community.general/plugins/modules/xml.py

#!/usr/bin/python
# -*- coding: utf-8 -*-

# Copyright (c) 2014, Red Hat, Inc.
# Copyright (c) 2014, Tim Bielawa <tbielawa@redhat.com>
# Copyright (c) 2014, Magnus Hedemark <mhedemar@redhat.com>
# Copyright (c) 2017, Dag Wieers <dag@wieers.com>
# GNU General Public License v3.0+ (see LICENSES/GPL-3.0-or-later.txt or https://www.gnu.org/licenses/gpl-3.0.txt)
# SPDX-License-Identifier: GPL-3.0-or-later

from __future__ import (absolute_import, division, print_function)
__metaclass__ = type

DOCUMENTATION = r'''
---
module: xml
short_description: Manage bits and pieces of XML files or strings
description:
- A CRUD-like interface to managing bits of XML files.
options:
  path:
    description:
    - Path to the file to operate on.
    - This file must exist ahead of time.
    - This parameter is required, unless I(xmlstring) is given.
    type: path
    aliases: [ dest, file ]
  xmlstring:
    description:
    - A string containing XML on which to operate.
    - This parameter is required, unless I(path) is given.
    type: str
  xpath:
    description:
    - A valid XPath expression describing the item(s) you want to manipulate.
    - Operates on the document root, C(/), by default.
    type: str
  namespaces:
    description:
    - The namespace C(prefix:uri) mapping for the XPath expression.
    - Needs to be a C(dict), not a C(list) of items.
    type: dict
    default: {}
  state:
    description:
    - Set or remove an xpath selection (node(s), attribute(s)).
    type: str
    choices: [ absent, present ]
    default: present
    aliases: [ ensure ]
  attribute:
    description:
    - The attribute to select when using parameter I(value).
    - This is a string, not prepended with C(@).
    type: raw
  value:
    description:
    - Desired state of the selected attribute.
    - Either a string, or to unset a value, the Python C(None) keyword (YAML Equivalent, C(null)).
    - Elements default to no value (but present).
    - Attributes default to an empty string.
    type: raw
  add_children:
    description:
    - Add additional child-element(s) to a selected element for a given I(xpath).
    - Child elements must be given in a list and each item may be either a string
      (eg. C(children=ansible) to add an empty C(<ansible/>) child element),
      or a hash where the key is an element name and the value is the element value.
    - This parameter requires I(xpath) to be set.
    type: list
    elements: raw
  set_children:
    description:
    - Set the child-element(s) of a selected element for a given I(xpath).
    - Removes any existing children.
    - Child elements must be specified as in I(add_children).
    - This parameter requires I(xpath) to be set.
    type: list
    elements: raw
  count:
    description:
    - Search for a given I(xpath) and provide the count of any matches.
    - This parameter requires I(xpath) to be set.
    type: bool
    default: false
  print_match:
    description:
    - Search for a given I(xpath) and print out any matches.
    - This parameter requires I(xpath) to be set.
    type: bool
    default: false
  pretty_print:
    description:
    - Pretty print XML output.
    type: bool
    default: false
  content:
    description:
    - Search for a given I(xpath) and get content.
    - This parameter requires I(xpath) to be set.
    type: str
    choices: [ attribute, text ]
  input_type:
    description:
    - Type of input for I(add_children) and I(set_children).
    type: str
    choices: [ xml, yaml ]
    default: yaml
  backup:
    description:
      - Create a backup file including the timestamp information so you can get
        the original file back if you somehow clobbered it incorrectly.
    type: bool
    default: false
  strip_cdata_tags:
    description:
      - Remove CDATA tags surrounding text values.
      - Note that this might break your XML file if text values contain characters that could be interpreted as XML.
    type: bool
    default: false
  insertbefore:
    description:
      - Add additional child-element(s) before the first selected element for a given I(xpath).
      - Child elements must be given in a list and each item may be either a string
        (eg. C(children=ansible) to add an empty C(<ansible/>) child element),
        or a hash where the key is an element name and the value is the element value.
      - This parameter requires I(xpath) to be set.
    type: bool
    default: false
  insertafter:
    description:
      - Add additional child-element(s) after the last selected element for a given I(xpath).
      - Child elements must be given in a list and each item may be either a string
        (eg. C(children=ansible) to add an empty C(<ansible/>) child element),
        or a hash where the key is an element name and the value is the element value.
      - This parameter requires I(xpath) to be set.
    type: bool
    default: false
requirements:
- lxml >= 2.3.0
notes:
- Use the C(--check) and C(--diff) options when testing your expressions.
- The diff output is automatically pretty-printed, so may not reflect the actual file content, only the file structure.
- This module does not handle complicated xpath expressions, so limit xpath selectors to simple expressions.
- Beware that in case your XML elements are namespaced, you need to use the I(namespaces) parameter, see the examples.
- Namespaces prefix should be used for all children of an element where namespace is defined, unless another namespace is defined for them.
seealso:
- name: Xml module development community wiki
  description: More information related to the development of this xml module.
  link: https://github.com/ansible/community/wiki/Module:-xml
- name: Introduction to XPath
  description: A brief tutorial on XPath (w3schools.com).
  link: https://www.w3schools.com/xml/xpath_intro.asp
- name: XPath Reference document
  description: The reference documentation on XSLT/XPath (developer.mozilla.org).
  link: https://developer.mozilla.org/en-US/docs/Web/XPath
author:
- Tim Bielawa (@tbielawa)
- Magnus Hedemark (@magnus919)
- Dag Wieers (@dagwieers)
'''

