xml: characters which are not white space in XML are treated as white space

[serhiy-storchaka]

XML defines white space as only #x20, #x9, #xD and #xA (XML 1.0, 2.3), but xml.dom and xml.etree.ElementTree use str.strip(), which also strips other characters, such as U+00A0 or U+000C. Such characters are content, and treating them as white space loses them.

ElementTree.indent() overwrites them:

>>> tree = ET.fromstring("<a>\xa0<b>x</b>\xa0</a>")
>>> ET.indent(tree, space="  ")
>>> ET.tostring(tree, encoding="unicode")
'<a>\n  <b>x</b>\n</a>'

canonicalize(strip_text=True) strips them, which changes the canonical form of a document.

In xml.dom, Text.isWhitespaceInElementContent reports such a node as ignorable white space, and the node is removed from the document when parsing with the whitespace-in-element-content feature turned off.

Linked PRs

• gh-156658: Only XML white space characters are treated as white space #156659
• [3.15] gh-156658: Only XML white space characters are treated as white space (GH-156659) #156825
• [3.14] gh-156658: Only XML white space characters are treated as white space (GH-156659) #156826
• [3.13] gh-156658: Only XML white space characters are treated as white space (GH-156659) #156827