mbfilter.c

php-4.4.7学习linux时下载的源代码

			filter1 = mbfl_convert_filter_new(string->no_encoding, mbfl_no_encoding_wchar, (int (*)(int, void*))filter2->filter_function, NULL, filter2);
			if (filter1 == NULL) {
				mbfl_convert_filter_delete(filter2);
			}
		}
	}
	if (filter1 == NULL) {
		return NULL;
	}

	if (filter2 != NULL) {
		filter2->illegal_mode = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
		filter2->illegal_substchar = 0x3f;		/* '?' */
	}

	mbfl_memory_device_init(&device, string->len, (string->len >> 2) + 8);

	/* feed data */
	n = string->len;
	p = string->val;
	if (p != NULL) {
		while (n > 0) {
			if ((*filter1->filter_function)(*p++, filter1) < 0) {
				break;
			}
			n--;
		}
	}

	mbfl_convert_filter_flush(filter1);
	mbfl_convert_filter_delete(filter1);
	if (filter2 != NULL) {
		mbfl_convert_filter_flush(filter2);
		mbfl_convert_filter_delete(filter2);
	}

	return mbfl_memory_device_result(&device, result);
}


/*
 * identify encoding
 */
const mbfl_encoding *
mbfl_identify_encoding(mbfl_string *string, enum mbfl_no_encoding *elist, int elistsz, int strict)
{
	int i, n, num, bad;
	unsigned char *p;
	mbfl_identify_filter *flist, *filter;
	const mbfl_encoding *encoding;

	/* flist is an array of mbfl_identify_filter instances */
	flist = (mbfl_identify_filter *)mbfl_calloc(elistsz, sizeof(mbfl_identify_filter));
	if (flist == NULL) {
		return NULL;
	}

	num = 0;
	if (elist != NULL) {
		for (i = 0; i < elistsz; i++) {
			if (!mbfl_identify_filter_init(&flist[num], elist[i])) {
				num++;
			}
		}
	}

	/* feed data */
	n = string->len;
	p = string->val;

	if (p != NULL) {
		bad = 0;
		while (n > 0) {
			for (i = 0; i < num; i++) {
				filter = &flist[i];
				if (!filter->flag) {
					(*filter->filter_function)(*p, filter);
					if (filter->flag) {
						bad++;
					}
				}
			}
			if ((num - 1) <= bad && !strict) {
				break;
			}
			p++;
			n--;
		}
	}

	/* judge */
	encoding = NULL;

	for (i = 0; i < num; i++) {
		filter = &flist[i];
		if (!filter->flag) {
			if (strict && filter->status) {
				continue;
			}
			encoding = filter->encoding;
			break;
		}
	}

	/* fall-back judge */
	if (!encoding) {
		for (i = 0; i < num; i++) {
			filter = &flist[i];
			if (!filter->flag) {
				encoding = filter->encoding;
				break;
			}
		}
	}

	/* cleanup */
	/* dtors should be called in reverse order */
	i = num; while (--i >= 0) {
		mbfl_identify_filter_cleanup(&flist[i]);
	}

	mbfl_free((void *)flist);

	return encoding;
}

const char*
mbfl_identify_encoding_name(mbfl_string *string, enum mbfl_no_encoding *elist, int elistsz, int strict)
{
	const mbfl_encoding *encoding;

	encoding = mbfl_identify_encoding(string, elist, elistsz, strict);
	if (encoding != NULL &&
	    encoding->no_encoding > mbfl_no_encoding_charset_min &&
	    encoding->no_encoding < mbfl_no_encoding_charset_max) {
		return encoding->name;
	} else {
		return NULL;
	}
}

enum mbfl_no_encoding
mbfl_identify_encoding_no(mbfl_string *string, enum mbfl_no_encoding *elist, int elistsz, int strict)
{
	const mbfl_encoding *encoding;

	encoding = mbfl_identify_encoding(string, elist, elistsz, strict);
	if (encoding != NULL &&
	    encoding->no_encoding > mbfl_no_encoding_charset_min &&
	    encoding->no_encoding < mbfl_no_encoding_charset_max) {
		return encoding->no_encoding;
	} else {
		return mbfl_no_encoding_invalid;
	}
}


