diff --git a/include/llvm/ConvertUTF.h b/include/llvm/ConvertUTF.h new file mode 100644 index 0000000000..5b025331b8 --- /dev/null +++ b/include/llvm/ConvertUTF.h @@ -0,0 +1,255 @@ +/*===--- ConvertUTF.h - Universal Character Names conversions ---------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is distributed under the University of Illinois Open Source + * License. See LICENSE.TXT for details. + * + *==------------------------------------------------------------------------==*/ +/* + * Copyright 2001-2004 Unicode, Inc. + * + * Disclaimer + * + * This source code is provided as is by Unicode, Inc. No claims are + * made as to fitness for any particular purpose. No warranties of any + * kind are expressed or implied. The recipient agrees to determine + * applicability of information provided. If this file has been + * purchased on magnetic or optical media from Unicode, Inc., the + * sole remedy for any claim will be exchange of defective media + * within 90 days of receipt. + * + * Limitations on Rights to Redistribute This Code + * + * Unicode, Inc. hereby grants the right to freely use the information + * supplied in this file in the creation of products supporting the + * Unicode Standard, and to make copies of this file in any form + * for internal or external distribution as long as this notice + * remains attached. + */ + +/* --------------------------------------------------------------------- + + Conversions between UTF32, UTF-16, and UTF-8. Header file. + + Several funtions are included here, forming a complete set of + conversions between the three formats. UTF-7 is not included + here, but is handled in a separate source file. + + Each of these routines takes pointers to input buffers and output + buffers. The input buffers are const. + + Each routine converts the text between *sourceStart and sourceEnd, + putting the result into the buffer between *targetStart and + targetEnd. Note: the end pointers are *after* the last item: e.g. + *(sourceEnd - 1) is the last item. + + The return result indicates whether the conversion was successful, + and if not, whether the problem was in the source or target buffers. + (Only the first encountered problem is indicated.) + + After the conversion, *sourceStart and *targetStart are both + updated to point to the end of last text successfully converted in + the respective buffers. + + Input parameters: + sourceStart - pointer to a pointer to the source buffer. + The contents of this are modified on return so that + it points at the next thing to be converted. + targetStart - similarly, pointer to pointer to the target buffer. + sourceEnd, targetEnd - respectively pointers to the ends of the + two buffers, for overflow checking only. + + These conversion functions take a ConversionFlags argument. When this + flag is set to strict, both irregular sequences and isolated surrogates + will cause an error. When the flag is set to lenient, both irregular + sequences and isolated surrogates are converted. + + Whether the flag is strict or lenient, all illegal sequences will cause + an error return. This includes sequences such as: , , + or in UTF-8, and values above 0x10FFFF in UTF-32. Conformant code + must check for illegal sequences. + + When the flag is set to lenient, characters over 0x10FFFF are converted + to the replacement character; otherwise (when the flag is set to strict) + they constitute an error. + + Output parameters: + The value "sourceIllegal" is returned from some routines if the input + sequence is malformed. When "sourceIllegal" is returned, the source + value will point to the illegal value that caused the problem. E.g., + in UTF-8 when a sequence is malformed, it points to the start of the + malformed sequence. + + Author: Mark E. Davis, 1994. + Rev History: Rick McGowan, fixes & updates May 2001. + Fixes & updates, Sept 2001. + +------------------------------------------------------------------------ */ + +#ifndef LLVM_SUPPORT_CONVERTUTF_H +#define LLVM_SUPPORT_CONVERTUTF_H + +/* --------------------------------------------------------------------- + The following 4 definitions are compiler-specific. + The C standard does not guarantee that wchar_t has at least + 16 bits, so wchar_t is no less portable than unsigned short! + All should be unsigned values to avoid sign extension during + bit mask & shift operations. +------------------------------------------------------------------------ */ + +typedef unsigned int UTF32; /* at least 32 bits */ +typedef unsigned short UTF16; /* at least 16 bits */ +typedef unsigned char UTF8; /* typically 8 bits */ +typedef bool Boolean; /* 0 or 1 */ + +/* Some fundamental constants */ +#define UNI_REPLACEMENT_CHAR (UTF32)0x0000FFFD +#define