Learn about using XSLT to produce EPUB2 and EPUB3 books from XML or XHTML input as well as PDF for print with CSS or XSL-FO. You don't need previous XSLT experience but a willingness to look at a little code will really help. XSLT can be a functional programming language or a simple template language or anywhere between. We’ll focus on the newer features of XSLT, after a quick overview. March 18, 2019 ebookcraft.booknetcanada.ca #EbookCraft

1. XSLT 3 for EPUB and Print
Liam Quin
Delightful Computing

2. ebookcraft 2019/Logistics
– Restrooms
– Breaks
– Refreshments

3. ebookcraft 2019/Schedule
●
This is all about using XSLT 3;
●
It’s a lot of fun
●
Are you in the right place? Of course you are!

4. Agenda
●
Introductions;
●
Background: XML, XPath, XSLT
●
Version Two: XQuery, XSD, Maturity
●
Version Three: Floods of Joy
●
Helpful Resources
●
Discussion & Doing Together

5. Introductions
●
I’m Liam Quin, former XML Activity lead at
W3C, background in digital typography,
programming, computer science, information
architecture. See www.delightfulcomputing.com
.
●
More important: who are you? Who is using
XSLT? Who isn’t? Who is making ebooks?

6. What’s XSLT
●
Functional Programming Language
●
Super Easy and Fun
●
Domain-Specific Language for Trees
●
Text Processing without programming
●
Batch way to convert tons of stuff to other stuff!
●
W3C Standard: books, tutorials, people

7. Functional Programming Language
●
Don’t be intimidated, you can ignore this slide.
●
Buzzword compliance: referential transparency,
lambda expressions, declarative, implicit
dispatch, domain-specific language…
●
The only important one: declarative…

8. Declarative Programming
●
It’s about telling the computer the result you
want, and letting the computer figure out how to
get there all by itself!
●
As opposed to imperative languages (Python,
Java, C, JavaScript, …) where you have to say
how to get there but never describe the result.

9. The Computer is Your Happy Slave
●
Is it really programming if the computer does
the hard parts?
●
Yes, but you don’t have to be a programmer!
●
It is as powerful as an imperative language?
●
Yes, and it’s ideally suited for ebooks!

10. Time for an example
<xsl:stylesheet
xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
version="1.0">
<xsl:template match="book">
<hello>Hello, EBookCraft!</hello>
</xsl:template>
</xsl:stylesheet>
<xsl:stylesheet
xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
version="1.0">
<xsl:template match="book">
<hello>Hello, EBookCraft!</hello>
</xsl:template>
</xsl:stylesheet>

11. Eg01.xsl
<xsl:stylesheet
xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
version="1.0">
It’s a stylesheet; you can call it xsl:transform if you prefer.
XSLT was originally designed for formatting and publishing.
</xsl:stylesheet>
<xsl:stylesheet
xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
version="1.0">
It’s a stylesheet; you can call it xsl:transform if you prefer.
XSLT was originally designed for formatting and publishing.
</xsl:stylesheet>

12. Eg01.xsl
<xsl:template match="book">
<hello>Hello, EBookCraft!</hello>
</xsl:template>
Any elements in the input called book should be
transformed into “hello” elements in the output.
<xsl:template match="book">
<hello>Hello, EBookCraft!</hello>
</xsl:template>
Any elements in the input called book should be
transformed into “hello” elements in the output.

13. Run the example

14. Domain-specific language
●
XSLT uses an XML syntax overall;
●
Expressions are in a W3C language called
XPath;
●
XPath has functions, data types, operators,
if/else, but most importantly it’s a domain-
specific language itself! For navigating trees!
●
This is so awesome!

15. Some sample XPath expressions
(: this is a comment, between smiles! :)
/book/chapter[3]/footnote
A list of all footnotes in the third chapter element
(//table//td[. = 300]/following-sibling::td)[1]
The first in a list of td elements with the same parent
as, and following, a td with content of 300
somewhere in a table!

16. Talk about Trees

17. Computer Science Trees
●
Computer scientists all live in basements with no
windows, and we think trees grow down from the
ground, with the roots at the top and the leaves
far, far, beneath our feet.
●
A tree is a way of representing information in a
computer: as a hierarchy.
●
Maybe computer scientists can’t spell hier… um?

18. A Document Tree

19. Eg01.xml

20. Eg01 with tree

21. eg01.xml tree

22. It’s all about Relationships
●
XPath axes are not about “Turtle Graphics” but
about the relationships between nodes in the
tree.
●
Our tree came from XML. In XSLT 3 it might
have come from JSON or even RDF.
●
There can be many trees in a forest.