/*
 *  strlen
 */
static int
filter_count_output(int c, void *data)
{
	(*(int *)data)++;
	return c;
}

int
mbfl_strlen(mbfl_string *string)
{
	int len, n, m, k;
	unsigned char *p;
	const unsigned char *mbtab;
	const mbfl_encoding *encoding;

	encoding = mbfl_no2encoding(string->no_encoding);
	if (encoding == NULL || string == NULL) {
		return -1;
	}

	len = 0;
	if (encoding->flag & MBFL_ENCTYPE_SBCS) {
		len = string->len;
	} else if (encoding->flag & (MBFL_ENCTYPE_WCS2BE | MBFL_ENCTYPE_WCS2LE)) {
		len = string->len/2;
	} else if (encoding->flag & (MBFL_ENCTYPE_WCS4BE | MBFL_ENCTYPE_WCS4LE)) {
		len = string->len/4;
	} else if (encoding->mblen_table != NULL) {
		mbtab = encoding->mblen_table;
		n = 0;
		p = string->val;
		k = string->len;
		/* count */
		if (p != NULL) {
			while (n < k) {
				m = mbtab[*p];
				n += m;
				p += m;
				len++;
			};
		}
	} else {
		/* wchar filter */
		mbfl_convert_filter *filter = mbfl_convert_filter_new(
		  string->no_encoding, 
		  mbfl_no_encoding_wchar,
		  filter_count_output, 0, &len);
		if (filter == NULL) {
			return -1;
		}
		/* count */
		n = string->len;
		p = string->val;
		if (p != NULL) {
			while (n > 0) {
				(*filter->filter_function)(*p++, filter);
				n--;
			}
		}
		mbfl_convert_filter_delete(filter);
	}

	return len;
}

 
/*
 *  strpos
 */
struct collector_strpos_data {
	mbfl_convert_filter *next_filter;
	mbfl_wchar_device needle;
	int needle_len;
	int start;
	int output;
	int found_pos;
	int needle_pos;
	int matched_pos;
};

static int
collector_strpos(int c, void* data)
{
	int *p, *h, *m, n;
	struct collector_strpos_data *pc = (struct collector_strpos_data*)data;

	if (pc->output >= pc->start) {
		if (c == (int)pc->needle.buffer[pc->needle_pos]) {
			if (pc->needle_pos == 0) {
				pc->found_pos = pc->output;			/* found position */
			}
			pc->needle_pos++;						/* needle pointer */
			if (pc->needle_pos >= pc->needle_len) {
				pc->matched_pos = pc->found_pos;	/* matched position */
				pc->needle_pos--;
				goto retry;
			}
		} else if (pc->needle_pos != 0) {
retry:
			h = (int *)pc->needle.buffer;
			h++;
			for (;;) {
				pc->found_pos++;
				p = h;
				m = pc->needle.buffer;
				n = pc->needle_pos - 1;
				while (n > 0 && *p == *m) {
					n--;
					p++;
					m++;
				}
				if (n <= 0) {
					if (*m != c) {
						pc->needle_pos = 0;
					}
					break;
				} else {
					h++;
					pc->needle_pos--;
				}
			}
		}
	}

	pc->output++;
	return c;
}

/*
 *	oddlen
 */
int 
mbfl_oddlen(mbfl_string *string)
{
	int len, n, m, k;
	unsigned char *p;
	const unsigned char *mbtab;
	const mbfl_encoding *encoding;


	if (string == NULL) {
		return -1;
	}
	encoding = mbfl_no2encoding(string->no_encoding);
	if (encoding == NULL) {
		return -1;
	}

	len = 0;
	if (encoding->flag & MBFL_ENCTYPE_SBCS) {
		return 0;
	} else if (encoding->flag & (MBFL_ENCTYPE_WCS2BE | MBFL_ENCTYPE_WCS2LE)) {
		return len % 2;
	} else if (encoding->flag & (MBFL_ENCTYPE_WCS4BE | MBFL_ENCTYPE_WCS4LE)) {
		return len % 4;
	} else if (encoding->mblen_table != NULL) {
 		mbtab = encoding->mblen_table;
 		n = 0;
		p = string->val;
		k = string->len;
		/* count */
		if (p != NULL) {
			while (n < k) {
				m = mbtab[*p];
				n += m;
				p += m;
			};
		}
		return n-k;
	} else {
		/* how can i do ? */
		return 0;
	}
	/* NOT REACHED */
}

int
mbfl_strpos(
    mbfl_string *haystack,
    mbfl_string *needle,
    int offset,
    int reverse)
{
	int n, result;
	unsigned char *p;
	mbfl_convert_filter *filter;
	struct collector_strpos_data pc;