UNI_MAX_BMP (UTF32)0x0000FFFF +#define UNI_MAX_UTF16 (UTF32)0x0010FFFF +#define UNI_MAX_UTF32 (UTF32)0x7FFFFFFF +#define UNI_MAX_LEGAL_UTF32 (UTF32)0x0010FFFF + +#define UNI_MAX_UTF8_BYTES_PER_CODE_POINT 4 + +#define UNI_UTF16_BYTE_ORDER_MARK_NATIVE 0xFEFF +#define UNI_UTF16_BYTE_ORDER_MARK_SWAPPED 0xFFFE + +typedef enum { + conversionOK, /* conversion successful */ + sourceExhausted, /* partial character in source, but hit end */ + targetExhausted, /* insuff. room in target for conversion */ + sourceIllegal /* source sequence is illegal/malformed */ +} ConversionResult; + +typedef enum { + strictConversion = 0, + lenientConversion +} ConversionFlags; + +/* This is for C++ and does no harm in C */ +#ifdef __cplusplus +extern "C" { +#endif + +ConversionResult ConvertUTF8toUTF16 ( + const UTF8** sourceStart, const UTF8* sourceEnd, + UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags); + +/** + * Convert a partial UTF8 sequence to UTF32. If the sequence ends in an + * incomplete code unit sequence, returns \c sourceExhausted. + */ +ConversionResult ConvertUTF8toUTF32Partial( + const UTF8** sourceStart, const UTF8* sourceEnd, + UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags); + +/** + * Convert a partial UTF8 sequence to UTF32. If the sequence ends in an + * incomplete code unit sequence, returns \c sourceIllegal. + */ +ConversionResult ConvertUTF8toUTF32( + const UTF8** sourceStart, const UTF8* sourceEnd, + UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags); + +ConversionResult ConvertUTF16toUTF8 ( + const UTF16** sourceStart, const UTF16* sourceEnd, + UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags); + +ConversionResult ConvertUTF32toUTF8 ( + const UTF32** sourceStart, const UTF32* sourceEnd, + UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags); + +ConversionResult ConvertUTF16toUTF32 ( + const UTF16** sourceStart, const UTF16* sourceEnd, + UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags); + +ConversionResult ConvertUTF32toUTF16 ( + const UTF32** sourceStart, const UTF32* sourceEnd, + UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags); + +Boolean isLegalUTF8Sequence(const UTF8 *source, const UTF8 *sourceEnd); + +Boolean isLegalUTF8String(const UTF8 **source, const UTF8 *sourceEnd); + +unsigned getNumBytesForUTF8(UTF8 firstByte); + +#ifdef __cplusplus +} + +/*************************************************************************/ +/* Below are LLVM-specific wrappers of the functions above. */ + +#include "llvm/ArrayRef.h" +#include "llvm/StringRef.h" + +namespace llvm { + +/** + * Convert an Unicode code point to UTF8 sequence. + * + * \param Source a Unicode code point. + * \param [in,out] ResultPtr pointer to the output buffer, needs to be at least + * \c UNI_MAX_UTF8_BYTES_PER_CODE_POINT bytes. On success \c ResultPtr is + * updated one past end of the converted sequence. + * + * \returns true on success. + */ +bool ConvertCodePointToUTF8(unsigned Source, char *&ResultPtr); + +/** + * Convert the first UTF8 sequence in the given source buffer to a UTF32 + * code point. + * + * \param [in,out] source A pointer to the source buffer. If the conversion + * succeeds, this pointer will be updated to point to the byte just past the + * end of the converted sequence. + * \param sourceEnd A pointer just past the end of the source buffer. + * \param [out] target The converted code + * \param flags Whether the conversion is strict or lenient. + * + * \returns conversionOK on success + * + * \sa ConvertUTF8toUTF32 + */ +static inline ConversionResult convertUTF8Sequence(const UTF8 **source, + const UTF8 *sourceEnd, + UTF32 *target, + ConversionFlags flags) { + if (*source == sourceEnd) + return sourceExhausted; + unsigned size = getNumBytesForUTF8(**source); + if ((ptrdiff_t)size > sourceEnd - *source) + return sourceExhausted; + return ConvertUTF8toUTF32(source, *source + size, &target, target + 1, flags); +} + +/** + * Returns true if a blob of text starts with a UTF-16 big or little endian byte + * order mark. + */ +bool hasUTF16ByteOrderMark(ArrayRef SrcBytes); + +/** + * Converts a UTF-16 string into a UTF-8 string. + * + * \returns true on success + */ +bool convertUTF16ToUTF8String(ArrayRef SrcUTF16, + SmallVectorImpl &DstUTF8); + +/** + * Converts a UTF-8 string into a UTF-16 string with native endianness. + * + * \returns true on success + */ +bool convertUTF8ToUTF16String(StringRef SrcUTF8, + SmallVectorImpl &DstUTF16); + +} /* end namespace llvm */ + +#endif + +/* --------------------------------------------------------------------- */ + +#endif diff --git a/java/lib/NetworkTablesJNI.cpp b/java/lib/NetworkTablesJNI.cpp index f425433811..8520cd5117 100644 --- a/java/lib/NetworkTablesJNI.cpp +++ b/java/lib/NetworkTablesJNI.cpp @@ -11,6 +11,9 @@ #include "ntcore.h" #include "atomic_static.h" #include "SafeThread.h" +#include "llvm/ConvertUTF.h" +#include "llvm/SmallString.h" +#include "llvm/SmallVector.h" // // Globals and load/unload @@ -221,20 +224,19 @@ class JavaWeakGlobal { class JavaStringRef { public: - JavaStringRef(JNIEnv *env, jstring str) - : m_env(env), - m_jstr(str), - m_str(env->GetStringUTFChars(str, nullptr)) {} - ~JavaStringRef() { m_env->ReleaseStringUTFChars(m_jstr, m_str); } + JavaStringRef(JNIEnv *env, jstring str) { + jsize size = env->GetStringLength(str); + const jchar *chars = env->GetStringChars(str, nullptr); + llvm::convertUTF16ToUTF8String(llvm::makeArrayRef(chars, size), m_str); + env->ReleaseStringChars(str, chars); + } - operator nt::StringRef() const { return nt::StringRef(m_str); } - nt::StringRef str() const { return nt::StringRef(m_str); } - const char *c_str() const { return m_str; } + operator nt::StringRef() const { return m_str; } + nt::StringRef str() const { return m_str; } + const char* c_str() const { return m_str.data(); } private: - JNIEnv *m_env; - jstring m_jstr; - const char *m_str; + llvm::SmallString<128> m_str; }; class JavaByteRef { @@ -344,11 +346,9 @@ std::shared_ptr FromJavaStringArray(JNIEnv *env, jobjectArray jarr) { // static inline jstring ToJavaString(JNIEnv *env, nt::StringRef str) { - // fastpath if string already null terminated; if not, need to make a copy - if (str.data()[str.size()] == '\0') - return env->NewStringUTF(str.data()); - else - return env->NewStringUTF(str.str().c_str()); + llvm::SmallVector chars; + llvm::convertUTF8ToUTF16String(str, chars); + return env->NewString(chars.begin(), chars.size()); } static jbyteArray ToJavaByteArray(JNIEnv *env, nt::StringRef str) { diff --git a/src/llvm/ConvertUTF.cpp b/src/llvm/ConvertUTF.cpp new file mode 100644 index 0000000000..8774c471e3 --- /dev/null +++ b/src/llvm/ConvertUTF.cpp @@ -0,0 +1,710 @@ +/*===--- ConvertUTF.c - Universal Character Names conversions ---------------=== + * + * The LLVM Compiler Infrastructure + * + * This file is distributed under the University of Illinois Open Source + * License. See LICENSE.TXT for details. + * + *===------------------------------------------------------------------------=*/ +/* + * Copyright 2001-2004 Unicode, Inc. + * + * Disclaimer + * + * This source code is provided as is by Unicode, Inc. No claims are + * made as to fitness for any particular purpose. No warranties of any + * kind are expressed or implied. The recipient agrees to determine + * applicability of information provided. If this file has been + * purchased on magnetic or optical media from Unicode, Inc., the + * sole remedy for any claim will be exchange of defective media + * within 90 days of receipt. + * + * Limitations on Rights to Redistribute This Code + * + * Unicode, Inc. hereby grants the right to freely use the information + * supplied in this file in the creation of products supporting the + * Unicode Standard, and to make copies of this file in any form + * for internal or external distribution as long as this notice + * remains attached. + */ + +/* --------------------------------------------------------------------- + + Conversions between UTF32, UTF-16, and UTF-8. Source code file. + Author: Mark E. Davis, 1994. + Rev History: Rick McGowan, fixes & updates May 2001. + Sept 2001: fixed const & error conditions per + mods suggested by S. Parent & A. Lillich. + June 2002: Tim Dodd added detection and handling of incomplete + source sequences, enhanced error detection, added casts + to eliminate compiler warnings. + July 2003: slight mods to back out aggressive FFFE detection. + Jan 2004: updated switches in from-UTF8 conversions. + Oct 2004: updated to use UNI_MAX_LEGAL_UTF32 in UTF-32 conversions. + + See the header file "ConvertUTF.h" for complete documentation. + +------------------------------------------------------------------------ */ + + +#include "llvm/ConvertUTF.h" +#ifdef CVTUTF_DEBUG +#include +#endif +#include + +static const int halfShift = 10; /* used for shifting by 10 bits */ + +static const UTF32 halfBase = 0x0010000UL; +static const UTF32 halfMask = 0x3FFUL; + +#define UNI_SUR_HIGH_START (UTF32)0xD800 +#define UNI_SUR_HIGH_END (UTF32)0xDBFF +#define UNI_SUR_LOW_START (UTF32)0xDC00 +#define UNI_SUR_LOW_END (UTF32)0xDFFF +#define false 0 +#define true 1 + +/* --------------------------------------------------------------------- */ + +/* + * Index into the table below with the first byte of a UTF-8 sequence to + * get the number of trailing bytes that are supposed to follow it. + * Note that *legal* UTF-8 values can't have 4 or 5-bytes. The table is + * left as-is for