Introduction
Perl's ``Plain Old Documentation'' format (Pod) is not, by any means, a perfect markup language. It lacks XML's robust features of well-formedness and unambiguous syntax. In fact, it was never intended to be anything more than a simple format for embedding documentation inside programs, as Larry Wall's comment in the perlpod man page alludes:
"...I'm not at all claiming this to be sufficient for producing a book."
Yet many Pod People --- er, I mean, authors --- want to do just that. Pod is, after all, the darling of Perl hackers everywhere. O'Reilly decided to extend it into the realm of large, complex documents, such as the typical book. This extension, called PseudoPod (thanks to Jason McIntosh for the excellent name), adds some inline tags, block structures, and rules to make processing work smoothly.
Before you go any further, you should go and read the perlpod man page, if you haven't already. It lays the foundation for PseudoPod, establishing some important rules for syntax and semantics that we rely on.
Terminology
Before we dive into the nitty-gritty details of PseudoPod, we should recap the basic concepts of Pod. The following table covers just about everything:- Block
-
A region of text that contains one or more lines, such as a
paragraph, head, code listing, or figure.
=begin foo Fa la la la la. =end foo
- Inline Segment
-
A region of text that contained inside a block, usually
within one line, with localized style properties or semantics.
I I<loved> Proust's T<Remembance of Things Past>.
- Simple Paragraph
-
Just a plain, ordinary block of text. Always unindented.
Hello there.
- Verbatim Paragraph
-
A section of code, screen output, or an equation whose space and
characters need to be preserved, verbatim. Always indented.
x + y = z - Command Paragraph
-
Start a new block style of one or more lines. Multi-line commands may
require another command to end the block.
=head1 What I Did On My Summer Vacation
- Begin Block
-
A block that contains other blocks, usually to create a
special structure, such as a figure or sidebar.
=begin note It's not a good idea to stick metal objects into a wall socket. =end note
- For Block
-
A block that denotes a special style or semantic, for
example to generate a visible comment in text.
=for production In the following paragraph, please use the Mayan petroglyph for corn in place of [corn]. =end for
Simple Paragraphs
Simple paragraphs are well-named, for they have no special formatting requirements other than to separate them with a blank line. Here are two simple paras living together in a file:
A bland, uninteresting, boring, garden-variety, nondescript, rudimentary, common, run-of-the-mill, low-key, unassuming, plain old paragraph. Another bland, uninteresting, boring, garden-variety, nondescript, rudimentary, common, run-of-the-mill, low-key, unassuming, plain old paragraph.
Simple paragraphs will not preserve special spacing. Pod formatters are allowed to throw away line breaks, adjust line width, and make any other changes to beautify the paragraphs. If you need to preserve spacing for any reason, you should use a verbatim paragraph.
Note that simple paragraphs may not be indented. Any indented line will be treated as a line of code in a separate block.
Verbatim Paragraphs
A verbatim paragraph will, unlike the simple paragraph type, preserve all spaces and linebreaks. It's distinguished by indenting every line at least one space. The amount doesn't matter, as long as subsequent lines are indented as least as much as the first line. The Pod parser will measure the first line's indentation and subtract it from following lines.Here's an example:
# generate a magic number
sub floopy {
my $x = shift;
if( $x >= 9 ) {
return ($x << 3) * 7;
} else {
return 0;
}
}
Blank lines in code are okay. The verbatim block will continue until a simple paragraph or command paragraph interrupt it.
Please do not use tabs! Tab characters are unpredictible. We will try to convert them into strings of 8 spaces, but that might not be what you want. So, to be unambiguous about it, please only use spaces.
Command Paragraphs
Command paragraphs are used for all other structures in Pod. The command takes the form of an equals sign (=) followed by some string of alphabetic characters and either a newline or a space and an optional string of data. It looks like this:
=COMMAND DATA
The equals sign must be the first character on the line or it will be interpreted as something else. If you need to place an equals sign at the beginning of a line, you should use an inline tag like "=" to encapsulate it. It's good style, but not required, to precede and follow a command paragraph with blank lines.
