gen_wc8bit.c

Axis 221 camera embedded programing interface

#define _GNU_SOURCE
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <locale.h>
#include <stddef.h>
#include <wctype.h>
#include <limits.h>

#ifndef _CTYPE_H
#define _CTYPE_H
#endif
#ifndef _WCTYPE_H
#define _WCTYPE_H
#endif
#include "../../libc/sysdeps/linux/common/bits/uClibc_ctype.h"


/*  #define CTYPE_PACKED */
#define UPLOW_IDX_SHIFT		3
/* best if 2 unpacked or 3 packed */
#define CTYPE_IDX_SHIFT		3
/* 3 or 4 are very similar */
#define C2WC_IDX_SHIFT		3

#define CTYPE_IDX_LEN		(128 >> (CTYPE_IDX_SHIFT))
#define UPLOW_IDX_LEN		(128 >> (UPLOW_IDX_SHIFT))
#define C2WC_IDX_LEN		(128 >> (C2WC_IDX_SHIFT))

/*  #ifdef CTYPE_PACKED */
/*  #define CTYPE_ROW_LEN		(1 << ((CTYPE_IDX_SHIFT)-1)) */
/*  #else */
#define CTYPE_ROW_LEN		(1 << (CTYPE_IDX_SHIFT))
/*  #endif */
#define UPLOW_ROW_LEN		(1 << (UPLOW_IDX_SHIFT))
#define C2WC_ROW_LEN		(1 << (C2WC_IDX_SHIFT))



#define MAX_WCHAR	(0x2600-1)

static unsigned char ctype_tbl[256 * CTYPE_ROW_LEN];
static unsigned char uplow_tbl[256 * UPLOW_ROW_LEN];
#ifdef DO_WIDE_CHAR
static unsigned short c2wc_tbl[256 * C2WC_ROW_LEN];
#endif
static unsigned char tt[MAX_WCHAR+1];
static unsigned char ti[MAX_WCHAR+1];
static unsigned char xi[MAX_WCHAR+1];

static int n_ctype_rows;
static int n_uplow_rows;
#ifdef DO_WIDE_CHAR
static int n_c2wc_rows;
#endif
static int tt_num;
static int ti_num;

#define RANGE MAX_WCHAR

#define TT_SHIFT 4
#define TI_SHIFT 4

#define II_LEN		((MAX_WCHAR+1) >> (TT_SHIFT+TI_SHIFT))

typedef struct {
	unsigned long c2w[256];
	unsigned char w2c[MAX_WCHAR];
	unsigned char ii[II_LEN];
	unsigned char ctype_idx[CTYPE_IDX_LEN];
	unsigned char uplow_idx[UPLOW_IDX_LEN];
	unsigned char c2wc_idx[C2WC_IDX_LEN];
} charset_data;

int main(int argc, char **argv)
{
	FILE *fp;
	FILE *out;
	charset_data csd[30];
	unsigned long max_wchar;
	unsigned char *p;
	int numsets;
	int i;
	int j;
	char buf[80];
	unsigned char row[256];
#ifdef DO_WIDE_CHAR
	unsigned short wrow[256];
#endif
	char codeset_list[500];
	char codeset_index[30];
	int codeset_list_end = 0;
	int total_size = 0;

	if (!setlocale(LC_CTYPE, "en_US.UTF-8")) {
		printf("setlocale(LC_CTYPE,\"en_US.UTF-8\") failed!\n");
		return EXIT_FAILURE;
	}

	if (!(out = fopen("c8tables.h","w"))) {
		printf("error: couldn't open file \"c8tables.h\"\n");
		return EXIT_FAILURE;
	}

#if 0
	if (argc == 1) {
		/* User requested 8-bit codesets, but didn't list any... */
		/* Allow to build, just so this feature can be left on in config. */
		fprintf(out, "#ifdef __CTYPE_HAS_8_BIT_LOCALES\n");
		fprintf(out, "#warning ignoring 8 bit codesets request"
				" as no codesets specified.\n");
		fprintf(out, "#endif\n");
		fprintf(out, "#undef __CTYPE_HAS_8_BIT_LOCALES\n\n");

		fprintf(out, "#define __LOCALE_DATA_NUM_CODESETS\t\t0\n");
		fprintf(out, "#define __LOCALE_DATA_CODESET_LIST\t\t\"\"\n");
		fclose(out);
		return EXIT_SUCCESS;
	}

