iconv.c

samba最新软件

			continue;
		}
		/* it's a hex character */
		if (*inbytesleft < 5) {
			errno = EINVAL;
			return -1;
		}
		
		if (sscanf(&(*inbuf)[1], "%04x", &v) != 1) {
			errno = EILSEQ;
			return -1;
		}

		(*outbuf)[0] = v&0xff;
		(*outbuf)[1] = v>>8;
		(*inbytesleft)  -= 5;
		(*outbytesleft) -= 2;
		(*inbuf)  += 5;
		(*outbuf) += 2;
	}

	if (*inbytesleft > 0) {
		errno = E2BIG;
		return -1;
	}
	
	return 0;
}

static size_t ucs2hex_push(void *cd, const char **inbuf, size_t *inbytesleft,
			   char **outbuf, size_t *outbytesleft)
{
	while (*inbytesleft >= 2 && *outbytesleft >= 1) {
		char buf[6];

		if ((*inbuf)[1] == 0 && 
		    ((*inbuf)[0] & 0x80) == 0 &&
		    (*inbuf)[0] != '@') {
			(*outbuf)[0] = (*inbuf)[0];
			(*inbytesleft)  -= 2;
			(*outbytesleft) -= 1;
			(*inbuf)  += 2;
			(*outbuf) += 1;
			continue;
		}
		if (*outbytesleft < 5) {
			errno = E2BIG;
			return -1;
		}
		snprintf(buf, 6, "@%04x", SVAL(*inbuf, 0));
		memcpy(*outbuf, buf, 5);
		(*inbytesleft)  -= 2;
		(*outbytesleft) -= 5;
		(*inbuf)  += 2;
		(*outbuf) += 5;
	}

	if (*inbytesleft == 1) {
		errno = EINVAL;
		return -1;
	}

	if (*inbytesleft > 1) {
		errno = E2BIG;
		return -1;
	}
	
	return 0;
}

static size_t iconv_swab(void *cd, const char **inbuf, size_t *inbytesleft,
			 char **outbuf, size_t *outbytesleft)
{
	int n;

	n = MIN(*inbytesleft, *outbytesleft);

	swab(*inbuf, *outbuf, (n&~1));
	if (n&1) {
		(*outbuf)[n-1] = 0;
	}

	(*inbytesleft) -= n;
	(*outbytesleft) -= n;
	(*inbuf) += n;
	(*outbuf) += n;

	if (*inbytesleft > 0) {
		errno = E2BIG;
		return -1;
	}

	return 0;
}


static size_t iconv_copy(void *cd, const char **inbuf, size_t *inbytesleft,
			 char **outbuf, size_t *outbytesleft)
{
	int n;

	n = MIN(*inbytesleft, *outbytesleft);

	memmove(*outbuf, *inbuf, n);

	(*inbytesleft) -= n;
	(*outbytesleft) -= n;
	(*inbuf) += n;
	(*outbuf) += n;

	if (*inbytesleft > 0) {
		errno = E2BIG;
		return -1;
	}

	return 0;
}

static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft,
			 char **outbuf, size_t *outbytesleft)
{
	size_t in_left=*inbytesleft, out_left=*outbytesleft;
	const uint8_t *c = (const uint8_t *)*inbuf;
	uint8_t *uc = (uint8_t *)*outbuf;

	while (in_left >= 1 && out_left >= 2) {
		if ((c[0] & 0x80) == 0) {
			uc[0] = c[0];
			uc[1] = 0;
			c  += 1;
			in_left  -= 1;
			out_left -= 2;
			uc += 2;
			continue;
		}

		if ((c[0] & 0xe0) == 0xc0) {
			if (in_left < 2 ||
			    (c[1] & 0xc0) != 0x80) {
				errno = EILSEQ;
				goto error;
			}
			uc[1] = (c[0]>>2) & 0x7;
			uc[0] = (c[0]<<6) | (c[1]&0x3f);
			c  += 2;
			in_left  -= 2;
			out_left -= 2;
			uc += 2;
			continue;
		}

		if ((c[0] & 0xf0) == 0xe0) {
			if (in_left < 3 ||
			    (c[1] & 0xc0) != 0x80 || 
			    (c[2] & 0xc0) != 0x80) {
				errno = EILSEQ;
				goto error;
			}
			uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF);
			uc[0] = (c[1]<<6) | (c[2]&0x3f);
			c  += 3;
			in_left  -= 3;
			out_left -= 2;
			uc += 2;
			continue;
		}

