michael@0: /*
michael@0: ******************************************************************************
michael@0: *
michael@0: *   Copyright (C) 1998-2012, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *
michael@0: ******************************************************************************
michael@0: *
michael@0: * File ustring.cpp
michael@0: *
michael@0: * Modification History:
michael@0: *
michael@0: *   Date        Name        Description
michael@0: *   12/07/98    bertrand    Creation.
michael@0: ******************************************************************************
michael@0: */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/putil.h"
michael@0: #include "unicode/ustring.h"
michael@0: #include "unicode/utf16.h"
michael@0: #include "cstring.h"
michael@0: #include "cwchar.h"
michael@0: #include "cmemory.h"
michael@0: #include "ustr_imp.h"
michael@0: 
michael@0: /* ANSI string.h - style functions ------------------------------------------ */
michael@0: 
michael@0: /* U+ffff is the highest BMP code point, the highest one that fits into a 16-bit UChar */
michael@0: #define U_BMP_MAX 0xffff
michael@0: 
michael@0: /* Forward binary string search functions ----------------------------------- */
michael@0: 
michael@0: /*
michael@0:  * Test if a substring match inside a string is at code point boundaries.
michael@0:  * All pointers refer to the same buffer.
michael@0:  * The limit pointer may be NULL, all others must be real pointers.
michael@0:  */
michael@0: static inline UBool
michael@0: isMatchAtCPBoundary(const UChar *start, const UChar *match, const UChar *matchLimit, const UChar *limit) {
michael@0:     if(U16_IS_TRAIL(*match) && start!=match && U16_IS_LEAD(*(match-1))) {
michael@0:         /* the leading edge of the match is in the middle of a surrogate pair */
michael@0:         return FALSE;
michael@0:     }
michael@0:     if(U16_IS_LEAD(*(matchLimit-1)) && match!=limit && U16_IS_TRAIL(*matchLimit)) {
michael@0:         /* the trailing edge of the match is in the middle of a surrogate pair */
michael@0:         return FALSE;
michael@0:     }
michael@0:     return TRUE;
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_strFindFirst(const UChar *s, int32_t length,
michael@0:                const UChar *sub, int32_t subLength) {
michael@0:     const UChar *start, *p, *q, *subLimit;
michael@0:     UChar c, cs, cq;
michael@0: 
michael@0:     if(sub==NULL || subLength<-1) {
michael@0:         return (UChar *)s;
michael@0:     }
michael@0:     if(s==NULL || length<-1) {
michael@0:         return NULL;
michael@0:     }
michael@0: 
michael@0:     start=s;
michael@0: 
michael@0:     if(length<0 && subLength<0) {
michael@0:         /* both strings are NUL-terminated */
michael@0:         if((cs=*sub++)==0) {
michael@0:             return (UChar *)s;
michael@0:         }
michael@0:         if(*sub==0 && !U16_IS_SURROGATE(cs)) {
michael@0:             /* the substring consists of a single, non-surrogate BMP code point */
michael@0:             return u_strchr(s, cs);
michael@0:         }
michael@0: 
michael@0:         while((c=*s++)!=0) {
michael@0:             if(c==cs) {
michael@0:                 /* found first substring UChar, compare rest */
michael@0:                 p=s;
michael@0:                 q=sub;
michael@0:                 for(;;) {
michael@0:                     if((cq=*q)==0) {
michael@0:                         if(isMatchAtCPBoundary(start, s-1, p, NULL)) {
michael@0:                             return (UChar *)(s-1); /* well-formed match */
michael@0:                         } else {
michael@0:                             break; /* no match because surrogate pair is split */
michael@0:                         }
michael@0:                     }
michael@0:                     if((c=*p)==0) {
michael@0:                         return NULL; /* no match, and none possible after s */
michael@0:                     }
michael@0:                     if(c!=cq) {
michael@0:                         break; /* no match */
michael@0:                     }
michael@0:                     ++p;
michael@0:                     ++q;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         /* not found */
michael@0:         return NULL;
michael@0:     }
michael@0: 
michael@0:     if(subLength<0) {
michael@0:         subLength=u_strlen(sub);
michael@0:     }
michael@0:     if(subLength==0) {
michael@0:         return (UChar *)s;
michael@0:     }
michael@0: 
michael@0:     /* get sub[0] to search for it fast */
michael@0:     cs=*sub++;
michael@0:     --subLength;
michael@0:     subLimit=sub+subLength;
michael@0: 
michael@0:     if(subLength==0 && !U16_IS_SURROGATE(cs)) {
michael@0:         /* the substring consists of a single, non-surrogate BMP code point */
michael@0:         return length<0 ? u_strchr(s, cs) : u_memchr(s, cs, length);
michael@0:     }
michael@0: 
michael@0:     if(length<0) {
michael@0:         /* s is NUL-terminated */
michael@0:         while((c=*s++)!=0) {
michael@0:             if(c==cs) {
michael@0:                 /* found first substring UChar, compare rest */
michael@0:                 p=s;
michael@0:                 q=sub;
michael@0:                 for(;;) {
michael@0:                     if(q==subLimit) {
michael@0:                         if(isMatchAtCPBoundary(start, s-1, p, NULL)) {
michael@0:                             return (UChar *)(s-1); /* well-formed match */
michael@0:                         } else {
michael@0:                             break; /* no match because surrogate pair is split */
michael@0:                         }
michael@0:                     }
michael@0:                     if((c=*p)==0) {
michael@0:                         return NULL; /* no match, and none possible after s */
michael@0:                     }
michael@0:                     if(c!=*q) {
michael@0:                         break; /* no match */
michael@0:                     }
michael@0:                     ++p;
michael@0:                     ++q;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:     } else {
michael@0:         const UChar *limit, *preLimit;
michael@0: 
michael@0:         /* subLength was decremented above */
michael@0:         if(length<=subLength) {
michael@0:             return NULL; /* s is shorter than sub */
michael@0:         }
michael@0: 
michael@0:         limit=s+length;
michael@0: 
michael@0:         /* the substring must start before preLimit */
michael@0:         preLimit=limit-subLength;
michael@0: 
michael@0:         while(s!=preLimit) {
michael@0:             c=*s++;
michael@0:             if(c==cs) {
michael@0:                 /* found first substring UChar, compare rest */
michael@0:                 p=s;
michael@0:                 q=sub;
michael@0:                 for(;;) {
michael@0:                     if(q==subLimit) {
michael@0:                         if(isMatchAtCPBoundary(start, s-1, p, limit)) {
michael@0:                             return (UChar *)(s-1); /* well-formed match */
michael@0:                         } else {
michael@0:                             break; /* no match because surrogate pair is split */
michael@0:                         }
michael@0:                     }
michael@0:                     if(*p!=*q) {
michael@0:                         break; /* no match */
michael@0:                     }
michael@0:                     ++p;
michael@0:                     ++q;
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* not found */
michael@0:     return NULL;
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_strstr(const UChar *s, const UChar *substring) {
michael@0:     return u_strFindFirst(s, -1, substring, -1);
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_strchr(const UChar *s, UChar c) {
michael@0:     if(U16_IS_SURROGATE(c)) {
michael@0:         /* make sure to not find half of a surrogate pair */
michael@0:         return u_strFindFirst(s, -1, &c, 1);
michael@0:     } else {
michael@0:         UChar cs;
michael@0: 
michael@0:         /* trivial search for a BMP code point */
michael@0:         for(;;) {
michael@0:             if((cs=*s)==c) {
michael@0:                 return (UChar *)s;
michael@0:             }
michael@0:             if(cs==0) {
michael@0:                 return NULL;
michael@0:             }
michael@0:             ++s;
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_strchr32(const UChar *s, UChar32 c) {
michael@0:     if((uint32_t)c<=U_BMP_MAX) {
michael@0:         /* find BMP code point */
michael@0:         return u_strchr(s, (UChar)c);
michael@0:     } else if((uint32_t)c<=UCHAR_MAX_VALUE) {
michael@0:         /* find supplementary code point as surrogate pair */
michael@0:         UChar cs, lead=U16_LEAD(c), trail=U16_TRAIL(c);
