tomwalters@268: /*
tomwalters@268:  * Copyright 2001-2004 Unicode, Inc.
tomwalters@268:  * 
tomwalters@268:  * Disclaimer
tomwalters@268:  * 
tomwalters@268:  * This source code is provided as is by Unicode, Inc. No claims are
tomwalters@268:  * made as to fitness for any particular purpose. No warranties of any
tomwalters@268:  * kind are expressed or implied. The recipient agrees to determine
tomwalters@268:  * applicability of information provided. If this file has been
tomwalters@268:  * purchased on magnetic or optical media from Unicode, Inc., the
tomwalters@268:  * sole remedy for any claim will be exchange of defective media
tomwalters@268:  * within 90 days of receipt.
tomwalters@268:  * 
tomwalters@268:  * Limitations on Rights to Redistribute This Code
tomwalters@268:  * 
tomwalters@268:  * Unicode, Inc. hereby grants the right to freely use the information
tomwalters@268:  * supplied in this file in the creation of products supporting the
tomwalters@268:  * Unicode Standard, and to make copies of this file in any form
tomwalters@268:  * for internal or external distribution as long as this notice
tomwalters@268:  * remains attached.
tomwalters@268:  */
tomwalters@268: 
tomwalters@268: /* ---------------------------------------------------------------------
tomwalters@268: 
tomwalters@268:     Conversions between UTF32, UTF-16, and UTF-8. Source code file.
tomwalters@268:     Author: Mark E. Davis, 1994.
tomwalters@268:     Rev History: Rick McGowan, fixes & updates May 2001.
tomwalters@268:     Sept 2001: fixed const & error conditions per
tomwalters@268: 	mods suggested by S. Parent & A. Lillich.
tomwalters@268:     June 2002: Tim Dodd added detection and handling of incomplete
tomwalters@268: 	source sequences, enhanced error detection, added casts
tomwalters@268: 	to eliminate compiler warnings.
tomwalters@268:     July 2003: slight mods to back out aggressive FFFE detection.
tomwalters@268:     Jan 2004: updated switches in from-UTF8 conversions.
tomwalters@268:     Oct 2004: updated to use UNI_MAX_LEGAL_UTF32 in UTF-32 conversions.
tomwalters@268: 
tomwalters@268:     See the header file "ConvertUTF.h" for complete documentation.
tomwalters@268: 
tomwalters@268: ------------------------------------------------------------------------ */
tomwalters@268: 
tomwalters@268: 
tomwalters@268: #include "ConvertUTF.h"
tomwalters@268: #ifdef CVTUTF_DEBUG
tomwalters@268: #include <stdio.h>
tomwalters@268: #endif
tomwalters@268: 
tomwalters@268: static const int halfShift  = 10; /* used for shifting by 10 bits */
tomwalters@268: 
tomwalters@268: static const UTF32 halfBase = 0x0010000UL;
tomwalters@268: static const UTF32 halfMask = 0x3FFUL;
tomwalters@268: 
tomwalters@268: #define UNI_SUR_HIGH_START  (UTF32)0xD800
tomwalters@268: #define UNI_SUR_HIGH_END    (UTF32)0xDBFF
tomwalters@268: #define UNI_SUR_LOW_START   (UTF32)0xDC00
tomwalters@268: #define UNI_SUR_LOW_END     (UTF32)0xDFFF
tomwalters@268: #define false	   0
tomwalters@268: #define true	    1
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: ConversionResult ConvertUTF32toUTF16 (
tomwalters@268: 	const UTF32** sourceStart, const UTF32* sourceEnd, 
tomwalters@268: 	UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags) {
tomwalters@268:     ConversionResult result = conversionOK;
tomwalters@268:     const UTF32* source = *sourceStart;
tomwalters@268:     UTF16* target = *targetStart;
tomwalters@268:     while (source < sourceEnd) {
tomwalters@268: 	UTF32 ch;