anyone who may want to do such conversion, which was + * allowed in earlier algorithms. + */ +static const char trailingBytesForUTF8[256] = { + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, + 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, + 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, 3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5 +}; + +/* + * Magic values subtracted from a buffer value during UTF8 conversion. + * This table contains as many values as there might be trailing bytes + * in a UTF-8 sequence. + */ +static const UTF32 offsetsFromUTF8[6] = { 0x00000000UL, 0x00003080UL, 0x000E2080UL, + 0x03C82080UL, 0xFA082080UL, 0x82082080UL }; + +/* + * Once the bits are split out into bytes of UTF-8, this is a mask OR-ed + * into the first byte, depending on how many bytes follow. There are + * as many entries in this table as there are UTF-8 sequence types. + * (I.e., one byte sequence, two byte... etc.). Remember that sequencs + * for *legal* UTF-8 will be 4 or fewer bytes total. + */ +static const UTF8 firstByteMark[7] = { 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC }; + +/* --------------------------------------------------------------------- */ + +/* The interface converts a whole buffer to avoid function-call overhead. + * Constants have been gathered. Loops & conditionals have been removed as + * much as possible for efficiency, in favor of drop-through switches. + * (See "Note A" at the bottom of the file for equivalent code.) + * If your compiler supports it, the "isLegalUTF8" call can be turned + * into an inline function. + */ + +extern "C" { +/* --------------------------------------------------------------------- */ + +ConversionResult ConvertUTF32toUTF16 ( + const UTF32** sourceStart, const UTF32* sourceEnd, + UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags) { + ConversionResult result = conversionOK; + const UTF32* source = *sourceStart; + UTF16* target = *targetStart; + while (source < sourceEnd) { + UTF32 ch; + if (target >= targetEnd) { + result = targetExhausted; break; + } + ch = *source++; + if (ch <= UNI_MAX_BMP) { /* Target is a character <= 0xFFFF */ + /* UTF-16 surrogate values are illegal in UTF-32; 0xffff or 0xfffe are both reserved values */ + if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) { + if (flags == strictConversion) { + --source; /* return to the illegal value itself */ + result = sourceIllegal; + break; + } else { + *target++ = UNI_REPLACEMENT_CHAR; + } + } else { + *target++ = (UTF16)ch; /* normal case */ + } + } else if (ch > UNI_MAX_LEGAL_UTF32) { + if (flags == strictConversion) { + result = sourceIllegal; + } else { + *target++ = UNI_REPLACEMENT_CHAR; + } + } else { + /* target is a character in range 0xFFFF - 0x10FFFF. */ + if (target + 1 >= targetEnd) { + --source; /* Back up source pointer! */ + result = targetExhausted; break; + } + ch -= halfBase; + *target++ = (UTF16)((ch >> halfShift) + UNI_SUR_HIGH_START); + *target++ = (UTF16)((ch & halfMask) + UNI_SUR_LOW_START); + } + } + *sourceStart = source; + *targetStart = target; + return result; +} + +/* --------------------------------------------------------------------- */ + +ConversionResult ConvertUTF16toUTF32 ( + const UTF16** sourceStart, const UTF16* sourceEnd, + UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags) { + ConversionResult result = conversionOK; + const UTF16* source = *sourceStart; + UTF32* target = *targetStart; + UTF32 ch, ch2; + while (source < sourceEnd) { + const UTF16* oldSource = source; /* In case we have to back up because of target overflow. */ + ch = *source++; + /* If we have a surrogate pair, convert to UTF32 first. */ + if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_HIGH_END) { + /* If the 16 bits following the high surrogate are in the source buffer... */ + if (source < sourceEnd) { + ch2 = *source; + /* If it's a low surrogate, convert to UTF32. */ + if (ch2 >= UNI_SUR_LOW_START && ch2 <= UNI_SUR_LOW_END) { + ch = ((ch - UNI_SUR_HIGH_START) << halfShift) + + (ch2 - UNI_SUR_LOW_START) + halfBase; + ++source; + } else if (flags == strictConversion) { /* it's an unpaired high surrogate */ + --source; /* return to the illegal value itself */ + result = sourceIllegal; + break; + } + } else { /* We don't have the 16 bits following the high surrogate. */ + --source; /* return to the high surrogate */ + result = sourceExhausted; + break; + } + } else if (flags == strictConversion) { + /* UTF-16 surrogate values are illegal in UTF-32 */ + if (ch >= UNI_SUR_LOW_START && ch <= UNI_SUR_LOW_END) { + --source; /* return to the illegal value itself */ + result = sourceIllegal; + break; + } + } + if (target >= targetEnd) { + source = oldSource; /* Back up