ConvertUTF.h revision 263508 
1/*===--- ConvertUTF.h - Universal Character Names conversions ---------------===
2 *
3 *                     The LLVM Compiler Infrastructure
4 *
5 * This file is distributed under the University of Illinois Open Source
6 * License. See LICENSE.TXT for details.
7 *
8 *==------------------------------------------------------------------------==*/
9/*
10 * Copyright 2001-2004 Unicode, Inc.
11 *
12 * Disclaimer
13 *
14 * This source code is provided as is by Unicode, Inc. No claims are
15 * made as to fitness for any particular purpose. No warranties of any
16 * kind are expressed or implied. The recipient agrees to determine
17 * applicability of information provided. If this file has been
18 * purchased on magnetic or optical media from Unicode, Inc., the
19 * sole remedy for any claim will be exchange of defective media
20 * within 90 days of receipt.
21 *
22 * Limitations on Rights to Redistribute This Code
23 *
24 * Unicode, Inc. hereby grants the right to freely use the information
25 * supplied in this file in the creation of products supporting the
26 * Unicode Standard, and to make copies of this file in any form
27 * for internal or external distribution as long as this notice
28 * remains attached.
29 */
30
31/* ---------------------------------------------------------------------
32
33    Conversions between UTF32, UTF-16, and UTF-8.  Header file.
34
35    Several funtions are included here, forming a complete set of
36    conversions between the three formats.  UTF-7 is not included
37    here, but is handled in a separate source file.
38
39    Each of these routines takes pointers to input buffers and output
40    buffers.  The input buffers are const.
41
42    Each routine converts the text between *sourceStart and sourceEnd,
43    putting the result into the buffer between *targetStart and
44    targetEnd. Note: the end pointers are *after* the last item: e.g.
45    *(sourceEnd - 1) is the last item.
46
47    The return result indicates whether the conversion was successful,
48    and if not, whether the problem was in the source or target buffers.
49    (Only the first encountered problem is indicated.)
50
51    After the conversion, *sourceStart and *targetStart are both
52    updated to point to the end of last text successfully converted in
53    the respective buffers.
54
55    Input parameters:
56        sourceStart - pointer to a pointer to the source buffer.
57                The contents of this are modified on return so that
58                it points at the next thing to be converted.
59        targetStart - similarly, pointer to pointer to the target buffer.
60        sourceEnd, targetEnd - respectively pointers to the ends of the
61                two buffers, for overflow checking only.
62
63    These conversion functions take a ConversionFlags argument. When this
64    flag is set to strict, both irregular sequences and isolated surrogates
65    will cause an error.  When the flag is set to lenient, both irregular
66    sequences and isolated surrogates are converted.
67
68    Whether the flag is strict or lenient, all illegal sequences will cause
69    an error return. This includes sequences such as: <F4 90 80 80>, <C0 80>,
70    or <A0> in UTF-8, and values above 0x10FFFF in UTF-32. Conformant code
71    must check for illegal sequences.
72
73    When the flag is set to lenient, characters over 0x10FFFF are converted
74    to the replacement character; otherwise (when the flag is set to strict)
75    they constitute an error.
76
77    Output parameters:
78        The value "sourceIllegal" is returned from some routines if the input
79        sequence is malformed.  When "sourceIllegal" is returned, the source
80        value will point to the illegal value that caused the problem. E.g.,
81        in UTF-8 when a sequence is malformed, it points to the start of the
82        malformed sequence.
83
84    Author: Mark E. Davis, 1994.
85    Rev History: Rick McGowan, fixes & updates May 2001.
86         Fixes & updates, Sept 2001.
87
88------------------------------------------------------------------------ */
89
90#ifndef LLVM_SUPPORT_CONVERTUTF_H
91#define LLVM_SUPPORT_CONVERTUTF_H
92
93/* ---------------------------------------------------------------------
94    The following 4 definitions are compiler-specific.
95    The C standard does not guarantee that wchar_t has at least
96    16 bits, so wchar_t is no less portable than unsigned short!
97    All should be unsigned values to avoid sign extension during
98    bit mask & shift operations.