tomwalters@268: 	if (target >= targetEnd) {
tomwalters@268: 	    result = targetExhausted; break;
tomwalters@268: 	}
tomwalters@268: 	ch = *source++;
tomwalters@268: 	if (ch <= UNI_MAX_BMP) { /* Target is a character <= 0xFFFF */
tomwalters@268: 	    /* UTF-16 surrogate values are illegal in UTF-32; 0xffff or 0xfffe are both reserved values */
tomwalters@268: 	    if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) {
tomwalters@268: 		if (flags == strictConversion) {
tomwalters@268: 		    --source; /* return to the illegal value itself */
tomwalters@268: 		    result = sourceIllegal;
tomwalters@268: 		    break;
tomwalters@268: 		} else {
tomwalters@268: 		    *target++ = UNI_REPLACEMENT_CHAR;
tomwalters@268: 		}
tomwalters@268: 	    } else {
tomwalters@268: 		*target++ = (UTF16)ch; /* normal case */
tomwalters@268: 	    }
tomwalters@268: 	} else if (ch > UNI_MAX_LEGAL_UTF32) {
tomwalters@268: 	    if (flags == strictConversion) {
tomwalters@268: 		result = sourceIllegal;
tomwalters@268: 	    } else {
tomwalters@268: 		*target++ = UNI_REPLACEMENT_CHAR;
tomwalters@268: 	    }
tomwalters@268: 	} else {
tomwalters@268: 	    /* target is a character in range 0xFFFF - 0x10FFFF. */
tomwalters@268: 	    if (target + 1 >= targetEnd) {
tomwalters@268: 		--source; /* Back up source pointer! */
tomwalters@268: 		result = targetExhausted; break;
tomwalters@268: 	    }
tomwalters@268: 	    ch -= halfBase;
tomwalters@268: 	    *target++ = (UTF16)((ch >> halfShift) + UNI_SUR_HIGH_START);
tomwalters@268: 	    *target++ = (UTF16)((ch & halfMask) + UNI_SUR_LOW_START);
tomwalters@268: 	}
tomwalters@268:     }
tomwalters@268:     *sourceStart = source;
tomwalters@268:     *targetStart = target;
tomwalters@268:     return result;
tomwalters@268: }
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: ConversionResult ConvertUTF16toUTF32 (
tomwalters@268: 	const UTF16** sourceStart, const UTF16* sourceEnd, 
tomwalters@268: 	UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags) {
tomwalters@268:     ConversionResult result = conversionOK;
tomwalters@268:     const UTF16* source = *sourceStart;
tomwalters@268:     UTF32* target = *targetStart;
tomwalters@268:     UTF32 ch, ch2;
tomwalters@268:     while (source < sourceEnd) {
tomwalters@268: 	const UTF16* oldSource = source; /*  In case we have to back up because of target overflow. */
tomwalters@268: 	ch = *source++;
tomwalters@268: 	/* If we have a surrogate pair, convert to UTF32 first. */
tomwalters@268: 	if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_HIGH_END) {
tomwalters@268: 	    /* If the 16 bits following the high surrogate are in the source buffer... */
tomwalters@268: 	    if (source < sourceEnd) {
tomwalters@268: 		ch2 = *source;
tomwalters@268: 		/* If it's a low surrogate, convert to UTF32. */
tomwalters@268: 		if (ch2 >= UNI_SUR_LOW_START && ch2 <= UNI_SUR_LOW_END) {
tomwalters@268: 		    ch = ((ch - UNI_SUR_HIGH_START) << halfShift)
tomwalters@268: 			+ (ch2 - UNI_SUR_LOW_START) + halfBase;
tomwalters@268: 		    ++source;
tomwalters@268: 		} else if (flags == strictConversion) { /* it's an unpaired high surrogate */
tomwalters@268: 		    --source; /* return to the illegal value itself */
tomwalters@268: 		    result = sourceIllegal;
tomwalters@268: 		    break;
tomwalters@268: 		}
tomwalters@268: 	    } else { /* We don't have the 16 bits following the high surrogate. */
tomwalters@268: 		--source; /* return to the high surrogate */
tomwalters@268: 		result = sourceExhausted;
tomwalters@268: 		break;
