Lightning Talk @ OSCON06: XGAWK

Whats all about

• Extensible GNU Awk, with an XML parsing extension
• essential: replace the line stream by an XML event stream
• events (like start-, or endelement) are defined by global variables (XMLEVENT, XMLNAME, ...)
• nice oneliners for XML processing

Example code

 # example for 'core' features
 BEGIN { XMLMODE=1 } # switch to XML processing
 XMLSTARTELEM {
   printf("XMLSTARTELEM %s\t", XMLSTARTELEM)
   for (i in XMLATTR)
     printf(" %s='%s'", i, XMLATTR[i])
   print ""
 }
 XMLENDELEM {
   print "XMLENDELEM", XMLENDELEM
 }
 XMLCHARDATA {
   print "XMLCHARDATA", "'" $0 "'"
 }
 #...

 # books.xml
 <?xml version="1.0" encoding="ISO-8859-1"?>
 <books>
   <book publisher="O'Reilly" on-loan="Sanjay">
     <title>XML in a Nutshell</title>
     <author>Elliotte Rusty Harold, W. Scott Means</author>
   </book>
   <book publisher="Addison-Wesley">
     <title>The Mythical Man Month</title>
     <author>Frederick Brooks</author>
   </book>
   <!-- more to come here -->
 </books>

 # example for using the xmllib.awk
 xgawk -f xmllib.awk --source ' # or shorter: xmlgawk
  EE == "title"  { title  = CDATA }
  EE == "author" { author = CDATA }
  EE == "book" && loaner = ATTR[PATH"@on-loan"] {
     print loaner, "was loaned", title, "by", author
  }' books2.xml

 # same in XQuery
 for $b in document("books2.xml")/books/book[@on-loan]
     return (string($b/@on-loan),
             " was loaned ", $b/title/text(),
             " by ", $b/author/text())

Goodies

• can handle streams (multiple documents in one file)
• xgawk: fast and simple core implementation (expat based)
• a library with convenience functions (xmllib.awk) for shorter scripts (xmlgawk)
• many XML encodings supported (code taken over from Perl ;-)
• dynamic loading of shared libs/extensions now possible
• GNU Awk @-commands now builtin
• a Postgres and MPFR extension

Implementation

• code branch from GNU Awk 3.1.5 (Arnold Robbins supports the effort)
• update follows roughly six month behind the official GNU Awk releases
• available at http://sourceforge.net/projects/xmlgawk/

Room for improvement

• documentation!
• XML namespace support
• DTD and Schema validation; still in discussion whether this is seen as essential
• usage of many special variables in contrast to one array is still discussed

Developers

• Jürgen Kahrs
• Andrew Schorr
• Stefan Tramm (me :-)
• and some others