source pointer! */ + result = targetExhausted; break; + } + *target++ = ch; + } + *sourceStart = source; + *targetStart = target; +#ifdef CVTUTF_DEBUG +if (result == sourceIllegal) { + fprintf(stderr, "ConvertUTF16toUTF32 illegal seq 0x%04x,%04x\n", ch, ch2); + fflush(stderr); +} +#endif + return result; +} +ConversionResult ConvertUTF16toUTF8 ( + const UTF16** sourceStart, const UTF16* sourceEnd, + UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags) { + ConversionResult result = conversionOK; + const UTF16* source = *sourceStart; + UTF8* target = *targetStart; + while (source < sourceEnd) { + UTF32 ch; + unsigned short bytesToWrite = 0; + const UTF32 byteMask = 0xBF; + const UTF32 byteMark = 0x80; + const UTF16* oldSource = source; /* In case we have to back up because of target overflow. */ + ch = *source++; + /* If we have a surrogate pair, convert to UTF32 first. */ + if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_HIGH_END) { + /* If the 16 bits following the high surrogate are in the source buffer... */ + if (source < sourceEnd) { + UTF32 ch2 = *source; + /* If it's a low surrogate, convert to UTF32. */ + if (ch2 >= UNI_SUR_LOW_START && ch2 <= UNI_SUR_LOW_END) { + ch = ((ch - UNI_SUR_HIGH_START) << halfShift) + + (ch2 - UNI_SUR_LOW_START) + halfBase; + ++source; + } else if (flags == strictConversion) { /* it's an unpaired high surrogate */ + --source; /* return to the illegal value itself */ + result = sourceIllegal; + break; + } + } else { /* We don't have the 16 bits following the high surrogate. */ + --source; /* return to the high surrogate */ + result = sourceExhausted; + break; + } + } else if (flags == strictConversion) { + /* UTF-16 surrogate values are illegal in UTF-32 */ + if (ch >= UNI_SUR_LOW_START && ch <= UNI_SUR_LOW_END) { + --source; /* return to the illegal value itself */ + result = sourceIllegal; + break; + } + } + /* Figure out how many bytes the result will require */ + if (ch < (UTF32)0x80) { bytesToWrite = 1; + } else if (ch < (UTF32)0x800) { bytesToWrite = 2; + } else if (ch < (UTF32)0x10000) { bytesToWrite = 3; + } else if (ch < (UTF32)0x110000) { bytesToWrite = 4; + } else { bytesToWrite = 3; + ch = UNI_REPLACEMENT_CHAR; + } + + target += bytesToWrite; + if (target > targetEnd) { + source = oldSource; /* Back up source pointer! */ + target -= bytesToWrite; result = targetExhausted; break; + } + switch (bytesToWrite) { /* note: everything falls through. */ + case 4: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6; + case 3: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6; + case 2: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6; + case 1: *--target = (UTF8)(ch | firstByteMark[bytesToWrite]); + } + target += bytesToWrite; + } + *sourceStart = source; + *targetStart = target; + return result; +} + +/* --------------------------------------------------------------------- */ + +ConversionResult ConvertUTF32toUTF8 ( + const UTF32** sourceStart, const UTF32* sourceEnd, + UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags) { + ConversionResult result = conversionOK; + const UTF32* source = *sourceStart; + UTF8* target = *targetStart; + while (source < sourceEnd) { + UTF32 ch; + unsigned short bytesToWrite = 0; + const UTF32 byteMask = 0xBF; + const UTF32 byteMark = 0x80; + ch = *source++; + if (flags == strictConversion ) { + /* UTF-16 surrogate values are illegal in UTF-32 */ + if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) { + --source; /* return to the illegal value itself */ + result = sourceIllegal; + break; + } + } + /* + * Figure out how many bytes the result will require. Turn any + * illegally large UTF32 things (> Plane 17) into replacement chars. + */ + if (ch < (UTF32)0x80) { bytesToWrite = 1; + } else if (ch < (UTF32)0x800) { bytesToWrite = 2; + } else if (ch < (UTF32)0x10000) { bytesToWrite = 3; + } else if (ch <= UNI_MAX_LEGAL_UTF32) { bytesToWrite = 4; + } else { bytesToWrite = 3; + ch = UNI_REPLACEMENT_CHAR; + result = sourceIllegal; + } + + target += bytesToWrite; + if (target > targetEnd) { + --source; /* Back up source pointer! */ + target -= bytesToWrite; result = targetExhausted; break; + } + switch (bytesToWrite) { /* note: everything falls through. */ + case 4: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6; + case 3: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6; + case 2: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6; + case 1: *--target = (UTF8) (ch | firstByteMark[bytesToWrite]); + } + target += bytesToWrite; + } + *sourceStart = source; + *targetStart = target; + return result; +} + +/* --------------------------------------------------------------------- */ + +/* + * Utility routine to tell whether a sequence of bytes