michael@0: 
michael@0:         while((cs=*s++)!=0) {
michael@0:             if(cs==lead && *s==trail) {
michael@0:                 return (UChar *)(s-1);
michael@0:             }
michael@0:         }
michael@0:         return NULL;
michael@0:     } else {
michael@0:         /* not a Unicode code point, not findable */
michael@0:         return NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_memchr(const UChar *s, UChar c, int32_t count) {
michael@0:     if(count<=0) {
michael@0:         return NULL; /* no string */
michael@0:     } else if(U16_IS_SURROGATE(c)) {
michael@0:         /* make sure to not find half of a surrogate pair */
michael@0:         return u_strFindFirst(s, count, &c, 1);
michael@0:     } else {
michael@0:         /* trivial search for a BMP code point */
michael@0:         const UChar *limit=s+count;
michael@0:         do {
michael@0:             if(*s==c) {
michael@0:                 return (UChar *)s;
michael@0:             }
michael@0:         } while(++s!=limit);
michael@0:         return NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_memchr32(const UChar *s, UChar32 c, int32_t count) {
michael@0:     if((uint32_t)c<=U_BMP_MAX) {
michael@0:         /* find BMP code point */
michael@0:         return u_memchr(s, (UChar)c, count);
michael@0:     } else if(count<2) {
michael@0:         /* too short for a surrogate pair */
michael@0:         return NULL;
michael@0:     } else if((uint32_t)c<=UCHAR_MAX_VALUE) {
michael@0:         /* find supplementary code point as surrogate pair */
michael@0:         const UChar *limit=s+count-1; /* -1 so that we do not need a separate check for the trail unit */
michael@0:         UChar lead=U16_LEAD(c), trail=U16_TRAIL(c);
michael@0: 
michael@0:         do {
michael@0:             if(*s==lead && *(s+1)==trail) {
michael@0:                 return (UChar *)s;
michael@0:             }
michael@0:         } while(++s!=limit);
michael@0:         return NULL;
michael@0:     } else {
michael@0:         /* not a Unicode code point, not findable */
michael@0:         return NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* Backward binary string search functions ---------------------------------- */
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_strFindLast(const UChar *s, int32_t length,
michael@0:               const UChar *sub, int32_t subLength) {
michael@0:     const UChar *start, *limit, *p, *q, *subLimit;
michael@0:     UChar c, cs;
michael@0: 
michael@0:     if(sub==NULL || subLength<-1) {
michael@0:         return (UChar *)s;
michael@0:     }
michael@0:     if(s==NULL || length<-1) {
michael@0:         return NULL;
michael@0:     }
michael@0: 
michael@0:     /*
michael@0:      * This implementation is more lazy than the one for u_strFindFirst():
michael@0:      * There is no special search code for NUL-terminated strings.
michael@0:      * It does not seem to be worth it for searching substrings to
michael@0:      * search forward and find all matches like in u_strrchr() and similar.
michael@0:      * Therefore, we simply get both string lengths and search backward.
michael@0:      *
michael@0:      * markus 2002oct23
michael@0:      */
michael@0: 
michael@0:     if(subLength<0) {
michael@0:         subLength=u_strlen(sub);
michael@0:     }
michael@0:     if(subLength==0) {
michael@0:         return (UChar *)s;
michael@0:     }
michael@0: 
michael@0:     /* get sub[subLength-1] to search for it fast */
michael@0:     subLimit=sub+subLength;
michael@0:     cs=*(--subLimit);
michael@0:     --subLength;
michael@0: 
michael@0:     if(subLength==0 && !U16_IS_SURROGATE(cs)) {
michael@0:         /* the substring consists of a single, non-surrogate BMP code point */
michael@0:         return length<0 ? u_strrchr(s, cs) : u_memrchr(s, cs, length);
michael@0:     }
michael@0: 
michael@0:     if(length<0) {
michael@0:         length=u_strlen(s);
michael@0:     }
michael@0: 
michael@0:     /* subLength was decremented above */
michael@0:     if(length<=subLength) {
michael@0:         return NULL; /* s is shorter than sub */
michael@0:     }
michael@0: 
michael@0:     start=s;
michael@0:     limit=s+length;
michael@0: 
michael@0:     /* the substring must start no later than s+subLength */
michael@0:     s+=subLength;
michael@0: 
michael@0:     while(s!=limit) {
michael@0:         c=*(--limit);
michael@0:         if(c==cs) {
michael@0:             /* found last substring UChar, compare rest */
michael@0:             p=limit;
michael@0:             q=subLimit;
michael@0:             for(;;) {
michael@0:                 if(q==sub) {
michael@0:                     if(isMatchAtCPBoundary(start, p, limit+1, start+length)) {
michael@0:                         return (UChar *)p; /* well-formed match */
michael@0:                     } else {
michael@0:                         break; /* no match because surrogate pair is split */
michael@0:                     }
michael@0:                 }
michael@0:                 if(*(--p)!=*(--q)) {
michael@0:                     break; /* no match */
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* not found */
michael@0:     return NULL;
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_strrstr(const UChar *s, const UChar *substring) {
michael@0:     return u_strFindLast(s, -1, substring, -1);
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_strrchr(const UChar *s, UChar c) {
michael@0:     if(U16_IS_SURROGATE(c)) {
michael@0:         /* make sure to not find half of a surrogate pair */
michael@0:         return u_strFindLast(s, -1, &c, 1);
michael@0:     } else {
michael@0:         const UChar *result=NULL;
michael@0:         UChar cs;
michael@0: 
michael@0:         /* trivial search for a BMP code point */
michael@0:         for(;;) {
michael@0:             if((cs=*s)==c) {
michael@0:                 result=s;
michael@0:             }
michael@0:             if(cs==0) {
michael@0:                 return (UChar *)result;
michael@0:             }
michael@0:             ++s;
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_strrchr32(const UChar *s, UChar32 c) {
michael@0:     if((uint32_t)c<=U_BMP_MAX) {
michael@0:         /* find BMP code point */
michael@0:         return u_strrchr(s, (UChar)c);
michael@0:     } else if((uint32_t)c<=UCHAR_MAX_VALUE) {
michael@0:         /* find supplementary code point as surrogate pair */
michael@0:         const UChar *result=NULL;
michael@0:         UChar cs, lead=U16_LEAD(c), trail=U16_TRAIL(c);
michael@0: 
michael@0:         while((cs=*s++)!=0) {
michael@0:             if(cs==lead && *s==trail) {
michael@0:                 result=s-1;
michael@0:             }
michael@0:         }
michael@0:         return (UChar *)result;
michael@0:     } else {
michael@0:         /* not a Unicode code point, not findable */
michael@0:         return NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_memrchr(const UChar *s, UChar c, int32_t count) {
michael@0:     if(count<=0) {
michael@0:         return NULL; /* no string */
michael@0:     } else if(U16_IS_SURROGATE(c)) {
michael@0:         /* make sure to not find half of a surrogate pair */
michael@0:         return u_strFindLast(s, count, &c, 1);
michael@0:     } else {
michael@0:         /* trivial search for a BMP code point */
michael@0:         const UChar *limit=s+count;
michael@0:         do {
michael@0:             if(*(--limit)==c) {
michael@0:                 return (UChar *)limit;
michael@0:             }
michael@0:         } while(s!=limit);
michael@0:         return NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_memrchr32(const UChar *s, UChar32 c, int32_t count) {
michael@0:     if((uint32_t)c<=U_BMP_MAX) {
michael@0:         /* find BMP code point */
michael@0:         return u_memrchr(s, (UChar)c, count);
michael@0:     } else if(count<2) {
michael@0:         /* too short for a surrogate pair */
michael@0:         return NULL;
michael@0:     } else if((uint32_t)c<=UCHAR_MAX_VALUE) {
michael@0:         /* find supplementary code point as surrogate pair */
michael@0:         const UChar *limit=s+count-1;
michael@0:         UChar lead=U16_LEAD(c), trail=U16_TRAIL(c);
michael@0: 
michael@0:         do {
michael@0:             if(*limit==trail && *(limit-1)==lead) {
michael@0:                 return (UChar *)(limit-1);
michael@0:             }
michael@0:         } while(s!=--limit);
michael@0:         return NULL;
michael@0:     } else {
michael@0:         /* not a Unicode code point, not findable */
michael@0:         return NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* Tokenization functions --------------------------------------------------- */
michael@0: 
michael@0: /*
michael@0:  * Match each code point in a string against each code point in the matchSet.