tomwalters@268: 	    }
tomwalters@268: 	} else if (flags == strictConversion) {
tomwalters@268: 	    /* UTF-16 surrogate values are illegal in UTF-32 */
tomwalters@268: 	    if (ch >= UNI_SUR_LOW_START && ch <= UNI_SUR_LOW_END) {
tomwalters@268: 		--source; /* return to the illegal value itself */
tomwalters@268: 		result = sourceIllegal;
tomwalters@268: 		break;
tomwalters@268: 	    }
tomwalters@268: 	}
tomwalters@268: 	if (target >= targetEnd) {
tomwalters@268: 	    source = oldSource; /* Back up source pointer! */
tomwalters@268: 	    result = targetExhausted; break;
tomwalters@268: 	}
tomwalters@268: 	*target++ = ch;
tomwalters@268:     }
tomwalters@268:     *sourceStart = source;
tomwalters@268:     *targetStart = target;
tomwalters@268: #ifdef CVTUTF_DEBUG
tomwalters@268: if (result == sourceIllegal) {
tomwalters@268:     fprintf(stderr, "ConvertUTF16toUTF32 illegal seq 0x%04x,%04x\n", ch, ch2);
tomwalters@268:     fflush(stderr);
tomwalters@268: }
tomwalters@268: #endif
tomwalters@268:     return result;
tomwalters@268: }
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: /*
tomwalters@268:  * Index into the table below with the first byte of a UTF-8 sequence to
tomwalters@268:  * get the number of trailing bytes that are supposed to follow it.
tomwalters@268:  * Note that *legal* UTF-8 values can't have 4 or 5-bytes. The table is
tomwalters@268:  * left as-is for anyone who may want to do such conversion, which was
tomwalters@268:  * allowed in earlier algorithms.
tomwalters@268:  */
tomwalters@268: static const char trailingBytesForUTF8[256] = {
tomwalters@268:     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,
tomwalters@268:     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,
tomwalters@268:     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,
tomwalters@268:     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,
tomwalters@268:     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,
tomwalters@268:     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,
tomwalters@268:     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,
tomwalters@268:     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
tomwalters@268: };
tomwalters@268: 
tomwalters@268: /*
tomwalters@268:  * Magic values subtracted from a buffer value during UTF8 conversion.
tomwalters@268:  * This table contains as many values as there might be trailing bytes
tomwalters@268:  * in a UTF-8 sequence.
tomwalters@268:  */
tomwalters@268: static const UTF32 offsetsFromUTF8[6] = { 0x00000000UL, 0x00003080UL, 0x000E2080UL, 
tomwalters@268: 		     0x03C82080UL, 0xFA082080UL, 0x82082080UL };
tomwalters@268: 
tomwalters@268: /*
tomwalters@268:  * Once the bits are split out into bytes of UTF-8, this is a mask OR-ed
tomwalters@268:  * into the first byte, depending on how many bytes follow.  There are
tomwalters@268:  * as many entries in this table as there are UTF-8 sequence types.
tomwalters@268:  * (I.e., one byte sequence, two byte... etc.). Remember that sequencs
tomwalters@268:  * for *legal* UTF-8 will be 4 or fewer bytes total.
tomwalters@268:  */
tomwalters@268: static const UTF8 firstByteMark[7] = { 0x00, 0x00, 0xC0, 0xE0, 0xF0, 0xF8, 0xFC };
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: /* The interface converts a whole buffer to avoid function-call overhead.
tomwalters@268:  * Constants have been gathered. Loops & conditionals have been removed as
tomwalters@268:  * much as possible for efficiency, in favor of drop-through switches.
