pxsl-tools: Parsimonious XML Shorthand Language--to-XML compiler

[ program, xml ] [ Propose Tags ]

The package contains the pxslcc compiler, which converts Parsimonious XML Shorthand Language (PXSL) into XML. PXSL ("pixel") is a convenient shorthand for writing markup-heavy XML documents. It provides XML authors and programmers with a simple, concise syntax that they can use to create XML documents. For more advanced users, PXSL offers customizable shortcuts and sophisticated refactoring tools like functional macros that can markedly reduce the size and complexity of markup-dense XML documents.

[Skip to Readme]


Maintainer's Corner

Package maintainers

For package maintainers and hackage trustees


  • No Candidates
Versions [RSS] 1.0.1
Dependencies base, containers, mtl (>=1.0), parsec (>=2.0) [details]
License LicenseRef-GPL
Copyright 2003-2008 Thomas G. Moertel <tom@moertel.com>
Author Tom Moertel <tom@moertel.com>
Maintainer Tom Moertel <tom@moertel.com>
Category XML
Home page http://community.moertel.com/ss/space/PXSL
Uploaded by TomMoertel at 2008-08-25T02:47:44Z
Distributions NixOS:1.0.1
Reverse Dependencies 1 direct, 0 indirect [details]
Executables pxslcc
Downloads 1425 total (5 in the last 30 days)
Rating (no votes yet) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Docs not available [build log]
Last success reported on 2017-01-03 [all 8 reports]

Readme for pxsl-tools-1.0.1

[back to package description]
                        ____  _  _  ___  __
                       (  _ \( \/ )/ __)(  )
                        )___/ )  ( \__ \ )(__
                       (__)  (_/\_)(___/(____)


                         Updated 2008-02-16

PXSL ("pixel") is a convenient shorthand for writing markup-heavy XML
documents.  The following document explains why PXSL is needed and
shows you how to use it.  For additional information, such as the FAQ
list, visit the community site:


You'll get more out of this document if you read it from start to
finish, but you can stop anywhere after the "Gentle Introduction to
PXSL" and be able to take advantage of PXSL in your documents.  The
later sections explain PXSL's advanced features.  If you're willing to
invest some time in learning them, you will have at your disposal new
and powerful ways to create and refactor XML documents.  The advanced
features are more complicated to master, but they can greatly reduce
the complexity of your documents.

* Table of Contents

  * Getting PXSL
  * Getting help
  * License
  * Getting or building the PXSL tools
  * Gentle Introduction to PXSL
    - Why PXSL ?
    - A closer look at PXSL
    - Using PXSL documents
  * Advanced topics
    - Element defaults provide convenient, customizable shortcuts
        Using element defaults to create attribute shortcuts
        Using element defaults to create virtual elements
        Making and using your own element defaults
        Built-in element defaults for XSLT stylesheets
    - Advanced quoting with << >> and <{ }>
    - Macro facility
    - Tip: store frequently used macros in reusable .pxsl files
    - Advanced macros and passing parameters with the <( )> delimiters
    - More advanced macros and functional programming
    - Automatic PXSL-to-XML conversion via Make
  * Reference:  pxlscc
  * Reference:  PXSL syntax
  * Authors

* Getting PXSL

The most-recent official version of the PXSL tools can always be found


The PXSL tools have been also packaged for Debian (thanks Kari Pahula)
and Red Hat / Fedora.

By the way, you pronounce PXSL like "pixel".

* Getting help

If you need help with PXSL, there is a discussion site for PXSL
users and developers.  Feel free to ask questions and leave your

    PXSL Community Forum



Copyright (C) 2003--2008 Thomas Moertel & Bill Hubauer.

The PXSL toolkit is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License as
published by the Free Software Foundation; either version 2 of
the License, or (at your option) any later version.

The text of the GNU GPL may be found in the LICENSE file,
included with this software.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
GNU General Public License for more details.

Except as provided for under the terms of the GNU GPL, all rights
are reserved worldwide.

* Getting or building the PXSL tools

If you don't want to build the PXSL tools from source code, you can
download one of the pre-built binary packages on the PXSL web site.
The PXSL tools have been also packaged for Debian (thanks Kari Pahula)
and Red Hat / Fedora.  You might want to search your local package
repositories before building from source.

