[IMP] define xml_utils.cleanup_xml_node

Intended to cleanup qweb xmls: - remove blank (empty) nodes and/or nodes with whitespace text - fix indentation - remove indentation (needed for some xml signatures) closes odoo/odoo#91006 X-original-commit: b7d7adb4f5a9873350511c227d5fbf54c8f38ce8 Signed-off-by: William André (wan) <wan@odoo.com>
2022-05-11 09:21:59 +02:00
parent 2490982fce
commit ac175c98e1
3 changed files with 207 additions and 0 deletions
@@ -2,3 +2,4 @@
 from . import test_methods
 from . import test_lxml
 from . import test_form_impl
+from . import test_xml_tools
@@ -0,0 +1,153 @@
+# -*- coding: utf-8 -*-
+# Part of Odoo. See LICENSE file for full copyright and licensing details.
+
+
+from lxml import etree
+from odoo.tests import common
+from odoo.tools.xml_utils import cleanup_xml_node
+
+
+class TestXMLTools(common.TransactionCase):
+
+    def setUp(self):
+        super(TestXMLTools, self).setUp()
+        self.qweb_poor = self.env()['ir.ui.view'].create({
+            'type': 'qweb',
+            'arch_db': """
+    <h1>
+            <h2/>
+                <h2>text</h2>
+        \t<h2><h3/></h2>
+            <h2>            </h2>
+<!-- comment removed by qweb -->
+</h1>"""})
+
+    def test_cleanup_xml_node_no_modif(self):
+        # Qweb removes comments and any whitespace before first tag, no other content affected
+        expected = """<h1>
+            <h2/>
+                <h2>text</h2>
+        \t<h2><h3/></h2>
+            <h2>            </h2>
+
+</h1>"""
+        qweb = self.env['ir.qweb']._render(self.qweb_poor.id)
+        self.check_xml_cleanup_result_is_as_expected(qweb, expected, remove_blank_text=False, remove_blank_nodes=False, indent_level=-1)
+
+    def test_cleanup_xml_node_indent_level(self):
+        # Indentation level and spacing works as expected, nothing else affected
+        # (quirk: first tag not indented because indent is actually previous tag's tail)
+        expected = """<h1>
+__<h2/>
+__<h2>text</h2>
+__<h2>
+___<h3/>
+__</h2>
+__<h2>            </h2>
+_</h1>
+"""
+        qweb = self.env['ir.qweb']._render(self.qweb_poor.id)
+        self.check_xml_cleanup_result_is_as_expected(qweb, expected, remove_blank_text=False, remove_blank_nodes=False, indent_level=1, indent_space="_")
+
+    def test_cleanup_xml_node_keep_blank_text(self):
+        # Blank nodes are removed but not nodes containing blank text
+        expected = """<h1>
+  <h2>text</h2>
+  <h2>            </h2>
+</h1>
+"""
+        qweb = self.env['ir.qweb']._render(self.qweb_poor.id)
+        self.check_xml_cleanup_result_is_as_expected(qweb, expected, remove_blank_text=False)
+
+    def test_cleanup_xml_node_keep_blank_nodes(self):
+        # Blank text is removed but blank (empty) nodes remain
+        expected = """<h1>
+  <h2/>
+  <h2>text</h2>
+  <h2>
+    <h3/>
+  </h2>
+  <h2></h2>
+</h1>
+"""
+        qweb = self.env['ir.qweb']._render(self.qweb_poor.id)
+        self.check_xml_cleanup_result_is_as_expected(qweb, expected, remove_blank_nodes=False)
+
+    def test_cleanup_xml_t_call_indent(self):
+        # Indentation is fixed after t-call (which keeps indentation of called template)
+        template_1 = self.env['ir.ui.view'].create({
+            'type': 'qweb',
+            'arch_db': '''<h1>
+    <content>This is content!</content>
+</h1>
+'''})
+        template_2 = self.env['ir.ui.view'].create({
+            'name': 'test',
+            'type': 'qweb',
+            'arch_db': f'''<odoo>
+    <data>
+        <t t-call="{template_1.id}"/>
+    </data>
+</odoo>
+'''})
+        expected = """<odoo>
+  <data>
+    <h1>
+      <content>This is content!</content>
+    </h1>
+  </data>
+</odoo>
+"""
+        qweb = self.env['ir.qweb']._render(template_2.id)
+        self.check_xml_cleanup_result_is_as_expected(qweb, expected)
+
+    def test_qweb_render_values_empty_nodes(self):
+        # Indentation is fixed and empty nodes are removed after conditional rendering
+        template_addresses = self.env['ir.ui.view'].create({
+            'type': 'qweb',
+            'arch_db': '''<t>
+    <street t-esc="address.get('street')"/>
+    <number t-esc="address.get('number')"/>
+    <city t-esc="address.get('city')"/>
+</t>
+'''})
+        template_main = self.env['ir.ui.view'].create({