tomwalters@268:  * (See "Note A" at the bottom of the file for equivalent code.)
tomwalters@268:  * If your compiler supports it, the "isLegalUTF8" call can be turned
tomwalters@268:  * into an inline function.
tomwalters@268:  */
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: ConversionResult ConvertUTF16toUTF8 (
tomwalters@268: 	const UTF16** sourceStart, const UTF16* sourceEnd, 
tomwalters@268: 	UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags) {
tomwalters@268:     ConversionResult result = conversionOK;
tomwalters@268:     const UTF16* source = *sourceStart;
tomwalters@268:     UTF8* target = *targetStart;
tomwalters@268:     while (source < sourceEnd) {
tomwalters@268: 	UTF32 ch;
tomwalters@268: 	unsigned short bytesToWrite = 0;
tomwalters@268: 	const UTF32 byteMask = 0xBF;
tomwalters@268: 	const UTF32 byteMark = 0x80; 
tomwalters@268: 	const UTF16* oldSource = source; /* In case we have to back up because of target overflow. */
tomwalters@268: 	ch = *source++;
tomwalters@268: 	/* If we have a surrogate pair, convert to UTF32 first. */
tomwalters@268: 	if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_HIGH_END) {
tomwalters@268: 	    /* If the 16 bits following the high surrogate are in the source buffer... */
tomwalters@268: 	    if (source < sourceEnd) {
tomwalters@268: 		UTF32 ch2 = *source;
tomwalters@268: 		/* If it's a low surrogate, convert to UTF32. */
tomwalters@268: 		if (ch2 >= UNI_SUR_LOW_START && ch2 <= UNI_SUR_LOW_END) {
tomwalters@268: 		    ch = ((ch - UNI_SUR_HIGH_START) << halfShift)
tomwalters@268: 			+ (ch2 - UNI_SUR_LOW_START) + halfBase;
tomwalters@268: 		    ++source;
tomwalters@268: 		} else if (flags == strictConversion) { /* it's an unpaired high surrogate */
tomwalters@268: 		    --source; /* return to the illegal value itself */
tomwalters@268: 		    result = sourceIllegal;
tomwalters@268: 		    break;
tomwalters@268: 		}
tomwalters@268: 	    } else { /* We don't have the 16 bits following the high surrogate. */
tomwalters@268: 		--source; /* return to the high surrogate */
tomwalters@268: 		result = sourceExhausted;
tomwalters@268: 		break;
tomwalters@268: 	    }
tomwalters@268: 	} else if (flags == strictConversion) {
tomwalters@268: 	    /* UTF-16 surrogate values are illegal in UTF-32 */
tomwalters@268: 	    if (ch >= UNI_SUR_LOW_START && ch <= UNI_SUR_LOW_END) {
tomwalters@268: 		--source; /* return to the illegal value itself */
tomwalters@268: 		result = sourceIllegal;
tomwalters@268: 		break;
tomwalters@268: 	    }
tomwalters@268: 	}
tomwalters@268: 	/* Figure out how many bytes the result will require */
tomwalters@268: 	if (ch < (UTF32)0x80) {	     bytesToWrite = 1;
tomwalters@268: 	} else if (ch < (UTF32)0x800) {     bytesToWrite = 2;
tomwalters@268: 	} else if (ch < (UTF32)0x10000) {   bytesToWrite = 3;
tomwalters@268: 	} else if (ch < (UTF32)0x110000) {  bytesToWrite = 4;
tomwalters@268: 	} else {			    bytesToWrite = 3;
tomwalters@268: 					    ch = UNI_REPLACEMENT_CHAR;
tomwalters@268: 	}
tomwalters@268: 
tomwalters@268: 	target += bytesToWrite;
tomwalters@268: 	if (target > targetEnd) {
tomwalters@268: 	    source = oldSource; /* Back up source pointer! */
tomwalters@268: 	    target -= bytesToWrite; result = targetExhausted; break;
tomwalters@268: 	}
tomwalters@268: 	switch (bytesToWrite) { /* note: everything falls through. */
tomwalters@268: 	    case 4: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
tomwalters@268: 	    case 3: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
tomwalters@268: 	    case 2: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
tomwalters@268: 	    case 1: *--target =  (UTF8)(ch | firstByteMark[bytesToWrite]);
tomwalters@268: 	}
tomwalters@268: 	target += bytesToWrite;
tomwalters@268:     }
tomwalters@268:     *sourceStart = source;
tomwalters@268:     *targetStart = target;
tomwalters@268:     return result;
tomwalters@268: }
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: /*
tomwalters@268:  * Utility routine to tell whether a sequence of bytes is legal UTF-8.