If a binary package isn't available for your computing platform of
choice, you can use the following procedure to build the PXSL tools
for your platform.

In order to build the tools you will need the following:

  - A PXSL source tarball (typically named "pxsl-tools-{version}.tar.gz")

  - A Haskell compiler supporting the Cabal build system.
    I use GHC:  http://www.haskell.org/ghc/

Just uncompress the tarball and build the project using the following
commands.  If you want to install a personal copy of pxslcc instead of
the doing the default, system-wide installation, uncomment the extra
command-line flags on the third command.

    $ tar zxvf pxsl-tools-{version}.tar.gz
    $ cd pxsl-tools-{version}
    $ runhaskell Setup.lhs configure # --user --prefix=$HOME
    $ runhaskell Setup.lhs build
    $ runhaskell Setup.lhs install

(Replace {version} with the version of PXSL that you downloaded.)

That's it.  You should now have a fully functional version of pxslcc.

** RPMs

If you are on a Red Hat or Fedora  Linux system (or a similar
RPM-based distribution), you can probably find RPMS and SRPMS here:


Otherwise, you can build custom RPMs using the cabal-rpm tool:


* Gentle Introduction to PXSL

PXSL ("pixel") is a convenient shorthand for writing markup-heavy XML
documents.  This introduction assumes that you are familiar with XML.
If you want a refresher, see the introductions on XML available here:


** Why PXSL ?

XML is a descendant of the markup language SGML and inherits its
ancestor's historical bias toward marking up textual documents.
However, XML is becoming an increasingly popular medium for the
representation of non-textual information such as metadata (RSS, XSD,
RELAX-NG), remote procedure calls (SOAP), and even information
that looks much like programming languages (XSLT, SVG, MathML).
For these uses, XML's text-centric syntax gets in the way.

Consider, for example, this snippet of MathML:

    MathML in XML

    <declare type="fn">
      <ci> f </ci>
        <bvar><ci> x </ci></bvar>
            <ci> x </ci>
            <cn> 2 </cn>
          <ci> x </ci>
          <cn> 3 </cn>

Notice something about MathML's structure: There is more markup than
text.  In fact, the only text in the snippet is "f x x2 x3"; the rest
is markup.  As you can see above, XML's document-centric style of
markup, in which the markup is delimited from the flow of surrounding
text, becomes a hindrance when markup is in the majority.

PXSL, in contrast, was designed specifically to handle this case well.
It makes dense markup easy because it assumes that everything is
markup to begin with.  You need only delimit the few portions of text
that are mixed into the flow of surrounding markup.

In other words, PXSL is what you get when you turn XML inside out:

    XML                             PXSL

    <markup>text</markup>           markup <<text>>

Let's see how this inside-out transformation simplifies our MathML
example from above:

    MathML in XML                   MathML in PXSL

    <declare type="fn">             declare -type=fn
      <ci> f </ci>                    ci << f >>
      <lambda>                        lambda
        <bvar><ci> x </ci></bvar>       bvar
        <apply>                           ci << x >>
          <plus/>                       apply
          <apply>                         plus
            <power/>                      apply
            <ci> x </ci>                    power
            <cn> 2 </cn>                    ci << x >>
          </apply>                          cn << 2 >>
          <ci> x </ci>                    ci << x >>
          <cn> 3 </cn>                    cn << 3 >>

There are two things to notice about the PXSL version in comparison
to the XML version.  First, the PXSL version is shorter.  Second, and
most important, PXSL is comparatively free of noisy characters like < >
/ and ".  In PXSL, noise is the exception rather than the rule.

** A closer look at PXSL

Writing PXSL is simple.  If you know how to write XML, you can write
PXSL.  In fact, PXSL is XML, just written in a different, inside-out
syntax.  Let's see how it works by way of comparison.

First, every XML document and hence every PXSL document has a root
element.  Here is a tiny document that has a root element and nothing

    XML                             PXSL

    <doc/>                          doc

If the document contains other elements, they are simply placed
underneath the root element, but indented to indicate that the root
element contains them.  In XML this indenting is optional, but most
people do it anyway because it is an established practice that makes
documents easier to understand.  In PXSL however, the indenting is
mandatory because indentation determines which elements contain
others.  (This requirement is what enables PXSL to do away with the
closing tags that XML uses to determine containment.)

