]> code.citadel.org Git - citadel.git/blobdiff - libcitadel/lib/stringbuf.c
* add StrBufSmash() so functions demanding a non-const char* can better cooperate.
[citadel.git] / libcitadel / lib / stringbuf.c
index 1af664ef1af1b4f7040d1449a175b8b01d286f81..6afbb16626a68362244f3fa28908c7e17b2aa8c1 100644 (file)
@@ -27,6 +27,8 @@ int ZEXPORT compress_gzip(Bytef * dest, size_t * destLen,
 #endif
 int BaseStrBufSize = 64;
 
+const char *StrBufNOTNULL = ((char*) NULL) - 1;
+
 /**
  * Private Structure for the Stringbuffer
  */
@@ -117,6 +119,9 @@ static int IncreaseBuf(StrBuf *Buf, int KeepOriginal, int DestSize)
                        NewSize *= 2;
 
        NewBuf= (char*) malloc(NewSize);
+       if (NewBuf == NULL)
+               return -1;
+
        if (KeepOriginal && (Buf->BufUsed > 0))
        {
                memcpy(NewBuf, Buf->buf, Buf->BufUsed);
@@ -406,6 +411,55 @@ void FreeStrBuf (StrBuf **FreeMe)
        *FreeMe = NULL;
 }
 
+/**
+ * \brief flatten a Buffer to the Char * we return 
+ * Its a double pointer, so it can NULL your pointer
+ * so fancy SIG11 appear instead of random results
+ * The Calle then owns the buffer and is responsible for freeing it.
+ * \param SmashMe Pointer Pointer to the buffer to release Buf from and free
+ * \return the pointer of the buffer; Callee owns the memory thereafter.
+ */
+char *SmashStrBuf (StrBuf **SmashMe)
+{
+       char *Ret;
+
+       if (*SmashMe == NULL)
+               return NULL;
+#ifdef SIZE_DEBUG
+       if (hFreeDbglog == -1){
+               pid_t pid = getpid();
+               char path [SIZ];
+               snprintf(path, SIZ, "/tmp/libcitadel_strbuf_realloc.log.%d", pid);
+               hFreeDbglog = open(path, O_APPEND|O_CREAT|O_WRONLY);
+       }
+       if ((*SmashMe)->nIncreases > 0)
+       {
+               char buf[SIZ * 3];
+               long n;
+               n = snprintf(buf, SIZ * 3, "S+|%ld|%ld|%ld|%s|%s|\n",
+                            (*SmashMe)->nIncreases,
+                            (*SmashMe)->BufUsed,
+                            (*SmashMe)->BufSize,
+                            (*SmashMe)->bt,
+                            (*SmashMe)->bt_lastinc);
+               n = write(hFreeDbglog, buf, n);
+       }
+       else
+       {
+               char buf[128];
+               long n;
+               n = snprintf(buf, 128, "S_|0|%ld%ld|\n",
+                            (*SmashMe)->BufUsed,
+                            (*SmashMe)->BufSize);
+               n = write(hFreeDbglog, buf, n);
+       }
+#endif
+       Ret = (*SmashMe)->buf;
+       free(*SmashMe);
+       *SmashMe = NULL;
+       return Ret;
+}
+
 /**
  * \brief Release the buffer
  * If you want put your StrBuf into a Hash, use this as Destructor.
@@ -564,6 +618,23 @@ void StrBufAppendBufPlain(StrBuf *Buf, const char *AppendBuf, long AppendSize, u
        Buf->buf[Buf->BufUsed] = '\0';
 }
 
+/**
+ * \brief Callback for cURL to append the webserver reply to a buffer
+ * params pre-defined by the cURL API; see man 3 curl for mre info
+ */
+size_t CurlFillStrBuf_callback(void *ptr, size_t size, size_t nmemb, void *stream)
+{
+
+       StrBuf *Target;
+
+       Target = stream;
+       if (ptr == NULL)
+               return 0;
+
+       StrBufAppendBufPlain(Target, ptr, size * nmemb, 0);
+       return size * nmemb;
+}
+
 
 /** 
  * \brief Escape a string for feeding out as a URL while appending it to a Buffer
@@ -746,7 +817,7 @@ long StrEscAppend(StrBuf *Target, const StrBuf *Source, const char *PlainIn, int
  * \param Source       source buffer; set to NULL if you just have a C-String
  * \param PlainIn       Plain-C string to append; set to NULL if unused
  */
-void StrMsgEscAppend(StrBuf *Target, StrBuf *Source, const char *PlainIn)
+void StrMsgEscAppend(StrBuf *Target, const StrBuf *Source, const char *PlainIn)
 {
        const char *aptr, *eiptr;
        char *tptr, *eptr;
@@ -803,6 +874,77 @@ void StrMsgEscAppend(StrBuf *Target, StrBuf *Source, const char *PlainIn)
        *tptr = '\0';
 }
 