EXAMPLES = r'''
# Consider the following XML file:
#
# <business type="bar">
#   <name>Tasty Beverage Co.</name>
#     <beers>
#       <beer>Rochefort 10</beer>
#       <beer>St. Bernardus Abbot 12</beer>
#       <beer>Schlitz</beer>
#    </beers>
#   <rating subjective="true">10</rating>
#   <website>
#     <mobilefriendly/>
#     <address>http://tastybeverageco.com</address>
#   </website>
# </business>

- name: Remove the 'subjective' attribute of the 'rating' element
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/rating/@subjective
    state: absent

- name: Set the rating to '11'
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/rating
    value: 11

# Retrieve and display the number of nodes
- name: Get count of 'beers' nodes
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/beers/beer
    count: true
  register: hits

- ansible.builtin.debug:
    var: hits.count

# Example where parent XML nodes are created automatically
- name: Add a 'phonenumber' element to the 'business' element
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/phonenumber
    value: 555-555-1234

- name: Add several more beers to the 'beers' element
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/beers
    add_children:
    - beer: Old Rasputin
    - beer: Old Motor Oil
    - beer: Old Curmudgeon

- name: Add several more beers to the 'beers' element and add them before the 'Rochefort 10' element
  community.general.xml:
    path: /foo/bar.xml
    xpath: '/business/beers/beer[text()="Rochefort 10"]'
    insertbefore: true
    add_children:
    - beer: Old Rasputin
    - beer: Old Motor Oil
    - beer: Old Curmudgeon

# NOTE: The 'state' defaults to 'present' and 'value' defaults to 'null' for elements
- name: Add a 'validxhtml' element to the 'website' element
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/website/validxhtml

- name: Add an empty 'validatedon' attribute to the 'validxhtml' element
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/website/validxhtml/@validatedon

- name: Add or modify an attribute, add element if needed
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/website/validxhtml
    attribute: validatedon
    value: 1976-08-05

# How to read an attribute value and access it in Ansible
- name: Read an element's attribute values
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/website/validxhtml
    content: attribute
  register: xmlresp

- name: Show an attribute value
  ansible.builtin.debug:
    var: xmlresp.matches[0].validxhtml.validatedon

- name: Remove all children from the 'website' element (option 1)
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/website/*
    state: absent

- name: Remove all children from the 'website' element (option 2)
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business/website
    set_children: []

# In case of namespaces, like in below XML, they have to be explicitly stated.
#
# <foo xmlns="http://x.test" xmlns:attr="http://z.test">
#   <bar>
#     <baz xmlns="http://y.test" attr:my_namespaced_attribute="true" />
#   </bar>
# </foo>

# NOTE: There is the prefix 'x' in front of the 'bar' element, too.
- name: Set namespaced '/x:foo/x:bar/y:baz/@z:my_namespaced_attribute' to 'false'
  community.general.xml:
    path: foo.xml
    xpath: /x:foo/x:bar/y:baz
    namespaces:
      x: http://x.test
      y: http://y.test
      z: http://z.test
    attribute: z:my_namespaced_attribute
    value: 'false'