/*  	fprintf(out, "#define __CTYPE_HAS_8_BIT_LOCALES\t1\n\n"); */
	fprintf(out, "#ifdef __CTYPE_HAS_8_BIT_LOCALES\n\n");
#endif

	if (argc == 1) {
		fprintf(out, "#undef __CTYPE_HAS_8_BIT_LOCALES\n\n");

		fprintf(out, "#define __LOCALE_DATA_NUM_CODESETS\t\t0\n");
		fprintf(out, "#define __LOCALE_DATA_CODESET_LIST\t\t\"\"\n");
	} else {
		fprintf(out, "#define __CTYPE_HAS_8_BIT_LOCALES\t\t1\n\n");
	}

	fprintf(out, "#define __LOCALE_DATA_Cctype_IDX_SHIFT\t%d\n", CTYPE_IDX_SHIFT);
	fprintf(out, "#define __LOCALE_DATA_Cctype_IDX_LEN\t\t%d\n", CTYPE_IDX_LEN);
#ifdef CTYPE_PACKED
	fprintf(out, "#define __LOCALE_DATA_Cctype_ROW_LEN\t\t%d\n", CTYPE_ROW_LEN >> 1);
	fprintf(out, "#define __LOCALE_DATA_Cctype_PACKED\t\t1\n");
#else
	fprintf(out, "#define __LOCALE_DATA_Cctype_ROW_LEN\t\t%d\n", CTYPE_ROW_LEN);
	fprintf(out, "#undef __LOCALE_DATA_Cctype_PACKED\n");
#endif

	fprintf(out, "\n#define __LOCALE_DATA_Cuplow_IDX_SHIFT\t%d\n", UPLOW_IDX_SHIFT);
	fprintf(out, "#define __LOCALE_DATA_Cuplow_IDX_LEN\t\t%d\n", UPLOW_IDX_LEN);
	fprintf(out, "#define __LOCALE_DATA_Cuplow_ROW_LEN\t\t%d\n", UPLOW_ROW_LEN);

#ifdef DO_WIDE_CHAR
	fprintf(out, "\n#define __LOCALE_DATA_Cc2wc_IDX_LEN\t\t%d\n", C2WC_IDX_LEN);
	fprintf(out, "#define __LOCALE_DATA_Cc2wc_IDX_SHIFT\t\t%d\n", C2WC_IDX_SHIFT);
	fprintf(out, "#define __LOCALE_DATA_Cc2wc_ROW_LEN\t\t%d\n", C2WC_ROW_LEN);
#endif

	fprintf(out, "\ntypedef struct {\n");
	fprintf(out, "\tunsigned char idx8ctype[%d];\n", CTYPE_IDX_LEN);
	fprintf(out, "\tunsigned char idx8uplow[%d];\n", UPLOW_IDX_LEN);
#ifdef DO_WIDE_CHAR
	fprintf(out, "\tunsigned char idx8c2wc[%d];\n", C2WC_IDX_LEN);
	fprintf(out, "\tunsigned char idx8wc2c[%d];\n", II_LEN);
#endif
	fprintf(out, "} __codeset_8_bit_t;\n\n");

	fprintf(out, "#ifdef WANT_DATA\n\n");
	fprintf(out, "static const __codeset_8_bit_t codeset_8_bit[%d] = {\n", argc-1);

	max_wchar = 0x7f;
	numsets = 0;
	codeset_index[0] = 0;
	while (--argc) {
		if (!(fp = fopen(*++argv,"r"))) {
			printf("error: couldn't open file \"%s\"\n", *argv);
			return EXIT_FAILURE;
		}
		printf("processing %s... ", *argv);

		{
			char *s0;
			char *s1;
			int n;
			
			s0 = strrchr(*argv, '/');
			if (!s0) {
				s0 = *argv;
			} else {
				++s0;
			}
			s1 = strchr(s0, '.');
			if (!s1) {
				n = strlen(s0);
			} else {
				n = s1 - s0;
			}

/*  			if ((numsets == 0) && strncmp("ASCII", s0, n)) { */
/*  				printf("error - first codeset isn't ASCII!\n"); */
/*  				return EXIT_FAILURE; */
/*  			} */

			if (numsets >= sizeof(codeset_index)) {
				printf("error - too many codesets!\n");
				return EXIT_FAILURE;
			}

			if (codeset_list_end + n + 1 + numsets + 1 + 1 >= 256) {
				printf("error - codeset list to big!\n");
				return EXIT_FAILURE;
			}

			codeset_index[numsets+1] = codeset_index[numsets] + n+1;
			strncpy(codeset_list + codeset_list_end, s0, n);
			codeset_list_end += (n+1);
			codeset_list[codeset_list_end - 1] = 0;

			fprintf(out, "\t{ /* %.*s */", n, s0);
		}

		memset(&csd[numsets],sizeof(charset_data),0);
		memset(xi, sizeof(xi), 0);
		{
			unsigned long c, wc;
			int lines;
			lines = 0;
			while (fgets(buf,sizeof(buf),fp)) {
				if ((2 != sscanf(buf, "{ %lx , %lx", &c, &wc))
					|| (c >= 256) || (wc > MAX_WCHAR)) {
					printf("error: scanf failure! \"%s\"\n", buf);
					return EXIT_FAILURE;
				}

