add unicodedata

unicodedata.go

@@ -0,0 +1,98 @@
+package main
+
+import (
+	"bufio"
+	"bytes"
+	_ "embed"
+	"flag"
+	"fmt"
+	"os"
+	"strconv"
+	"strings"
+)
+
+//go:embed UnicodeData.txt
+var unicodedata []byte
+
+func die(msg string) {
+	fmt.Fprintln(os.Stderr, msg)
+	fmt.Fprintln(os.Stderr, "USAGE:", os.Args[0], "[-ranges RANGES]")
+	flag.PrintDefaults()
+	os.Exit(-1)
+}
+
+type rangeInterval struct {
+	from int64
+	to   int64
+}
+
+func (r *rangeInterval) contains(i int64) bool {
+	return r.from <= i && i <= r.to
+}
+
+func parseHex(s string) (int64, error) {
+	return strconv.ParseInt(s, 16, 32)
+}
+
+func parseRanges(ranges string) []rangeInterval {
+	intervals := []rangeInterval{}
+	if strings.TrimSpace(ranges) != "" {
+		splitranges := strings.Split(ranges, " ")
+		for _, r := range splitranges {
+			fromto := strings.Split(r, "..")
+			if len(fromto) != 2 {
+				die("invalid ranges")
+			}
+			from, err := parseHex(fromto[0])
+			if err != nil {
+				die("invalid ranges")
+			}
+			to, err := parseHex(fromto[1])
+			if err != nil {
+				die("invalid ranges")
+			}
+			intervals = append(intervals, rangeInterval{from, to})
+		}
+	}
+	return intervals
+}
+
+func printCodepoint(i int64, description string) {
+	var ch rune = rune(i)
+	var s string = string([]rune{ch})
+	fmt.Println(s, description)
+}
+
+func main() {
+	var ranges string
+	flag.StringVar(&ranges, "ranges", "", "define multiple codepoint ranges with \"N1..N2 N3..N4\"")
+	flag.Parse()
+	rangeIntervals := parseRanges(ranges)
+
+	scanner := bufio.NewScanner(bytes.NewReader(unicodedata))
+	for scanner.Scan() {
+		line := scanner.Text()
+		splitline := strings.Split(line, ";")
+		if len(splitline) < 2 || len(splitline[1]) == 0 {
+			continue
+		}
+		codepoint, description := splitline[0], splitline[1]
+		if (description[0] == '<' && description[len(description)-1] == '>') || strings.Contains(description, "MODIFIER") || strings.Contains(description, "VARIATION SELECTOR") {
+			continue
+		}
+
+		i, err := parseHex(codepoint)
+		if err != nil {
+			continue
+		}
+		if len(rangeIntervals) == 0 {
+			printCodepoint(i, description)
+		}
+		for _, r := range rangeIntervals {
+			if r.contains(i) {
+				printCodepoint(i, description)
+				break
+			}
+		}
+	}
+}