- name: Adding building nodes with floor subnodes from a YAML variable
  community.general.xml:
    path: /foo/bar.xml
    xpath: /business
    add_children:
      - building:
          # Attributes
          name: Scumm bar
          location: Monkey island
          # Subnodes
          _:
            - floor: Pirate hall
            - floor: Grog storage
            - construction_date: "1990"  # Only strings are valid
      - building: Grog factory

# Consider this XML for following example -
#
# <config>
#   <element name="test1">
#     <text>part to remove</text>
#   </element>
#   <element name="test2">
#     <text>part to keep</text>
#   </element>
# </config>

- name: Delete element node based upon attribute
  community.general.xml:
    path: bar.xml
    xpath: /config/element[@name='test1']
    state: absent
'''

RETURN = r'''
actions:
    description: A dictionary with the original xpath, namespaces and state.
    type: dict
    returned: success
    sample: {xpath: xpath, namespaces: [namespace1, namespace2], state=present}
backup_file:
    description: The name of the backup file that was created
    type: str
    returned: when I(backup=true)
    sample: /path/to/file.xml.1942.2017-08-24@14:16:01~
count:
    description: The count of xpath matches.
    type: int
    returned: when parameter 'count' is set
    sample: 2
matches:
    description: The xpath matches found.
    type: list
    returned: when parameter 'print_match' is set
msg:
    description: A message related to the performed action(s).
    type: str
    returned: always
xmlstring:
    description: An XML string of the resulting output.
    type: str
    returned: when parameter 'xmlstring' is set
'''

import copy
import json
import os
import re
import traceback

from io import BytesIO

from ansible_collections.community.general.plugins.module_utils.version import LooseVersion

LXML_IMP_ERR = None
try:
    from lxml import etree, objectify
    HAS_LXML = True
except ImportError:
    LXML_IMP_ERR = traceback.format_exc()
    HAS_LXML = False

from ansible.module_utils.basic import AnsibleModule, json_dict_bytes_to_unicode, missing_required_lib
from ansible.module_utils.six import iteritems, string_types
from ansible.module_utils.common.text.converters import to_bytes, to_native
from ansible.module_utils.common._collections_compat import MutableMapping

_IDENT = r"[a-zA-Z-][a-zA-Z0-9_\-\.]*"
_NSIDENT = _IDENT + "|" + _IDENT + ":" + _IDENT
# Note: we can't reasonably support the 'if you need to put both ' and " in a string, concatenate
# strings wrapped by the other delimiter' XPath trick, especially as simple XPath.
_XPSTR = "('(?:.*)'|\"(?:.*)\")"

_RE_SPLITSIMPLELAST = re.compile("^(.*)/(" + _NSIDENT + ")$")
_RE_SPLITSIMPLELASTEQVALUE = re.compile("^(.*)/(" + _NSIDENT + ")/text\\(\\)=" + _XPSTR + "$")
_RE_SPLITSIMPLEATTRLAST = re.compile("^(.*)/(@(?:" + _NSIDENT + "))$")
_RE_SPLITSIMPLEATTRLASTEQVALUE = re.compile("^(.*)/(@(?:" + _NSIDENT + "))=" + _XPSTR + "$")
_RE_SPLITSUBLAST = re.compile("^(.*)/(" + _NSIDENT + ")\\[(.*)\\]$")
_RE_SPLITONLYEQVALUE = re.compile("^(.*)/text\\(\\)=" + _XPSTR + "$")


def has_changed(doc):
    orig_obj = etree.tostring(objectify.fromstring(etree.tostring(orig_doc)))
    obj = etree.tostring(objectify.fromstring(etree.tostring(doc)))
    return (orig_obj != obj)


def do_print_match(module, tree, xpath, namespaces):
    match = tree.xpath(xpath, namespaces=namespaces)
    match_xpaths = []
    for m in match:
        match_xpaths.append(tree.getpath(m))
    match_str = json.dumps(match_xpaths)
    msg = "selector '%s' match: %s" % (xpath, match_str)
    finish(module, tree, xpath, namespaces, changed=False, msg=msg)


def count_nodes(module, tree, xpath, namespaces):
    """ Return the count of nodes matching the xpath """
    hits = tree.xpath("count(/%s)" % xpath, namespaces=namespaces)
    msg = "found %d nodes" % hits
    finish(module, tree, xpath, namespaces, changed=False, msg=msg, hitcount=int(hits))


def is_node(tree, xpath, namespaces):
    """ Test if a given xpath matches anything and if that match is a node.

