xmlsort - sorts 'records' in XML files

  xmlsort -r=<recordname> [ <other options> ] [ <filename> ]

  Options:
   -r <name>  name of the elements to be sorted
   -k <keys>  child nodes to be used as sort keys
   -i         ignore case when sorting
   -s         normalise whitespace when comparing sort keys
   -t <dir>   buffer records to named directory rather than in memory
   -m <bytes> set memory chunk size for disk buffering
   -h         help - display the full documentation

  Example:
   xmlsort -r 'person' -k 'lastname;firstname' -i -s in.xml >out.xml

This script takes an XML document either on STDIN or from a named file and writes a sorted version of the file to STDOUT. The CW-r option should be used to identify 'records' in the document - the bits you want sorted. Elements before and after the records will be unaffected by the sort.

Here is a brief summary of the command line options (and the XML::Filter::Sort options which they correspond to). For more details see XML::Filter::Sort.

-r <recordname> (Record)

The name of the elements to be sorted. This can be a simple element name like CW'person' or a pathname like CW'employees/person' (only person elements contained directly within an employees element).

-k <keys> (Keys)

Semicolon separated list of elements (or attributes) within a record which should be used as sort keys. Each key can optionally be followed by 'alpha' or 'num' to indicate alphanumeric of numeric sorting and 'asc' or 'desc' for ascending or descending order (eg: -k 'lastname;firstname;age,n,d').

-i (IgnoreCase)

This option makes sort comparisons case insensitive.

-s (NormaliseKeySpace)

By default all whitespace in the sort key elements is considered significant. Specifying -s will case leading and trailing whitespace to be stripped and internal whitespace runs to be collapsed to a single space.

-t <directory> (TempDir)

When sorting large documents, it may be prudent to use disk buffering rather than memory buffering. This option allows you to specify where temporary files should be written.

-m <bytes> (MaxMem)

If you use the -t option to enable disk buffering, records will be collected in memory in 'chunks' of up to about 10 megabytes before being sorted and spooled to temporary files. This option allows you to specify a larger chunk size. A suffix of K or M indicates kilobytes or megabytes respectively.

This script uses the following modules:

  XML::SAX::ParserFactory
  XML::Filter::Sort
  XML::SAX::Writer

Grant McLean <grantm@cpan.org>

Copyright (c) 2002 Grant McLean. All rights reserved. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.