	if (haystack == NULL || needle == NULL) {
		return -8;
	}
	/* needle is converted into wchar */
	mbfl_wchar_device_init(&pc.needle);
	filter = mbfl_convert_filter_new(
	  needle->no_encoding,
	  mbfl_no_encoding_wchar,
	  mbfl_wchar_device_output, 0, &pc.needle);
	if (filter == NULL) {
		return -4;
	}
	p = needle->val;
	n = needle->len;
	if (p != NULL) {
		while (n > 0) {
			if ((*filter->filter_function)(*p++, filter) < 0) {
				break;
			}
			n--;
		}
	}
	mbfl_convert_filter_flush(filter);
	mbfl_convert_filter_delete(filter);
	pc.needle_len = pc.needle.pos;
	if (pc.needle.buffer == NULL) {
		return -4;
	}
	if (pc.needle_len <= 0) {
		mbfl_wchar_device_clear(&pc.needle);
		return -2;
	}
	/* initialize filter and collector data */
	filter = mbfl_convert_filter_new(
	  haystack->no_encoding,
	  mbfl_no_encoding_wchar,
	  collector_strpos, 0, &pc);
	if (filter == NULL) {
		mbfl_wchar_device_clear(&pc.needle);
		return -4;
	}
	pc.start = offset;
	pc.output = 0;
	pc.needle_pos = 0;
	pc.found_pos = 0;
	pc.matched_pos = -1;

	/* feed data */
	p = haystack->val;
	n = haystack->len;
	if (p != NULL) {
		while (n > 0) {
			if ((*filter->filter_function)(*p++, filter) < 0) {
				pc.matched_pos = -4;
				break;
			}
			if (pc.matched_pos >= 0 && !reverse) {
				break;
			}
			n--;
		}
	}
	mbfl_convert_filter_flush(filter);
	result = pc.matched_pos;
	mbfl_convert_filter_delete(filter);
	mbfl_wchar_device_clear(&pc.needle);

	return result;
}

/*
 *  substr_count
 */

int
mbfl_substr_count(
    mbfl_string *haystack,
    mbfl_string *needle
   )
{
	int n, result = 0;
	unsigned char *p;
	mbfl_convert_filter *filter;
	struct collector_strpos_data pc;

	if (haystack == NULL || needle == NULL) {
		return -8;
	}
	/* needle is converted into wchar */
	mbfl_wchar_device_init(&pc.needle);
	filter = mbfl_convert_filter_new(
	  needle->no_encoding,
	  mbfl_no_encoding_wchar,
	  mbfl_wchar_device_output, 0, &pc.needle);
	if (filter == NULL) {
		return -4;
	}
	p = needle->val;
	n = needle->len;
	if (p != NULL) {
		while (n > 0) {
			if ((*filter->filter_function)(*p++, filter) < 0) {
				break;
			}
			n--;
		}
	}
	mbfl_convert_filter_flush(filter);
	mbfl_convert_filter_delete(filter);
	pc.needle_len = pc.needle.pos;
	if (pc.needle.buffer == NULL) {
		return -4;
	}
	if (pc.needle_len <= 0) {
		mbfl_wchar_device_clear(&pc.needle);
		return -2;
	}
	/* initialize filter and collector data */
	filter = mbfl_convert_filter_new(
	  haystack->no_encoding,
	  mbfl_no_encoding_wchar,
	  collector_strpos, 0, &pc);
	if (filter == NULL) {
		mbfl_wchar_device_clear(&pc.needle);
		return -4;
	}
	pc.start = 0;
	pc.output = 0;
	pc.needle_pos = 0;
	pc.found_pos = 0;
	pc.matched_pos = -1;

	/* feed data */
	p = haystack->val;
	n = haystack->len;
	if (p != NULL) {
		while (n > 0) {
			if ((*filter->filter_function)(*p++, filter) < 0) {
				pc.matched_pos = -4;
				break;
			}
			if (pc.matched_pos >= 0) {
				++result;
				pc.matched_pos = -1;
			}
			n--;
		}
	}
	mbfl_convert_filter_flush(filter);
	mbfl_convert_filter_delete(filter);
	mbfl_wchar_device_clear(&pc.needle);

	return result;
}

/*
 *  substr
 */
struct collector_substr_data {
	mbfl_convert_filter *next_filter;
	int start;
	int stop;
	int output;
};

static int
collector_substr(int c, void* data)
{
	struct collector_substr_data *pc = (struct collector_substr_data*)data;