[FIX] tools: classify \uFEFF as whitespace for translations

Versions
--------
- 15.0
- 16.0
- 17.0
- saas-17.1
- saas-17.2

Issue
-----
Commit 9426ee5 introduced the \uFEFF character to web_editor as a
zero-width non-breaking whitespace. When this gets added to a HTML node,
and processed for translation, it throws an "empty document" error.

Cause
-----
When passed to the `get_text_content` function, the call to
`html.fromstring('\uFEFF').text_content()` throws an error.
\uFEFF is not technically classified as whitespace, so the `nonspace`
function which attempts to prevent processing empty documents doesn't
catch it.

Solution
--------
Instead of the `isspace` method, use a regex which matches on all
whitespace as well as \uFEFF.

To be applied on stable versions while the origins of stray ZWNBSPs
get tackled on master.

opw-3957259

closes odoo#171578

X-original-commit: 5e70361
Signed-off-by: Raphael Collet <[email protected]>
Signed-off-by: Levi Siuzdak <[email protected]>

odoo/tools/translate.py

@@ -176,6 +176,7 @@ def translate_attrib_value(node):
 )
 
 avoid_pattern = re.compile(r"\s*<!DOCTYPE", re.IGNORECASE | re.MULTILINE | re.UNICODE)
+space_pattern = re.compile(r"[\s\uFEFF]*")  # web_editor uses \uFEFF as ZWNBSP
 
 
 def translate_xml_node(node, callback, parse, serialize):
@@ -189,7 +190,7 @@ def translate_xml_node(node, callback, parse, serialize):
 
     def nonspace(text):
         """ Return whether ``text`` is a string with non-space characters. """
-        return bool(text) and not text.isspace()
+        return bool(text) and not space_pattern.fullmatch(text)
 
     def translatable(node):
         """ Return whether the given node can be translated as a whole. """