1 files changed, 125 insertions, 0 deletions
diff --git a/tools/make_charset_table.c b/tools/make_charset_table.c
new file mode 100644
index 0000000..27d921a
--- /dev/null
+++ b/tools/make_charset_table.c
@@ -0,0 +1,125 @@
+/* make_charset_table.c
+ * sample program to generate tables for charsets.c using iconv
+ *
+ * public domain
+ */
+
+#include <stdio.h>
+#include <stdint.h>
+#include <errno.h>
+#include <iconv.h>
+
+#define UNREPL 0xFFFD
+
+int main(int argc, char **argv) {
+	/* for now only UCS-2 */
+	uint16_t table[0x100];
+
+	iconv_t conv;
+	const char *charset;
+	int i, j;
+
+	/* 0x00 ... 0x7F same as ASCII? */
+	int ascii_based = 1;
+	/* 0x00 ... 0x9F same as ISO? */
+	int iso_based = 1;
+
+	if (argc != 2) {
+		printf("usage: %s <charset>\n", argv[0]);
+		return 1;
+	}
+
+	charset = argv[1];
+
+	conv = iconv_open("UCS-2", charset);
+	if (conv == (iconv_t) -1) {
+		perror("iconv_open");
+		return 2;
+	}
+	iconv_close(conv);
+
+	for (i = 0x00; i < 0x100; i++) {
+		unsigned char in[1], out[2];
+		size_t inlen = 1, outlen = 2;
+
+		char *inbuf = (char *) in;
+		char *outbuf = (char *) out;
+
+		size_t ret;
+
+		in[0] = i;
+
+		conv = iconv_open("UCS-2BE", charset);
+
+		if (conv == (iconv_t) -1) {
+			/* shouldn't fail now */
+			perror("iconv_open");
+			return 2;
+		}
+
+		ret = iconv(conv, &inbuf, &inlen, &outbuf, &outlen);
+
+		if (ret == (size_t) -1 && errno == EILSEQ) {
+			table[i] = UNREPL;
+			iconv_close(conv);
+			continue;
+		}
+
+		if (ret == (size_t) -1) {
+			perror("iconv");
+			iconv_close(conv);
+			return 4;
+		}
+
+		iconv_close(conv);
+
+		if (ret != 0 || inlen != 0 || outlen != 0) {
+			fprintf(stderr, "%d: smth went wrong: %zu %zu %zu\n", i, ret, inlen, outlen);
+			return 3;
+		}
+
+		if (i < 0x80 && (out[0] != 0 || out[1] != i))
+			ascii_based = 0;
+
+		if (i < 0xA0 && (out[0] != 0 || out[1] != i))
+			iso_based = 0;
+			
+		table[i] = (out[0] << 8) | out[1];
+	}
+
+	/* iso_based not supported */
+	iso_based = 0;
+
+	printf("/* generated by %s %s */\n", argv[0], charset);
+
+	if (iso_based)
+		i = 0xA0;
+	else if (ascii_based)
+		i = 0x80;
+	else
+		i = 0;
+
+	printf("const gunichar2 charset_table_%s[0x%x] = {\n", charset, 0x100 - i);
+	while (i < 0x100) {
+		int start = i;
+
+		printf("    ");
+
+		for (j = 0; j < 8; j++, i++) {
+			if (table[i] == UNREPL)
+				printf("UNREPL, ");
+			else
+				printf("0x%.4x, ", table[i]);
+		}
+
+		if ((start & 0xf) == 0)
+			printf("       /* 0x%.2X -      */", start);
+		else
+			printf("       /*      - 0x%.2X */", i - 1);
+
+		printf("\n");
+	}
+	printf("};\n");
+
+	return 0;
+}