[FIX] core: fix check for text-only translated terms

`get_text_content` will transform contiguous space chars into single
spaces, plus translate special HTML elements
```
>>> " ".join(html.fromstring(f"a\n    b & c").text_content().split())
'a b & c'
```

In order the correctly verify if a term is text-only we need to use the
HTML parser. Note that to be resilient against bad XML, but valid HTML,
we cannot use the default XML parser.

Part-of: odoo/odoo#150152

odoo/fields.py

@@ -1853,14 +1853,15 @@ class _String(Field):
             for term in new_terms:
                 text2terms[self.get_text_content(term)].append(term)
 
+            is_text = self.translate.is_text if hasattr(self.translate, 'is_text') else lambda term: True
             for old_term in list(translation_dictionary.keys()):
                 if old_term not in new_terms:
                     old_term_text = self.get_text_content(old_term)
                     matches = get_close_matches(old_term_text, text2terms, 1, 0.9)
                     if matches:
                         closest_term = get_close_matches(old_term, text2terms[matches[0]], 1, 0)[0]
-                        old_is_text = old_term == self.get_text_content(old_term)
-                        closest_is_text = closest_term == self.get_text_content(closest_term)
+                        old_is_text = is_text(old_term)
+                        closest_is_text = is_text(closest_term)
                         if old_is_text or not closest_is_text:
                             translation_dictionary[closest_term] = translation_dictionary.pop(old_term)
             # pylint: disable=not-callable

odoo/tools/translate.py

@@ -367,12 +367,19 @@ def get_text_content(term):
     content = html.fromstring(term).text_content()
     return " ".join(content.split())
 
+def is_text(term):
+    """ Return whether the term has only text. """
+    return len(html.fromstring(f"<_>{term}</_>")) == 0
+
 xml_translate.get_text_content = get_text_content
 html_translate.get_text_content = get_text_content
 
 xml_translate.term_converter = xml_term_converter
 html_translate.term_converter = html_term_converter
 
+xml_translate.is_text = is_text
+html_translate.is_text = is_text
+
 def translate_sql_constraint(cr, key, lang):
     cr.execute("""
         SELECT COALESCE(c.message->>%s, c.message->>'en_US') as message