    For now we just assume you're only searching for one specific thing."""
    if xpath_matches(tree, xpath, namespaces):
        # OK, it found something
        match = tree.xpath(xpath, namespaces=namespaces)
        if isinstance(match[0], etree._Element):
            return True

    return False


def is_attribute(tree, xpath, namespaces):
    """ Test if a given xpath matches and that match is an attribute

    An xpath attribute search will only match one item"""
    if xpath_matches(tree, xpath, namespaces):
        match = tree.xpath(xpath, namespaces=namespaces)
        if isinstance(match[0], etree._ElementStringResult):
            return True
        elif isinstance(match[0], etree._ElementUnicodeResult):
            return True
    return False


def xpath_matches(tree, xpath, namespaces):
    """ Test if a node exists """
    if tree.xpath(xpath, namespaces=namespaces):
        return True
    return False


def delete_xpath_target(module, tree, xpath, namespaces):
    """ Delete an attribute or element from a tree """
    changed = False
    try:
        for result in tree.xpath(xpath, namespaces=namespaces):
            changed = True
            # Get the xpath for this result
            if is_attribute(tree, xpath, namespaces):
                # Delete an attribute
                parent = result.getparent()
                # Pop this attribute match out of the parent
                # node's 'attrib' dict by using this match's
                # 'attrname' attribute for the key
                parent.attrib.pop(result.attrname)
            elif is_node(tree, xpath, namespaces):
                # Delete an element
                result.getparent().remove(result)
            else:
                raise Exception("Impossible error")
    except Exception as e:
        module.fail_json(msg="Couldn't delete xpath target: %s (%s)" % (xpath, e))
    else:
        finish(module, tree, xpath, namespaces, changed=changed)


def replace_children_of(children, match):
    for element in list(match):
        match.remove(element)
    match.extend(children)


def set_target_children_inner(module, tree, xpath, namespaces, children, in_type):
    matches = tree.xpath(xpath, namespaces=namespaces)

    # Create a list of our new children
    children = children_to_nodes(module, children, in_type)
    children_as_string = [etree.tostring(c) for c in children]

    changed = False

    # xpaths always return matches as a list, so....
    for match in matches:
        # Check if elements differ
        if len(list(match)) == len(children):
            for idx, element in enumerate(list(match)):
                if etree.tostring(element) != children_as_string[idx]:
                    replace_children_of(children, match)
                    changed = True
                    break
        else:
            replace_children_of(children, match)
            changed = True

    return changed


def set_target_children(module, tree, xpath, namespaces, children, in_type):
    changed = set_target_children_inner(module, tree, xpath, namespaces, children, in_type)
    # Write it out
    finish(module, tree, xpath, namespaces, changed=changed)


def add_target_children(module, tree, xpath, namespaces, children, in_type, insertbefore, insertafter):
    if is_node(tree, xpath, namespaces):
        new_kids = children_to_nodes(module, children, in_type)
        if insertbefore or insertafter:
            insert_target_children(tree, xpath, namespaces, new_kids, insertbefore, insertafter)
        else:
            for node in tree.xpath(xpath, namespaces=namespaces):
                node.extend(new_kids)
        finish(module, tree, xpath, namespaces, changed=True)
    else:
        finish(module, tree, xpath, namespaces)


def insert_target_children(tree, xpath, namespaces, children, insertbefore, insertafter):
    """
    Insert the given children before or after the given xpath. If insertbefore is True, it is inserted before the
    first xpath hit, with insertafter, it is inserted after the last xpath hit.
    """
    insert_target = tree.xpath(xpath, namespaces=namespaces)
    loc_index = 0 if insertbefore else -1
    index_in_parent = insert_target[loc_index].getparent().index(insert_target[loc_index])
    parent = insert_target[0].getparent()
    if insertafter:
        index_in_parent += 1
    for child in children:
        parent.insert(index_in_parent, child)
        index_in_parent += 1


