michael@0: /*
michael@0:  * Copyright (C) 2005 The Android Open Source Project
michael@0:  *
michael@0:  * Licensed under the Apache License, Version 2.0 (the "License");
michael@0:  * you may not use this file except in compliance with the License.
michael@0:  * You may obtain a copy of the License at
michael@0:  *
michael@0:  *      http://www.apache.org/licenses/LICENSE-2.0
michael@0:  *
michael@0:  * Unless required by applicable law or agreed to in writing, software
michael@0:  * distributed under the License is distributed on an "AS IS" BASIS,
michael@0:  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
michael@0:  * See the License for the specific language governing permissions and
michael@0:  * limitations under the License.
michael@0:  */
michael@0: 
michael@0: #ifndef ANDROID_UNICODE_H
michael@0: #define ANDROID_UNICODE_H
michael@0: 
michael@0: #include <sys/types.h>
michael@0: #include <stdint.h>
michael@0: 
michael@0: extern "C" {
michael@0: 
michael@0: #if !defined(__cplusplus) || __cplusplus == 199711L // C or C++98
michael@0: typedef uint32_t char32_t;
michael@0: typedef uint16_t char16_t;
michael@0: #endif
michael@0: 
michael@0: // Standard string functions on char16_t strings.
michael@0: int strcmp16(const char16_t *, const char16_t *);
michael@0: int strncmp16(const char16_t *s1, const char16_t *s2, size_t n);
michael@0: size_t strlen16(const char16_t *);
michael@0: size_t strnlen16(const char16_t *, size_t);
michael@0: char16_t *strcpy16(char16_t *, const char16_t *);
michael@0: char16_t *strncpy16(char16_t *, const char16_t *, size_t);
michael@0: 
michael@0: // Version of comparison that supports embedded nulls.
michael@0: // This is different than strncmp() because we don't stop
michael@0: // at a nul character and consider the strings to be different
michael@0: // if the lengths are different (thus we need to supply the
michael@0: // lengths of both strings).  This can also be used when
michael@0: // your string is not nul-terminated as it will have the
michael@0: // equivalent result as strcmp16 (unlike strncmp16).
michael@0: int strzcmp16(const char16_t *s1, size_t n1, const char16_t *s2, size_t n2);
michael@0: 
michael@0: // Version of strzcmp16 for comparing strings in different endianness.
michael@0: int strzcmp16_h_n(const char16_t *s1H, size_t n1, const char16_t *s2N, size_t n2);
michael@0: 
michael@0: // Standard string functions on char32_t strings.
michael@0: size_t strlen32(const char32_t *);
michael@0: size_t strnlen32(const char32_t *, size_t);
michael@0: 
michael@0: /**
michael@0:  * Measure the length of a UTF-32 string in UTF-8. If the string is invalid
michael@0:  * such as containing a surrogate character, -1 will be returned.
michael@0:  */
michael@0: ssize_t utf32_to_utf8_length(const char32_t *src, size_t src_len);
michael@0: 
michael@0: /**
michael@0:  * Stores a UTF-8 string converted from "src" in "dst", if "dst_length" is not
michael@0:  * large enough to store the string, the part of the "src" string is stored
michael@0:  * into "dst" as much as possible. See the examples for more detail.
michael@0:  * Returns the size actually used for storing the string.
michael@0:  * dst" is not null-terminated when dst_len is fully used (like strncpy).
michael@0:  *
michael@0:  * Example 1
michael@0:  * "src" == \u3042\u3044 (\xE3\x81\x82\xE3\x81\x84)
michael@0:  * "src_len" == 2
michael@0:  * "dst_len" >= 7
michael@0:  * ->
michael@0:  * Returned value == 6
michael@0:  * "dst" becomes \xE3\x81\x82\xE3\x81\x84\0
michael@0:  * (note that "dst" is null-terminated)
michael@0:  *
michael@0:  * Example 2
michael@0:  * "src" == \u3042\u3044 (\xE3\x81\x82\xE3\x81\x84)
michael@0:  * "src_len" == 2
michael@0:  * "dst_len" == 5
michael@0:  * ->
michael@0:  * Returned value == 3
michael@0:  * "dst" becomes \xE3\x81\x82\0
michael@0:  * (note that "dst" is null-terminated, but \u3044 is not stored in "dst"
michael@0:  * since "dst" does not have enough size to store the character)
michael@0:  *
michael@0:  * Example 3
michael@0:  * "src" == \u3042\u3044 (\xE3\x81\x82\xE3\x81\x84)
michael@0:  * "src_len" == 2
michael@0:  * "dst_len" == 6
michael@0:  * ->
michael@0:  * Returned value == 6
michael@0:  * "dst" becomes \xE3\x81\x82\xE3\x81\x84
michael@0:  * (note that "dst" is NOT null-terminated, like strncpy)
michael@0:  */
michael@0: void utf32_to_utf8(const char32_t* src, size_t src_len, char* dst);
michael@0: 
michael@0: /**
michael@0:  * Returns the unicode value at "index".
