blob: 33d1662ca66e6e6e74802bf65cb4994b2a8b23ba (
plain) (
tree)
|
|
#!/bin/sh
cache()
{
name="/tmp/librune/rtype/$(basename "$1")"
if test ! -f "$name"
then
mkdir -p /tmp/librune/rtype
wget -q "$1" -O "$name"
fi
}
set -e
cd "${0%/*}/.."
exec >include/internal/rtype/gc.h
readonly URL='https://www.unicode.org/Public/UCD/latest/ucd/UnicodeData.txt'
cache "$URL"
cat <<C
/* This file is autogenerated by gen/rtype-gc; DO NOT EDIT. */
#ifndef RUNE_INTERNAL_RTYPE_GC_H
#define RUNE_INTERNAL_RTYPE_GC_H
/* IWYU pragma: private */
/* clang-format off */
#include "../types.h"
#include "../../rtype.h"
#include "../../rune.h"
static const rprop_gc_bf rtype_gc_lat1_tbl[] = {
C
gawk '
BEGIN {
FS = ";"
}
{
s = "GC_" toupper($3)
lo = strtonum("0X" $1)
if ($2 ~ /First/) {
getline
hi = strtonum("0X" $1)
} else
hi = lo
for (i = lo; i <= hi; i++)
props[i] = s
}
END {
for (i = 0; i <= 0xFF; i++)
print props[i] ","
}
' /tmp/librune/rtype/UnicodeData.txt \
| paste -d' ' - - - - - - - - \
| sed 's/^/\t/'
cat <<C
};
static const struct {
rune lo, hi;
rprop_gc_bf val;
} rtype_gc_tbl[] = {
C
gawk '
BEGIN {
FS = ";"
}
{
s = "GC_" toupper($3)
lo = strtonum("0X" $1)
if ($2 ~ /First/) {
getline
hi = strtonum("0X" $1)
} else
hi = lo
for (i = lo; i <= hi; i++)
props[i] = s
}
END {
for (i = 0x100; i <= 0x10FFFF; i++) {
if (!props[i])
continue
lo = i
while (props[lo] == props[i + 1])
i++
printf "\t{RUNE_C(0x%06X), RUNE_C(0x%06X), %s},\n", lo, i, props[lo]
}
}
' /tmp/librune/rtype/UnicodeData.txt | sort
cat <<C
};
#endif /* !RUNE_INTERNAL_RTYPE_GC_H */
C
|