workflows: do not feed ascii control characters to FTS (#88716)

tests/workflow/test_register_comment.py

@@ -607,3 +607,22 @@ def test_register_comment_to_with_attachment(pub):
     assert 'to-role.txt' in display_parts()[2]
     assert 'to-submitter.txt' in display_parts()[4]
     assert 'to-role-or-submitter.txt' in display_parts()[6]
+
+
+def test_register_comment_fts(pub):
+    pub.substitutions.feed(MockSubstitutionVariables())
+
+    formdef = FormDef()
+    formdef.name = 'baz'
+    formdef.fields = []
+    formdef.store()
+
+    formdata = formdef.data_class()()
+    formdata.just_created()
+    formdata.store()
+
+    item = RegisterCommenterWorkflowStatusItem()
+    item.comment = 'Hello\x00\nworld'
+    item.perform(formdata)
+    assert formdata.evolution[-1].parts[-1].content == '<p>Hello\x00\nworld</p>'  # kept
+    assert formdata.evolution[-1].parts[-1].render_for_fts() == 'Hello  world'  # not kept

wcs/workflows.py

@@ -360,7 +360,8 @@ class EvolutionPart:
         if not self.view or self.to:
             # don't include parts with no content or restricted visibility
             return ''
-        return misc.html2text(self.view() or '')
+        illegal_fts_chars = re.compile(r'[\x00-\x1F]')
+        return illegal_fts_chars.sub(' ', misc.html2text(self.view() or ''))
 
 
 class AttachmentEvolutionPart(EvolutionPart):