23. XSLT 3 Inputs
●
Default input is still a single XML file;
●
A different SAX parser might read e.g. HTML;
●
Can now read binary file, text and JSON;
●
EXPath extension functions can read ZIP;
●
Can make a tree from XML in a ZIP archive!
●
A W E S O M E N E S S I N C A R N A T E

24. Reading an EPUB file
<xsl:variable name="thebook"
select="file:read-binary($inputfile)" />
<xsl:variable name="container"
select="archive:extract-text(
$thebook, 'META-INF/container.xml')" />
<xsl:variable name="cxml" as="document-node()"
select="parse-xml($container)" />

25. Reminder

26. Finding the OPF file
<xsl:variable name="where-to-look" as="xs:string"
select="$cxml//*:rootfile/@full-path" />
<xsl:variable name="the-opf" as="document-node()"
select="parse-xml(archive:extract-text(
$thebook, $where-to-look))" />
<xsl:value-of select="$the-opf//dc:title" />

27. What have we done?
●
We read an epub zip file into memory;
●
We extracted container.xml from it;
●
We looked inside container.xml and found the
name of the content.opf file;
●
We looked in that and found the title of the book
that’s in the epub zip file!

28. What else could we do?
●
We can write to a zip archive too;
●
We can update an existing zip archive;
●
We can merge zip archives, so that we can
handle different compression for different files;
●
With this, we can write an epub file!

29. Asweome Moreness!
The single most important new feature in
XSLT 3
Is called . . .

30. Are You Ready?

31. fn:transform()
●
Run another transformation from within XSLT:
<xsl:sequence select='transform(
map {
"stylesheet-location" : "mkcards.xsl",
"source-node" : $toc
}
)?output' />

32. Maps
●
Before we can understand fn:tranform we need
to grok maps and arrays
●
Arrays are in XPath 3.1, maps in 3.0 and 3.1
●
Array: let $a := [ 4, 5, 6 ] return $a(1)
●
Let $b := map { "venue" : "Toronto", "coolness" :
11 } return $b?”venue”

33. Fn:transform return value
●
The fn:transform() function returns a map
whose keys are URIs (filenames!) with
corresponding values being the file contents.
●
The special key “output” gives the direct result
of the stylesheet.
●
Xsl:result-document goes to the map; file:write
happens directly.

34. So What?
●
XSLT 3 can read an epub file, write an epub file,
read and write JSON, XML, HTML, text,
images, zip archives…
●
You don’t need external scripts or pipelines to
make ebooks any more! One stylesheet! And
it’s not even very complicated.

35. All-XSLT advantages
●
Character encodings are handled in the Web
way at every stage;
●
Namespace URIs, XHTML syntax, consistent
treatment of text..
●
Same person can maintain entire workflow;
●
XSLT is a domain-specific language for markup!

36. So can we really…

37. …write zip files?
●
First, read a zip file:
<xsl:variable name="base-archive"
as="xs:base64Binary"
select="file:read-binary('mimetype.zip')" />

38. …zip ties are fun…
●
So now $base-archive contains a zip archive.
Let’s add some more files to it:
arch:update(
$archive as xs:base64Binary,
$entries as xs:string*,
$new as xs:base64Binary*) as xs:base64Binary

39. Zippitty doohdah
<xsl:variable name="the-zip-archive"
as="xs:base64Binary"
select="archive:update($base-archive,
$filenames,
$contents-of-those-files-in-base64
)" />

40. Zippitty doohdah
<xsl:variable name="the-zip-archive"
as="xs:base64Binary"
select="archive:update($base-archive,
$name-value-list[ ( position() mod 2) eq 1 ],
$name-value-list[ ( position() mod 2) eq 0 ]
)" />

41. Big Fat Zip
●
So now $the-zip-archive is a zip archive full of
happy stuff. Let’s write it out:
<xsl:value-of select="file:write-binary(
'hairyplodder.epub', $the-zip-archive)" />

42. Time to rest for a moment

43. XSLT 3 More Betterness
●
Parse-xml(), parse-json(), serialize();
●
Lots of new functions;
●
More regular expression support;
●
Contains-token() for HTML class attributes;
●
HTML 5 output (read HTML e.g. with TagSoup)
●
Parse-date() helps with metadata

44. not(xslt = boring)
●
R