The data can be anything, and it depends on the context what will be done with it. A =head1 command would use data as a section title. This line would begin a table and use everything after the word ``graphic'' as the title of the table:
=begin table graphic Using array references.
So the data may contain other commands with data, several levels deep. In general, we try to keep it pretty simple, however.
Heads and Document Structure
The most common command paragraph type is the head. A head takes the form of =headN TITLE where N is a single digit from 0 to 4, inversely proportional to its level of significance and TITLE is the text to be used in the head. Level 0 is for the title of a chapter, while level 4 is a sub-sub-sub-section (also called a D-head). An A-head, or level 1 section, would look like this:
=head1 Lizard Feeding Tips
Pod books typically are split up into multiple files, each containing a single chapter, appendix, or part page (intro to a part of the book). Every PseudoPod file should follow this pattern:
=head0 CHAPTER TITLE
INTRO PARAGRAPH.
=head1 SECTION TITLE
INTRO PARAGRAPH.
=head2 SUBSECTION TITLE
INTRO PARAGRAPH.
PARAGRAPH.
line of code
line of code
line of code
line of code
line of code
...
The file should always start with a =head0 which corresponds to a chapter or appendix title. There should be only one per file. Following that is a =head1 which starts a new section. This may be followed by a =head2 and so on down to =head4, but no further than that. Please be careful to nest section levels properly. It's an error to have something line this:
=head1 A Happy Section Blah blah blah blah. =head3 A Misplaced Section! This section doesn't belong here.
So never let a =head3 follow a =head1 without an intervening =head2 or the parser will likely burst into flames. And it's bad style too.
Inline Character Tagging
Inline character tags, also known as interior sequences in Pod parlance, are a special syntactic form that delineates special treatment for character sequences inside a block. For example, to mark a word so that it has italic font style, you would do this:
This I<word> should be italicized.
In this example, the word ``word'' will be rendered in italics, while all the other words will be treated normally.
The general form of an inline tag is a single character (A-Z are currently supported), followed by a start delimiter and end delimiter. The simplest delimiters are angle brackets (< and >). But sometimes these aren't enough. If you need to enclose the character >, for example, then you have to use an alternate delimiter set so the parser won't be confused. In that case, you can use multiples of < and >, as long as the number of braces on the left match the number on the right. You can also add space between brackets and data, which will be stripped. These all do the same thing:
C<foo> C<<foo>> C<< foo >> C<<<<< foo >>>>>
We have tried to preserve all the inline tags defined in the original Pod spec. The set used by PseudoPod adds a bunch more. The following table lays out the tags and their meanings:
Tag Meaning Example
A, L Cross reference to an end See A<sect-fooby>.
point declared with Z<>
B Bold You make me B<very>
angry.
C Constant Width Set the data using the
method C<setData()>.
E Entity reference The product is x
E<times> y.
F Filename Edit the file F<.cshrc>.
G Superscript E = MCG<2>
H Subscript HH<2>O
I Italic Do I<not> eat that.
M First occuring term This phenomenon is
called M<granulation>.
N Footnote TheoreticallyN<Meaning,
"not really>, it's possible.
R Replaceable thing C< R<n> + 2 >, where
R<n> is the number of pages
S Text with non-breaking spaces
T Citation Read my book, T<Eating Meat
and Loving It>.
U URL Download the module from
U<http://www.cpan.org/>.
X Index term X<chicken, recipes for>
Z A cross reference endpoint =begin figure My Bedroom
Z<fig-br>
Some notes about the above table:
- 1.
- The "A" tag will be replaced with some generated text that references another object. Its data is a unique indentifier string that matches a "Z" tag elsewhere in the document. So, if there is an A head followed by a "Z<floof>", then you can insert "A<floof>" anywhere in the document. It will be replaced with something like ``Section 3.4, 'Gloppy Drainpipes''' or whatever makes sense in that context. This will work with sections, chapters, tables, examples, or figures.
- 2.
- In standard Pod, "E" means any escaped character, but we have taken this further. In PseudoPod, the data in "E" is the name of an XML entity, as defined in ISO-8879. So, translated into XML it would wind up as × and when translated into Unicode, it would become the multiplication symbol X. For a complete list of these entities, consult DocBook, The Definitive Guide by Walsh and Muellner, which contains a handy table in an appendix.