michael@0:  * Returns -1 when the index is invalid (equals to or more than "src_len").
michael@0:  * If returned value is positive, it is able to be converted to char32_t, which
michael@0:  * is unsigned. Then, if "next_index" is not NULL, the next index to be used is
michael@0:  * stored in "next_index". "next_index" can be NULL.
michael@0:  */
michael@0: int32_t utf32_from_utf8_at(const char *src, size_t src_len, size_t index, size_t *next_index);
michael@0: 
michael@0: 
michael@0: /**
michael@0:  * Returns the UTF-8 length of UTF-16 string "src".
michael@0:  */
michael@0: ssize_t utf16_to_utf8_length(const char16_t *src, size_t src_len);
michael@0: 
michael@0: /**
michael@0:  * Converts a UTF-16 string to UTF-8. The destination buffer must be large
michael@0:  * enough to fit the UTF-16 as measured by utf16_to_utf8_length with an added
michael@0:  * NULL terminator.
michael@0:  */
michael@0: void utf16_to_utf8(const char16_t* src, size_t src_len, char* dst);
michael@0: 
michael@0: /**
michael@0:  * Returns the length of "src" when "src" is valid UTF-8 string.
michael@0:  * Returns 0 if src is NULL or 0-length string. Returns -1 when the source
michael@0:  * is an invalid string.
michael@0:  *
michael@0:  * This function should be used to determine whether "src" is valid UTF-8
michael@0:  * characters with valid unicode codepoints. "src" must be null-terminated.
michael@0:  *
michael@0:  * If you are going to use other utf8_to_... functions defined in this header
michael@0:  * with string which may not be valid UTF-8 with valid codepoint (form 0 to
michael@0:  * 0x10FFFF), you should use this function before calling others, since the
michael@0:  * other functions do not check whether the string is valid UTF-8 or not.
michael@0:  *
michael@0:  * If you do not care whether "src" is valid UTF-8 or not, you should use
michael@0:  * strlen() as usual, which should be much faster.
michael@0:  */
michael@0: ssize_t utf8_length(const char *src);
michael@0: 
michael@0: /**
michael@0:  * Measure the length of a UTF-32 string.
michael@0:  */
michael@0: size_t utf8_to_utf32_length(const char *src, size_t src_len);
michael@0: 
michael@0: /**
michael@0:  * Stores a UTF-32 string converted from "src" in "dst". "dst" must be large
michael@0:  * enough to store the entire converted string as measured by
michael@0:  * utf8_to_utf32_length plus space for a NULL terminator.
michael@0:  */
michael@0: void utf8_to_utf32(const char* src, size_t src_len, char32_t* dst);
michael@0: 
michael@0: /**
michael@0:  * Returns the UTF-16 length of UTF-8 string "src".
michael@0:  */
michael@0: ssize_t utf8_to_utf16_length(const uint8_t* src, size_t srcLen);
michael@0: 
michael@0: /**
michael@0:  * Convert UTF-8 to UTF-16 including surrogate pairs.
michael@0:  * Returns a pointer to the end of the string (where a null terminator might go
michael@0:  * if you wanted to add one).
michael@0:  */
michael@0: char16_t* utf8_to_utf16_no_null_terminator(const uint8_t* src, size_t srcLen, char16_t* dst);
michael@0: 
michael@0: /**
michael@0:  * Convert UTF-8 to UTF-16 including surrogate pairs. The destination buffer
michael@0:  * must be large enough to hold the result as measured by utf8_to_utf16_length
michael@0:  * plus an added NULL terminator.
michael@0:  */
michael@0: void utf8_to_utf16(const uint8_t* src, size_t srcLen, char16_t* dst);
michael@0: 
michael@0: }
michael@0: 
michael@0: #endif