				/* don't put in w2c... dynamicly build tt instead. */

				if (c <= 0x7f) { /* check the 7bit entries but don't store */
					if (c != wc) {
						printf("error: c != wc in %s\n", buf);
						return EXIT_FAILURE;
					}
					csd[numsets].c2w[c] = wc;
					csd[numsets].w2c[wc] = 0; /* ignore */
					if (wc > max_wchar) {
						max_wchar = wc;
					}
				} else {
					csd[numsets].c2w[c] = wc;
					csd[numsets].w2c[wc] = c;
					if (wc > max_wchar) {
						max_wchar = wc;
					}
				}
				++lines;
			}
			printf("%d lines  ", lines);

			for (i = 0 ; i <= MAX_WCHAR ; i += (1 << TT_SHIFT)) {
				p = &csd[numsets].w2c[i];
				for (j = 0 ; j < tt_num ; j++) {
					if (!memcmp(p, &tt[j << TT_SHIFT], (1 << TT_SHIFT))) {
						break;
					}
				}
				if (j == tt_num) { /* new entry */
					memcpy(&tt[j << TT_SHIFT], p, (1 << TT_SHIFT));
					++tt_num;
				}
				xi[i >> TT_SHIFT] = j;
			}

			for (i = 0 ; i <= (MAX_WCHAR >> TT_SHIFT)  ; i += (1 << TI_SHIFT)) {
				p = &xi[i];
				for (j = 0 ; j < ti_num ; j++) {
					if (!memcmp(p, &ti[j << TI_SHIFT], (1 << TI_SHIFT))) {
						break;
					}
				}
				if (j == ti_num) { /* new entry */
					memcpy(&ti[j << TI_SHIFT], p, (1 << TI_SHIFT));
					++ti_num;
				}
				csd[numsets].ii[i >> TI_SHIFT] = j;
/*  				printf("%d ", i >> TI_SHIFT); */
			}

#if 1
			fprintf(out, "\n\t\t/* idx8ctype data */\n\t\t{");
			for (i = 128 ; i < 256 ; i++) {
				wchar_t c;
				unsigned int d;

/*  				if (!(i & 0x7)) { */
/*  					fprintf(out, "\n"); */
/*  				} */

				c = csd[numsets].c2w[i];

				if (c == 0) {	/* non-existant char in codeset */
					d = __CTYPE_unclassified;
				} else if (iswdigit(c)) {
					d = __CTYPE_digit;
				} else if (iswalpha(c)) {
					d = __CTYPE_alpha_nonupper_nonlower;
					if (iswlower(c)) {
						d = __CTYPE_alpha_lower;
						if (iswupper(c)) {
							d = __CTYPE_alpha_upper_lower;
						}
					} else if (iswupper(c)) {
						d = __CTYPE_alpha_upper;
					}
				} else if (iswpunct(c)) {
					d = __CTYPE_punct;
				} else if (iswgraph(c)) {
					d = __CTYPE_graph;
				} else if (iswprint(c)) {
					d = __CTYPE_print_space_nonblank;
					if (iswblank(c)) {
						d = __CTYPE_print_space_blank;
					}
				} else if (iswspace(c) && !iswcntrl(c)) {
					d = __CTYPE_space_nonblank_noncntrl;
					if (iswblank(c)) {
						d = __CTYPE_space_blank_noncntrl;
					}
				} else if (iswcntrl(c)) {
					d = __CTYPE_cntrl_nonspace;
					if (iswspace(c)) {
						d = __CTYPE_cntrl_space_nonblank;
						if (iswblank(c)) {
							d = __CTYPE_cntrl_space_blank;
						}
					}
				} else {
					d = __CTYPE_unclassified;
				}

#if 1
				row[i & (CTYPE_ROW_LEN-1)] = d;
				if ((i & (CTYPE_ROW_LEN-1)) == (CTYPE_ROW_LEN-1)) {
					p = ctype_tbl;
					for (j=0 ; j < n_ctype_rows ; j++) {
						if (!memcmp(p, row, CTYPE_ROW_LEN)) {
							break;
						}
						p += CTYPE_ROW_LEN;
					}
					if (j == n_ctype_rows) { /* new entry */
						if (++n_ctype_rows > 256) {
							printf("error -- to many ctype rows!\n");
							return EXIT_FAILURE;
						}
						memcpy(p, row, CTYPE_ROW_LEN);
					}
					csd[numsets].ctype_idx[i >> CTYPE_IDX_SHIFT] = j;
					if (!((i >> CTYPE_IDX_SHIFT) & 0x7)
						&& (i != (127 + CTYPE_ROW_LEN))
						) {
						fprintf(out, "\n\t\t ");
					}