tomwalters@268:  * This must be called with the length pre-determined by the first byte.
tomwalters@268:  * If not calling this from ConvertUTF8to*, then the length can be set by:
tomwalters@268:  *  length = trailingBytesForUTF8[*source]+1;
tomwalters@268:  * and the sequence is illegal right away if there aren't that many bytes
tomwalters@268:  * available.
tomwalters@268:  * If presented with a length > 4, this returns false.  The Unicode
tomwalters@268:  * definition of UTF-8 goes up to 4-byte sequences.
tomwalters@268:  */
tomwalters@268: 
tomwalters@268: static Boolean isLegalUTF8(const UTF8 *source, int length) {
tomwalters@268:     UTF8 a;
tomwalters@268:     const UTF8 *srcptr = source+length;
tomwalters@268:     switch (length) {
tomwalters@268:     default: return false;
tomwalters@268: 	/* Everything else falls through when "true"... */
tomwalters@268:     case 4: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
tomwalters@268:     case 3: if ((a = (*--srcptr)) < 0x80 || a > 0xBF) return false;
tomwalters@268:     case 2: if ((a = (*--srcptr)) > 0xBF) return false;
tomwalters@268: 
tomwalters@268: 	switch (*source) {
tomwalters@268: 	    /* no fall-through in this inner switch */
tomwalters@268: 	    case 0xE0: if (a < 0xA0) return false; break;
tomwalters@268: 	    case 0xED: if (a > 0x9F) return false; break;
tomwalters@268: 	    case 0xF0: if (a < 0x90) return false; break;
tomwalters@268: 	    case 0xF4: if (a > 0x8F) return false; break;
tomwalters@268: 	    default:   if (a < 0x80) return false;
tomwalters@268: 	}
tomwalters@268: 
tomwalters@268:     case 1: if (*source >= 0x80 && *source < 0xC2) return false;
tomwalters@268:     }
tomwalters@268:     if (*source > 0xF4) return false;
tomwalters@268:     return true;
tomwalters@268: }
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: /*
tomwalters@268:  * Exported function to return whether a UTF-8 sequence is legal or not.
tomwalters@268:  * This is not used here; it's just exported.
tomwalters@268:  */
tomwalters@268: Boolean isLegalUTF8Sequence(const UTF8 *source, const UTF8 *sourceEnd) {
tomwalters@268:     int length = trailingBytesForUTF8[*source]+1;
tomwalters@268:     if (source+length > sourceEnd) {
tomwalters@268: 	return false;
tomwalters@268:     }
tomwalters@268:     return isLegalUTF8(source, length);
tomwalters@268: }
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: ConversionResult ConvertUTF8toUTF16 (
tomwalters@268: 	const UTF8** sourceStart, const UTF8* sourceEnd, 
tomwalters@268: 	UTF16** targetStart, UTF16* targetEnd, ConversionFlags flags) {
tomwalters@268:     ConversionResult result = conversionOK;
tomwalters@268:     const UTF8* source = *sourceStart;
tomwalters@268:     UTF16* target = *targetStart;
tomwalters@268:     while (source < sourceEnd) {
tomwalters@268: 	UTF32 ch = 0;
tomwalters@268: 	unsigned short extraBytesToRead = trailingBytesForUTF8[*source];
tomwalters@268: 	if (source + extraBytesToRead >= sourceEnd) {
tomwalters@268: 	    result = sourceExhausted; break;
tomwalters@268: 	}
tomwalters@268: 	/* Do this check whether lenient or strict */
tomwalters@268: 	if (! isLegalUTF8(source, extraBytesToRead+1)) {
tomwalters@268: 	    result = sourceIllegal;
tomwalters@268: 	    break;
tomwalters@268: 	}
tomwalters@268: 	/*
tomwalters@268: 	 * The cases all fall through. See "Note A" below.
