Simplify test_ident using str.isidentifier()

I have checked that this gives the same results as the previous method.

tools/gen_latex_symbols.py

@@ -1,87 +1,84 b''
 # coding: utf-8
 
 # This script autogenerates `IPython.core.latex_symbols.py`, which contains a
 # single dict , named `latex_symbols`. The keys in this dict are latex symbols,
 # such as `\\alpha` and the values in the dict are the unicode equivalents for
 # those. Most importantly, only unicode symbols that are valid identifers in
 # Python 3 are included. 
 
 # 
 # The original mapping of latex symbols to unicode comes from the `latex_symbols.jl` files from Julia.
 
 from __future__ import print_function
 import os, sys
 
 if not sys.version_info[0] == 3:
     print("This script must be run with Python 3, exiting...")
     sys.exit(1)
 
 # Import the Julia LaTeX symbols
 print('Importing latex_symbols.js from Julia...')
 import requests
 url = 'https://raw.githubusercontent.com/JuliaLang/julia/master/base/latex_symbols.jl'
 r = requests.get(url)
 
 
 # Build a list of key, value pairs
 print('Building a list of (latex, unicode) key-vaule pairs...')
 lines = r.text.splitlines()[60:]
 lines = [line for line in lines if '=>' in line]
 lines = [line.replace('=>',':') for line in lines]
 
 def line_to_tuple(line):
     """Convert a single line of the .jl file to a 2-tuple of strings like ("\\alpha", "α")"""
     kv = line.split(',')[0].split(':')
 #     kv = tuple(line.strip(', ').split(':'))
     k, v = kv[0].strip(' "'), kv[1].strip(' "')
 #     if not test_ident(v):
 #         print(line)
     return k, v
 
 assert line_to_tuple('    "\\sqrt" : "\u221A",') == ('\\sqrt', '\u221A')
 lines = [line_to_tuple(line) for line in lines]
 
 
 # Filter out non-valid identifiers
 print('Filtering out characters that are not valid Python 3 identifiers')
 
 def test_ident(i):
-    """Is the unicode string a valid Python 3 identifer."""
-    try:
-        exec('a%s = 10' % i, {}, {})
-    except SyntaxError:
-        return False
-    else:
-        return True
+    """Is the unicode string valid in a Python 3 identifer."""
+    # Some characters are not valid at the start of a name, but we still want to
+    # include them. So prefix with 'a', which is valid at the start.
+    return ('a' + i).isidentifier()
 
 assert test_ident("α")
 assert not test_ident('‴')
 
 valid_idents = [line for line in lines if test_ident(line[1])]
 
 
 # Write the `latex_symbols.py` module in the cwd
 
 s = """# encoding: utf-8
 
 # DO NOT EDIT THIS FILE BY HAND.
 
 # To update this file, run the script /tools/gen_latex_symbols.py using Python 3
 
 # This file is autogenerated from the file:
 # https://raw.githubusercontent.com/JuliaLang/julia/master/base/latex_symbols.jl
 # This original list is filtered to remove any unicode characters that are not valid
 # Python identifiers.
 
 latex_symbols = {\n
 """
 for line in valid_idents:
     s += '    "%s" : "%s",\n' % (line[0], line[1])
 s += "}\n"
 
 fn = os.path.join('..','IPython','core','latex_symbols.py')
 print("Writing the file: %s" % fn)
 with open(fn, 'w', encoding='utf-8') as f:
     f.write(s)