    <doc>                           doc
      <title/>                        title
      <body/>                         body

If an element has attributes, they are written in the form of
-name=value in PXSL.

    <doc>                           doc
      <title/>                        title
      <body id="db13"/>               body -id=db13

If an attribute value contains whitespace, it must be quoted within
the literal delimiters << and >>.

    <doc keywords="x y z">          doc -keywords=<<x y z>>
      <title/>                        title
      <body id="db13"/>               body -id=db13

Now let's consider text.  If an element contains text, the text is
quoted in << and >> delimiters and indented underneath the element
that owns the text.

    <doc keywords="x y z">          doc -keywords=<<x y z>>
      <title/>                        title
      <body id="db13">                body -id=db13
        This is text.                   <<This is text.>>

The << and >> delimiters are powerful.  The text within them,
including all whitespace, is quoted verbatim.  The text can span
multiple lines and even stray outside of the outline-like indentation
hierarchy.  If you place sections of quoted text next to one another
<<like>> <<so>> they effectively become one section <<likeso>>.

    <doc keywords="x y z">          doc -keywords=<<x y z>>
      <title>                         title
        My title                        <<My title>>
      </title>                        body -id=db13
      <body id="db13">                  <<This is multi-
        This is multi-                  line text.>>
        line text.

If you want to add an XML comment, introduce it with the -- delimiter.
The comment extends to the end of the line.

    <!-- my document -->            -- my document
    <doc keywords="x y z">          doc -keywords=<<x y z>>
      <title>                         title
        My title                        <<My title>>
      </title>                        body -id=db13
      <body id="db13">                  <<This is multi-
        This is multi-                  line text.>>
        line text.

You can also use the # delimiter, which creates a PXSL comment that is
invisible in XML:

    <!-- my document -->            -- my document
    <doc keywords="x y z">          doc -keywords=<<x y z>>
      <title>                         title
        My title                        <<My title>>
      </title>                        body -id=db13
      <body id="db13">                  <<This is multi-
        This is multi-                  line text.>>
        line text.
      </body>                       # hidden comment, for
    </doc>                          # PXSL readers only

That's it.  You now know everything necessary to create PXSL

PXSL lets you do more, however, and if you want to take full advantage
of it, read the Advanced Topics section.  For now, though, let's
consider how to use PXSL documents with your existing XML-based

** Using PXSL documents

Using PXSL documents is easy because they are really XML documents in
disguise.  (In fact, you may wish to consider PXSL as a convenient
shorthand for writing XML.)  Any program that can read XML can handle
PXSL.  All you need to do is remove the disguise first so that the
programs will recognize your documents for what they are.

The included tool pxlscc (short for PXSL conversion compiler) performs
this task.  Just feed it a PXSL document, and it returns the
equivalent plain-old XML document:

    $ pxlscc document.pxsl > document.xml

You can then use the returned document in your XML-aware programs.

If you know how to use Make or Ant or similar tools, you can easily
automate this process so that your PXSL files are automagically
converted into XML when needed.

NOTE:  The pxslcc program expects UTF-8 encoded input and emits UTF-8
encoded output.

* Advanced topics

The following sections describe the more advanced capabilities of PXSL
that can make your life easier.  The element defaults, in particular,
can significantly reduce markup burdens.

** Element defaults provide convenient, customizable shortcuts

Most XML documents conform to established vocabularies.  Once you
become familiar with your documents' vocabularies, you'll probably
find that certain elements and attributes always or often occur
together -- to the point where typing them becomes repetitive.  For
example, in XHTML, almost all img elements take the following form:

    <img src="..." alt="..." [ additional attributes here ] />

Or, in PXSL:

    img -src=... -alt=... [ additional attributes here ]

So, why should you have to type in the repetitive src="" and alt=""
every time you use an img element?  With PXSL's element defaults,
you don't need to.