michael@0:  * Return the index of the first string code point that
michael@0:  * is (polarity==TRUE) or is not (FALSE) contained in the matchSet.
michael@0:  * Return -(string length)-1 if there is no such code point.
michael@0:  */
michael@0: static int32_t
michael@0: _matchFromSet(const UChar *string, const UChar *matchSet, UBool polarity) {
michael@0:     int32_t matchLen, matchBMPLen, strItr, matchItr;
michael@0:     UChar32 stringCh, matchCh;
michael@0:     UChar c, c2;
michael@0: 
michael@0:     /* first part of matchSet contains only BMP code points */
michael@0:     matchBMPLen = 0;
michael@0:     while((c = matchSet[matchBMPLen]) != 0 && U16_IS_SINGLE(c)) {
michael@0:         ++matchBMPLen;
michael@0:     }
michael@0: 
michael@0:     /* second part of matchSet contains BMP and supplementary code points */
michael@0:     matchLen = matchBMPLen;
michael@0:     while(matchSet[matchLen] != 0) {
michael@0:         ++matchLen;
michael@0:     }
michael@0: 
michael@0:     for(strItr = 0; (c = string[strItr]) != 0;) {
michael@0:         ++strItr;
michael@0:         if(U16_IS_SINGLE(c)) {
michael@0:             if(polarity) {
michael@0:                 for(matchItr = 0; matchItr < matchLen; ++matchItr) {
michael@0:                     if(c == matchSet[matchItr]) {
michael@0:                         return strItr - 1; /* one matches */
michael@0:                     }
michael@0:                 }
michael@0:             } else {
michael@0:                 for(matchItr = 0; matchItr < matchLen; ++matchItr) {
michael@0:                     if(c == matchSet[matchItr]) {
michael@0:                         goto endloop;
michael@0:                     }
michael@0:                 }
michael@0:                 return strItr - 1; /* none matches */
michael@0:             }
michael@0:         } else {
michael@0:             /*
michael@0:              * No need to check for string length before U16_IS_TRAIL
michael@0:              * because c2 could at worst be the terminating NUL.
michael@0:              */
michael@0:             if(U16_IS_SURROGATE_LEAD(c) && U16_IS_TRAIL(c2 = string[strItr])) {
michael@0:                 ++strItr;
michael@0:                 stringCh = U16_GET_SUPPLEMENTARY(c, c2);
michael@0:             } else {
michael@0:                 stringCh = c; /* unpaired trail surrogate */
michael@0:             }
michael@0: 
michael@0:             if(polarity) {
michael@0:                 for(matchItr = matchBMPLen; matchItr < matchLen;) {
michael@0:                     U16_NEXT(matchSet, matchItr, matchLen, matchCh);
michael@0:                     if(stringCh == matchCh) {
michael@0:                         return strItr - U16_LENGTH(stringCh); /* one matches */
michael@0:                     }
michael@0:                 }
michael@0:             } else {
michael@0:                 for(matchItr = matchBMPLen; matchItr < matchLen;) {
michael@0:                     U16_NEXT(matchSet, matchItr, matchLen, matchCh);
michael@0:                     if(stringCh == matchCh) {
michael@0:                         goto endloop;
michael@0:                     }
michael@0:                 }
michael@0:                 return strItr - U16_LENGTH(stringCh); /* none matches */
michael@0:             }
michael@0:         }
michael@0: endloop:
michael@0:         /* wish C had continue with labels like Java... */;
michael@0:     }
michael@0: 
michael@0:     /* Didn't find it. */
michael@0:     return -strItr-1;
michael@0: }
michael@0: 
michael@0: /* Search for a codepoint in a string that matches one of the matchSet codepoints. */
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_strpbrk(const UChar *string, const UChar *matchSet)
michael@0: {
michael@0:     int32_t idx = _matchFromSet(string, matchSet, TRUE);
michael@0:     if(idx >= 0) {
michael@0:         return (UChar *)string + idx;
michael@0:     } else {
michael@0:         return NULL;
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* Search for a codepoint in a string that matches one of the matchSet codepoints. */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_strcspn(const UChar *string, const UChar *matchSet)
michael@0: {
michael@0:     int32_t idx = _matchFromSet(string, matchSet, TRUE);
michael@0:     if(idx >= 0) {
michael@0:         return idx;
michael@0:     } else {
michael@0:         return -idx - 1; /* == u_strlen(string) */
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* Search for a codepoint in a string that does not match one of the matchSet codepoints. */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_strspn(const UChar *string, const UChar *matchSet)
michael@0: {
michael@0:     int32_t idx = _matchFromSet(string, matchSet, FALSE);
michael@0:     if(idx >= 0) {
michael@0:         return idx;
michael@0:     } else {
michael@0:         return -idx - 1; /* == u_strlen(string) */
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* ----- Text manipulation functions --- */
michael@0: 
michael@0: U_CAPI UChar* U_EXPORT2
michael@0: u_strtok_r(UChar    *src, 
michael@0:      const UChar    *delim,
michael@0:            UChar   **saveState)
michael@0: {
michael@0:     UChar *tokSource;
michael@0:     UChar *nextToken;
michael@0:     uint32_t nonDelimIdx;
michael@0: 
michael@0:     /* If saveState is NULL, the user messed up. */
michael@0:     if (src != NULL) {
michael@0:         tokSource = src;
michael@0:         *saveState = src; /* Set to "src" in case there are no delimiters */
michael@0:     }
michael@0:     else if (*saveState) {
michael@0:         tokSource = *saveState;
michael@0:     }
michael@0:     else {
michael@0:         /* src == NULL && *saveState == NULL */
michael@0:         /* This shouldn't happen. We already finished tokenizing. */
michael@0:         return NULL;
michael@0:     }
michael@0: 
michael@0:     /* Skip initial delimiters */
michael@0:     nonDelimIdx = u_strspn(tokSource, delim);
michael@0:     tokSource = &tokSource[nonDelimIdx];
michael@0: 
michael@0:     if (*tokSource) {
michael@0:         nextToken = u_strpbrk(tokSource, delim);
michael@0:         if (nextToken != NULL) {
michael@0:             /* Create a token */
michael@0:             *(nextToken++) = 0;
michael@0:             *saveState = nextToken;
michael@0:             return tokSource;
michael@0:         }
michael@0:         else if (*saveState) {
michael@0:             /* Return the last token */
michael@0:             *saveState = NULL;
michael@0:             return tokSource;
michael@0:         }
michael@0:     }
michael@0:     else {
michael@0:         /* No tokens were found. Only delimiters were left. */
michael@0:         *saveState = NULL;
michael@0:     }
michael@0:     return NULL;
michael@0: }
michael@0: 
michael@0: /* Miscellaneous functions -------------------------------------------------- */
michael@0: 
michael@0: U_CAPI UChar* U_EXPORT2
michael@0: u_strcat(UChar     *dst, 