- 3.
- As is also true for all other inlines, the "N" tag, for footnotes, cannot contain multiple paragraphs.
- 4.
- The "X" tag will not be displayed where it's invoked. Instead, the parser will stash it away to use in building the index later. (Technically, it will still be located there after conversion, but in another form that is also invisible.) Separate primary, secondary, and tertiary terms with a comma. Start the data with the words ``See'' or ``See also'' to create an index entry that redirects to another term. Only one "X" tag per entry is allowed.
- 5.
- The "Z" tag always follows a command paragraph to which it lends its data as an identifier. Elsewhere in the file, an "A" tag will contain the same data, setting up a cross reference to that structure.
The most common type of error we see in Pod files is imbalanced delimiters in inline tags. Be wary of this! These are all errors:
C<>> C<< x - y > z > T<The Marsupial Handbook, by Joeseph Skrim, is a great...
Cross References
Cross references are easy, if you remember where to put the link endpoints. We've named the two inline tag types for internal cross references to be easy to remember. Just think, ``from A to Z''. "A<DATA>" starts a cross reference to an object with an identifer ``DATA''. It can appear inside any simple paragraph or list. "Z<DATA>" completes a cross reference by labelling the object that contains it ``DATA''. It always appears inside a structure, right after the command paragraph that starts it, like a figure or section.The following is an example, with a paragraph containing a cross reference that points to a figure:
Our escape route takes us underneath the prison wall, out into an
old apple orchard. The map is detailed in A<escape-route>.
=begin figure picture Tunnel Trajectory
Z<escape-route>
... picture here ...
=end figure
It doesn't matter if the "A" tag comes before or after the "Z" tag, nor how many times the "Z" is referenced. However, every "A" must reference an existing "Z", and no two "Z" tags can contain the same identifier. It's not a fatal error, but will cause the parser to complain and be unable to complete the link.
If you want to reference something in another file, it works the same way. However, all the files share the same namespace for identifiers. Make sure that you don't use the same identifier twice in different files or cross references will behave unpredictibly.
Lists
A list always begins with a =over command and ends with a =back command. List items start with =item and can take several forms, depending on the kind of list:- bulleted list
-
Place a star after the =item like this:
=over =item * popsicle =item * ice cream =back
Note that it doesn't matter if the item text continues on the same line as the star or begins on a new line.
- numbered list
-
Place a number after the =item. The actual value doesn't
matter, as the formatter will number items automatically, so
typically people just use the number ``1'':
=over =item 1 mount bicycle =item 1 balance on seat =item 1 put feet on pedals =item 1 pedal quickly so you don't fall over =back
Again, it doesn't matter if you continue on the same line with item text. Longer paras will probably be more readable if you use a new line.
- term-definition list
-
The term immediately follows the =item command, with definition
on a new line:
=over =item food A thing to ingest that gives you energy and tastes yummy. =item mud A thing you play with and makes your clothes all dirty. =back
Lists can be nested, and each list item can hold multiple paragraphs. But please, no tables or figures inside lists. They're icky and make the parser sad. Here's a complex example:
If you need sugar in a hurry, this list will provide some suggestions: =over =item 1 Donuts There are three principle varieties of this confection: =over =item 1 Crullers =item 1 Roundies =item 1 Jellies =back =item 1 Candy bars Even more carbohydrates packed into a convenient, tiny package. Some of my favorites are: =over =item 1 Payday (nutty) =item 1 Snickers (nougaty) =item 1 Zagnut (coconutty) =back =back
Examples
An example is simply a wrapper for something to give it a title and hook for cross references. The usual candidate for inclusion is a code listing. Here's an example:
=begin listing A Frightening Subroutine
Z<ex-scary>
for( my $i=0; $i<10; $i++ ) {
print "BOO!\n";
}
=end listing
Tables
A table is a great way to convey complex information in a compact way. Unfortunately, tables are themselves complex when it comes to markup. We realize that authors don't like to be constrained in complex markup scenarios, so we offer three ways to markup tables:- As a Pod formatted table
- As verbatim text
- As an HTML-tagged table
Here's the basic form of a table:
=begin table TYPE TITLE Z<IDENTIFIER> CONTENT =end table
The preferred method is the PseudoPod table format which has =row and =cell tags, as well as =headrow to mark the heading row and =bodyrows to mark the start of the main body of the table:
=begin table An Example Table =headrow =row =cell Header for first column (row 1, col 1) =cell Header for 2nd column (row 1, col 2) =bodyrows =row =cell Cell for row 2, col 1 =cell Cell for row 2, col 2 =row =cell Cell for row 3, col 1 =cell Cell for row 3, col 2 =end table
The second method is a freestyle, anything-goes format. Make your own columns and headers any way you want. The plus side is: infinite creativity! The downside is: we will end up recoding the whole table ourselves in production which slows us down a bit. Just try to make it clear what is in which column and on what row, and it will be okay with us.