*** Using element defaults to create attribute shortcuts

Element defaults are shortcuts that are defined in a separate file
using a simple syntax.  (For the specifics of creating and loading
these files, see the Reference section on pxslcc.)  For example:

    img = img src alt

This shortcut allows you optionally to leave off the -src= and -alt=
part whenever you write the PXSL markup for an img element.  For
example, with this definition in place, all three of these PXSL
statements mean the exact same thing:

    img -src=/images/logo.png -alt=Logo
    img /images/logo.png -alt=Logo
    img /images/logo.png Logo

All of them convert into the same XHTML:

    <img src="/images/logo.png" alt="Logo"/>
    <img src="/images/logo.png" alt="Logo"/>
    <img src="/images/logo.png" alt="Logo"/>

In other words, shortcuts let you pass attribute values by position
instead of by the -name=value syntax.  You provide only the values,
and the shortcut provides the corresponding -name= parts behind the

But there are a couple of restrictions to keep in mind.  First,
attribute values passed by position must come first, before any values
passed using the -name=value syntax, and they must occur in the same
order as declared in the shortcut definition.

Second, you can only pass values this way if they do not contain
whitespace.  If a value contains whitespace, you must use the
-name=value syntax and quote the value: -name=<<my value>> (There is
an advanced feature, the <( )> delimiters, that overcome this
restriction.  They are described in the section on advanced macros,
later in this document.)

*** Using element defaults to create virtual elements

You can also use the element defaults to create your own virtual
elements.  If you work in XHTML, you have probably noticed that the
<a> element is used to create both hypertext links and anchors.  For

   <a name="anchor-name">Anchored text</a>
   <a href="#anchor-name">Link to anchored text</a>

Why not make these two uses more obviously distinct while cutting down
on markup at the same time?  Let's create virtual "anchor" and "hlink"
elements that do just that:

   anchor = a name
   hlink = a href

Now we can use these elements in PXSL to express the above XHTML more

   anchor anchor-name <<Anchored text>>
   hlink #anchor-name <<Link to anchored text>>

(Notice that we used << and >> in an advanced way that lets us put
quoted text on the same line as the element that contains it.  This is
discussed further in the "Advanced quoting" section.)

When we convert the above PXSL into XML, it results in exactly the
same XHTML that we discussed earlier:

   <a name="anchor-name">Anchored text</a>
   <a href="#anchor-name">Link to anchored text</a>

*** Making and using your own element defaults

Making your own shortcuts is easy.  Just create a file that contains
lines of this form:

    element-name = preferred-element-name opt-attr-1 opt-attr-2 ...

It's a good idea to extend the file's name with a suffix of ".edf",
which is short for "element defaults," but feel free to ignore this
convention.  (Note: element defaults are *not* PXSL macros.  If you
want to create a file that contains commonly used macros, just save
them in a regular .pxsl file and include it by mentioning it on the
command line; the .edf suffix is for element defaults only.  See "Tip:
store frequently used macros in separate .pxsl files" for more.)

For example, we might create a "xhtml-shortcuts.edf" file to capture
our shortcuts from above:

    # File: xhtml-shortcuts.edf

    anchor = a name
    hlink = a href

(Notice that you can place comment lines in your .edf files by
starting them with a "#" character.)

To use the shortcuts, tell pxslcc to --add them to the set of active
element defaults that are used when processing your PXSL files:

    $ pxslcc --add=xhtml-shortcuts.edf my-doc.pxsl > my-doc.xhtml

You can --add more than one set of defaults, and pxslcc will use them

*** Built-in element defaults for XSLT stylesheets

PXSL was originally created to reduce the verbosity of XSLT
stylesheets.  As a result, pxslcc has a built-in set of element
defaults for XSLT that you can enable by passing the --xslt

    $ pxslcc --xslt stylesheet.pxsl > stylesheet.xsl

The built-in defaults provide two benefits: First, you can use element
names from within the XSLT namespace without having to use the xsl:
prefix.  Second, you can pass common required attributes like "select"
and "match" by position.

Together, these benefits result in massive markup reductions, making
your life as an XSLT author much easier.  Compare the following
snippet of XSLT in XML

    <xsl:template match="/">
      <xsl:for-each select="//*/@src|//*/@href">
        <xsl:value-of select="."/>

with the same snippet rewritten in PXSL (using --xslt defaults):

    template /
      for-each //*/@src|//*/@href
        value-of .
        text <<&#10;>>

Among the many XSLT shortcuts enabled by the --xslt flag, the above
PXSL snippet uses the following:

    template = xsl:template match name
    for-each = xsl:for-each select xml:space
    value-of = xsl:value-of select disable-output-escaping
    text     = xsl:text disable-output-escaping

To see the complete list of XSLT shortcuts, --export them:

    $ pxslcc --xslt --export

** Advanced quoting with the << >> and <{ }> delimiters

PXSL has two kinds of quoting delimiters that can be used to quote
mixed and text-only content.  Both are described in this section.