def _extract_xpstr(g):
    return g[1:-1]


def split_xpath_last(xpath):
    """split an XPath of the form /foo/bar/baz into /foo/bar and baz"""
    xpath = xpath.strip()
    m = _RE_SPLITSIMPLELAST.match(xpath)
    if m:
        # requesting an element to exist
        return (m.group(1), [(m.group(2), None)])
    m = _RE_SPLITSIMPLELASTEQVALUE.match(xpath)
    if m:
        # requesting an element to exist with an inner text
        return (m.group(1), [(m.group(2), _extract_xpstr(m.group(3)))])

    m = _RE_SPLITSIMPLEATTRLAST.match(xpath)
    if m:
        # requesting an attribute to exist
        return (m.group(1), [(m.group(2), None)])
    m = _RE_SPLITSIMPLEATTRLASTEQVALUE.match(xpath)
    if m:
        # requesting an attribute to exist with a value
        return (m.group(1), [(m.group(2), _extract_xpstr(m.group(3)))])

    m = _RE_SPLITSUBLAST.match(xpath)
    if m:
        content = [x.strip() for x in m.group(3).split(" and ")]
        return (m.group(1), [('/' + m.group(2), content)])

    m = _RE_SPLITONLYEQVALUE.match(xpath)
    if m:
        # requesting a change of inner text
        return (m.group(1), [("", _extract_xpstr(m.group(2)))])
    return (xpath, [])


def nsnameToClark(name, namespaces):
    if ":" in name:
        (nsname, rawname) = name.split(":")
        # return "{{%s}}%s" % (namespaces[nsname], rawname)
        return "{{{0}}}{1}".format(namespaces[nsname], rawname)

    # no namespace name here
    return name


def check_or_make_target(module, tree, xpath, namespaces):
    (inner_xpath, changes) = split_xpath_last(xpath)
    if (inner_xpath == xpath) or (changes is None):
        module.fail_json(msg="Can't process Xpath %s in order to spawn nodes! tree is %s" %
                             (xpath, etree.tostring(tree, pretty_print=True)))
        return False

    changed = False

    if not is_node(tree, inner_xpath, namespaces):
        changed = check_or_make_target(module, tree, inner_xpath, namespaces)

    # we test again after calling check_or_make_target
    if is_node(tree, inner_xpath, namespaces) and changes:
        for (eoa, eoa_value) in changes:
            if eoa and eoa[0] != '@' and eoa[0] != '/':
                # implicitly creating an element
                new_kids = children_to_nodes(module, [nsnameToClark(eoa, namespaces)], "yaml")
                if eoa_value:
                    for nk in new_kids:
                        nk.text = eoa_value

                for node in tree.xpath(inner_xpath, namespaces=namespaces):
                    node.extend(new_kids)
                    changed = True
                # module.fail_json(msg="now tree=%s" % etree.tostring(tree, pretty_print=True))
            elif eoa and eoa[0] == '/':
                element = eoa[1:]
                new_kids = children_to_nodes(module, [nsnameToClark(element, namespaces)], "yaml")
                for node in tree.xpath(inner_xpath, namespaces=namespaces):
                    node.extend(new_kids)
                    for nk in new_kids:
                        for subexpr in eoa_value:
                            # module.fail_json(msg="element=%s subexpr=%s node=%s now tree=%s" %
                            #                      (element, subexpr, etree.tostring(node, pretty_print=True), etree.tostring(tree, pretty_print=True))
                            check_or_make_target(module, nk, "./" + subexpr, namespaces)
                    changed = True

                # module.fail_json(msg="now tree=%s" % etree.tostring(tree, pretty_print=True))
            elif eoa == "":
                for node in tree.xpath(inner_xpath, namespaces=namespaces):
                    if (node.text != eoa_value):
                        node.text = eoa_value
                        changed = True

            elif eoa and eoa[0] == '@':
                attribute = nsnameToClark(eoa[1:], namespaces)

                for element in tree.xpath(inner_xpath, namespaces=namespaces):
                    changing = (attribute not in element.attrib or element.attrib[attribute] != eoa_value)

                    if changing:
                        changed = changed or changing
                        if eoa_value is None:
                            value = ""
                        else:
                            value = eoa_value
                        element.attrib[attribute] = value

