I do it from gedit. In gedit, you can add any script, in particular a Python script, as an External Tool. The script reads data from stdin and writes output to stdout, so it may be used as a stand-alone program. It layouts XML and sorts child nodes.
#!/usr/bin/env python
# encoding: utf-8
"""
This is a gedit plug-in to sort and layout XML.
In gedit, to add this tool, open: menu -- Tools -- Manage External Tools...
Create a new tool: click [+] under the list of tools, type in "Sort XML" as tool name,
paste the whole text from this file in the "Edit:" box, then
configure the tool:
Input: Current selection
Output: Replace current selection
In gedit, to run this tool,
FIRST SELECT THE XML,
then open: menu -- Tools -- External Tools > -- Sort XML
"""
from lxml import etree
import sys
import io
def headerFirst(node):
"""Return the sorting key prefix, so that 'header' will go before any other node
"""
nodetag=('%s' % node.tag).lower()
if nodetag.endswith('}header') or nodetag == 'header':
return '0'
else:
return '1'
def get_node_key(node, attr=None):
"""Return the sorting key of an xml node
using tag and attributes
"""
if attr is None:
return '%s' % node.tag + ':'.join([node.get(attr)
for attr in sorted(node.attrib)])
if attr in node.attrib:
return '%s:%s' % (node.tag, node.get(attr))
return '%s' % node.tag
def sort_children(node, attr=None):
""" Sort children along tag and given attribute.
if attr is None, sort along all attributes"""
if not isinstance(node.tag, str): # PYTHON 2: use basestring instead
# not a TAG, it is comment or DATA
# no need to sort
return
# sort child along attr
node[:] = sorted(node, key=lambda child: (headerFirst(child) + get_node_key(child, attr)))
# and recurse
for child in node:
sort_children(child, attr)
def sort(unsorted_stream, sorted_stream, attr=None):
"""Sort unsorted xml file and save to sorted_file"""
parser = etree.XMLParser(remove_blank_text=True)
tree = etree.parse(unsorted_stream,parser=parser)
root = tree.getroot()
sort_children(root, attr)
sorted_unicode = etree.tostring(tree, pretty_print=True, xml_declaration=True, encoding="UTF-8")
sorted_stream.write('%s' % sorted_unicode)
#we could do this,
#sort(sys.stdin, sys.stdout)
#but we want to check selection:
inputstr = ''
for line in sys.stdin:
inputstr += line
if not inputstr:
sys.stderr.write('no XML selected!')
exit(100)
sort(io.BytesIO(inputstr), sys.stdout)
There are two tricky things:
parser = etree.XMLParser(remove_blank_text=True)
tree = etree.parse(unsorted_stream,parser=parser)
By default, the spaces are not ignored, which may produce a strange result.
sorted_unicode = etree.tostring(tree, pretty_print=True, xml_declaration=True, encoding="UTF-8")
Again, by default there is no pretty-printing either.
I configure this tool to work on the current selection and replace the current selection because usually there are HTTP headers in the same file, YMMV.
$ python --version
Python 2.7.6
$ lsb_release -a
Distributor ID: Ubuntu
Description: Ubuntu 14.04.5 LTS
Release: 14.04
Codename: trusty
If you do not need child node sorting, just comment the corresponding line out.
Links: here, here
UPDATE v2 places header in front of anything else; fixed spaces
UPDATE getting lxml
on Ubuntu 18.04.3 LTS bionic:
sudo apt install python-pip
pip install --upgrade lxml
$ python --version
Python 2.7.15+