tomwalters@268: 	 */
tomwalters@268: 	switch (extraBytesToRead) {
tomwalters@268: 	    case 5: ch += *source++; ch <<= 6; /* remember, illegal UTF-8 */
tomwalters@268: 	    case 4: ch += *source++; ch <<= 6; /* remember, illegal UTF-8 */
tomwalters@268: 	    case 3: ch += *source++; ch <<= 6;
tomwalters@268: 	    case 2: ch += *source++; ch <<= 6;
tomwalters@268: 	    case 1: ch += *source++; ch <<= 6;
tomwalters@268: 	    case 0: ch += *source++;
tomwalters@268: 	}
tomwalters@268: 	ch -= offsetsFromUTF8[extraBytesToRead];
tomwalters@268: 
tomwalters@268: 	if (target >= targetEnd) {
tomwalters@268: 	    source -= (extraBytesToRead+1); /* Back up source pointer! */
tomwalters@268: 	    result = targetExhausted; break;
tomwalters@268: 	}
tomwalters@268: 	if (ch <= UNI_MAX_BMP) { /* Target is a character <= 0xFFFF */
tomwalters@268: 	    /* UTF-16 surrogate values are illegal in UTF-32 */
tomwalters@268: 	    if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) {
tomwalters@268: 		if (flags == strictConversion) {
tomwalters@268: 		    source -= (extraBytesToRead+1); /* return to the illegal value itself */
tomwalters@268: 		    result = sourceIllegal;
tomwalters@268: 		    break;
tomwalters@268: 		} else {
tomwalters@268: 		    *target++ = UNI_REPLACEMENT_CHAR;
tomwalters@268: 		}
tomwalters@268: 	    } else {
tomwalters@268: 		*target++ = (UTF16)ch; /* normal case */
tomwalters@268: 	    }
tomwalters@268: 	} else if (ch > UNI_MAX_UTF16) {
tomwalters@268: 	    if (flags == strictConversion) {
tomwalters@268: 		result = sourceIllegal;
tomwalters@268: 		source -= (extraBytesToRead+1); /* return to the start */
tomwalters@268: 		break; /* Bail out; shouldn't continue */
tomwalters@268: 	    } else {
tomwalters@268: 		*target++ = UNI_REPLACEMENT_CHAR;
tomwalters@268: 	    }
tomwalters@268: 	} else {
tomwalters@268: 	    /* target is a character in range 0xFFFF - 0x10FFFF. */
tomwalters@268: 	    if (target + 1 >= targetEnd) {
tomwalters@268: 		source -= (extraBytesToRead+1); /* Back up source pointer! */
tomwalters@268: 		result = targetExhausted; break;
tomwalters@268: 	    }
tomwalters@268: 	    ch -= halfBase;
tomwalters@268: 	    *target++ = (UTF16)((ch >> halfShift) + UNI_SUR_HIGH_START);
tomwalters@268: 	    *target++ = (UTF16)((ch & halfMask) + UNI_SUR_LOW_START);
tomwalters@268: 	}
tomwalters@268:     }
tomwalters@268:     *sourceStart = source;
tomwalters@268:     *targetStart = target;
tomwalters@268:     return result;
tomwalters@268: }
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: ConversionResult ConvertUTF32toUTF8 (
tomwalters@268: 	const UTF32** sourceStart, const UTF32* sourceEnd, 
tomwalters@268: 	UTF8** targetStart, UTF8* targetEnd, ConversionFlags flags) {
tomwalters@268:     ConversionResult result = conversionOK;
tomwalters@268:     const UTF32* source = *sourceStart;
tomwalters@268:     UTF8* target = *targetStart;
tomwalters@268:     while (source < sourceEnd) {
tomwalters@268: 	UTF32 ch;
tomwalters@268: 	unsigned short bytesToWrite = 0;
tomwalters@268: 	const UTF32 byteMask = 0xBF;
tomwalters@268: 	const UTF32 byteMark = 0x80; 