michael@0:     const UChar     *src)
michael@0: {
michael@0:     UChar *anchor = dst;            /* save a pointer to start of dst */
michael@0: 
michael@0:     while(*dst != 0) {              /* To end of first string          */
michael@0:         ++dst;
michael@0:     }
michael@0:     while((*(dst++) = *(src++)) != 0) {     /* copy string 2 over              */
michael@0:     }
michael@0: 
michael@0:     return anchor;
michael@0: }
michael@0: 
michael@0: U_CAPI UChar*  U_EXPORT2
michael@0: u_strncat(UChar     *dst, 
michael@0:      const UChar     *src, 
michael@0:      int32_t     n ) 
michael@0: {
michael@0:     if(n > 0) {
michael@0:         UChar *anchor = dst;            /* save a pointer to start of dst */
michael@0: 
michael@0:         while(*dst != 0) {              /* To end of first string          */
michael@0:             ++dst;
michael@0:         }
michael@0:         while((*dst = *src) != 0) {     /* copy string 2 over              */
michael@0:             ++dst;
michael@0:             if(--n == 0) {
michael@0:                 *dst = 0;
michael@0:                 break;
michael@0:             }
michael@0:             ++src;
michael@0:         }
michael@0: 
michael@0:         return anchor;
michael@0:     } else {
michael@0:         return dst;
michael@0:     }
michael@0: }
michael@0: 
michael@0: /* ----- Text property functions --- */
michael@0: 
michael@0: U_CAPI int32_t   U_EXPORT2
michael@0: u_strcmp(const UChar *s1, 
michael@0:     const UChar *s2) 
michael@0: {
michael@0:     UChar  c1, c2;
michael@0: 
michael@0:     for(;;) {
michael@0:         c1=*s1++;
michael@0:         c2=*s2++;
michael@0:         if (c1 != c2 || c1 == 0) {
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0:     return (int32_t)c1 - (int32_t)c2;
michael@0: }
michael@0: 
michael@0: U_CFUNC int32_t U_EXPORT2
michael@0: uprv_strCompare(const UChar *s1, int32_t length1,
michael@0:                 const UChar *s2, int32_t length2,
michael@0:                 UBool strncmpStyle, UBool codePointOrder) {
michael@0:     const UChar *start1, *start2, *limit1, *limit2;
michael@0:     UChar c1, c2;
michael@0: 
michael@0:     /* setup for fix-up */
michael@0:     start1=s1;
michael@0:     start2=s2;
michael@0: 
michael@0:     /* compare identical prefixes - they do not need to be fixed up */
michael@0:     if(length1<0 && length2<0) {
michael@0:         /* strcmp style, both NUL-terminated */
michael@0:         if(s1==s2) {
michael@0:             return 0;
michael@0:         }
michael@0: 
michael@0:         for(;;) {
michael@0:             c1=*s1;
michael@0:             c2=*s2;
michael@0:             if(c1!=c2) {
michael@0:                 break;
michael@0:             }
michael@0:             if(c1==0) {
michael@0:                 return 0;
michael@0:             }
michael@0:             ++s1;
michael@0:             ++s2;
michael@0:         }
michael@0: 
michael@0:         /* setup for fix-up */
michael@0:         limit1=limit2=NULL;
michael@0:     } else if(strncmpStyle) {
michael@0:         /* special handling for strncmp, assume length1==length2>=0 but also check for NUL */
michael@0:         if(s1==s2) {
michael@0:             return 0;
michael@0:         }
michael@0: 
michael@0:         limit1=start1+length1;
michael@0: 
michael@0:         for(;;) {
michael@0:             /* both lengths are same, check only one limit */
michael@0:             if(s1==limit1) {
michael@0:                 return 0;
michael@0:             }
michael@0: 
michael@0:             c1=*s1;
michael@0:             c2=*s2;
michael@0:             if(c1!=c2) {
michael@0:                 break;
michael@0:             }
michael@0:             if(c1==0) {
michael@0:                 return 0;
michael@0:             }
michael@0:             ++s1;
michael@0:             ++s2;
michael@0:         }
michael@0: 
michael@0:         /* setup for fix-up */
michael@0:         limit2=start2+length1; /* use length1 here, too, to enforce assumption */
michael@0:     } else {
michael@0:         /* memcmp/UnicodeString style, both length-specified */
michael@0:         int32_t lengthResult;
michael@0: 
michael@0:         if(length1<0) {
michael@0:             length1=u_strlen(s1);
michael@0:         }
michael@0:         if(length2<0) {
michael@0:             length2=u_strlen(s2);
michael@0:         }
michael@0: 
michael@0:         /* limit1=start1+min(lenght1, length2) */
michael@0:         if(length1<length2) {
michael@0:             lengthResult=-1;
michael@0:             limit1=start1+length1;
michael@0:         } else if(length1==length2) {
michael@0:             lengthResult=0;
michael@0:             limit1=start1+length1;
michael@0:         } else /* length1>length2 */ {
michael@0:             lengthResult=1;
michael@0:             limit1=start1+length2;
michael@0:         }
michael@0: 
michael@0:         if(s1==s2) {
michael@0:             return lengthResult;
michael@0:         }
michael@0: 
michael@0:         for(;;) {
michael@0:             /* check pseudo-limit */
michael@0:             if(s1==limit1) {
michael@0:                 return lengthResult;
michael@0:             }
michael@0: 
michael@0:             c1=*s1;
michael@0:             c2=*s2;
michael@0:             if(c1!=c2) {
michael@0:                 break;
michael@0:             }
michael@0:             ++s1;
michael@0:             ++s2;
michael@0:         }
michael@0: 
michael@0:         /* setup for fix-up */
michael@0:         limit1=start1+length1;
michael@0:         limit2=start2+length2;
michael@0:     }
michael@0: 
michael@0:     /* if both values are in or above the surrogate range, fix them up */
michael@0:     if(c1>=0xd800 && c2>=0xd800 && codePointOrder) {
michael@0:         /* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
michael@0:         if(
michael@0:             (c1<=0xdbff && (s1+1)!=limit1 && U16_IS_TRAIL(*(s1+1))) ||
michael@0:             (U16_IS_TRAIL(c1) && start1!=s1 && U16_IS_LEAD(*(s1-1)))
michael@0:         ) {
michael@0:             /* part of a surrogate pair, leave >=d800 */
michael@0:         } else {
michael@0:             /* BMP code point - may be surrogate code point - make <d800 */
michael@0:             c1-=0x2800;
michael@0:         }
michael@0: 
michael@0:         if(
michael@0:             (c2<=0xdbff && (s2+1)!=limit2 && U16_IS_TRAIL(*(s2+1))) ||
michael@0:             (U16_IS_TRAIL(c2) && start2!=s2 && U16_IS_LEAD(*(s2-1)))
michael@0:         ) {
michael@0:             /* part of a surrogate pair, leave >=d800 */
michael@0:         } else {
michael@0:             /* BMP code point - may be surrogate code point - make <d800 */
michael@0:             c2-=0x2800;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* now c1 and c2 are in the requested (code unit or code point) order */
michael@0:     return (int32_t)c1-(int32_t)c2;
michael@0: }
michael@0: 
michael@0: /*
michael@0:  * Compare two strings as presented by UCharIterators.