is legal UTF-8. + * This must be called with the length pre-determined by the first byte. + * If not calling this from ConvertUTF8to*, then the length can be set by: + * length = trailingBytesForUTF8[*source]+1; + * and the sequence is illegal right away if there aren't that many bytes + * available. + * If presented with a length > 4, this returns false. The Unicode + * definition of UTF-8 goes up to 4-byte sequences. + */ + +static Boolean isLegalUTF8(const UTF8 *source, int length) { + UTF8 a; + const UTF8 *srcptr = source+length; + switch (length) { + default: return false; + /* Everything else falls through when "true"... */ + case 4: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false; + case 3: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false; + case 2: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false; + + switch (*source) { + /* no fall-through in this inner switch */ + case 0xE0: if (a < 0xA0) return false; break; + case 0xED: if (a > 0x9F) return false; break; + case 0xF0: if (a < 0x90) return false; break; + case 0xF4: if (a > 0x8F) return false; break; + default: if (a < 0x80) return false; + } + + case 1: if (*source >= 0x80 && *source < 0xC2) return false; + } + if (*source > 0xF4) return false; + return true; +} + +/* --------------------------------------------------------------------- */ + +/* + * Exported function to return whether a UTF-8 sequence is legal or not. + * This is not used here; it's just exported. + */ +Boolean isLegalUTF8Sequence(const UTF8 *source, const UTF8 *sourceEnd) { + int length = trailingBytesForUTF8[*source]+1; + if (length > sourceEnd - source) { + return false; + } + return isLegalUTF8(source, length); +} + +/* --------------------------------------------------------------------- */ + +static unsigned +findMaximalSubpartOfIllFormedUTF8Sequence(const UTF8 *source, + const UTF8 *sourceEnd) { + UTF8 b1, b2, b3; + + assert(!isLegalUTF8Sequence(source, sourceEnd)); + + /* + * Unicode 6.3.0, D93b: + * + * Maximal subpart of an ill-formed subsequence: The longest code unit + * subsequence starting at an unconvertible offset that is either: + * a. the initial subsequence of a well-formed code unit sequence, or + * b. a subsequence of length one. + */ + + if (source == sourceEnd) + return 0; + + /* + * Perform case analysis. See Unicode 6.3.0, Table 3-7. Well-Formed UTF-8 + * Byte Sequences. + */ + + b1 = *source; + ++source; + if (b1 >= 0xC2 && b1 <= 0xDF) { + /* + * First byte is valid, but we know that this code unit sequence is + * invalid, so the maximal subpart has to end after the first byte. + */ + return 1; + } + + if (source == sourceEnd) + return 1; + + b2 = *source; + ++source; + + if (b1 == 0xE0) { + return (b2 >= 0xA0 && b2 <= 0xBF) ? 2 : 1; + } + if (b1 >= 0xE1 && b1 <= 0xEC) { + return (b2 >= 0x80 && b2 <= 0xBF) ? 2 : 1; + } + if (b1 == 0xED) { + return (b2 >= 0x80 && b2 <= 0x9F) ? 2 : 1; + } + if (b1 >= 0xEE && b1 <= 0xEF) { + return (b2 >= 0x80 && b2 <= 0xBF) ? 2 : 1; + } + if (b1 == 0xF0) { + if (b2 >= 0x90 && b2 <= 0xBF) { + if (source == sourceEnd) + return 2; + + b3 = *source; + return (b3 >= 0x80 && b3 <= 0xBF) ? 3 : 2; + } + return 1; + } + if (b1 >= 0xF1 && b1 <= 0xF3) { + if (b2 >= 0x80 && b2 <= 0xBF) { + if (source == sourceEnd) + return 2; + + b3 = *source; + return (b3 >= 0x80 && b3 <= 0xBF) ? 3 : 2; + } + return 1; + } + if (b1 == 0xF4) { + if (b2 >= 0x80 && b2 <= 0x8F) { + if (source == sourceEnd) + return 2; + + b3 = *source; + return (b3 >= 0x80 && b3 <= 0xBF) ? 3 : 2; + } + return 1; + } + + assert((b1 >= 0x80 && b1 <= 0xC1) || b1 >= 0xF5); + /* + * There are no valid sequences that start with these bytes. Maximal subpart + * is defined to have length 1 in these cases. + */ + return 1; +} + +/* --------------------------------------------------------------------- */ + +/* + * Exported function to return the total number of bytes in a codepoint + * represented in UTF-8, given the value of the first byte. + */ +unsigned getNumBytesForUTF8(UTF8 first) { + return trailingBytesForUTF8[first] + 1; +} + +/* --------------------------------------------------------------------- */ + +/* + * Exported function to return whether a UTF-8 string is legal or not. + * This is not used here; it's just exported. + */ +Boolean isLegalUTF8String(const UTF8 **source, const UTF8 *sourceEnd) { + while (*source != sourceEnd) { + int length = trailingBytesForUTF8[**source] + 1; + if (length > sourceEnd - *source || !isLegalUTF8(*source, length)) + return false; + *source += length; + } + return true; +} + +/* --------------------------------------------------------------------- */ + +ConversionResult ConvertUTF8toUTF16 ( + const UTF8** sourceStart, const