Add script to generate the new hex numbers on Python update

2 files changed, 45 insertions, 17 deletions

diff --git a/dev/generate_new_unicode_numbers.py b/dev/generate_new_unicode_numbers.py
new file mode 100755
index 0000000..c59f4e4
--- /dev/null
+++ b/dev/generate_new_unicode_numbers.py
@@ -0,0 +1,45 @@
+#! /usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+Generate the numeric hex list of unicode numerals
+"""
+import os
+import os.path
+import sys
+import unicodedata
+
+# This is intended to be called from project root. Enforce this.
+this_file = os.path.abspath(__file__)
+this_base = os.path.basename(this_file)
+cwd = os.path.abspath(os.getcwd())
+desired_this_file = os.path.join(cwd, "dev", this_base)
+if this_file != desired_this_file:
+    sys.exit(f"{this_base} must be called from project root")
+
+# We will write the new numeric hex collection to a natsort package file.
+target = os.path.join(cwd, "natsort", "unicode_numeric_hex.py")
+with open(target, "w") as fl:
+    print(
+        '''# -*- coding: utf-8 -*-
+"""
+Contains all possible non-ASCII unicode numbers.
+"""
+
+# Rather than determine what unicode characters are numeric on the fly which
+# would incur a startup runtime penalty, the hex values are hard-coded below.
+numeric_hex = (''',
+        file=fl,
+    )
+
+    # Write out each individual hex value.
+    for i in range(0x110000):
+        try:
+            a = chr(i)
+        except ValueError:
+            break
+        if a in "0123456789":
+            continue
+        if unicodedata.numeric(a, None) is not None:
+            print(f"    0x{i:X},", file=fl)
+
+    print(")", file=fl)
diff --git a/natsort/unicode_numeric_hex.py b/natsort/unicode_numeric_hex.py
index ba5841b..c1e789d 100644
--- a/natsort/unicode_numeric_hex.py
+++ b/natsort/unicode_numeric_hex.py
@@ -1859,20 +1859,3 @@ numeric_hex = (
     0x2626D,
     0x2F890,
 )
-
-# Some code that can be used to create the above list of hex numbers.
-if __name__ == "__main__":
-    import unicodedata
-
-    hex_chars = []
-    for i in range(0x110000):
-        try:
-            a = chr(i)
-        except ValueError:
-            break
-        if a in "0123456789":
-            continue
-        if unicodedata.numeric(a, None) is not None:
-            hex_chars.append(i)
-
-    print(", ".join(["0X{:X}".format(i) for i in hex_chars]))