michael@0:  * Use code unit or code point order.
michael@0:  * When the function returns, it is undefined where the iterators
michael@0:  * have stopped.
michael@0:  */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_strCompareIter(UCharIterator *iter1, UCharIterator *iter2, UBool codePointOrder) {
michael@0:     UChar32 c1, c2;
michael@0: 
michael@0:     /* argument checking */
michael@0:     if(iter1==NULL || iter2==NULL) {
michael@0:         return 0; /* bad arguments */
michael@0:     }
michael@0:     if(iter1==iter2) {
michael@0:         return 0; /* identical iterators */
michael@0:     }
michael@0: 
michael@0:     /* reset iterators to start? */
michael@0:     iter1->move(iter1, 0, UITER_START);
michael@0:     iter2->move(iter2, 0, UITER_START);
michael@0: 
michael@0:     /* compare identical prefixes - they do not need to be fixed up */
michael@0:     for(;;) {
michael@0:         c1=iter1->next(iter1);
michael@0:         c2=iter2->next(iter2);
michael@0:         if(c1!=c2) {
michael@0:             break;
michael@0:         }
michael@0:         if(c1==-1) {
michael@0:             return 0;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* if both values are in or above the surrogate range, fix them up */
michael@0:     if(c1>=0xd800 && c2>=0xd800 && codePointOrder) {
michael@0:         /* subtract 0x2800 from BMP code points to make them smaller than supplementary ones */
michael@0:         if(
michael@0:             (c1<=0xdbff && U16_IS_TRAIL(iter1->current(iter1))) ||
michael@0:             (U16_IS_TRAIL(c1) && (iter1->previous(iter1), U16_IS_LEAD(iter1->previous(iter1))))
michael@0:         ) {
michael@0:             /* part of a surrogate pair, leave >=d800 */
michael@0:         } else {
michael@0:             /* BMP code point - may be surrogate code point - make <d800 */
michael@0:             c1-=0x2800;
michael@0:         }
michael@0: 
michael@0:         if(
michael@0:             (c2<=0xdbff && U16_IS_TRAIL(iter2->current(iter2))) ||
michael@0:             (U16_IS_TRAIL(c2) && (iter2->previous(iter2), U16_IS_LEAD(iter2->previous(iter2))))
michael@0:         ) {
michael@0:             /* part of a surrogate pair, leave >=d800 */
michael@0:         } else {
michael@0:             /* BMP code point - may be surrogate code point - make <d800 */
michael@0:             c2-=0x2800;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* now c1 and c2 are in the requested (code unit or code point) order */
michael@0:     return (int32_t)c1-(int32_t)c2;
michael@0: }
michael@0: 
michael@0: #if 0
michael@0: /*
michael@0:  * u_strCompareIter() does not leave the iterators _on_ the different units.
michael@0:  * This is possible but would cost a few extra indirect function calls to back
michael@0:  * up if the last unit (c1 or c2 respectively) was >=0.
michael@0:  *
michael@0:  * Consistently leaving them _behind_ the different units is not an option
michael@0:  * because the current "unit" is the end of the string if that is reached,
michael@0:  * and in such a case the iterator does not move.
michael@0:  * For example, when comparing "ab" with "abc", both iterators rest _on_ the end
michael@0:  * of their strings. Calling previous() on each does not move them to where
michael@0:  * the comparison fails.
michael@0:  *
michael@0:  * So the simplest semantics is to not define where the iterators end up.
michael@0:  *
michael@0:  * The following fragment is part of what would need to be done for backing up.
michael@0:  */
michael@0: void fragment {
michael@0:         /* iff a surrogate is part of a surrogate pair, leave >=d800 */
michael@0:         if(c1<=0xdbff) {
michael@0:             if(!U16_IS_TRAIL(iter1->current(iter1))) {
michael@0:                 /* lead surrogate code point - make <d800 */
michael@0:                 c1-=0x2800;
michael@0:             }
michael@0:         } else if(c1<=0xdfff) {
michael@0:             int32_t idx=iter1->getIndex(iter1, UITER_CURRENT);
michael@0:             iter1->previous(iter1); /* ==c1 */
michael@0:             if(!U16_IS_LEAD(iter1->previous(iter1))) {
michael@0:                 /* trail surrogate code point - make <d800 */
michael@0:                 c1-=0x2800;
michael@0:             }
michael@0:             /* go back to behind where the difference is */
michael@0:             iter1->move(iter1, idx, UITER_ZERO);
michael@0:         } else /* 0xe000<=c1<=0xffff */ {
michael@0:             /* BMP code point - make <d800 */
michael@0:             c1-=0x2800;
michael@0:         }
michael@0: }
michael@0: #endif
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_strCompare(const UChar *s1, int32_t length1,
michael@0:              const UChar *s2, int32_t length2,
michael@0:              UBool codePointOrder) {
michael@0:     /* argument checking */
michael@0:     if(s1==NULL || length1<-1 || s2==NULL || length2<-1) {
michael@0:         return 0;
michael@0:     }
michael@0:     return uprv_strCompare(s1, length1, s2, length2, FALSE, codePointOrder);
michael@0: }
michael@0: 
michael@0: /* String compare in code point order - u_strcmp() compares in code unit order. */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_strcmpCodePointOrder(const UChar *s1, const UChar *s2) {
michael@0:     return uprv_strCompare(s1, -1, s2, -1, FALSE, TRUE);
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t   U_EXPORT2
michael@0: u_strncmp(const UChar     *s1, 
michael@0:      const UChar     *s2, 
michael@0:      int32_t     n) 
michael@0: {
michael@0:     if(n > 0) {
michael@0:         int32_t rc;
michael@0:         for(;;) {
michael@0:             rc = (int32_t)*s1 - (int32_t)*s2;
michael@0:             if(rc != 0 || *s1 == 0 || --n == 0) {
michael@0:                 return rc;
michael@0:             }
michael@0:             ++s1;
michael@0:             ++s2;
michael@0:         }
michael@0:     } else {
michael@0:         return 0;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_strncmpCodePointOrder(const UChar *s1, const UChar *s2, int32_t n) {
michael@0:     return uprv_strCompare(s1, n, s2, n, TRUE, TRUE);
michael@0: }
michael@0: 
michael@0: U_CAPI UChar* U_EXPORT2
michael@0: u_strcpy(UChar     *dst, 
michael@0:     const UChar     *src) 
michael@0: {
michael@0:     UChar *anchor = dst;            /* save a pointer to start of dst */
michael@0: 
michael@0:     while((*(dst++) = *(src++)) != 0) {     /* copy string 2 over              */
michael@0:     }