*** XML quoting << >> delimiters

The << and >> delimiters not only let you insert text into your PXSL
documents, but also let you insert raw, full-featured XML.  This works
great for those times when it's just easier to write a bit of XML than
its PXSL equivalent.  For example, if you're writing an XSLT
stylesheet that generates XHTML output, you'll certainly want to use
PXSL to express the markup-dense xsl:stylesheet directives.  But, if
you need to drop in some XHTML boilerplate that a designer gave you to
use in the page footer, just copy-and-paste it using << and >>:

       <div class="footer">
       Copyright (C) 2003 Blah, Blah, Blah, Inc.
       <!--  lots more boilerplate ... -->

Another great use for the << >> delimiters is to drop XML specials
like processing instructions into your code:

    <<<?xml version="1.0" encoding="ISO-8859-1"?>>>

The above PXSL is equivalent to the following XML:

    <?xml version="1.0" encoding="ISO-8859-1"?>

Because the << >> delimiters quote XML, you must follow XML's
syntactical rules when using them.  That means that if you
want to include literal less-than "<" and ampersand "&"
characters, you must use character entity references:

    << less-than: &lt; >>
    << ampersand: &amp; >>

*** Verbatim text <{ }> delimiters (CDATA)

When copy-and-pasting blocks of text from outside sources, you must be
careful to "escape" any literal "<" and "&" characters that may be
within.  This can be annoying, especially for large blocks of text.
Another place where this requirement is burdensome is in mathematical
expressions that sometimes occur in XSLT markup:

    xsl:test -when=<< $i &lt; 5 >>

For this reason, PXSL provides the verbatim-text delimiters <{ and }>
that perform the same function as XML's more verbose CDATA delimiters:

    XML                                   PXSL

    <![CDATA[ toast & jelly ]]>           <{ toast & jelly }>

Any characters that you place inside of <{ }> will come out as a
character literals.  PXSL will take care of any escaping that is
necessary to prevent XML from misinterpreting your text as markup.
For example, we can rewrite the above XSLT snippet more clearly
using the verbatim-text delimiters:

    xsl:test -when=<{ $i < 5 }>

These delimiters are especially handy for including examples of XML
markup in your documents.  Like << >>, <{ }> can handle large blocks
of multi-line text and preserves whitespace and indentation.

*** Text-content shortcut

As you may have noticed from the MathML example at the beginning of
this document, if an element contains text, you can declare the
text on the same line as the element.  This saves space and often
reads more easily:

    NORMAL                         SHORTCUT

    h1                             h1 <<Chapter 1>>
      <<Chapter 1>>                h2 <{Sections 1 & 2}>
      <{Sections 1 & 2}>

** Macro facility

PXSL has a simple macro facility that you can use to reorganize your
markup and "factor out" boilerplate.  A macro is defined with a
leading comma and a trailing equals sign, like so:

    ,name =

where "name" is the name of the macro and "body-of-the-macro" can be
any series of elements and text.  Macros can be defined at any level
of nesting within a PXSL document, but they are only visible (i.e.,
available for use) at the level where they were defined and at levels
nested underneath.  (If two macros with the same name are visible at
the same time, the deepest one will hide the other, or if both are on
the same level, the one defined latest in the document will hide the
earlier.  In other words, the policy is "the last, deepest one wins.")

*** Using macros (i.e., macro expansion)

To use a macro, simply refer to it by name anywhere that an element
can be used:

    ,hello =

        <<Hello! Again!>>

When processed with pxslcc (using the --indent flag), this is the

      <body>Hello! Again!</body>

Note that the macro definition has been removed and that the reference
to the macro inside of the "title" element has been replaced by the
macro's body.  This is called macro expansion.