99------------------------------------------------------------------------ */
100
101typedef unsigned int    UTF32;  /* at least 32 bits */
102typedef unsigned short  UTF16;  /* at least 16 bits */
103typedef unsigned char   UTF8;   /* typically 8 bits */
104typedef unsigned char   Boolean; /* 0 or 1 */
105
106/* Some fundamental constants */
107#define UNI_REPLACEMENT_CHAR (UTF32)0x0000FFFD
108#define UNI_MAX_BMP (UTF32)0x0000FFFF
109#define UNI_MAX_UTF16 (UTF32)0x0010FFFF
110#define UNI_MAX_UTF32 (UTF32)0x7FFFFFFF
111#define UNI_MAX_LEGAL_UTF32 (UTF32)0x0010FFFF
112
113#define UNI_MAX_UTF8_BYTES_PER_CODE_POINT 4
114
115#define UNI_UTF16_BYTE_ORDER_MARK_NATIVE  0xFEFF
116#define UNI_UTF16_BYTE_ORDER_MARK_SWAPPED 0xFFFE
117
118typedef enum {
119  conversionOK,           /* conversion successful */
120  sourceExhausted,        /* partial character in source, but hit end */
121  targetExhausted,        /* insuff. room in target for conversion */
122  sourceIllegal           /* source sequence is illegal/malformed */
123} ConversionResult;
124
125typedef enum {
126  strictConversion = 0,
127  lenientConversion
128} ConversionFlags;
129
130/* This is for C++ and does no harm in C */
131#ifdef __cplusplus
132extern "C" {
133#endif
134
135ConversionResult ConvertUTF8toUTF16 (
136  const UTF8** sourceStart, const UTF8* sourceEnd,
137  UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags);
138
139ConversionResult ConvertUTF8toUTF32 (
140  const UTF8** sourceStart, const UTF8* sourceEnd,
141  UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags);
142
143ConversionResult ConvertUTF16toUTF8 (
144  const UTF16** sourceStart, const UTF16* sourceEnd,
145  UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags);
146
147ConversionResult ConvertUTF32toUTF8 (
148  const UTF32** sourceStart, const UTF32* sourceEnd,
149  UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags);
150
151ConversionResult ConvertUTF16toUTF32 (
152  const UTF16** sourceStart, const UTF16* sourceEnd,
153  UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags);
154
155ConversionResult ConvertUTF32toUTF16 (
156  const UTF32** sourceStart, const UTF32* sourceEnd,
157  UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags);
158
159Boolean isLegalUTF8Sequence(const UTF8 *source, const UTF8 *sourceEnd);
160
161Boolean isLegalUTF8String(const UTF8 **source, const UTF8 *sourceEnd);
162
163unsigned getNumBytesForUTF8(UTF8 firstByte);
164
165#ifdef __cplusplus
166}
167
168/*************************************************************************/
169/* Below are LLVM-specific wrappers of the functions above. */
170
171#include "llvm/ADT/ArrayRef.h"
172#include "llvm/ADT/StringRef.h"
173
174namespace llvm {
175
176/**
177 * Convert an UTF8 StringRef to UTF8, UTF16, or UTF32 depending on
178 * WideCharWidth. The converted data is written to ResultPtr, which needs to
179 * point to at least WideCharWidth * (Source.Size() + 1) bytes. On success,
180 * ResultPtr will point one after the end of the copied string. On failure,
181 * ResultPtr will not be changed, and ErrorPtr will be set to the location of
182 * the first character which could not be converted.
183 * \return true on success.
184 */
185bool ConvertUTF8toWide(unsigned WideCharWidth, llvm::StringRef Source,
186                       char *&ResultPtr, const UTF8 *&ErrorPtr);
187
188/**
189 * Convert an Unicode code point to UTF8 sequence.
190 *
191 * \param Source a Unicode code point.
192 * \param [in,out] ResultPtr pointer to the output buffer, needs to be at least
193 * \c UNI_MAX_UTF8_BYTES_PER_CODE_POINT bytes.  On success \c ResultPtr is
194 * updated one past end of the converted sequence.
195 *
196 * \returns true on success.
197 */
198bool ConvertCodePointToUTF8(unsigned Source, char *&ResultPtr);
199
200/**
201 * Convert the first UTF8 sequence in the given source buffer to a UTF32
202 * code point.
203 *
204 * \param [in,out] source A pointer to the source buffer. If the conversion
205 * succeeds, this pointer will be updated to point to the byte just past the
206 * end of the converted sequence.
207 * \param sourceEnd A pointer just past the end of the source buffer.
208 * \param [out] target The converted code
209 * \param flags Whether the conversion is strict or lenient.
210 *
211 * \returns conversionOK on success
212 *
213 * \sa ConvertUTF8toUTF32
214 */
215static inline ConversionResult convertUTF8Sequence(const UTF8 **source,
216                                                   const UTF8 *sourceEnd,
217                                                   UTF32 *target,
218                                                   ConversionFlags flags) {
219  if (*source == sourceEnd)
220    return sourceExhausted;
221  unsigned size = getNumBytesForUTF8(**source);
222  if ((ptrdiff_t)size > sourceEnd - *source)
223    return sourceExhausted;
224  return ConvertUTF8toUTF32(source, *source + size, &target, target + 1, flags);
225}
226
227/**
228 * Returns true if a blob of text starts with a UTF-16 big or little endian byte
229 * order mark.
230 */
231bool hasUTF16ByteOrderMark(ArrayRef<char> SrcBytes);
232
233/**
234 * Converts a stream of raw bytes assumed to be UTF16 into a UTF8 std::string.
235 *
236 * \param [in] SrcBytes A buffer of what is assumed to be UTF-16 encoded text.
237 * \param [out] Out Converted UTF-8 is stored here on success.
238 * \returns true on success
239 */
240bool convertUTF16ToUTF8String(ArrayRef<char> SrcBytes, std::string &Out);
241
242} /* end namespace llvm */
243
244#endif
245
246/* --------------------------------------------------------------------- */
247
248#endif
249