		if ((c[0] & 0xf8) == 0xf0) {
			unsigned int codepoint;
			if (in_left < 4 ||
			    (c[1] & 0xc0) != 0x80 || 
			    (c[2] & 0xc0) != 0x80 ||
			    (c[3] & 0xc0) != 0x80) {
				errno = EILSEQ;
				goto error;
			}
			codepoint = 
				(c[3]&0x3f) | 
				((c[2]&0x3f)<<6) | 
				((c[1]&0x3f)<<12) |
				((c[0]&0x7)<<18);
			if (codepoint < 0x10000) {
				/* accept UTF-8 characters that are not
				   minimally packed, but pack the result */
				uc[0] = (codepoint & 0xFF);
				uc[1] = (codepoint >> 8);
				c += 4;
				in_left -= 4;
				out_left -= 2;
				uc += 2;
				continue;
			}

			codepoint -= 0x10000;

			if (out_left < 4) {
				errno = E2BIG;
				goto error;
			}

			uc[0] = (codepoint>>10) & 0xFF;
			uc[1] = (codepoint>>18) | 0xd8;
			uc[2] = codepoint & 0xFF;
			uc[3] = ((codepoint>>8) & 0x3) | 0xdc;
			c  += 4;
			in_left  -= 4;
			out_left -= 4;
			uc += 4;
			continue;
		}

		/* we don't handle 5 byte sequences */
		errno = EINVAL;
		goto error;
	}

	if (in_left > 0) {
		errno = E2BIG;
		goto error;
	}

	*inbytesleft = in_left;
	*outbytesleft = out_left;
	*inbuf = (const char *)c;
	*outbuf = (char *)uc;
	return 0;

error:
	*inbytesleft = in_left;
	*outbytesleft = out_left;
	*inbuf = (const char *)c;
	*outbuf = (char *)uc;
	return -1;
}

static size_t utf8_push(void *cd, const char **inbuf, size_t *inbytesleft,
			char **outbuf, size_t *outbytesleft)
{
	size_t in_left=*inbytesleft, out_left=*outbytesleft;
	uint8_t *c = (uint8_t *)*outbuf;
	const uint8_t *uc = (const uint8_t *)*inbuf;

	while (in_left >= 2 && out_left >= 1) {
		unsigned int codepoint;

		if (uc[1] == 0 && !(uc[0] & 0x80)) {
			/* simplest case */
			c[0] = uc[0];
			in_left  -= 2;
			out_left -= 1;
			uc += 2;
			c  += 1;
			continue;
		}

		if ((uc[1]&0xf8) == 0) {
			/* next simplest case */
			if (out_left < 2) {
				errno = E2BIG;
				goto error;
			}
			c[0] = 0xc0 | (uc[0]>>6) | (uc[1]<<2);
			c[1] = 0x80 | (uc[0] & 0x3f);
			in_left  -= 2;
			out_left -= 2;
			uc += 2;
			c  += 2;
			continue;
		}

		if ((uc[1] & 0xfc) == 0xdc) {
			/* its the second part of a 4 byte sequence. Illegal */
			if (in_left < 4) {
				errno = EINVAL;
			} else {
				errno = EILSEQ;
			}
			goto error;
		}

		if ((uc[1] & 0xfc) != 0xd8) {
			codepoint = uc[0] | (uc[1]<<8);
			if (out_left < 3) {
				errno = E2BIG;
				goto error;
			}
			c[0] = 0xe0 | (codepoint >> 12);
			c[1] = 0x80 | ((codepoint >> 6) & 0x3f);
			c[2] = 0x80 | (codepoint & 0x3f);
			
			in_left  -= 2;
			out_left -= 3;
			uc  += 2;
			c   += 3;
			continue;
		}

		/* its the first part of a 4 byte sequence */
		if (in_left < 4) {
			errno = EINVAL;
			goto error;
		}
		if ((uc[3] & 0xfc) != 0xdc) {
			errno = EILSEQ;
			goto error;
		}
		codepoint = 0x10000 + (uc[2] | ((uc[3] & 0x3)<<8) | 
				       (uc[0]<<10) | ((uc[1] & 0x3)<<18));
		
		if (out_left < 4) {
			errno = E2BIG;
			goto error;
		}
		c[0] = 0xf0 | (codepoint >> 18);
		c[1] = 0x80 | ((codepoint >> 12) & 0x3f);
		c[2] = 0x80 | ((codepoint >> 6) & 0x3f);
		c[3] = 0x80 | (codepoint & 0x3f);
		
		in_left  -= 4;
		out_left -= 4;
		uc       += 4;
		c        += 4;
	}

	if (in_left == 1) {
		errno = EINVAL;
		goto error;
	}

	if (in_left > 1) {
		errno = E2BIG;
		goto error;
	}

	*inbytesleft = in_left;
	*outbytesleft = out_left;
	*inbuf  = (const char *)uc;
	*outbuf = (char *)c;
	
	return 0;

error:
	*inbytesleft = in_left;
	*outbytesleft = out_left;
	*inbuf  = (const char *)uc;
	*outbuf = (char *)c;
	return -1;
}