                    # module.fail_json(msg="arf %s changing=%s as curval=%s changed tree=%s" %
                    #       (xpath, changing, etree.tostring(tree, changing, element[attribute], pretty_print=True)))

            else:
                module.fail_json(msg="unknown tree transformation=%s" % etree.tostring(tree, pretty_print=True))

    return changed


def ensure_xpath_exists(module, tree, xpath, namespaces):
    changed = False

    if not is_node(tree, xpath, namespaces):
        changed = check_or_make_target(module, tree, xpath, namespaces)

    finish(module, tree, xpath, namespaces, changed)


def set_target_inner(module, tree, xpath, namespaces, attribute, value):
    changed = False

    try:
        if not is_node(tree, xpath, namespaces):
            changed = check_or_make_target(module, tree, xpath, namespaces)
    except Exception as e:
        missing_namespace = ""
        # NOTE: This checks only the namespaces defined in root element!
        # TODO: Implement a more robust check to check for child namespaces' existence
        if tree.getroot().nsmap and ":" not in xpath:
            missing_namespace = "XML document has namespace(s) defined, but no namespace prefix(es) used in xpath!\n"
        module.fail_json(msg="%sXpath %s causes a failure: %s\n  -- tree is %s" %
                             (missing_namespace, xpath, e, etree.tostring(tree, pretty_print=True)), exception=traceback.format_exc())

    if not is_node(tree, xpath, namespaces):
        module.fail_json(msg="Xpath %s does not reference a node! tree is %s" %
                             (xpath, etree.tostring(tree, pretty_print=True)))

    for element in tree.xpath(xpath, namespaces=namespaces):
        if not attribute:
            changed = changed or (element.text != value)
            if element.text != value:
                element.text = value
        else:
            changed = changed or (element.get(attribute) != value)
            if ":" in attribute:
                attr_ns, attr_name = attribute.split(":")
                # attribute = "{{%s}}%s" % (namespaces[attr_ns], attr_name)
                attribute = "{{{0}}}{1}".format(namespaces[attr_ns], attr_name)
            if element.get(attribute) != value:
                element.set(attribute, value)

    return changed


def set_target(module, tree, xpath, namespaces, attribute, value):
    changed = set_target_inner(module, tree, xpath, namespaces, attribute, value)
    finish(module, tree, xpath, namespaces, changed)


def get_element_text(module, tree, xpath, namespaces):
    if not is_node(tree, xpath, namespaces):
        module.fail_json(msg="Xpath %s does not reference a node!" % xpath)

    elements = []
    for element in tree.xpath(xpath, namespaces=namespaces):
        elements.append({element.tag: element.text})

    finish(module, tree, xpath, namespaces, changed=False, msg=len(elements), hitcount=len(elements), matches=elements)


def get_element_attr(module, tree, xpath, namespaces):
    if not is_node(tree, xpath, namespaces):
        module.fail_json(msg="Xpath %s does not reference a node!" % xpath)

    elements = []
    for element in tree.xpath(xpath, namespaces=namespaces):
        child = {}
        for key in element.keys():
            value = element.get(key)
            child.update({key: value})
        elements.append({element.tag: child})

    finish(module, tree, xpath, namespaces, changed=False, msg=len(elements), hitcount=len(elements), matches=elements)


def child_to_element(module, child, in_type):
    if in_type == 'xml':
        infile = BytesIO(to_bytes(child, errors='surrogate_or_strict'))

        try:
            parser = etree.XMLParser()
            node = etree.parse(infile, parser)
            return node.getroot()
        except etree.XMLSyntaxError as e:
            module.fail_json(msg="Error while parsing child element: %s" % e)
    elif in_type == 'yaml':
        if isinstance(child, string_types):
            return etree.Element(child)
        elif isinstance(child, MutableMapping):
            if len(child) > 1:
                module.fail_json(msg="Can only create children from hashes with one key")