Macros don't need to be defined before they are expanded, as long as
they are visible from the sites (locations) where they are expanded.
Also, macros can call other macros:

    ,hello =


      ,head =

      ,body =
        body <<Hello! Again!>>

This snippet results in exactly the same XML as the one above.
Nevertheless, we have made a number of organizational changes.
The "head" and "body" within the "html" element have been
factored out into the macros ,head and ,body and relocated within
the document.  These macros are defined within the "html"
element, after the sites where they are expanded.  Note that the
,head macro calls upon the ,hello macro that we defined earlier.

Although contrived in this small example, factoring out blocks of
markup makes the structure of large documents easier to understand and
manage because you are free to move them around, subdivide them
further, and reuse them in many locations.

*** Tip: store frequently used macros in reusable .pxsl files

If you use certain macros frequently in your PXSL documents, you might
benefit from placing the macros into a separate .pxsl file that you
can reuse.  For example, you could place your macros into a file
macros.pxsl and then use them when processing several documents:

    $ pxslcc macros.pxsl doc1.pxsl > doc1.xml
    $ pxslcc macros.pxsl doc2.pxsl > doc2.xml
    $ pxslcc macros.pxsl doc3.pxsl > doc3.xml

*** Parameterized macros

Macros can take any number of parameters, which allows you to customize
their definitions.

**** Using named parameters

For example, we could customize the definition of the ,head macro that
we used above to accept the title as a parameter:

    ,make-head my-title =

Now we can use it to create a head element that contains any title
that we want:

    ,make-head -my-title=<<This is my title.>>

Note that we pass parameters to a macro just like we pass attributes
to an element definition.

**** Using the magic, implicit BODY parameter

But what if we want to pass more than strings?  What if we want to
pass large sections of documents as parameters?  We can do this using
the special BODY parameter that all macros have implicitly:

    ,make-section title =
        -- start of section
        -- end of section

(Note that the BODY parameter must be spelled exactly "BODY" and in
all caps.)  The BODY parameter accepts any content defined underneath
the macro-expansion site (i.e., the body of the macro-expansion

    ,make-section -title=<<This is my title>>
      p <<This is a paragraph.>>
      p <<And another.>>
      p <<And so on.>>

The result of calling this macro is the following XML:

      <!-- start of section -->
      <title>This is my title</title>
      <p>This is a paragraph.</p>
      <p>And another.</p>
      <p>And so on.</p>
      <!-- end of section -->

*** Advanced macros and passing parameters with the <( )> delimiters

As we showed earlier, one way of passing document fragments to macros
is via the implicit BODY parameter that all macros have.  Another is
to pass them as normal arguments using the <( )> delimiters, which let
you group PXSL document fragments into chunks that you can pass as

For example, let's redefine the make-section macro we defined above to
accept the body of the section as a normal parameter:

    ,make-section title body =
        -- start of section
        -- end of section

Now we can call it like so:

    ,make-section -title=<<This is my title>> \
        p <<This is a paragraph.>>
        p <<And another.>>
        p <<And so on.>>

(Note the use of the backslash in between parameters to continue the
parameter list to the next line.  This useful trick also works to
continue attribute lists when you are creating elements.)

Because the <( )> delimiters can be used only to pass arguments, you
can use them to "quote" arguments that otherwise could not be passed
via position, e.g., a fragment of text that contains whitespace:

    ,make-section <( <<This is my title>> )> \
        p <<This is a paragraph.>>
        p <<And another.>>
        p <<And so on.>>

You can even use the <( )> delimiters to pass the results of macro
calls to elements and other macros:

    ,h1 x =
      -- level one heading

    ,bang x =

    ,h1 <( <<Hello, >>
           ,bang World )>

The above produces the following XML:

    <!-- level one heading -->
    <h1>Hello, World!</h1>

*** More advanced macros and functional programming

Like functions in functional programming languages, macros in PXSL are
first-class values that can be created, bound to parameters, and
passed to other macros.  While this might initially seem like a
programming-language curiosity, it is actually a simple yet immensely
powerful tool that you can use to reduce the size and complexity of
your XML documents.  In particular, this tool lets you "factor out"
and reuse common, boilerplate portions of your documents.

To see how this works, consider the following XML document that
represents an address book:

        <preferred>Joe Smith</preferred>
        <preferred>John Doe</preferred>
      <!-- ... more persons ... -->

The address book contains a long list of persons, each of which has a
first and last name and a "preferred name" that is usually the first
and last named joined together (but might be something else).

