unicodedata functionsisxidstart() andisxidcontinue() were added in#129117. But it added also corresponded methods inunicodedata.ucd_3_2_0 which are identical to module level functions except that they return False for codes not assigned in Unicode 3.2.0.

unicodedata.ucd_3_2_0 exists solely for implementation of obsolete IDNA2003 (RFC 3490 and RFC 3491) in theidna module.isxidstart() andisxidcontinue() are not needed for this. I am not even sure they return correct values for Unicode 3.2.0, because the XID_Start and XID_Continue properties can be added to assigned codes (at least this happened with other properties).

So, I think thatisxidstart() andisxidcontinue() should only be exposed asunicodedata functions, not asunicodedata.ucd_3_2_0 methods. There is already a precedence -- functions related to grapheme cluster breaking are only exposed at the module level. This is because the grapheme cluster break algorithm was completely different in older versions of Unicode, and many properties that are used now did not even exist in older versions of Unicode.

Linked PRs

• gh-143897: Remove the isxidstart() and isxidcontinue() methods of unicodedata.ucd_3_2_0 #143898