mbfilter_jis.c

php-4.4.7学习linux时下载的源代码

			s = -1;
		}
	}
	if (s >= 0) {
		if (s < 0x80) { /* ASCII */
			if ((filter->status & 0xff00) != 0) {
				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
			}
			filter->status = 0;
			CK((*filter->output_function)(s, filter->data));
		} else if (s < 0x100) { /* kana */
			if ((filter->status & 0xff00) != 0x100) {
				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
				CK((*filter->output_function)(0x49, filter->data));		/* 'I' */
			}
			filter->status = 0x100;
			CK((*filter->output_function)(s & 0x7f, filter->data));
		} else if (s < 0x8080) { /* X 0208 */
			if ((filter->status & 0xff00) != 0x200) {
				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
				CK((*filter->output_function)(0x24, filter->data));		/* '$' */
				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
			}
			filter->status = 0x200;
			CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
			CK((*filter->output_function)(s & 0x7f, filter->data));
		} else if (s < 0x10000) { /* X 0212 */
			if ((filter->status & 0xff00) != 0x300) {
				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
				CK((*filter->output_function)(0x24, filter->data));		/* '$' */
				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
				CK((*filter->output_function)(0x44, filter->data));		/* 'D' */
			}
			filter->status = 0x300;
			CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
			CK((*filter->output_function)(s & 0x7f, filter->data));
		} else { /* X 0201 latin */
			if ((filter->status & 0xff00) != 0x400) {
				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
				CK((*filter->output_function)(0x4a, filter->data));		/* 'J' */
			}
			filter->status = 0x400;
			CK((*filter->output_function)(s & 0x7f, filter->data));
		}
	} else {
		if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
			CK(mbfl_filt_conv_illegal_output(c, filter));
		}
	}

	return c;
}


/*
 * wchar => ISO-2022-JP
 */
int
mbfl_filt_conv_wchar_2022jp(int c, mbfl_convert_filter *filter)
{
	int s;

	s = 0;
	if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) {
		s = ucs_a1_jis_table[c - ucs_a1_jis_table_min];
	} else if (c >= ucs_a2_jis_table_min && c < ucs_a2_jis_table_max) {
		s = ucs_a2_jis_table[c - ucs_a2_jis_table_min];
	} else if (c >= ucs_i_jis_table_min && c < ucs_i_jis_table_max) {
		s = ucs_i_jis_table[c - ucs_i_jis_table_min];
	} else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) {
		s = ucs_r_jis_table[c - ucs_r_jis_table_min];
	}
	if (s <= 0) {
		if (c == 0xa5) {			/* YEN SIGN */
			s = 0x1005c;
		} else if (c == 0x203e) {	/* OVER LINE */
			s = 0x1007e;
		} else if (c == 0xff3c) {	/* FULLWIDTH REVERSE SOLIDUS */
			s = 0x2140;
		} else if (c == 0xff5e) {	/* FULLWIDTH TILDE */
			s = 0x2141;
		} else if (c == 0x2225) {	/* PARALLEL TO */
			s = 0x2142;
		} else if (c == 0xff0d) {	/* FULLWIDTH HYPHEN-MINUS */
			s = 0x215d;
		} else if (c == 0xffe0) {	/* FULLWIDTH CENT SIGN */
			s = 0x2171;
		} else if (c == 0xffe1) {	/* FULLWIDTH POUND SIGN */
			s = 0x2172;
		} else if (c == 0xffe2) {	/* FULLWIDTH NOT SIGN */
			s = 0x224c;
		}
		if (c == 0) {
			s = 0;
		} else if (s <= 0) {
			s = -1;
		}
	} else if ((s >= 0x80 && s < 0x2121) || (s > 0x8080)) {
		s = -1;
	}
	if (s >= 0) {
		if (s < 0x80) { /* ASCII */
			if ((filter->status & 0xff00) != 0) {
				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
			}
			filter->status = 0;
			CK((*filter->output_function)(s, filter->data));
		} else if (s < 0x10000) { /* X 0208 */
			if ((filter->status & 0xff00) != 0x200) {
				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
				CK((*filter->output_function)(0x24, filter->data));		/* '$' */
				CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
			}
			filter->status = 0x200;
			CK((*filter->output_function)((s >> 8) & 0x7f, filter->data));
			CK((*filter->output_function)(s & 0x7f, filter->data));
		} else { /* X 0201 latin */
			if ((filter->status & 0xff00) != 0x400) {
				CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
				CK((*filter->output_function)(0x28, filter->data));		/* '(' */
				CK((*filter->output_function)(0x4a, filter->data));		/* 'J' */
			}
			filter->status = 0x400;
			CK((*filter->output_function)(s & 0x7f, filter->data));
		}
	} else {
		if (filter->illegal_mode != MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
			CK(mbfl_filt_conv_illegal_output(c, filter));
		}
	}