tomwalters@268: 	ch = *source++;
tomwalters@268: 	if (flags == strictConversion ) {
tomwalters@268: 	    /* UTF-16 surrogate values are illegal in UTF-32 */
tomwalters@268: 	    if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) {
tomwalters@268: 		--source; /* return to the illegal value itself */
tomwalters@268: 		result = sourceIllegal;
tomwalters@268: 		break;
tomwalters@268: 	    }
tomwalters@268: 	}
tomwalters@268: 	/*
tomwalters@268: 	 * Figure out how many bytes the result will require. Turn any
tomwalters@268: 	 * illegally large UTF32 things (> Plane 17) into replacement chars.
tomwalters@268: 	 */
tomwalters@268: 	if (ch < (UTF32)0x80) {	     bytesToWrite = 1;
tomwalters@268: 	} else if (ch < (UTF32)0x800) {     bytesToWrite = 2;
tomwalters@268: 	} else if (ch < (UTF32)0x10000) {   bytesToWrite = 3;
tomwalters@268: 	} else if (ch <= UNI_MAX_LEGAL_UTF32) {  bytesToWrite = 4;
tomwalters@268: 	} else {			    bytesToWrite = 3;
tomwalters@268: 					    ch = UNI_REPLACEMENT_CHAR;
tomwalters@268: 					    result = sourceIllegal;
tomwalters@268: 	}
tomwalters@268: 	
tomwalters@268: 	target += bytesToWrite;
tomwalters@268: 	if (target > targetEnd) {
tomwalters@268: 	    --source; /* Back up source pointer! */
tomwalters@268: 	    target -= bytesToWrite; result = targetExhausted; break;
tomwalters@268: 	}
tomwalters@268: 	switch (bytesToWrite) { /* note: everything falls through. */
tomwalters@268: 	    case 4: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
tomwalters@268: 	    case 3: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
tomwalters@268: 	    case 2: *--target = (UTF8)((ch | byteMark) & byteMask); ch >>= 6;
tomwalters@268: 	    case 1: *--target = (UTF8) (ch | firstByteMark[bytesToWrite]);
tomwalters@268: 	}
tomwalters@268: 	target += bytesToWrite;
tomwalters@268:     }
tomwalters@268:     *sourceStart = source;
tomwalters@268:     *targetStart = target;
tomwalters@268:     return result;
tomwalters@268: }
tomwalters@268: 
tomwalters@268: /* --------------------------------------------------------------------- */
tomwalters@268: 
tomwalters@268: ConversionResult ConvertUTF8toUTF32 (
tomwalters@268: 	const UTF8** sourceStart, const UTF8* sourceEnd, 
tomwalters@268: 	UTF32** targetStart, UTF32* targetEnd, ConversionFlags flags) {
tomwalters@268:     ConversionResult result = conversionOK;
tomwalters@268:     const UTF8* source = *sourceStart;
tomwalters@268:     UTF32* target = *targetStart;
tomwalters@268:     while (source < sourceEnd) {
tomwalters@268: 	UTF32 ch = 0;
tomwalters@268: 	unsigned short extraBytesToRead = trailingBytesForUTF8[*source];
tomwalters@268: 	if (source + extraBytesToRead >= sourceEnd) {
tomwalters@268: 	    result = sourceExhausted; break;
tomwalters@268: 	}
tomwalters@268: 	/* Do this check whether lenient or strict */
tomwalters@268: 	if (! isLegalUTF8(source, extraBytesToRead+1)) {
tomwalters@268: 	    result = sourceIllegal;
tomwalters@268: 	    break;
tomwalters@268: 	}
tomwalters@268: 	/*
tomwalters@268: 	 * The cases all fall through. See "Note A" below.