            (key, value) = next(iteritems(child))
            if isinstance(value, MutableMapping):
                children = value.pop('_', None)

                node = etree.Element(key, value)

                if children is not None:
                    if not isinstance(children, list):
                        module.fail_json(msg="Invalid children type: %s, must be list." % type(children))

                    subnodes = children_to_nodes(module, children)
                    node.extend(subnodes)
            else:
                node = etree.Element(key)
                node.text = value
            return node
        else:
            module.fail_json(msg="Invalid child type: %s. Children must be either strings or hashes." % type(child))
    else:
        module.fail_json(msg="Invalid child input type: %s. Type must be either xml or yaml." % in_type)


def children_to_nodes(module=None, children=None, type='yaml'):
    """turn a str/hash/list of str&hash into a list of elements"""
    children = [] if children is None else children

    return [child_to_element(module, child, type) for child in children]


def make_pretty(module, tree):
    xml_string = etree.tostring(tree, xml_declaration=True, encoding='UTF-8', pretty_print=module.params['pretty_print'])

    result = dict(
        changed=False,
    )

    if module.params['path']:
        xml_file = module.params['path']
        with open(xml_file, 'rb') as xml_content:
            if xml_string != xml_content.read():
                result['changed'] = True
                if not module.check_mode:
                    if module.params['backup']:
                        result['backup_file'] = module.backup_local(module.params['path'])
                    tree.write(xml_file, xml_declaration=True, encoding='UTF-8', pretty_print=module.params['pretty_print'])

    elif module.params['xmlstring']:
        result['xmlstring'] = xml_string
        # NOTE: Modifying a string is not considered a change !
        if xml_string != module.params['xmlstring']:
            result['changed'] = True

    module.exit_json(**result)


def finish(module, tree, xpath, namespaces, changed=False, msg='', hitcount=0, matches=tuple()):

    result = dict(
        actions=dict(
            xpath=xpath,
            namespaces=namespaces,
            state=module.params['state']
        ),
        changed=has_changed(tree),
    )

    if module.params['count'] or hitcount:
        result['count'] = hitcount

    if module.params['print_match'] or matches:
        result['matches'] = matches

    if msg:
        result['msg'] = msg

    if result['changed']:
        if module._diff:
            result['diff'] = dict(
                before=etree.tostring(orig_doc, xml_declaration=True, encoding='UTF-8', pretty_print=True),
                after=etree.tostring(tree, xml_declaration=True, encoding='UTF-8', pretty_print=True),
            )

        if module.params['path'] and not module.check_mode:
            if module.params['backup']:
                result['backup_file'] = module.backup_local(module.params['path'])
            tree.write(module.params['path'], xml_declaration=True, encoding='UTF-8', pretty_print=module.params['pretty_print'])

    if module.params['xmlstring']:
        result['xmlstring'] = etree.tostring(tree, xml_declaration=True, encoding='UTF-8', pretty_print=module.params['pretty_print'])

    module.exit_json(**result)


def main():
    module = AnsibleModule(
        argument_spec=dict(
            path=dict(type='path', aliases=['dest', 'file']),
            xmlstring=dict(type='str'),
            xpath=dict(type='str'),
            namespaces=dict(type='dict', default={}),
            state=dict(type='str', default='present', choices=['absent', 'present'], aliases=['ensure']),
            value=dict(type='raw'),
            attribute=dict(type='raw'),
            add_children=dict(type='list', elements='raw'),
            set_children=dict(type='list', elements='raw'),
            count=dict(type='bool', default=False),
            print_match=dict(type='bool', default=False),
            pretty_print=dict(type='bool', default=False),
            content=dict(type='str', choices=['attribute', 'text']),
            input_type=dict(type='str', default='yaml', choices=['xml', 'yaml']),
            backup=dict(type='bool', default=False),
            strip_cdata_tags=dict(type='bool', default=False),
            insertbefore=dict(type='bool', default=False),
            insertafter=dict(type='bool', default=False),
        ),
        supports_check_mode=True,
        required_by=dict(
            add_children=['xpath'],
            attribute=['value'],
            content=['xpath'],
            set_children=['xpath'],
            value=['xpath'],
        ),
        required_if=[
            ['count', True, ['xpath']],
            ['print_match', True, ['xpath']],
            ['insertbefore', True, ['xpath']],
            ['insertafter', True, ['xpath']],
        ],
        required_one_of=[
            ['path', 'xmlstring'],
            ['add_children', 'content', 'count', 'pretty_print', 'print_match', 'set_children', 'value'],
        ],
        mutually_exclusive=[
            ['add_children', 'content', 'count', 'print_match', 'set_children', 'value'],
            ['path', 'xmlstring'],
            ['insertbefore', 'insertafter'],
        ],
    )

