Add script and region constants

These are extracted from the Universal Interfaces Script.h file.

scripts/README.md

@@ -0,0 +1,5 @@
+# Mac OS Region Definitions
+
+This folder contains the script and region definitions for the Mac OS toolbox.
+
+These constants are extracted from the `Script.h` file in Universal Interfaces.

scripts/extract.py

@@ -0,0 +1,44 @@
+"""Extract script and region constants from Script.h."""
+import csv
+import re
+import sys
+
+from typing import List, Tuple
+
+Item = Tuple[str, int]
+
+def index_of(data: List[Item], key: str) -> int:
+    for i, (name, _) in enumerate(data):
+        if name == key:
+            return i
+    raise ValueError('missing value: {!r}'.format(key))
+
+def slice(data: List[Item], first: str, last: str) -> List[Item]:
+    return data[index_of(data, first):index_of(data, last)+1]
+
+def write_csv(fname: str, data: List[Item]) -> None:
+    with open(fname, 'w') as fp:
+        w = csv.writer(fp)
+        w.writerow(['Name', 'Value'])
+        for item in data:
+            w.writerow(item)
+
+def main(argv: List[str]) -> None:
+    if len(argv) != 1:
+        print('usage: script_gen.py <Script.h>', file=sys.stderr)
+        raise SystemExit(2)
+    with open(argv[0], 'rb') as fp:
+        data = fp.read()
+    scripts: List[Item] = []
+    regions: List[Item] = []
+    for item in re.finditer(rb'^\s*(\w+)\s*=\s*(\d+)', data, re.MULTILINE):
+        name, value = item.groups()
+        if name.startswith(b'sm'):
+            scripts.append((name.decode('ASCII'), int(value)))
+        elif name.startswith(b'ver'):
+            regions.append((name.decode('ASCII'), int(value)))
+    write_csv('script.csv', slice(scripts, 'smRoman', 'smUninterp'))
+    write_csv('region.csv', slice(regions, 'verUS', 'verGreenland'))
+
+if __name__ == '__main__':
+    main(sys.argv[1:])

scripts/region.csv

@@ -0,0 +1,101 @@
+Name,Value
+verUS,0
+verFrance,1
+verBritain,2
+verGermany,3
+verItaly,4
+verNetherlands,5
+verFlemish,6
+verSweden,7
+verSpain,8
+verDenmark,9
+verPortugal,10
+verFrCanada,11
+verNorway,12
+verIsrael,13
+verJapan,14
+verAustralia,15
+verArabic,16
+verFinland,17
+verFrSwiss,18
+verGrSwiss,19
+verGreece,20
+verIceland,21
+verMalta,22
+verCyprus,23
+verTurkey,24
+verYugoCroatian,25
+verNetherlandsComma,26
+verBelgiumLuxPoint,27
+verCanadaComma,28
+verCanadaPoint,29
+vervariantPortugal,30
+vervariantNorway,31
+vervariantDenmark,32
+verIndiaHindi,33
+verPakistanUrdu,34
+verTurkishModified,35
+verItalianSwiss,36
+verInternational,37
+verRomania,39
+verGreecePoly,40
+verLithuania,41
+verPoland,42
+verHungary,43
+verEstonia,44
+verLatvia,45
+verSami,46
+verFaroeIsl,47
+verIran,48
+verRussia,49
+verIreland,50
+verKorea,51
+verChina,52
+verTaiwan,53
+verThailand,54
+verScriptGeneric,55
+verCzech,56
+verSlovak,57
+verFarEastGeneric,58
+verMagyar,59
+verBengali,60
+verByeloRussian,61
+verUkraine,62
+verGreeceAlt,64
+verSerbian,65
+verSlovenian,66
+verMacedonian,67
+verCroatia,68
+verGermanReformed,70
+verBrazil,71
+verBulgaria,72
+verCatalonia,73
+verMultilingual,74
+verScottishGaelic,75
+verManxGaelic,76
+verBreton,77
+verNunavut,78
+verWelsh,79
+verIrishGaelicScript,81
+verEngCanada,82
+verBhutan,83
+verArmenian,84
+verGeorgian,85
+verSpLatinAmerica,86
+verTonga,88
+verFrenchUniversal,91
+verAustria,92
+verGujarati,94
+verPunjabi,95
+verIndiaUrdu,96
+verVietnam,97
+verFrBelgium,98
+verUzbek,99
+verSingapore,100
+verNynorsk,101
+verAfrikaans,102
+verEsperanto,103
+verMarathi,104
+verTibetan,105
+verNepal,106
+verGreenland,107

scripts/script.csv

@@ -0,0 +1,35 @@
+Name,Value
+smRoman,0
+smJapanese,1
+smTradChinese,2
+smKorean,3
+smArabic,4
+smHebrew,5
+smGreek,6
+smCyrillic,7
+smRSymbol,8
+smDevanagari,9
+smGurmukhi,10
+smGujarati,11
+smOriya,12
+smBengali,13
+smTamil,14
+smTelugu,15
+smKannada,16
+smMalayalam,17
+smSinhalese,18
+smBurmese,19
+smKhmer,20
+smThai,21
+smLao,22
+smGeorgian,23
+smArmenian,24
+smSimpChinese,25
+smTibetan,26
+smMongolian,27
+smEthiopic,28
+smGeez,28
+smCentralEuroRoman,29
+smVietnamese,30
+smExtArabic,31
+smUninterp,32