tomwalters@268: 	 */
tomwalters@268: 	switch (extraBytesToRead) {
tomwalters@268: 	    case 5: ch += *source++; ch <<= 6;
tomwalters@268: 	    case 4: ch += *source++; ch <<= 6;
tomwalters@268: 	    case 3: ch += *source++; ch <<= 6;
tomwalters@268: 	    case 2: ch += *source++; ch <<= 6;
tomwalters@268: 	    case 1: ch += *source++; ch <<= 6;
tomwalters@268: 	    case 0: ch += *source++;
tomwalters@268: 	}
tomwalters@268: 	ch -= offsetsFromUTF8[extraBytesToRead];
tomwalters@268: 
tomwalters@268: 	if (target >= targetEnd) {
tomwalters@268: 	    source -= (extraBytesToRead+1); /* Back up the source pointer! */
tomwalters@268: 	    result = targetExhausted; break;
tomwalters@268: 	}
tomwalters@268: 	if (ch <= UNI_MAX_LEGAL_UTF32) {
tomwalters@268: 	    /*
tomwalters@268: 	     * UTF-16 surrogate values are illegal in UTF-32, and anything
tomwalters@268: 	     * over Plane 17 (> 0x10FFFF) is illegal.
tomwalters@268: 	     */
tomwalters@268: 	    if (ch >= UNI_SUR_HIGH_START && ch <= UNI_SUR_LOW_END) {
tomwalters@268: 		if (flags == strictConversion) {
tomwalters@268: 		    source -= (extraBytesToRead+1); /* return to the illegal value itself */
tomwalters@268: 		    result = sourceIllegal;
tomwalters@268: 		    break;
tomwalters@268: 		} else {
tomwalters@268: 		    *target++ = UNI_REPLACEMENT_CHAR;
tomwalters@268: 		}
tomwalters@268: 	    } else {
tomwalters@268: 		*target++ = ch;
tomwalters@268: 	    }
tomwalters@268: 	} else { /* i.e., ch > UNI_MAX_LEGAL_UTF32 */
tomwalters@268: 	    result = sourceIllegal;
tomwalters@268: 	    *target++ = UNI_REPLACEMENT_CHAR;
tomwalters@268: 	}
tomwalters@268:     }
tomwalters@268:     *sourceStart = source;
tomwalters@268:     *targetStart = target;
tomwalters@268:     return result;
tomwalters@268: }
tomwalters@268: 
tomwalters@268: /* ---------------------------------------------------------------------
tomwalters@268: 
tomwalters@268:     Note A.
tomwalters@268:     The fall-through switches in UTF-8 reading code save a
tomwalters@268:     temp variable, some decrements & conditionals.  The switches
tomwalters@268:     are equivalent to the following loop:
tomwalters@268: 	{
tomwalters@268: 	    int tmpBytesToRead = extraBytesToRead+1;
tomwalters@268: 	    do {
tomwalters@268: 		ch += *source++;
tomwalters@268: 		--tmpBytesToRead;
tomwalters@268: 		if (tmpBytesToRead) ch <<= 6;
tomwalters@268: 	    } while (tmpBytesToRead > 0);
tomwalters@268: 	}
tomwalters@268:     In UTF-8 writing code, the switches on "bytesToWrite" are
tomwalters@268:     similarly unrolled loops.
tomwalters@268: 
tomwalters@268:    --------------------------------------------------------------------- */