michael@0: 
michael@0:     return anchor;
michael@0: }
michael@0: 
michael@0: U_CAPI UChar*  U_EXPORT2
michael@0: u_strncpy(UChar     *dst, 
michael@0:      const UChar     *src, 
michael@0:      int32_t     n) 
michael@0: {
michael@0:     UChar *anchor = dst;            /* save a pointer to start of dst */
michael@0: 
michael@0:     /* copy string 2 over */
michael@0:     while(n > 0 && (*(dst++) = *(src++)) != 0) {
michael@0:         --n;
michael@0:     }
michael@0: 
michael@0:     return anchor;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t   U_EXPORT2
michael@0: u_strlen(const UChar *s) 
michael@0: {
michael@0: #if U_SIZEOF_WCHAR_T == U_SIZEOF_UCHAR
michael@0:     return (int32_t)uprv_wcslen(s);
michael@0: #else
michael@0:     const UChar *t = s;
michael@0:     while(*t != 0) {
michael@0:       ++t;
michael@0:     }
michael@0:     return t - s;
michael@0: #endif
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_countChar32(const UChar *s, int32_t length) {
michael@0:     int32_t count;
michael@0: 
michael@0:     if(s==NULL || length<-1) {
michael@0:         return 0;
michael@0:     }
michael@0: 
michael@0:     count=0;
michael@0:     if(length>=0) {
michael@0:         while(length>0) {
michael@0:             ++count;
michael@0:             if(U16_IS_LEAD(*s) && length>=2 && U16_IS_TRAIL(*(s+1))) {
michael@0:                 s+=2;
michael@0:                 length-=2;
michael@0:             } else {
michael@0:                 ++s;
michael@0:                 --length;
michael@0:             }
michael@0:         }
michael@0:     } else /* length==-1 */ {
michael@0:         UChar c;
michael@0: 
michael@0:         for(;;) {
michael@0:             if((c=*s++)==0) {
michael@0:                 break;
michael@0:             }
michael@0:             ++count;
michael@0: 
michael@0:             /*
michael@0:              * sufficient to look ahead one because of UTF-16;
michael@0:              * safe to look ahead one because at worst that would be the terminating NUL
michael@0:              */
michael@0:             if(U16_IS_LEAD(c) && U16_IS_TRAIL(*s)) {
michael@0:                 ++s;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0:     return count;
michael@0: }
michael@0: 
michael@0: U_CAPI UBool U_EXPORT2
michael@0: u_strHasMoreChar32Than(const UChar *s, int32_t length, int32_t number) {
michael@0: 
michael@0:     if(number<0) {
michael@0:         return TRUE;
michael@0:     }
michael@0:     if(s==NULL || length<-1) {
michael@0:         return FALSE;
michael@0:     }
michael@0: 
michael@0:     if(length==-1) {
michael@0:         /* s is NUL-terminated */
michael@0:         UChar c;
michael@0: 
michael@0:         /* count code points until they exceed */
michael@0:         for(;;) {
michael@0:             if((c=*s++)==0) {
michael@0:                 return FALSE;
michael@0:             }
michael@0:             if(number==0) {
michael@0:                 return TRUE;
michael@0:             }
michael@0:             if(U16_IS_LEAD(c) && U16_IS_TRAIL(*s)) {
michael@0:                 ++s;
michael@0:             }
michael@0:             --number;
michael@0:         }
michael@0:     } else {
michael@0:         /* length>=0 known */
michael@0:         const UChar *limit;
michael@0:         int32_t maxSupplementary;
michael@0: 
michael@0:         /* s contains at least (length+1)/2 code points: <=2 UChars per cp */
michael@0:         if(((length+1)/2)>number) {
michael@0:             return TRUE;
michael@0:         }
michael@0: 
michael@0:         /* check if s does not even contain enough UChars */
michael@0:         maxSupplementary=length-number;
michael@0:         if(maxSupplementary<=0) {
michael@0:             return FALSE;
michael@0:         }
michael@0:         /* there are maxSupplementary=length-number more UChars than asked-for code points */
michael@0: 
michael@0:         /*
michael@0:          * count code points until they exceed and also check that there are
michael@0:          * no more than maxSupplementary supplementary code points (UChar pairs)
michael@0:          */
michael@0:         limit=s+length;
michael@0:         for(;;) {
michael@0:             if(s==limit) {
michael@0:                 return FALSE;
michael@0:             }
michael@0:             if(number==0) {
michael@0:                 return TRUE;
michael@0:             }
michael@0:             if(U16_IS_LEAD(*s++) && s!=limit && U16_IS_TRAIL(*s)) {
michael@0:                 ++s;
michael@0:                 if(--maxSupplementary<=0) {
michael@0:                     /* too many pairs - too few code points */
michael@0:                     return FALSE;
michael@0:                 }
michael@0:             }
michael@0:             --number;
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_memcpy(UChar *dest, const UChar *src, int32_t count) {
michael@0:     if(count > 0) {
michael@0:         uprv_memcpy(dest, src, count*U_SIZEOF_UCHAR);
michael@0:     }
michael@0:     return dest;
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_memmove(UChar *dest, const UChar *src, int32_t count) {
michael@0:     if(count > 0) {
michael@0:         uprv_memmove(dest, src, count*U_SIZEOF_UCHAR);
michael@0:     }
michael@0:     return dest;
michael@0: }
michael@0: 
michael@0: U_CAPI UChar * U_EXPORT2
michael@0: u_memset(UChar *dest, UChar c, int32_t count) {
michael@0:     if(count > 0) {
michael@0:         UChar *ptr = dest;
michael@0:         UChar *limit = dest + count;
michael@0: 
michael@0:         while (ptr < limit) {
michael@0:             *(ptr++) = c;
michael@0:         }
michael@0:     }
michael@0:     return dest;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_memcmp(const UChar *buf1, const UChar *buf2, int32_t count) {
michael@0:     if(count > 0) {
michael@0:         const UChar *limit = buf1 + count;
michael@0:         int32_t result;
michael@0: 
michael@0:         while (buf1 < limit) {
michael@0:             result = (int32_t)(uint16_t)*buf1 - (int32_t)(uint16_t)*buf2;
michael@0:             if (result != 0) {
michael@0:                 return result;
michael@0:             }
michael@0:             buf1++;
michael@0:             buf2++;
michael@0:         }
michael@0:     }
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_memcmpCodePointOrder(const UChar *s1, const UChar *s2, int32_t count) {
michael@0:     return uprv_strCompare(s1, count, s2, count, FALSE, TRUE);
michael@0: }
michael@0: 