    xml_file = module.params['path']
    xml_string = module.params['xmlstring']
    xpath = module.params['xpath']
    namespaces = module.params['namespaces']
    state = module.params['state']
    value = json_dict_bytes_to_unicode(module.params['value'])
    attribute = module.params['attribute']
    set_children = json_dict_bytes_to_unicode(module.params['set_children'])
    add_children = json_dict_bytes_to_unicode(module.params['add_children'])
    pretty_print = module.params['pretty_print']
    content = module.params['content']
    input_type = module.params['input_type']
    print_match = module.params['print_match']
    count = module.params['count']
    backup = module.params['backup']
    strip_cdata_tags = module.params['strip_cdata_tags']
    insertbefore = module.params['insertbefore']
    insertafter = module.params['insertafter']

    # Check if we have lxml 2.3.0 or newer installed
    if not HAS_LXML:
        module.fail_json(msg=missing_required_lib("lxml"), exception=LXML_IMP_ERR)
    elif LooseVersion('.'.join(to_native(f) for f in etree.LXML_VERSION)) < LooseVersion('2.3.0'):
        module.fail_json(msg='The xml ansible module requires lxml 2.3.0 or newer installed on the managed machine')
    elif LooseVersion('.'.join(to_native(f) for f in etree.LXML_VERSION)) < LooseVersion('3.0.0'):
        module.warn('Using lxml version lower than 3.0.0 does not guarantee predictable element attribute order.')

    # Check if the file exists
    if xml_string:
        infile = BytesIO(to_bytes(xml_string, errors='surrogate_or_strict'))
    elif os.path.isfile(xml_file):
        infile = open(xml_file, 'rb')
    else:
        module.fail_json(msg="The target XML source '%s' does not exist." % xml_file)

    # Parse and evaluate xpath expression
    if xpath is not None:
        try:
            etree.XPath(xpath)
        except etree.XPathSyntaxError as e:
            module.fail_json(msg="Syntax error in xpath expression: %s (%s)" % (xpath, e))
        except etree.XPathEvalError as e:
            module.fail_json(msg="Evaluation error in xpath expression: %s (%s)" % (xpath, e))

    # Try to parse in the target XML file
    try:
        parser = etree.XMLParser(remove_blank_text=pretty_print, strip_cdata=strip_cdata_tags)
        doc = etree.parse(infile, parser)
    except etree.XMLSyntaxError as e:
        module.fail_json(msg="Error while parsing document: %s (%s)" % (xml_file or 'xml_string', e))

    # Ensure we have the original copy to compare
    global orig_doc
    orig_doc = copy.deepcopy(doc)

    if print_match:
        do_print_match(module, doc, xpath, namespaces)

    if count:
        count_nodes(module, doc, xpath, namespaces)

    if content == 'attribute':
        get_element_attr(module, doc, xpath, namespaces)
    elif content == 'text':
        get_element_text(module, doc, xpath, namespaces)

    # File exists:
    if state == 'absent':
        # - absent: delete xpath target
        delete_xpath_target(module, doc, xpath, namespaces)

    # - present: carry on

    # children && value both set?: should have already aborted by now
    # add_children && set_children both set?: should have already aborted by now

    # set_children set?
    if set_children is not None:
        set_target_children(module, doc, xpath, namespaces, set_children, input_type)

    # add_children set?
    if add_children:
        add_target_children(module, doc, xpath, namespaces, add_children, input_type, insertbefore, insertafter)

    # No?: Carry on

    # Is the xpath target an attribute selector?
    if value is not None:
        set_target(module, doc, xpath, namespaces, attribute, value)

    # If an xpath was provided, we need to do something with the data
    if xpath is not None:
        ensure_xpath_exists(module, doc, xpath, namespaces)

    # Otherwise only reformat the xml data?
    if pretty_print:
        make_pretty(module, doc)

    module.fail_json(msg="Don't know what to do")


if __name__ == '__main__':
    main()