UTF8* sourceEnd, + UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags) { + ConversionResult result = conversionOK; + const UTF8* source = *sourceStart; + UTF16* target = *targetStart; + while (source < sourceEnd) { + UTF32 ch = 0; + unsigned short extraBytesToRead = trailingBytesForUTF8[*source]; + if (extraBytesToRead >= sourceEnd - source) { + result = sourceExhausted; break; + } + /* Do this check whether lenient or strict */ + if (!isLegalUTF8(source, extraBytesToRead+1)) { + result = sourceIllegal; + break; + } + /* + * The cases all fall through. See "Note A" below. + */ + switch (extraBytesToRead) { + case 5: ch += *source++; ch <<= 6; /* remember, illegal UTF-8 */ + case 4: ch += *source++; ch <<= 6; /* remember, illegal UTF-8 */ + case 3: ch += *source++; ch <<= 6; + case 2: ch += *source++; ch <<= 6; + case 1: ch += *source++; ch <<= 6; + case 0: ch += *source++; + } + ch -= offsetsFromUTF8[extraBytesToRead]; + + if (target >= targetEnd) { + source -= (extraBytesToRead+1); /* Back up source pointer! */ + result = targetExhausted; break; + } + if (ch <= UNI_MAX_BMP) { /* Target is a character <= 0xFFFF */ + /* UTF-16 surrogate values are illegal in UTF-32 */ + if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) { + if (flags == strictConversion) { + source -= (extraBytesToRead+1); /* return to the illegal value itself */ + result = sourceIllegal; + break; + } else { + *target++ = UNI_REPLACEMENT_CHAR; + } + } else { + *target++ = (UTF16)ch; /* normal case */ + } + } else if (ch > UNI_MAX_UTF16) { + if (flags == strictConversion) { + result = sourceIllegal; + source -= (extraBytesToRead+1); /* return to the start */ + break; /* Bail out; shouldn't continue */ + } else { + *target++ = UNI_REPLACEMENT_CHAR; + } + } else { + /* target is a character in range 0xFFFF - 0x10FFFF. */ + if (target + 1 >= targetEnd) { + source -= (extraBytesToRead+1); /* Back up source pointer! */ + result = targetExhausted; break; + } + ch -= halfBase; + *target++ = (UTF16)((ch >> halfShift) + UNI_SUR_HIGH_START); + *target++ = (UTF16)((ch & halfMask) + UNI_SUR_LOW_START); + } + } + *sourceStart = source; + *targetStart = target; + return result; +} + +/* --------------------------------------------------------------------- */ + +static ConversionResult ConvertUTF8toUTF32Impl( + const UTF8** sourceStart, const UTF8* sourceEnd, + UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags, + Boolean InputIsPartial) { + ConversionResult result = conversionOK; + const UTF8* source = *sourceStart; + UTF32* target = *targetStart; + while (source < sourceEnd) { + UTF32 ch = 0; + unsigned short extraBytesToRead = trailingBytesForUTF8[*source]; + if (extraBytesToRead >= sourceEnd - source) { + if (flags == strictConversion || InputIsPartial) { + result = sourceExhausted; + break; + } else { + result = sourceIllegal; + + /* + * Replace the maximal subpart of ill-formed sequence with + * replacement character. + */ + source += findMaximalSubpartOfIllFormedUTF8Sequence(source, + sourceEnd); + *target++ = UNI_REPLACEMENT_CHAR; + continue; + } + } + if (target >= targetEnd) { + result = targetExhausted; break; + } + + /* Do this check whether lenient or strict */ + if (!isLegalUTF8(source, extraBytesToRead+1)) { + result = sourceIllegal; + if (flags == strictConversion) { + /* Abort conversion. */ + break; + } else { + /* + * Replace the maximal subpart of ill-formed sequence with + * replacement character. + */ + source += findMaximalSubpartOfIllFormedUTF8Sequence(source, + sourceEnd); + *target++ = UNI_REPLACEMENT_CHAR; + continue; + } + } + /* + * The cases all fall through. See "Note A" below. + */ + switch (extraBytesToRead) { + case 5: ch += *source++; ch <<= 6; + case 4: ch += *source++; ch <<= 6; + case 3: ch += *source++; ch <<= 6; + case 2: ch += *source++; ch <<= 6; + case 1: ch += *source++; ch <<= 6; + case 0: ch += *source++; + } + ch -= offsetsFromUTF8[extraBytesToRead]; + + if (ch <= UNI_MAX_LEGAL_UTF32) { + /* + * UTF-16 surrogate values are illegal in UTF-32, and anything + * over Plane 17 (> 0x10FFFF) is illegal. + */ + if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) { + if (flags == strictConversion) { + source -= (extraBytesToRead+1); /* return to the illegal value itself */ + result = sourceIllegal; + break; + } else { + *target++ = UNI_REPLACEMENT_CHAR; + } + } else { + *target++ = ch; + } + } else { /* i.e., ch > UNI_MAX_LEGAL_UTF32 */ + result = sourceIllegal; + *target++ = UNI_REPLACEMENT_CHAR; + } + } + *sourceStart = source; + *targetStart = target; + return result; +} + +ConversionResult ConvertUTF8toUTF32Partial(const UTF8 **sourceStart, + const UTF8 *sourceEnd, + UTF32 **targetStart, + UTF32 *targetEnd, + ConversionFlags flags) { + return