The preferred method for rendering a
Document or single
is by using the pretty printing facility defined in Pretty.
Pretty-printing does not work well for cases, however, where the
formatting in the XML document is significant. Examples of this
case are XHTML's
<pre> tag, Docbook's
and many more.
Theoretically, the document author could avoid this problem by
wrapping the contents of these tags in a <![CDATA[...]]> section,
but often this is not practical, for instance when the
literal-layout section contains other elements. Finally, program
writers could manually format these elements by transforming them
literal string in their
CFliter, etc., but this is
annoying to do and prone to omissions and formatting errors.
- HaXml's parser eats line feeds between two tags.
Attributes should be formatted by making them an instance of
Verbatimas well, but since an
Attributeis just a tuple, not a full data type, the helper function
verbAttrmust be used instead.
CMiscis not yet supported.
Elements, which contain no content, are formatted as
<element-name/>, even if they were not defined as being of type
EMPTY. In XML this perfectly alright, but in SGML it is not. Those, who wish to use
verbatimto format parts of say an HTML page will have to (a) replace problematic elements by
literals before running
verbatimor (b) use a second search-and-replace stage to fix this.
This class promises that the function
verbatim knows how to
format this data type into a string without changing the
This is a helper function is required because Haskell does not
allow to make an ordinary tuple (like
Attribute) an instance of a
class. The resulting output will preface the actual attribute with
a single blank so that lists of
Attributes can be handled
implicitly by the definition for lists of
Verbatim data types.