The table markup is a wrapper, similar to examples, but with an extra keyword, TYPE, to denote the kind of table. If the keyword is ``html'', our parser will read it like an HTML table. If it's anything else, like ``graphic'' or ``picture'', it will be treated as an ASCII rendering. For example:
=begin table picture Comparing Camels to Horses.
Z<camel-horse-chart>
CAMEL HORSE
Lives in desert Lives in grassland
Bumpy Smooth
Spits Kicks
=end table
Figures
A figure holds some kind of picture, whether an imported graphic or an ASCII masterpiece you paint yourself. The general form is similar to a table's:
=begin figure TYPE TITLE Z<IDENTIFIER> CONTENT =end figure
If the TYPE is ``graphic'', then the parser will expect to see a reference to an external file, the name of a graphic to import. For example:
=begin figure graphic My Hairstyle Z<fig-hair> F<figs/myhair.gif> =end figure
For any other value of TYPE, the parser assumes you drew your own lovely diagram in text:
=begin figure pikchur My Hairstyle
Z<fig-hair>
\ | / //
\ \ - / \
\ | /
(o o) I look like a pineapple!
( < )
(===)
=end figure
Other Structures
The rest of this tutorial is a mixed bag of oddball stuff you can drop in your book.Comments
If you want to leave a comment in the manuscript for somebody to see, you can use the =for command. The data in the command specifies who the comment is for. Typical designations include ``production'', ``author'', and ``editor''. For instance:
=for editors My misspelling of 'antidisestablishmentarianism' in the following paragraph is intentional. =end for
Literal Layouts
If you want something to be treated like a verbatim paragraph, but not rendered in constant width font, then use =begin literal. Here's a bit of poetry done up like that:
=begin literal As I was going up the stair, I met a man who was not there. He wasn't there again today; I wish that man would go away. =end literal
Footnotes
Footnotes use the "N" inline tag to locate them inside blocks. Strangely, they also function like blocks in that they can be many lines long. The following example shows how you might use them:
O'Reilly books often contain footnotes,N<Perl books especially, it would seemN<Though I wouldn't go so far as to say they I<overdo> it.>.> though I believe that house style limits them to three per page.N<Uh oh.>
Epigraphs
For an epigraph, just wrap it up in a =begin epigraph command like this:
=begin epigraph Great art must be licked. --Jas W Felter, Mail Artist =end epigraph
Author Information
If you need to specify the author's name for a particular chapter or article, use the =author command:
=author Ferdinand Buscaglia
History
PseudoPod was originally the brainchild of Jason ``J-mac'' McIntosh who worked for O'Reilly's Publishing Tools Group. It was used for many venerable Perl books, including several revisions of the Camel and Llama. The Pod::PseudoPod modules were developed by Allison Randal while writing a book for O'Reilly as an easy-to-use and easy-to-install alternative to O'Reilly's internal PseudoPod parsing and formatting tools, to allow authors to check their formatting before submitting files.As of 2007, PsuedoPod is no longer used or maintained by O'Reilly, after they switched all their manuscripts to DocBook. Pod::PseudoPod is now the primary parser for PseudoPod, and includes Pod::PseudoPod::DocBook to generate DocBook output suitable for submission to O'Reilly.