+            'type': 'qweb',
+            'arch_db': f'''<data>
+    <item t-foreach="items" t-as="item" t-esc="item"/>
+    <addressSender t-call='{template_addresses.id}'>
+        <t t-set="address" t-value="addressSender"/>
+    </addressSender>
+    <addressRecipient t-call='{template_addresses.id}'>
+        <t t-set="address" t-value="addressRecipient"/>
+    </addressRecipient>
+</data>
+'''})
+        expected = """<data>
+  <item>1</item>
+  <item>2</item>
+  <item>Three</item>
+  <addressRecipient>
+    <street>Baker street</street>
+    <number>221B</number>
+    <city>London</city>
+  </addressRecipient>
+</data>
+"""
+        qweb = self.env['ir.qweb']._render(template_main.id, {
+            'items': [1, 2, "Three", False],
+            'addressRecipient': {
+                'number': '221B',
+                'street': 'Baker street',
+                'city': 'London',
+            },
+            'addressSender': {
+                'street': ' '
+            }
+        })
+        self.check_xml_cleanup_result_is_as_expected(qweb, expected)
+
+    def check_xml_cleanup_result_is_as_expected(self, original_string, expected_string, **kwargs):
+        result_string = etree.tostring(cleanup_xml_node(original_string, **kwargs)).decode()
+        self.assertEqual(expected_string, result_string)
+        self.assertNotEqual(expected_string, original_string)
@@ -86,3 +86,56 @@ def create_xml_node(parent_node, node_name, node_value=None):
    :rtype: etree._Element
    """
    return create_xml_node_chain(parent_node, [node_name], node_value)[0]
+
+
+def cleanup_xml_node(xml_node_or_string, remove_blank_text=True, remove_blank_nodes=True, indent_level=0, indent_space="  "):
+    """Clean up the sub-tree of the provided XML node.
+
+    If the provided XML node is of type:
+    - etree._Element, it is modified in-place.
+    - string/bytes, it is first parsed into an etree._Element
+    :param xml_node_or_string (etree._Element, str): XML node (or its string/bytes representation)
+    :param remove_blank_text (bool): if True, removes whitespace-only text from nodes
+    :param remove_blank_nodes (bool): if True, removes leaf nodes with no text (iterative, depth-first, done after remove_blank_text)
+    :param indent_level (int): depth or level of node within root tree (use -1 to leave indentation as-is)
+    :param indent_space (str): string to use for indentation (use '' to remove all indentation)
+    :returns (etree._Element): clean node, same instance that was received (if applicable)
+    """
+    xml_node = xml_node_or_string
+
+    # Convert str/bytes to etree._Element
+    if isinstance(xml_node, str):
+        xml_node = xml_node.encode()  # misnomer: fromstring actually reads bytes
+    if isinstance(xml_node, bytes):
+        xml_node = etree.fromstring(xml_node)
+
+    # Process leaf nodes iteratively
+    # Depth-first, so any inner node may become a leaf too (if children are removed)
+    def leaf_iter(parent_node, node, level):
+        for child_node in node:
+            leaf_iter(node, child_node, level if level < 0 else level + 1)
+
+        # Indentation
+        if level >= 0:
+            indent = '\n' + indent_space * level
+            if not node.tail or not node.tail.strip():
+                node.tail = '\n' if parent_node is None else indent
+            if len(node) > 0:
+                if not node.text or not node.text.strip():
+                    # First child's indentation is parent's text
+                    node.text = indent + indent_space
+                last_child = node[-1]
+                if last_child.tail == indent + indent_space:
+                    # Last child's tail is parent's closing tag indentation
+                    last_child.tail = indent
+
+        # Removal condition: node is leaf (not root nor inner node)
+        if parent_node is not None and len(node) == 0:
+            if remove_blank_text and node.text is not None and not node.text.strip():
+                # node.text is None iff node.tag is self-closing (text='' creates closing tag)
+                node.text = ''
+            if remove_blank_nodes and not (node.text or ''):
+                parent_node.remove(node)
+
+    leaf_iter(None, xml_node, indent_level)
+    return xml_node