	return c;
}

int
mbfl_filt_conv_any_jis_flush(mbfl_convert_filter *filter)
{
	/* back to latin */
	if ((filter->status & 0xff00) != 0) {
		CK((*filter->output_function)(0x1b, filter->data));		/* ESC */
		CK((*filter->output_function)(0x28, filter->data));		/* '(' */
		CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
	}
	filter->status &= 0xff;
	return 0;
}

static int mbfl_filt_ident_jis(int c, mbfl_identify_filter *filter)
{
retry:
	switch (filter->status & 0xf) {
/*	case 0x00:	 ASCII */
/*	case 0x10:	 X 0201 latin */
/*	case 0x20:	 X 0201 kana */
/*	case 0x80:	 X 0208 */
/*	case 0x90:	 X 0212 */
	case 0:
		if (c == 0x1b) {
			filter->status += 2;
		} else if (c == 0x0e) {			/* "kana in" */
			filter->status = 0x20;
		} else if (c == 0x0f) {			/* "kana out" */
			filter->status = 0;
		} else if ((filter->status == 0x80 || filter->status == 0x90) && c > 0x20 && c < 0x7f) {		/* kanji first char */
			filter->status += 1;
		} else if (c >= 0 && c < 0x80) {		/* latin, CTLs */
			;
		} else {
			filter->flag = 1;	/* bad */
		}
		break;

/*	case 0x81:	 X 0208 second char */
/*	case 0x91:	 X 0212 second char */
	case 1:
		filter->status &= ~0xf;
		if (c == 0x1b) {
			goto retry;
		} else if (c < 0x21 || c > 0x7e) {		/* bad */
			filter->flag = 1;
		}
		break;

	/* ESC */
	case 2:
		if (c == 0x24) {		/* '$' */
			filter->status++;
		} else if (c == 0x28) {		/* '(' */
			filter->status += 3;
		} else {
			filter->flag = 1;	/* bad */
			filter->status &= ~0xf;
			goto retry;
		}
		break;

	/* ESC $ */
	case 3:
		if (c == 0x40 || c == 0x42) {		/* '@' or 'B' */
			filter->status = 0x80;
		} else if (c == 0x28) {		/* '(' */
			filter->status++;
		} else {
			filter->flag = 1;	/* bad */
			filter->status &= ~0xf;
			goto retry;
		}
		break;

	/* ESC $ ( */
	case 4:
		if (c == 0x40 || c == 0x42) {		/* '@' or 'B' */
			filter->status = 0x80;
		} else if (c == 0x44) {		/* 'D' */
			filter->status = 0x90;
		} else {
			filter->flag = 1;	/* bad */
			filter->status &= ~0xf;
			goto retry;
		}
		break;

	/* ESC ( */
	case 5:
		if (c == 0x42 || c == 0x48) {		/* 'B' or 'H' */
			filter->status = 0;
		} else if (c == 0x4a) {		/* 'J' */
			filter->status = 0x10;
		} else if (c == 0x49) {		/* 'I' */
			filter->status = 0x20;
		} else {
			filter->flag = 1;	/* bad */
			filter->status &= ~0xf;
			goto retry;
		}
		break;

	default:
		filter->status = 0;
		break;
	}

	return c;
}

static int mbfl_filt_ident_2022jp(int c, mbfl_identify_filter *filter)
{
retry:
	switch (filter->status & 0xf) {
/*	case 0x00:	 ASCII */
/*	case 0x10:	 X 0201 latin */
/*	case 0x80:	 X 0208 */
	case 0:
		if (c == 0x1b) {
			filter->status += 2;
		} else if (filter->status == 0x80 && c > 0x20 && c < 0x7f) {		/* kanji first char */
			filter->status += 1;
		} else if (c >= 0 && c < 0x80) {		/* latin, CTLs */
			;
		} else {
			filter->flag = 1;	/* bad */
		}
		break;

/*	case 0x81:	 X 0208 second char */
	case 1:
		if (c == 0x1b) {
			filter->status++;
		} else {
			filter->status &= ~0xf;
			if (c < 0x21 || c > 0x7e) {		/* bad */
				filter->flag = 1;
			}
		}
		break;

	/* ESC */
	case 2:
		if (c == 0x24) {		/* '$' */
			filter->status++;
		} else if (c == 0x28) {		/* '(' */
			filter->status += 3;
		} else {
			filter->flag = 1;	/* bad */
			filter->status &= ~0xf;
			goto retry;
		}
		break;

	/* ESC $ */
	case 3:
		if (c == 0x40 || c == 0x42) {		/* '@' or 'B' */
			filter->status = 0x80;
		} else {
			filter->flag = 1;	/* bad */
			filter->status &= ~0xf;
			goto retry;
		}
		break;

	/* ESC ( */
	case 5:
		if (c == 0x42) {		/* 'B' */
			filter->status = 0;
		} else if (c == 0x4a) {		/* 'J' */
			filter->status = 0x10;
		} else {
			filter->flag = 1;	/* bad */
			filter->status &= ~0xf;
			goto retry;
		}
		break;

	default:
		filter->status = 0;
		break;
	}

	return c;
}