We might write the address book in PXSL like this:

        first <<Joe>>
        last <<Smith>>
        preferred <<Joe Smith>>
        first <<John>
        last <<Doe>>
        preferred <<John Doe>>
      -- ... more persons ...

But, seeing how repetitive that is, we might create a ,person macro
to make our lives easier:

    ,person first last =
          << >>

Now, with our new macro, we can simply write

      ,person Joe Smith
      ,person John Doe
      -- ... more persons ...

And, indeed, running the above PXSL code through pxslcc, yields the
identical XML:

        <preferred>Joe Smith</preferred>
        <preferred>John Doe</preferred>
      <!-- ... more persons ... -->

Already, we have saved a great deal of work, but let's say that the
situation is a little more complicated.  Let's say that in addition
to the address-book, we also need to make a roster of persons:

      <formal>Smith, Joe</formal>
      <formal>Doe, John</formal>
      <!-- ... more persons ... -->

and, most important, we need to keep the address-book and roster
synchronized.  In other words, we have one list of names and we
must use it in two places.

At this point, we might be tempted to put the list of names in a
separate XML document and write a small external program or a couple
of XSLT stylesheets to transform the document into the address-book
and roster.  After all, we don't want to have to keep the address-book
and roster synchronized by hand.

But we can do this without leaving PXSL.  All we have to do is create
a macro that builds things out of our list of people:

    ,build-from-people builder-macro =
        ,builder-macro Joe Smith
        ,builder-macro John Doe
        -- ... more persons ...

The interesting thing is that our ,build-from-people macro takes
another macro as a parameter and binds it to the name "builder-macro",
just like it would any other kind of parameter.  It uses this macro to
transform a first and last name into something else.  What that
something else is, is up to us: We simply tailor the ,builder-macro to
suit our purpose.

For example, to build an address book:

      ,build-from-people <( , first last =
                              ,person <(,first)> <(,last)> )>

or, to build a roster:

      ,build-from-people <( , first last =
                                <<, >>
                                ,first  )>

That's it.  We have just built an address book and a roster from our
list of people.

Now, you may have noticed something new in the above two snippets of
PXSL.  In each snippet, inside of the outer-most <( )> delimiters, we
created a macro on the fly -- an anonymous macro, so called because we
didn't give it a name.  (It doesn't need a name because we're using it
just this one time; nobody else will ever call it.)  We simply created
it right when we needed it and passed it to the ,build-from-people
macro, where it was bound to the name "builder-macro." Then
,build-from-people used it to construct "person" or "formal" elements
(depending on how we defined the anonymous macro).  It's a pretty neat

You can create anonymous macros using the familiar comma syntax --
just don't provide a name.  Note the space between the comma and the
start of the argument list:

    , arg1 arg2... =

To call an anonymous macro, of course, you'll first have to bind it to
a name.  The way you do this is to pass the anonymous macro to another
macro, just like we did earlier, causing the anonymous macro to be
bound to one of the other macro's parameters:

    ,some-other-macro <( , arg1 arg2... =
                           body  )>

Then that other macro can call it via the parameter's name:

    ,some-other-macro marco-arg =
      ,macro-arg -arg1=... -arg2=...

Here's another example, less practical but illustrative.  See if you
can figure out how the code works before reading the explanation
that follows.

    ,double str =
    ,single str =
    ,add-quotes quote-fn str =
       ,quote-fn <( ,str )>

    -- let's quote a couple of strings

    ,add-quotes <( , x = ,double <(,x)> )> -str=<<Quote Me>>
    << >>
    ,add-quotes <( , x = ,single <(,x)> )> Please!

Pxslcc compiles the above into the following output:

    <!-- let's quote a couple of strings -->

    "Quote Me" 'Please!'

In this example, the two calls to the ,add-quotes macro each pass in
an anonymous macro that performs the desired quoting operation.  The
anonymous macro is bound to "quote-fn" when the ,add-quotes macro is
called and expanded.  Thus, when ,add-quotes calls ,quote-fn, it is
really calling the anonymous macro that we passed to it.  This lets us
customize the behavior of ,add-quotes without having to rewrite it.

