src/gen-arabic-pua.py

#!/usr/bin/env python3

"""usage: ./gen-arabic-pua.py [--rust] ArabicPUASimplified.txt ArabicPUATraditional.txt
"""

import sys

import packTab

if len(sys.argv) > 1 and sys.argv[1] == "--rust":
    del sys.argv[1]
    language = packTab.languages["rust"]
else:
    language = packTab.languages["c"]

if len(sys.argv) != 3:
    sys.exit(__doc__)

paths = sys.argv[1:]

print ("/* == Start of generated table == */")
print ("/*")
print (" * The following table is generated by running:")
print (" *")
print (" *   ./gen-arabic-pua.py%s ArabicPUASimplified.txt ArabicPUATraditional.txt" %
       (" --rust" if language.name == "rust" else ""))
print (" *")
print (" */")
print ()
if language.name == "c":
    print ("#ifndef HB_OT_SHAPER_ARABIC_PUA_HH")
    print ("#define HB_OT_SHAPER_ARABIC_PUA_HH")
    print ()
    print ('#include "hb.hh"')
    print ()
else:
    print ("#![allow(clippy::unreadable_literal)]")
    print ()

code = packTab.Code('_hb_arabic_pua')

for p, name in zip(paths, ("simp_map", "trad_map")):
    with open (p, encoding='utf-8') as f:
        fields = [l.split('\t') for l in f if l[:1] != '#']
    data = {int(fs[1], 16):int(fs[0], 16) for fs in fields}
    sol = packTab.pack_table(data, compression=9)
    sol.genCode(
        code,
        name,
        language=language,
        private=language.name == "c",
    )

code.print_code(
    file=sys.stdout,
    language=language,
    private=language.name == "c",
)

if language.name == "c":
    print ()
    print ("#endif /* HB_OT_SHAPER_ARABIC_PUA_HH */")
print ()
print ("/* == End of generated table == */")