Unicode Variables

[NeilFraser]

Current behaviour

Blockly variables have their Unicode characters encoded. Thus the variable '中国' is turned into '_E4_B8_AD_E5_9B_BD'. Such JavaScript is legal, but completely unreadable.

Expected behaviour

It turns out that Unicode characters are allowed in JavaScript variable names.

var 中国 = 'China';
alert(中国);

Or:

var Hͫ̆̒̐ͣ̊̄ͯ͗͏̵̗̻̰̠̬͝ͅE̴̷̬͎̱̘͇͍̾ͦ͊͒͊̓̓̐_̫̠̱̩̭̤͈̑̎̋ͮͩ̒͑̾͋͘Ç̳͕̯̭̱̲̣̠̜͋̍O̴̦̗̯̹̼ͭ̐ͨ̊̈͘͠M̶̝̠̭̭̤̻͓͑̓̊ͣͤ̎͟͠E̢̞̮̹͍̞̳̣ͣͪ͐̈T̡̯̳̭̜̠͕͌̈́̽̿ͤ̿̅̑Ḧ̱̱̺̰̳̹̘̰́̏ͪ̂̽͂̀͠ = 'Zalgo';
alert(Hͫ̆̒̐ͣ̊̄ͯ͗͏̵̗̻̰̠̬͝ͅE̴̷̬͎̱̘͇͍̾ͦ͊͒͊̓̓̐_̫̠̱̩̭̤͈̑̎̋ͮͩ̒͑̾͋͘Ç̳͕̯̭̱̲̣̠̜͋̍O̴̦̗̯̹̼ͭ̐ͨ̊̈͘͠M̶̝̠̭̭̤̻͓͑̓̊ͣͤ̎͟͠E̢̞̮̹͍̞̳̣ͣͪ͐̈T̡̯̳̭̜̠͕͌̈́̽̿ͤ̿̅̑Ḧ̱̱̺̰̳̹̘̰́̏ͪ̂̽͂̀͠);

Here's documentation regarding what is and isn't allowed:
https://mathiasbynens.be/notes/javascript-identifiers

[rachel-fenichel]

[AnmAtAnm]

This doesn't appear to be occuring at a JavaScript specific level. (/me sneaks away for it can be reassigned.)

[NeilFraser]

Yup, it's in Blockly.Names.prototype.safeName_ which states:

Given a proposed entity name, generate a name that conforms to the
[_A-Za-z][_A-Za-z0-9]* format that most languages consider legal for
variables.

A quick check of Dart, Lua and Python 2 shows that they do not support Unicode variables, though Python 3, JavaScript and PHP do.

There's a feasible way to implement this in JavaScript:

eval('"use strict"; var ' + x)

where x is the first letter then the first two letters, then the first three letters of the name. If it blows up, then encode the offending letter. Also, '/n' and '/r' and ';' need to be escaped up front to prevent injection attacks.

[cpcallen]

There's a feasible way to implement this in JavaScript:

eval('"use strict"; var ' + x)

Feasible, yes, but not exactly sensible, as the list of characters that would need to be escaped first is longer than just [\n\r;] (possibly much longer, and certainly not obvious).

A better approach is probably to use a regexp, even though the required regexp 8-10kB long and depends on what version of ECMAScript and Unicode one is targeting.

See: Valid JavaScript variable names in ES2015.