guacamole-spice-protocol/src/libguac/unicode.c
2016-03-28 20:39:19 -07:00

184 lines
4.0 KiB
C

/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
#include "config.h"
#include "unicode.h"
#include <stddef.h>
size_t guac_utf8_charsize(unsigned char c) {
/* Determine size in bytes of character */
if ((c | 0x7F) == 0x7F) return 1;
if ((c | 0x1F) == 0xDF) return 2;
if ((c | 0x0F) == 0xEF) return 3;
if ((c | 0x07) == 0xF7) return 4;
/* Default to one character */
return 1;
}
size_t guac_utf8_strlen(const char* str) {
/* The current length of the string */
int length = 0;
/* Number of characters before start of next character */
int skip = 0;
while (*str != 0) {
/* If skipping, then skip */
if (skip > 0) skip--;
/* Otherwise, determine next skip value, and increment length */
else {
/* Get next character */
unsigned char c = (unsigned char) *str;
/* Determine skip value (size in bytes of rest of character) */
skip = guac_utf8_charsize(c) - 1;
length++;
}
str++;
}
return length;
}
int guac_utf8_write(int codepoint, char* utf8, int length) {
int i;
int mask, bytes;
/* If not even one byte, cannot write */
if (length <= 0)
return 0;
/* Determine size and initial byte mask */
if (codepoint <= 0x007F) {
mask = 0x00;
bytes = 1;
}
else if (codepoint <= 0x7FF) {
mask = 0xC0;
bytes = 2;
}
else if (codepoint <= 0xFFFF) {
mask = 0xE0;
bytes = 3;
}
else if (codepoint <= 0x1FFFFF) {
mask = 0xF0;
bytes = 4;
}
/* Otherwise, invalid codepoint */
else {
*(utf8++) = '?';
return 1;
}
/* If not enough room, don't write anything */
if (bytes > length)
return 0;
/* Offset buffer by size */
utf8 += bytes - 1;
/* Add trailing bytes, if any */
for (i=1; i<bytes; i++) {
*(utf8--) = 0x80 | (codepoint & 0x3F);
codepoint >>= 6;
}
/* Set initial byte */
*utf8 = mask | codepoint;
/* Done */
return bytes;
}
int guac_utf8_read(const char* utf8, int length, int* codepoint) {
unsigned char initial;
int bytes;
int result;
int i;
/* If not even one byte, cannot read */
if (length <= 0)
return 0;
/* Read initial byte */
initial = (unsigned char) *(utf8++);
/* 0xxxxxxx */
if ((initial | 0x7F) == 0x7F) {
result = initial;
bytes = 1;
}
/* 110xxxxx 10xxxxxx */
else if ((initial | 0x1F) == 0xDF) {
result = initial & 0x1F;
bytes = 2;
}
/* 1110xxxx 10xxxxxx 10xxxxxx */
else if ((initial | 0x0F) == 0xEF) {
result = initial & 0x0F;
bytes = 3;
}
/* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
else if ((initial | 0x07) == 0xF7) {
result = initial & 0x07;
bytes = 4;
}
/* Otherwise, invalid codepoint */
else {
*codepoint = 0xFFFD; /* Replacement character */
return 1;
}
/* If not enough room, don't read anything */
if (bytes > length)
return 0;
/* Read trailing bytes, if any */
for (i=1; i<bytes; i++) {
result <<= 6;
result |= *(utf8++) & 0x3F;
}
*codepoint = result;
return bytes;
}