]> code.citadel.org Git - citadel.git/blobdiff - libcitadel/lib/stringbuf.c
* move utf8 handling stuff into strbuf, so we can be more exact about our buffer...
[citadel.git] / libcitadel / lib / stringbuf.c
index 4990429ca9e18dc19b48ff35da40829d5c125163..4bec25f3dde47740c3e4d931ee054b3c6c4331a3 100644 (file)
@@ -25,7 +25,7 @@
 int ZEXPORT compress_gzip(Bytef * dest, size_t * destLen,
                           const Bytef * source, uLong sourceLen, int level);
 #endif
-int BaseStrBufSize = SIZ;
+int BaseStrBufSize = 64;
 
 /**
  * Private Structure for the Stringbuffer
@@ -128,7 +128,7 @@ static int IncreaseBuf(StrBuf *Buf, int KeepOriginal, int DestSize)
        }
        free (Buf->buf);
        Buf->buf = NewBuf;
-       Buf->BufSize *= 2;
+       Buf->BufSize = NewSize;
 #ifdef SIZE_DEBUG
        Buf->nIncreases++;
 #if HAVE_BACKTRACE
@@ -138,6 +138,22 @@ static int IncreaseBuf(StrBuf *Buf, int KeepOriginal, int DestSize)
        return Buf->BufSize;
 }
 
+/**
+ * \brief shrink an _EMPTY_ buffer if its Buffer superseeds threshhold to NewSize. Buffercontent is thoroughly ignored and flushed.
+ * \param Buf Buffer to shrink (has to be empty)
+ * \param ThreshHold if the buffer is bigger then this, its readjusted
+ * \param NewSize if we Shrink it, how big are we going to be afterwards?
+ */
+void ReAdjustEmptyBuf(StrBuf *Buf, long ThreshHold, long NewSize)
+{
+       if (Buf->BufUsed > ThreshHold) {
+               free(Buf->buf);
+               Buf->buf = (char*) malloc(NewSize);
+               Buf->BufUsed = 0;
+               Buf->BufSize = NewSize;
+       }
+}
+
 /**
  * \brief shrink long term buffers to their real size so they don't waste memory
  * \param Buf buffer to shrink
@@ -827,13 +843,17 @@ long StrECMAEscAppend(StrBuf *Target, const StrBuf *Source, const char *PlainIn)
                        bptr = Target->buf + Target->BufUsed;
                }
                else if (*aptr == '"') {
-                       memcpy(bptr, "\\\"", 2);
-                       bptr += 2;
+                       *bptr = '\\';
+                       bptr ++;
+                       *bptr = '"';
+                       bptr ++;
                        Target->BufUsed += 2;
                } else if (*aptr == '\\') {
-                 memcpy(bptr, "\\\\", 2);
-                 bptr += 2;
-                 Target->BufUsed += 2;
+                       *bptr = '\\';
+                       bptr ++;
+                       *bptr = '\\';
+                       bptr ++;
+                       Target->BufUsed += 2;
                }
                else{
                        *bptr = *aptr;
@@ -843,7 +863,7 @@ long StrECMAEscAppend(StrBuf *Target, const StrBuf *Source, const char *PlainIn)
                aptr ++;
        }
        *bptr = '\0';
-       if ((bptr = eptr - 1 ) && !IsEmptyStr(aptr) )
+       if ((bptr == eptr - 1 ) && !IsEmptyStr(aptr) )
                return -1;
        return Target->BufUsed;
 }
@@ -906,6 +926,7 @@ void StrBufVAppendPrintf(StrBuf *Buf, const char *format, va_list ap)
                newused = Offset + nWritten;
                if (newused >= Buf->BufSize) {
                        IncreaseBuf(Buf, 1, newused);
+                       newused = Buf->BufSize + 1;
                }
                else {
                        Buf->BufUsed = Offset + nWritten;
@@ -938,6 +959,7 @@ void StrBufAppendPrintf(StrBuf *Buf, const char *format, ...)
                newused = Buf->BufUsed + nWritten;
                if (newused >= Buf->BufSize) {
                        IncreaseBuf(Buf, 1, newused);
+                       newused = Buf->BufSize + 1;
                }
                else {
                        Buf->BufUsed += nWritten;
@@ -962,9 +984,12 @@ void StrBufPrintf(StrBuf *Buf, const char *format, ...)
                va_start(arg_ptr, format);
                nWritten = vsnprintf(Buf->buf, Buf->BufSize, format, arg_ptr);
                va_end(arg_ptr);
-               Buf->BufUsed = nWritten ;
-               if (nWritten >= Buf->BufSize)
+               if (nWritten >= Buf->BufSize) {
                        IncreaseBuf(Buf, 0, 0);
+                       nWritten = Buf->BufSize + 1;
+                       continue;
+               }
+               Buf->BufUsed = nWritten ;
        }
 }
 
