[unicodedata] script and script_extension return names in different format

I just noticed that script function returns long script names (e.g. “Bengali”) whereas script_extension returns short ones (e.g. {"Beng"}). This reflects the fact that in Scripts.txt they use the long names, whereas in ScriptExtensions.txt the short ones.

The problem is with the current implementation of script_extension that delegates to script whenever a codepoint has not been explicitly assigned a script extension: it returns a set containing a single script name, however using the long format which is inconsistent with the other results from script_extension function.

So, we also need to parse the PropertyValueAliases.txt and load the mapping between long and short script names, and return one or the other, but not a mix of the two.

Would you prefer that both script and script_extension functions return the long version, or that both return the short version?

Or we keep the difference, and script returns the long version (“Bengali”), whereas the script_extension returns the short name (“Beng”).

[twardoch]

I think it would be best if the script and script_extension functions always returned the Unicode script codes (4-letter names). The Unicode script codes should be a reference for other functions.

• Scripts.py should get a NAMES dictionary which maps the script codes to the longer names (with _ kept, so it’s closer to UnicodeData)
• a script_name function could take a script code and would return the long name (with _ replaced by ); this would in a sense mirror the unicodadata.name function which returns the character name for a Unicode codepoint; people could query Scripts.NAMES for a unique list of the script codes

[anthrotype]

Thanks Adam. Ok, I’ll have both functions return the four-character script codes and add a script_name function as well.