ConvertUTF8toUTF32Impl(sourceStart, sourceEnd, targetStart, targetEnd, + flags, /*InputIsPartial=*/true); +} + +ConversionResult ConvertUTF8toUTF32(const UTF8 **sourceStart, + const UTF8 *sourceEnd, UTF32 **targetStart, + UTF32 *targetEnd, ConversionFlags flags) { + return ConvertUTF8toUTF32Impl(sourceStart, sourceEnd, targetStart, targetEnd, + flags, /*InputIsPartial=*/false); +} + +} + +/* --------------------------------------------------------------------- + + Note A. + The fall-through switches in UTF-8 reading code save a + temp variable, some decrements & conditionals. The switches + are equivalent to the following loop: + { + int tmpBytesToRead = extraBytesToRead+1; + do { + ch += *source++; + --tmpBytesToRead; + if (tmpBytesToRead) ch <<= 6; + } while (tmpBytesToRead > 0); + } + In UTF-8 writing code, the switches on "bytesToWrite" are + similarly unrolled loops. + + --------------------------------------------------------------------- */ diff --git a/src/llvm/ConvertUTFWrapper.cpp b/src/llvm/ConvertUTFWrapper.cpp new file mode 100644 index 0000000000..810c135166 --- /dev/null +++ b/src/llvm/ConvertUTFWrapper.cpp @@ -0,0 +1,122 @@ +//===-- ConvertUTFWrapper.cpp - Wrap ConvertUTF.h with clang data types -----=== +// +// The LLVM Compiler Infrastructure +// +// This file is distributed under the University of Illinois Open Source +// License. See LICENSE.TXT for details. +// +//===----------------------------------------------------------------------===// + +#include "llvm/ConvertUTF.h" +#include +#include + +namespace llvm { + +bool ConvertCodePointToUTF8(unsigned Source, char *&ResultPtr) { + const UTF32 *SourceStart = &Source; + const UTF32 *SourceEnd = SourceStart + 1; + UTF8 *TargetStart = reinterpret_cast(ResultPtr); + UTF8 *TargetEnd = TargetStart + 4; + ConversionResult CR = ConvertUTF32toUTF8(&SourceStart, SourceEnd, + &TargetStart, TargetEnd, + strictConversion); + if (CR != conversionOK) + return false; + + ResultPtr = reinterpret_cast(TargetStart); + return true; +} + +bool hasUTF16ByteOrderMark(ArrayRef S) { + return (S.size() >= 2 && + ((S[0] == '\xff' && S[1] == '\xfe') || + (S[0] == '\xfe' && S[1] == '\xff'))); +} + +bool convertUTF16ToUTF8String(ArrayRef SrcUTF16, + SmallVectorImpl &DstUTF8) { + assert(DstUTF8.empty()); + + // Avoid OOB by returning early on empty input. + if (SrcUTF16.empty()) + return true; + + const UTF16 *Src = SrcUTF16.begin(); + const UTF16 *SrcEnd = SrcUTF16.end(); + + // Byteswap if necessary. + std::vector ByteSwapped; + if (Src[0] == UNI_UTF16_BYTE_ORDER_MARK_SWAPPED) { + ByteSwapped.insert(ByteSwapped.end(), Src, SrcEnd); + for (unsigned I = 0, E = ByteSwapped.size(); I != E; ++I) + ByteSwapped[I] = (ByteSwapped[I] << 8) | (ByteSwapped[I] >> 8); + Src = &ByteSwapped[0]; + SrcEnd = &ByteSwapped[ByteSwapped.size() - 1] + 1; + } + + // Skip the BOM for conversion. + if (Src[0] == UNI_UTF16_BYTE_ORDER_MARK_NATIVE) + Src++; + + // Just allocate enough space up front. We'll shrink it later. Allocate + // enough that we can fit a null terminator without reallocating. + DstUTF8.resize(SrcUTF16.size() * UNI_MAX_UTF8_BYTES_PER_CODE_POINT + 1); + UTF8 *Dst = reinterpret_cast(&DstUTF8[0]); + UTF8 *DstEnd = Dst + DstUTF8.size(); + + ConversionResult CR = + ConvertUTF16toUTF8(&Src, SrcEnd, &Dst, DstEnd, strictConversion); + assert(CR != targetExhausted); + + if (CR != conversionOK) { + DstUTF8.clear(); + return false; + } + + DstUTF8.resize(reinterpret_cast(Dst) - &DstUTF8[0]); + DstUTF8.push_back(0); + DstUTF8.pop_back(); + return true; +} + +bool convertUTF8ToUTF16String(StringRef SrcUTF8, + SmallVectorImpl &DstUTF16) { + assert(DstUTF16.empty()); + + // Avoid OOB by returning early on empty input. + if (SrcUTF8.empty()) { + DstUTF16.push_back(0); + DstUTF16.pop_back(); + return true; + } + + const UTF8 *Src = reinterpret_cast(SrcUTF8.begin()); + const UTF8 *SrcEnd = reinterpret_cast(SrcUTF8.end()); + + // Allocate the same number of UTF-16 code units as UTF-8 code units. Encoding + // as UTF-16 should always require the same amount or less code units than the + // UTF-8 encoding. Allocate one extra byte for the null terminator though, + // so that someone calling DstUTF16.data() gets a null terminated string. + // We resize down later so we don't have to worry that this over allocates. + DstUTF16.resize(SrcUTF8.size()+1); + UTF16 *Dst = &DstUTF16[0]; + UTF16 *DstEnd = Dst + DstUTF16.size(); + + ConversionResult CR = + ConvertUTF8toUTF16(&Src, SrcEnd, &Dst, DstEnd, strictConversion); + assert(CR != targetExhausted); + + if (CR != conversionOK) { + DstUTF16.clear(); + return false; + } + + DstUTF16.resize(Dst - &DstUTF16[0]); + DstUTF16.push_back(0); + DstUTF16.pop_back(); + return true; +} + +} // end namespace llvm +