michael@0: /* u_unescape & support fns ------------------------------------------------- */
michael@0: 
michael@0: /* This map must be in ASCENDING ORDER OF THE ESCAPE CODE */
michael@0: static const UChar UNESCAPE_MAP[] = {
michael@0:     /*"   0x22, 0x22 */
michael@0:     /*'   0x27, 0x27 */
michael@0:     /*?   0x3F, 0x3F */
michael@0:     /*\   0x5C, 0x5C */
michael@0:     /*a*/ 0x61, 0x07,
michael@0:     /*b*/ 0x62, 0x08,
michael@0:     /*e*/ 0x65, 0x1b,
michael@0:     /*f*/ 0x66, 0x0c,
michael@0:     /*n*/ 0x6E, 0x0a,
michael@0:     /*r*/ 0x72, 0x0d,
michael@0:     /*t*/ 0x74, 0x09,
michael@0:     /*v*/ 0x76, 0x0b
michael@0: };
michael@0: enum { UNESCAPE_MAP_LENGTH = sizeof(UNESCAPE_MAP) / sizeof(UNESCAPE_MAP[0]) };
michael@0: 
michael@0: /* Convert one octal digit to a numeric value 0..7, or -1 on failure */
michael@0: static int8_t _digit8(UChar c) {
michael@0:     if (c >= 0x0030 && c <= 0x0037) {
michael@0:         return (int8_t)(c - 0x0030);
michael@0:     }
michael@0:     return -1;
michael@0: }
michael@0: 
michael@0: /* Convert one hex digit to a numeric value 0..F, or -1 on failure */
michael@0: static int8_t _digit16(UChar c) {
michael@0:     if (c >= 0x0030 && c <= 0x0039) {
michael@0:         return (int8_t)(c - 0x0030);
michael@0:     }
michael@0:     if (c >= 0x0041 && c <= 0x0046) {
michael@0:         return (int8_t)(c - (0x0041 - 10));
michael@0:     }
michael@0:     if (c >= 0x0061 && c <= 0x0066) {
michael@0:         return (int8_t)(c - (0x0061 - 10));
michael@0:     }
michael@0:     return -1;
michael@0: }
michael@0: 
michael@0: /* Parse a single escape sequence.  Although this method deals in
michael@0:  * UChars, it does not use C++ or UnicodeString.  This allows it to
michael@0:  * be used from C contexts. */
michael@0: U_CAPI UChar32 U_EXPORT2
michael@0: u_unescapeAt(UNESCAPE_CHAR_AT charAt,
michael@0:              int32_t *offset,
michael@0:              int32_t length,
michael@0:              void *context) {
michael@0: 
michael@0:     int32_t start = *offset;
michael@0:     UChar c;
michael@0:     UChar32 result = 0;
michael@0:     int8_t n = 0;
michael@0:     int8_t minDig = 0;
michael@0:     int8_t maxDig = 0;
michael@0:     int8_t bitsPerDigit = 4; 
michael@0:     int8_t dig;
michael@0:     int32_t i;
michael@0:     UBool braces = FALSE;
michael@0: 
michael@0:     /* Check that offset is in range */
michael@0:     if (*offset < 0 || *offset >= length) {
michael@0:         goto err;
michael@0:     }
michael@0: 
michael@0:     /* Fetch first UChar after '\\' */
michael@0:     c = charAt((*offset)++, context);
michael@0: 
michael@0:     /* Convert hexadecimal and octal escapes */
michael@0:     switch (c) {
michael@0:     case 0x0075 /*'u'*/:
michael@0:         minDig = maxDig = 4;
michael@0:         break;
michael@0:     case 0x0055 /*'U'*/:
michael@0:         minDig = maxDig = 8;
michael@0:         break;
michael@0:     case 0x0078 /*'x'*/:
michael@0:         minDig = 1;
michael@0:         if (*offset < length && charAt(*offset, context) == 0x7B /*{*/) {
michael@0:             ++(*offset);
michael@0:             braces = TRUE;
michael@0:             maxDig = 8;
michael@0:         } else {
michael@0:             maxDig = 2;
michael@0:         }
michael@0:         break;
michael@0:     default:
michael@0:         dig = _digit8(c);
michael@0:         if (dig >= 0) {
michael@0:             minDig = 1;
michael@0:             maxDig = 3;
michael@0:             n = 1; /* Already have first octal digit */
michael@0:             bitsPerDigit = 3;
michael@0:             result = dig;
michael@0:         }
michael@0:         break;
michael@0:     }
michael@0:     if (minDig != 0) {
michael@0:         while (*offset < length && n < maxDig) {
michael@0:             c = charAt(*offset, context);
michael@0:             dig = (int8_t)((bitsPerDigit == 3) ? _digit8(c) : _digit16(c));
michael@0:             if (dig < 0) {
michael@0:                 break;
michael@0:             }
michael@0:             result = (result << bitsPerDigit) | dig;
michael@0:             ++(*offset);
michael@0:             ++n;
michael@0:         }
michael@0:         if (n < minDig) {
michael@0:             goto err;
michael@0:         }
michael@0:         if (braces) {
michael@0:             if (c != 0x7D /*}*/) {
michael@0:                 goto err;
michael@0:             }
michael@0:             ++(*offset);
michael@0:         }
michael@0:         if (result < 0 || result >= 0x110000) {
michael@0:             goto err;
michael@0:         }
michael@0:         /* If an escape sequence specifies a lead surrogate, see if
michael@0:          * there is a trail surrogate after it, either as an escape or
michael@0:          * as a literal.  If so, join them up into a supplementary.
michael@0:          */
michael@0:         if (*offset < length && U16_IS_LEAD(result)) {
michael@0:             int32_t ahead = *offset + 1;
michael@0:             c = charAt(*offset, context);
michael@0:             if (c == 0x5C /*'\\'*/ && ahead < length) {
michael@0:                 c = (UChar) u_unescapeAt(charAt, &ahead, length, context);
michael@0:             }
michael@0:             if (U16_IS_TRAIL(c)) {
michael@0:                 *offset = ahead;
michael@0:                 result = U16_GET_SUPPLEMENTARY(result, c);
michael@0:             }
michael@0:         }
michael@0:         return result;
michael@0:     }
michael@0: 
michael@0:     /* Convert C-style escapes in table */
michael@0:     for (i=0; i<UNESCAPE_MAP_LENGTH; i+=2) {
michael@0:         if (c == UNESCAPE_MAP[i]) {
michael@0:             return UNESCAPE_MAP[i+1];
michael@0:         } else if (c < UNESCAPE_MAP[i]) {
michael@0:             break;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* Map \cX to control-X: X & 0x1F */
michael@0:     if (c == 0x0063 /*'c'*/ && *offset < length) {
michael@0:         c = charAt((*offset)++, context);
michael@0:         if (U16_IS_LEAD(c) && *offset < length) {
michael@0:             UChar c2 = charAt(*offset, context);
michael@0:             if (U16_IS_TRAIL(c2)) {
michael@0:                 ++(*offset);
michael@0:                 c = (UChar) U16_GET_SUPPLEMENTARY(c, c2); /* [sic] */
michael@0:             }
michael@0:         }
michael@0:         return 0x1F & c;
michael@0:     }
michael@0: 
michael@0:     /* If no special forms are recognized, then consider
michael@0:      * the backslash to generically escape the next character.
michael@0:      * Deal with surrogate pairs. */
michael@0:     if (U16_IS_LEAD(c) && *offset < length) {
michael@0:         UChar c2 = charAt(*offset, context);
michael@0:         if (U16_IS_TRAIL(c2)) {
michael@0:             ++(*offset);
michael@0:             return U16_GET_SUPPLEMENTARY(c, c2);
michael@0:         }
michael@0:     }
michael@0:     return c;
michael@0: 
michael@0:  err:
michael@0:     /* Invalid escape sequence */
michael@0:     *offset = start; /* Reset to initial value */
michael@0:     return (UChar32)0xFFFFFFFF;
michael@0: }
michael@0: 
michael@0: /* u_unescapeAt() callback to return a UChar from a char* */
michael@0: static UChar U_CALLCONV
michael@0: _charPtr_charAt(int32_t offset, void *context) {
michael@0:     UChar c16;
michael@0:     /* It would be more efficient to access the invariant tables
michael@0:      * directly but there is no API for that. */
michael@0:     u_charsToUChars(((char*) context) + offset, &c16, 1);
michael@0:     return c16;
michael@0: }
michael@0: 
michael@0: /* Append an escape-free segment of the text; used by u_unescape() */
michael@0: static void _appendUChars(UChar *dest, int32_t destCapacity,
michael@0:                           const char *src, int32_t srcLen) {
michael@0:     if (destCapacity < 0) {
michael@0:         destCapacity = 0;
michael@0:     }
michael@0:     if (srcLen > destCapacity) {
michael@0:         srcLen = destCapacity;
michael@0:     }
michael@0:     u_charsToUChars(src, dest, srcLen);
michael@0: }
michael@0: 
michael@0: /* Do an invariant conversion of char* -> UChar*, with escape parsing */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_unescape(const char *src, UChar *dest, int32_t destCapacity) {
michael@0:     const char *segment = src;
michael@0:     int32_t i = 0;
michael@0:     char c;
michael@0: 
michael@0:     while ((c=*src) != 0) {
michael@0:         /* '\\' intentionally written as compiler-specific
michael@0:          * character constant to correspond to compiler-specific
michael@0:          * char* constants. */
michael@0:         if (c == '\\') {
michael@0:             int32_t lenParsed = 0;
michael@0:             UChar32 c32;
michael@0:             if (src != segment) {
michael@0:                 if (dest != NULL) {
michael@0:                     _appendUChars(dest + i, destCapacity - i,
michael@0:                                   segment, (int32_t)(src - segment));
michael@0:                 }
michael@0:                 i += (int32_t)(src - segment);
michael@0:             }
michael@0:             ++src; /* advance past '\\' */
michael@0:             c32 = (UChar32)u_unescapeAt(_charPtr_charAt, &lenParsed, (int32_t)uprv_strlen(src), (void*)src);
michael@0:             if (lenParsed == 0) {
michael@0:                 goto err;
michael@0:             }
michael@0:             src += lenParsed; /* advance past escape seq. */
michael@0:             if (dest != NULL && U16_LENGTH(c32) <= (destCapacity - i)) {
michael@0:                 U16_APPEND_UNSAFE(dest, i, c32);
michael@0:             } else {
michael@0:                 i += U16_LENGTH(c32);
michael@0:             }
michael@0:             segment = src;
michael@0:         } else {
michael@0:             ++src;
michael@0:         }
michael@0:     }
michael@0:     if (src != segment) {
michael@0:         if (dest != NULL) {
michael@0:             _appendUChars(dest + i, destCapacity - i,
michael@0:                           segment, (int32_t)(src - segment));
michael@0:         }
michael@0:         i += (int32_t)(src - segment);
michael@0:     }
michael@0:     if (dest != NULL && i < destCapacity) {
michael@0:         dest[i] = 0;
michael@0:     }
michael@0:     return i;
michael@0: 
michael@0:  err:
michael@0:     if (dest != NULL && destCapacity > 0) {
michael@0:         *dest = 0;
michael@0:     }
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: /* NUL-termination of strings ----------------------------------------------- */
michael@0: 
michael@0: /**
michael@0:  * NUL-terminate a string no matter what its type.
michael@0:  * Set warning and error codes accordingly.
michael@0:  */
michael@0: #define __TERMINATE_STRING(dest, destCapacity, length, pErrorCode)      \
michael@0:     if(pErrorCode!=NULL && U_SUCCESS(*pErrorCode)) {                    \
michael@0:         /* not a public function, so no complete argument checking */   \
michael@0:                                                                         \
michael@0:         if(length<0) {                                                  \
michael@0:             /* assume that the caller handles this */                   \
michael@0:         } else if(length<destCapacity) {                                \
michael@0:             /* NUL-terminate the string, the NUL fits */                \
michael@0:             dest[length]=0;                                             \
michael@0:             /* unset the not-terminated warning but leave all others */ \
michael@0:             if(*pErrorCode==U_STRING_NOT_TERMINATED_WARNING) {          \
michael@0:                 *pErrorCode=U_ZERO_ERROR;                               \
michael@0:             }                                                           \
michael@0:         } else if(length==destCapacity) {                               \
michael@0:             /* unable to NUL-terminate, but the string itself fit - set a warning code */ \
michael@0:             *pErrorCode=U_STRING_NOT_TERMINATED_WARNING;                \
michael@0:         } else /* length>destCapacity */ {                              \
michael@0:             /* even the string itself did not fit - set an error code */ \
michael@0:             *pErrorCode=U_BUFFER_OVERFLOW_ERROR;                        \
michael@0:         }                                                               \
michael@0:     }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_terminateUChars(UChar *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode) {
michael@0:     __TERMINATE_STRING(dest, destCapacity, length, pErrorCode);
michael@0:     return length;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_terminateChars(char *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode) {
michael@0:     __TERMINATE_STRING(dest, destCapacity, length, pErrorCode);
michael@0:     return length;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_terminateUChar32s(UChar32 *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode) {
michael@0:     __TERMINATE_STRING(dest, destCapacity, length, pErrorCode);
michael@0:     return length;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: u_terminateWChars(wchar_t *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode) {
michael@0:     __TERMINATE_STRING(dest, destCapacity, length, pErrorCode);
michael@0:     return length;
michael@0: }
michael@0: 
michael@0: // Compute the hash code for a string -------------------------------------- ***
michael@0: 
michael@0: // Moved here from uhash.c so that UnicodeString::hashCode() does not depend
michael@0: // on UHashtable code.
michael@0: 
michael@0: /*
michael@0:   Compute the hash by iterating sparsely over about 32 (up to 63)
michael@0:   characters spaced evenly through the string.  For each character,
michael@0:   multiply the previous hash value by a prime number and add the new
michael@0:   character in, like a linear congruential random number generator,
michael@0:   producing a pseudorandom deterministic value well distributed over
michael@0:   the output range. [LIU]
michael@0: */
michael@0: 
michael@0: #define STRING_HASH(TYPE, STR, STRLEN, DEREF) \
michael@0:     int32_t hash = 0;                         \
michael@0:     const TYPE *p = (const TYPE*) STR;        \
michael@0:     if (p != NULL) {                          \
michael@0:         int32_t len = (int32_t)(STRLEN);      \
michael@0:         int32_t inc = ((len - 32) / 32) + 1;  \
michael@0:         const TYPE *limit = p + len;          \
michael@0:         while (p<limit) {                     \
michael@0:             hash = (hash * 37) + DEREF;       \
michael@0:             p += inc;                         \
michael@0:         }                                     \
michael@0:     }                                         \
michael@0:     return hash
michael@0: 
michael@0: /* Used by UnicodeString to compute its hashcode - Not public API. */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ustr_hashUCharsN(const UChar *str, int32_t length) {
michael@0:     STRING_HASH(UChar, str, length, *p);
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ustr_hashCharsN(const char *str, int32_t length) {
michael@0:     STRING_HASH(uint8_t, str, length, *p);
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ustr_hashICharsN(const char *str, int32_t length) {
michael@0:     STRING_HASH(char, str, length, (uint8_t)uprv_tolower(*p));
michael@0: }