@@ -1530,15 +1555,15 @@ int StrBufTCP_read_buffered_line(StrBuf *Line,
  * \returns numbers of chars read
  */
 int StrBufTCP_read_buffered_line_fast(StrBuf *Line, 
-                                     StrBuf *buf, 
+                                     StrBuf *IOBuf, 
                                      const char **Pos,
                                      int *fd, 
                                      int timeout, 
                                      int selectresolution, 
                                      const char **Error)
 {
-       const char *pche;
-       const char *pos;
+       const char *pche = NULL;
+       const char *pos = NULL;
        int len, rlen;
        int nSuccessLess = 0;
        fd_set rfds;
@@ -1547,11 +1572,11 @@ int StrBufTCP_read_buffered_line_fast(StrBuf *Line,
        struct timeval tv;
        
        pos = *Pos;
-       if ((buf->BufUsed > 0) && 
+       if ((IOBuf->BufUsed > 0) && 
            (pos != NULL) && 
-           (pos < buf->buf + buf->BufUsed)) 
+           (pos < IOBuf->buf + IOBuf->BufUsed)) 
        {
-               pche = buf->buf + buf->BufUsed;
+               pche = IOBuf->buf + IOBuf->BufUsed;
                pch = pos;
                while ((pch < pche) && (*pch != '\n'))
                        pch ++;
@@ -1564,25 +1589,29 @@ int StrBufTCP_read_buffered_line_fast(StrBuf *Line,
                        len = pch - pos;
                        if (len > 0 && (*(pch - 1) == '\r') )
                                rlen ++;
-                       StrBufSub(Line, buf, (pos - buf->buf), len - rlen);
+                       StrBufSub(Line, IOBuf, (pos - IOBuf->buf), len - rlen);
                        *Pos = pch + 1;
                        return len - rlen;
                }
        }
        
        if (pos != NULL) {
-               StrBufCutLeft(buf, (pos - buf->buf));
+               if (pos > pche)
+                       FlushStrBuf(IOBuf);
+               else 
+                       StrBufCutLeft(IOBuf, (pos - IOBuf->buf));
                *Pos = NULL;
        }
        
-       if (buf->BufSize - buf->BufUsed < 10) {
-               IncreaseBuf(buf, 1, -1);
+       if (IOBuf->BufSize - IOBuf->BufUsed < 10) {
+               IncreaseBuf(IOBuf, 1, -1);
        }
 
        fdflags = fcntl(*fd, F_GETFL);
        if ((fdflags & O_NONBLOCK) == O_NONBLOCK)
                return -1;
 
+       pch = NULL;
        while ((nSuccessLess < timeout) && (pch == NULL)) {
                tv.tv_sec = selectresolution;
                tv.tv_usec = 0;
@@ -1597,8 +1626,8 @@ int StrBufTCP_read_buffered_line_fast(StrBuf *Line,
                }               
                if (FD_ISSET(*fd, &rfds) != 0) {
                        rlen = read(*fd, 
-                                   &buf->buf[buf->BufUsed], 
-                                   buf->BufSize - buf->BufUsed - 1);
+                                   &IOBuf->buf[IOBuf->BufUsed], 
+                                   IOBuf->BufSize - IOBuf->BufUsed - 1);
                        if (rlen < 1) {
                                *Error = strerror(errno);
                                close(*fd);
@@ -1607,24 +1636,30 @@ int StrBufTCP_read_buffered_line_fast(StrBuf *Line,
                        }
                        else if (rlen > 0) {
                                nSuccessLess = 0;
-                               buf->BufUsed += rlen;
-                               buf->buf[buf->BufUsed] = '\0';
-                               if (buf->BufUsed + 10 > buf->BufSize) {
-                                       IncreaseBuf(buf, 1, -1);
+                               IOBuf->BufUsed += rlen;
+                               IOBuf->buf[IOBuf->BufUsed] = '\0';
+                               if (IOBuf->BufUsed + 10 > IOBuf->BufSize) {
+                                       IncreaseBuf(IOBuf, 1, -1);
                                }
-                               pch = strchr(buf->buf, '\n');
+
+                               pche = IOBuf->buf + IOBuf->BufUsed;
+                               pch = IOBuf->buf;
+                               while ((pch < pche) && (*pch != '\n'))
+                                       pch ++;
+                               if ((pch >= pche) || (*pch == '\0'))
+                                       pch = NULL;
                                continue;
                        }
                }
                nSuccessLess ++;
        }
        if (pch != NULL) {
-               pos = buf->buf;
+               pos = IOBuf->buf;
                rlen = 0;
                len = pch - pos;
                if (len > 0 && (*(pch - 1) == '\r') )
                        rlen ++;
-               StrBufSub(Line, buf, 0, len - rlen);
+               StrBufSub(Line, IOBuf, 0, len - rlen);
                *Pos = pos + len + 1;
                return len - rlen;
        }
@@ -1699,15 +1734,17 @@ int StrBufReadBLOB(StrBuf *Buf, int *fd, int append, long nBytes, const char **E
  * \param Error strerror() on error 
  * \returns numbers of chars read
  */
-int StrBufReadBLOBBuffered(StrBuf *Buf
+int StrBufReadBLOBBuffered(StrBuf *Blob
                           StrBuf *IOBuf, 
-                          const char **BufPos,
+                          const char **Pos,
                           int *fd, 
                           int append, 
                           long nBytes, 
                           int check, 
                           const char **Error)
 {
+       const char *pche;
+       const char *pos;
        int nSelects = 0;
        int SelRes;
         fd_set wset;
@@ -1716,50 +1753,62 @@ int StrBufReadBLOBBuffered(StrBuf *Buf,
        int rlen, slen;
        int nRead = 0;
        char *ptr;
+       const char *pch;
 
-       if ((Buf == NULL) || (*fd == -1) || (IOBuf == NULL) || (BufPos == NULL))
+       if ((Blob == NULL) || (*fd == -1) || (IOBuf == NULL) || (Pos == NULL))
                return -1;
        if (!append)
-               FlushStrBuf(Buf);
-       if (Buf->BufUsed + nBytes >= Buf->BufSize) 
-               IncreaseBuf(Buf, 1, Buf->BufUsed + nBytes);
+               FlushStrBuf(Blob);
+       if (Blob->BufUsed + nBytes >= Blob->BufSize) 
+               IncreaseBuf(Blob, append, Blob->BufUsed + nBytes);
        
+       pos = *Pos;
 
-       if (*BufPos > 0)
-               len = *BufPos - IOBuf->buf;
+       if (pos > 0)
+               len = pos - IOBuf->buf;
        rlen = IOBuf->BufUsed - len;
 
+
        if ((IOBuf->BufUsed > 0) && 
-           ((IOBuf->BufUsed - len > 0))) {
+           (pos != NULL) && 
+           (pos < IOBuf->buf + IOBuf->BufUsed)) 
+       {
+               pche = IOBuf->buf + IOBuf->BufUsed;
+               pch = pos;
+
                if (rlen < nBytes) {
-                       memcpy(Buf->buf + Buf->BufUsed, *BufPos, rlen);
-                       Buf->BufUsed += rlen;
-                       Buf->buf[Buf->BufUsed] = '\0';
+                       memcpy(Blob->buf + Blob->BufUsed, pos, rlen);
+                       Blob->BufUsed += rlen;
+                       Blob->buf[Blob->BufUsed] = '\0';
                        nRead = rlen;
-                       *BufPos = NULL; 
-                       FlushStrBuf(IOBuf);
+                       *Pos = NULL; 
                }
                if (rlen >= nBytes) {
-                       memcpy(Buf->buf + Buf->BufUsed, *BufPos, nBytes);
-                       Buf->BufUsed += nBytes;
-                       Buf->buf[Buf->BufUsed] = '\0';
+                       memcpy(Blob->buf + Blob->BufUsed, pos, nBytes);
+                       Blob->BufUsed += nBytes;
+                       Blob->buf[Blob->BufUsed] = '\0';
                        if (rlen == nBytes) {
-                               *BufPos = NULL; 
+                               *Pos = NULL; 
                                FlushStrBuf(IOBuf);
                        }
                        else 
-                               *BufPos += nBytes;
+                               *Pos += nBytes;
                        return nBytes;
                }
        }
 
-       ptr = IOBuf->buf + Buf->BufUsed;
+       FlushStrBuf(IOBuf);
+       if (IOBuf->BufSize < nBytes - nRead)
+               IncreaseBuf(IOBuf, 0, nBytes - nRead);
+       ptr = IOBuf->buf;
 
-       slen = len = Buf->BufUsed;
+       slen = len = Blob->BufUsed;
 
        fdflags = fcntl(*fd, F_GETFL);
 
        SelRes = 1;
+       nBytes -= nRead;
+       nRead = 0;
        while (nRead < nBytes) {
                if ((fdflags & O_NONBLOCK) == O_NONBLOCK) {
                         FD_ZERO(&wset);
@@ -1772,9 +1821,10 @@ int StrBufReadBLOBBuffered(StrBuf *Buf,
                }
                else if (SelRes) {
                        nSelects = 0;
-                       if ((rlen = read(*fd, 
-                                        ptr,
-                                        nBytes - nRead)) == -1) {
+                       rlen = read(*fd, 
+                                   ptr,
+                                   nBytes - nRead);
+                       if (rlen == -1) {
                                close(*fd);
                                *fd = -1;
                                *Error = strerror(errno);
@@ -1785,22 +1835,28 @@ int StrBufReadBLOBBuffered(StrBuf *Buf,
                        nSelects ++;
                        if ((check == NNN_TERM) && 
                            (nRead > 5) &&
-                           (strncmp(Buf->buf + Buf->BufUsed - 5, "\n000\n", 5) == 0)) 
+                           (strncmp(IOBuf->buf + IOBuf->BufUsed - 5, "\n000\n", 5) == 0)) 
                        {
-                               StrBufPlain(IOBuf, HKEY("\n000\n"));
-                               StrBufCutRight(Buf, 5);
-                               return Buf->BufUsed;
+                               StrBufPlain(Blob, HKEY("\n000\n"));
+                               StrBufCutRight(Blob, 5);
+                               return Blob->BufUsed;
                        }
                        if (nSelects > 10) {
-                               FlushStrBuf(Buf);
+                               FlushStrBuf(IOBuf);
                                return -1;
                        }
                }
-               nRead += rlen;
-               ptr += rlen;
-               Buf->BufUsed += rlen;
+               if (rlen > 0) {
+                       nRead += rlen;
+                       ptr += rlen;
+                       IOBuf->BufUsed += rlen;
+               }
        }
-       Buf->buf[Buf->BufUsed] = '\0';
+       if (nRead > nBytes) {
+               *Pos = IOBuf->buf + nBytes;
+       }
+       Blob->buf[Blob->BufUsed] = '\0';
+       StrBufAppendBufPlain(Blob, IOBuf->buf, nBytes, 0);
        return nRead;
 }
 
@@ -2585,23 +2641,97 @@ void StrBuf_RFC822_to_Utf8(StrBuf *Target, const StrBuf *DecodeMe, const StrBuf*
        FreeStrBuf(&ConvertBuf2);
 }
 
+/**
+ * \brief evaluate the length of an utf8 special character sequence
+ * \param Char the character to examine
+ * \returns width of utf8 chars in bytes
+ */
+static inline int Ctdl_GetUtf8SequenceLength(char *CharS, char *CharE)
+{
+       int n = 1;
+        char test = (1<<7);
+       
+       while ((n < 8) && ((test & *CharS) != 0)) {
+               test = test << 1;
+               n ++;
+       }
+       if ((n > 6) || ((CharE - CharS) > n))
+               n = 1;
+       return n;
+}
 
+/**
+ * \brief detect whether this char starts an utf-8 encoded char
+ * \param Char character to inspect
+ * \returns yes or no
+ */
+static inline int Ctdl_IsUtf8SequenceStart(char Char)
+{
+/** 11??.???? indicates an UTF8 Sequence. */
+       return ((Char & 0xC0) != 0);
+}
 
+/**
+ * \brief measure the number of glyphs in an UTF8 string...
+ * \param str string to measure
+ * \returns the length of str
+ */
 long StrBuf_Utf8StrLen(StrBuf *Buf)
 {
-       return Ctdl_Utf8StrLen(Buf->buf);
+       int n = 0;
+       int m = 0;
+       char *aptr, *eptr;
+
+       if ((Buf == NULL) || (Buf->BufUsed == 0))
+               return 0;
+       aptr = Buf->buf;
+       eptr = Buf->buf + Buf->BufUsed;
+       while ((aptr < eptr) && (*aptr != '\0')) {
+               if (Ctdl_IsUtf8SequenceStart(*aptr)){
+                       m = Ctdl_GetUtf8SequenceLength(aptr, eptr);
+                       while ((aptr < eptr) && (m-- > 0) && (*aptr++ != '\0'))
+                               n ++;
+               }
+               else {
+                       n++;
+                       aptr++;
+               }
+                       
+       }
+       return n;
 }
 
+/**
+ * \brief cuts a string after maxlen glyphs
+ * \param str string to cut to maxlen glyphs
+ * \param maxlen how long may the string become?
+ * \returns pointer to maxlen or the end of the string
+ */
 long StrBuf_Utf8StrCut(StrBuf *Buf, int maxlen)
 {
-       char *CutAt;
+       char *aptr, *eptr;
+       int n = 0, m = 0;
 
-       CutAt = Ctdl_Utf8StrCut(Buf->buf, maxlen);
-       if (CutAt != NULL) {
-               Buf->BufUsed = CutAt - Buf->buf;
-               Buf->buf[Buf->BufUsed] = '\0';
+       aptr = Buf->buf;
+       eptr = Buf->buf + Buf->BufUsed;
+       while ((aptr < eptr) && (*aptr != '\0')) {
+               if (Ctdl_IsUtf8SequenceStart(*aptr)){
+                       m = Ctdl_GetUtf8SequenceLength(aptr, eptr);
+                       while ((m-- > 0) && (*aptr++ != '\0'))
+                               n ++;
+               }
+               else {
+                       n++;
+                       aptr++;
+               }
+               if (n > maxlen) {
+                       *aptr = '\0';
+                       Buf->BufUsed = aptr - Buf->buf;
+                       return Buf->BufUsed;
+               }                       
        }
-       return Buf->BufUsed;    
+       return Buf->BufUsed;
+
 }