*** Real-world example

The examples above are contrived and don't do justice to the
usefulness of this tool.  This type of refactoring shines when dealing
with large, complicated, and repetitive data structures, but such
examples are too unwieldy to include in an introduction like this.
For this reason, I urge you to take a look at the
"xsl-menus-w-macros.pxsl" example, in examples directory.  It shows
one way that you can use anonymous macros to factor out common code in
production XSLT stylesheets.


** Automatic PXSL-to-XML conversion via Make

Most Make utilities allow you to define pattern rules that are then
used automatically to convert one class of documents into another.
Pattern rules can be used to automate the conversion of PXSL documents
into their XML counterparts.  For example, if you place the following
rule into a makefile (this is for GNU make),

    %.xml: %.pxsl
            pxlscc --indent=2 --header $< > $@

Make will automatically generate .xml documents from the
corresponding .pxsl documents whenever they are needed.  This
frees you to substitute .pxsl documents anywhere that your
project calls for .xml documents, knowing that make will keep all
of the .xml documents up to date, regenerating them as needed
when you update your .pxsl documents.

* Reference:  pxlscc

  Usage: pxslcc [OPTION...] [file...]
  -i[NUM]  --indent[=NUM]  Re-indent XML using NUM spaces per nesting level
  -h       --header        Insert edit-the-PXSL-instead header into output XML
  -x       --xslt          Add XSLT defaults
  -a FILE  --add=FILE      Add the given defaults file
           --export        Export (print) all of the active defaults
           --dump          Dump internal parse format (for debugging)

When you list more than one PXSL file on the command line, pxslcc will
join the files, in order, into one big PXSL document and process that
document.  You can use this feature to incorporate commonly used
macros into your documents:

    $ pxslcc macros1.pxsl macros2.pxsl doc.pxsl > doc.xml

In the example above, doc.pxsl can use the macros defined in
macros1.pxsl and macros2.pxsl.

The --header option requires some explanation.  It inserts the following
header comment into the output XML:


    NOTICE:  This XML document was generated from PXSL source.
             If you want to edit this file, you should probably
             edit the original PXSL source file instead.


It's a good idea to use the --header option all of the time.  This
prevents you (or somebody else) from accidentally editing an XML file
when you really ought to be editing the PXSL file from which the
XML file is generated.

[TODO: Expand this section]

* Reference:  PXSL syntax

The PXSL grammar, in EBNF-like notation:

    pxsl-document       ::= statement*, EOF

    statement           ::= pxsl-comment
                          | xml-comment
                          | literal-constructor
                          | element-constructor
                          | macro-def
                          | macro-app

    pxsl-comment        ::= '#',  all-text-until-newline, NEWLINE
    xml-comment         ::= "--", all-text-until-newline, NEWLINE
    literal-constructor ::= mixed-literal | cdata-literal
    element-constructor ::= xml-name, posn-args, nv-args, children
    macro-def           ::= ',', xml-name?, param-names, '=', macro-body
    macro-app           ::= ',', xml-name, posn-args, nv-args, children

    xml-name            ::= ( LETTER | '_' | ':' ),
                            ( LETTER | DIGIT | '_' | ':' | '.' | '-' )*
    posn-args           ::= expr-list
    nv-args             ::= ( line-continuation?, name-value-pair )*
    name-value-pair     ::= '-', xml-name, '=', expr
    children            ::= statement*     /* must be indented */
    macro-body          ::= children
    param-names         ::= ( line-continuation?, xml-name )*

    line-continuation   ::= '\', newline

    expr-list           ::= ( line-continuation?, arg-expr )*
    arg-expr            ::= expr    /* cannot start with '-' */
    expr                ::= expr-single | NON-WHITESPACE+
    expr-single         ::= mixed-literal | cdata-literal | pxsl-fragment
    mixed-literal       ::= "<<", all-text-until->>-delimiter, ">>"
    cdata-literal       ::= "<{", all-text-until-}>-delimiter, "}>"
    pxsl-fragment       ::= "<(", statement*, ")>"

* Authors

Tom Moertel <tom@moertel.com> http://blog.moertel.com/

Bill Hubauer <bill@hubauer.com>

* (For Emacs)

  Local Variables: