now, after the first paragraph of the paper, we have two sub-sections, each created by calling section to generate a sub-section declaration. The first sub-section has two paragraphs. The second section, as initiated by the result of the second section call, has a single paragraph.

Run the scribble command(s) from A First Example again. You may notice the curly double-quotes in the output, and the --- turned into an em dash.

1.3 Splitting the Document Source

As a document grows larger, it’s better to split sections into separate source files. The include-section operation incorporates a document defined by a ".scrbl" file into a larger document.

To split the example document into multiple files, change "mouse.scrbl" to just

#lang scribble/base
 
@title{On the Cookie-Eating Habits of Mice}
 
If you give a mouse a cookie, he's going to ask for a
glass of milk.
 
@include-section["milk.scrbl"]
@include-section["straw.scrbl"]

Create "milk.scrbl" and "straw.scrbl" in the same directory as "mouse.scrbl". In "milk.scrbl", put

#lang scribble/base
 
@title{The Consequences of Milk}
 
That ``squeak'' was the mouse asking for milk...

and in "straw.scbl", put

#lang scribble/base
 
@title{Not the Last Straw}
 
For now, to handle the milk moustache, ...

Notice that the new files both start with #lang, like the original document, and the sections from the original document become titles in the new documents. Both "milk.scrbl" and "straw.scrbl" are documents in their own right with their own titles, and they can be individually rendered using scribble. Running scribble on "mouse.scrbl", meanwhile, incorporates the smaller documents into one document that is the same as before.

1.4 Document Styles

Scribble currently supports only one form of HTML output. You can replace the "scribble.css" file for the generated pages, and that’s about it. (We expect to add more styles in the future.)

For Latex-based PDF output, Scribble includes support for multiple page-layout configurations. The "mouse.scrbl" example so far uses the default Latex style. If you plan on submitting the paper to a workshop on programming languages, then—well, you probably need a different topic. But you can start making the current content look right by changing the first line to

#lang scribble/sigplan

If you’re instead working toward Racket library documentation, try changing the first line to

#lang scribble/manual

which produces output with a separate title page, initial content on that page (intended as a brief orientation to the document), and top-level sections turned into chapters that each start on a new page. If you have split the document into multiple files, the first line of the main document file determines the output format.

Using scribble/sigplan or scribble/manual does not change the rendered HTML for a document—aside from scribble/manual adding a version number—but it changes the set of bindings available in the document body. For example, with scribble/sigplan, the introductory text can be marked as an abstract:

#lang scribble/sigplan
 
@title{On the Cookie-Eating Habits of Mice}
 
@abstract{If you give a mouse a cookie, he's going to
          ask for a glass of milk.}
 
@section{The Consequences of Milk}
 
....

When rendered as HTML, the abstract shows up as an inset paragraph. If you try to use abstract with the scribble/base or scribble/manual language, then you get an error, because abstract is not defined.

When a document is implemented across multiple files, changing the language of the main document can set the style for all of the parts, but it does not introduce bindings into the other part files. For example, if you change the language of "mouse.scrbl" to scribble/sigplan, then abstract becomes available in "mouse.scrbl" but not in "milk.scrbl" or "straw.scrbl". In other words, operator names are lexically scoped.

1.5 More Functions

The scribble/sigplan and scribble/manual languages are supersets of the scribble/base language, which provides a collection of basic operations. Many of the operations are style variations that you can apply to text:

He's a @smaller{small mouse}. The glass is too
@larger{big}---@bold{way @larger{too @larger{big}}}. So, he'll
@italic{probably} ask you for a straw.

which renders as

He’s a small mouse. The glass is too big—way too big. So, he’ll probably ask you for a straw.

As you would expect, calls to functions like smaller, larger, and bold can be nested in other calls. They can also be nested within calls to title or section:

@section{@italic{Not} the Last Straw}

The centered operation centers a flow of text:

If a mouse eats all your cookies, put up a sign that says
@centered{
  @bold{Cookies Wanted}
 
  @italic{Chocolate chip preferred!}
}
and see if anyone brings you more.

which renders as

If a mouse eats all your cookies, put up a sign that says

Cookies Wanted

Chocolate chip preferred!

and see if anyone brings you more.

The margin-note operation is used in a similar way, but the rendered text is moved to the margins.

If you use margin-note, then the content shows up over here.

The itemlist operation creates a sequence of bulleted text, where the item operation groups text to appear in a single bullet. The itemlist operation is different from the others that we have seen before, because it only accepts values produced by item instead of arbitrary text. This difference is reflected in the use of [...] for the arguments to itemlist instead of {...}:

@centered{@bold{Notice to Mice}}
 
@itemlist[@item{We have cookies for you.}
          @item{If you want to eat a cookie,
                you must bring your own straw.}]

which renders as

Notice to Mice

  • We have cookies for you.

  • If you want to eat a cookie, you must bring your own straw.

1.6 Text Mode vs. Racket Mode for Arguments

When [...] surrounds the arguments of an operation, the argument expressions are in Racket mode rather than text mode. Even in Racket mode, @ can be used to apply operations; once the @ syntax is enabled through a language like scribble/base (as opposed to racket), it behaves the same in both Racket mode and text mode.

One advantage of using Racket mode for the arguments to itemlist is that we can pass a keyword-tagged optional argument to itemlist. In particular, if you want a list with numbers instead of bullets, supply the 'ordered style to itemlist using the #:style keyword:

@itemlist[#:style 'ordered
          @item{Eat cookie.}
          @item{Drink milk.}
          @item{Wipe mouth.}
          @item{...}]

An operation doesn’t care whether it’s used with [...] or {...}. Roughly, {...} forms an argument that is a string. (Only roughly, though. Newlines or uses of @ within {...} complicate the picture, and we’ll get back to that soon.) So,

@italic{Yummy!}

is equivalent to

@italic["Yummy!"]

which is equivalent to the Racket expression

(italic "Yummy!")

These equivalences explain why Scribble functions are documented in Racket notation. If you’re reading this in HTML format, you can click italic above to access its documentation. The documentation won’t completely make sense, yet, but it will by the end of this chapter.

What if you want to provide arguments in text mode, but you also want to supply other optional arguments? You can use both [...] and {...} for an operation, as long as the [...] is first, and as long as no character separate the closing ] from the opening {. For example, calling italic is the same as using elem with the 'italic style:

@elem[#:style 'italic]{Yummy!}

You can also omit both [...] and {...}. In that case, the Racket expression after @ is used directly instead of applied as an operation. For example,

1 plus 2 is @(number->string (+ 1 2)).

renders as

1 plus 2 is 3.

The call to number->string is needed because a naked number is not valid as document content.

1.7 @ Syntax Basics

The @ notation provided by Scribble is just another way of writing Racket expressions. Scribble documents could be constructed using normal Racket notation, without using @ at all, but that would be inconvenient for most purposes. The @ notation makes dealing with textual content much easier.

Whether in text mode or Racket mode, @ in a document provides an escape to Racket mode. The basic syntax of @ is

@ ‹cmd› [ ‹datum›* ] { ‹text-body› }

where all three parts after @ are optional, but at least one must be present. No spaces are allowed between

A ‹cmd› or ‹datum› is normal Racket notation, while a ‹text-body› is itself in text mode. A ‹cmd› obviously must not start with [ or {, even though Racket forms could otherwise start with those characters.

The expansion of just @‹cmd› into Racket code is

‹cmd›

When either [ ] or { } are used, the expansion is

(‹cmd› ‹datum›* ‹parsed-body›*)

where ‹parsed-body›* is the parse result of the ‹text-body›. The ‹parsed-body›* part often turns out to be a sequence of Racket strings.

In practice, the ‹cmd› is normally a Racket identifier that is bound to a procedure or syntactic form. If the procedure or form expects further text to typeset, then {...} supplies the text. If the form expects other data, typically [...] is used to surround Racket arguments, instead. Even if an operation’s argument is a string, if the string is not used as content text (but instead used as, say, a hyperlink label), then the string is typically provided through [...] instead of {...}. Sometimes, both [...] and {...} are used, where the former surround Racket arguments that precede text to typeset. Finally, if a form is a purely Racket-level form with not typeset result, such as a require to import more operations, then typically just @ is used.

For example the text-mode stream

@(require scriblib/figure)
 
@section[#:tag "poetry"]{Of Mice and Cookies}
See @secref["milk"].
 
@section[#:tag "milk"]{@italic{Important} Milk Supplies}
@figure["straw" @elem{A straw}]{@image["straw.png"]}

is equivalent to the Racket-mode sequence

(require scriblib/figure) "\n"
"\n"
(section #:tag "poetry" "Of Mice and Cookies") "\n"
"See " (secref "milk") "." "\n"
"\n"
(section #:tag "milk" (italic "Important") " Milk Supplies") "\n"
(figure "straw" (elem "A straw") (image "straw.png")) "\n"

Besides showing how different argument conventions are used for different operations, the above example illustrates how whitespace is preserved in the Racket form of a text-mode stream—including newlines preserved as their own strings. Notice how the second section gets two arguments for its content, since the argument content for section in the source stream includes both the use of an operator and additional text. When an operation like section or italic accepts content to typeset, it normally accepts an arbitrary number of arguments that together form the content.

For more information on the syntax of @, see @ Syntax. The full syntax includes a few more details, such as brackets like |{...}| for text-mode arguments while disabling @ between the brackets.

1.8 Decoding Sequences

In a document that starts #lang scribble/base, the top level is a text-mode stream, just like the ‹text-body› in a @ form. As illustrated in the previous section, such a top-level sequence corresponds to a mixture of Racket-mode strings and operation applications. There’s an implicit operation, decode, that wraps the whole document to consume this mixture of strings and other values and turn them into a document description.

The decode operation implements flow decoding, which takes a document stream and breaks it up into sections and paragraphs. Blank lines delimit paragraphs, and the results of operations like title and section generate “here’s the title” or “a new section starts here” declarations that are recognized by decode.

A different but related content decoding takes place within a paragraph or section title. Content decoding is responsible for converting --- to an em dash or for converting " and ' to suitable curly quotes.

The decoding process for document’s stream is ultimately determined by the #lang line that starts the document. The scribble/base, scribble/manual, and scribble/sigplan languages all use the same decode operation. The scribble/text language, however, acts more like a plain-text preprocessor and it does not perform any such decoding rules. (For more on scribble/text, see Text Preprocessing.)

More precisely, languages like scribble/base apply decode only after lifting out all definitions and imports from the document stream.

When the flow decoder is used, after it breaks the input stream into paragraphs, it applies content decoding to strings within the paragraph. When content is wrapped with an operation, however, content decoding does not apply automatically. An operation is responsible for calling a content or flow decoder as it sees fit. Most operations call the decoder; for example, italic, bold, smaller, etc., all decode their arguments. Similarly, title and section decode the given content for the title or section name. The literal and verbatim operators, however, do not decode the given strings. For example,

@verbatim{---}

renders as

---

Don’t confuse decoding with the expansion of @ notation. The source form

@verbatim{@(number->string (+ 1 2))}

renders as

3

because the source is equivalent to

(verbatim (number->string (+ 1 2)))

where (number->string (+ 1 2)) is evaluated to produce the argument to verbatim. The |{...}| style of brackets is often used with verbatim, because |{...}| disables @ notation for arguments. For example,

@verbatim|{@(number->string (+ 1 2))}|

renders as

@(number->string (+ 1 2))

1.9 Next Steps

If your immediate goal is to document a Racket library or write literate programs, skip to Getting Started with Documentation, and then go back to @ Syntax and other chapters.

If you are more interested in producing documents unrelated to Racket, continue with @ Syntax and then High-Level Scribble API. Move on to Low-Level Scribble API when you need more power.

If you are interested in text preprocessing, contiue with @ Syntax, but then skip to Text Preprocessing.

  top← prev  up  next →