+
+
+/*
+ * \brief Append a string, escaping characters which have meaning in ICAL.  
+ * [\n,] 
+ * \param Target       target buffer
+ * \param Source       source buffer; set to NULL if you just have a C-String
+ * \param PlainIn       Plain-C string to append; set to NULL if unused
+ */
+void StrIcalEscAppend(StrBuf *Target, const StrBuf *Source, const char *PlainIn)
+{
+       const char *aptr, *eiptr;
+       char *tptr, *eptr;
+       long len;
+
+       if (((Source == NULL) && (PlainIn == NULL)) || (Target == NULL) )
+               return ;
+
+       if (PlainIn != NULL) {
+               aptr = PlainIn;
+               len = strlen(PlainIn);
+               eiptr = aptr + len;
+       }
+       else {
+               aptr = Source->buf;
+               eiptr = aptr + Source->BufUsed;
+               len = Source->BufUsed;
+       }
+
+       if (len == 0) 
+               return;
+
+       eptr = Target->buf + Target->BufSize - 8; 
+       tptr = Target->buf + Target->BufUsed;
+       
+       while (aptr < eiptr){
+               if(tptr + 3 >= eptr) {
+                       IncreaseBuf(Target, 1, -1);
+                       eptr = Target->buf + Target->BufSize - 8; 
+                       tptr = Target->buf + Target->BufUsed;
+               }
+              
+               if (*aptr == '\n') {
+                       *tptr = '\\';
+                       Target->BufUsed++;
+                       tptr++;
+                       *tptr = 'n';
+                       Target->BufUsed++;
+               }
+               else if (*aptr == '\r') {
+                       *tptr = '\\';
+                       Target->BufUsed++;
+                       tptr++;
+                       *tptr = 'r';
+                       Target->BufUsed++;
+               }
+               else if (*aptr == ',') {
+                       *tptr = '\\';
+                       Target->BufUsed++;
+                       tptr++;
+                       *tptr = ',';
+                       Target->BufUsed++;
+               } else {
+                       *tptr = *aptr;
+                       Target->BufUsed++;
+               }
+               tptr++; aptr++;
+       }
+       *tptr = '\0';
+}
+
 /*
  * \brief Append a string, escaping characters which have meaning in JavaScript strings .  
  *
@@ -842,7 +984,7 @@ long StrECMAEscAppend(StrBuf *Target, const StrBuf *Source, const char *PlainIn)
                        eptr = Target->buf + Target->BufSize - 3; 
                        bptr = Target->buf + Target->BufUsed;
                }
-               else if (*aptr == '"') {
+               if (*aptr == '"') {
                        *bptr = '\\';
                        bptr ++;
                        *bptr = '"';
@@ -1049,8 +1191,8 @@ int StrBufRemove_token(StrBuf *Source, int parmnum, char separator)
 
        /* Hack and slash */
        if (*s) {
-               memmove(d, s, Source->BufUsed - (s - Source->buf) + 1);
-               Source->BufUsed -= (ReducedBy + 1);
+               memmove(d, s, Source->BufUsed - (s - Source->buf));
+               Source->BufUsed += ReducedBy;
        }
        else if (d == Source->buf) {
                *d = 0;
@@ -1058,7 +1200,7 @@ int StrBufRemove_token(StrBuf *Source, int parmnum, char separator)
        }
        else {
                *--d = 0;
-               Source->BufUsed -= (ReducedBy + 1);
+               Source->BufUsed += ReducedBy;
        }
        /*
        while (*s) {
@@ -1106,8 +1248,10 @@ int StrBufExtract_token(StrBuf *dest, const StrBuf *Source, int parmnum, char se
                }
                if (len >= dest->BufSize) {
                        dest->BufUsed = len;
-                       if (!IncreaseBuf(dest, 1, -1))
+                       if (IncreaseBuf(dest, 1, -1) < 0) {
+                               dest->BufUsed --;
                                break;
+                       }
                }
                if ( (current_token == parmnum) && 
                     (*s != separator)) {
@@ -1212,77 +1356,131 @@ unsigned long StrBufExtract_unsigned_long(const StrBuf* Source, int parmnum, cha
 
 
 
+/**
+ * \briefa string tokenizer; Bounds checker
+ *  function to make shure whether StrBufExtract_NextToken and friends have reached the end of the string.
+ * \param Source our tokenbuffer
+ * \param pStart the token iterator pointer to inspect
+ * \returns whether the revolving pointer is inside of the search range
+ */
+int StrBufHaveNextToken(const StrBuf *Source, const char **pStart)
+{
+       if ((Source == NULL) || 
+           (*pStart == StrBufNOTNULL) ||
+           (Source->BufUsed == 0))
+       {
+               return 0;
+       }
+       if (*pStart == NULL)
+       {
+               return 1;
+       }
+       else if (*pStart > Source->buf + Source->BufUsed)
+       {
+               return 0;
+       }
+       else if (*pStart <= Source->buf)
+       {
+               return 0;
+       }
+
+       return 1;
+}
+
 /**
  * \brief a string tokenizer
  * \param dest Destination StringBuffer
  * \param Source StringBuffer to read into
- * \param pStart pointer to the end of the last token. Feed with NULL.
+ * \param pStart pointer to the end of the last token. Feed with NULL on start.
  * \param separator tokenizer param
  * \returns -1 if not found, else length of token.
  */
 int StrBufExtract_NextToken(StrBuf *dest, const StrBuf *Source, const char **pStart, char separator)
 {
-       const char *s, *EndBuffer;      //* source * /
-       int len = 0;                    //* running total length of extracted string * /
-       int current_token = 0;          //* token currently being processed * /
+       const char *s;          /* source */
+       const char *EndBuffer;  /* end stop of source buffer */
+       int current_token = 0;  /* token currently being processed */
+       int len = 0;            /* running total length of extracted string */
+
+       if ((Source          == NULL) || 
+           (Source->BufUsed == 0)      ) 
+       {
+               *pStart = StrBufNOTNULL;
+               return -1;
+       }
         
-       if (dest != NULL) {
+       EndBuffer = Source->buf + Source->BufUsed;
+
+       if (dest != NULL) 
+       {
                dest->buf[0] = '\0';
                dest->BufUsed = 0;
        }
        else
-               return(-1);
-
-       if ((Source == NULL) || 
-           (Source->BufUsed ==0)) {
-               return(-1);
+       {
+               *pStart = EndBuffer + 1;
+               return -1;
        }
-       if (*pStart == NULL)
-               *pStart = Source->buf;
-
-       EndBuffer = Source->buf + Source->BufUsed;
 
-       if ((*pStart < Source->buf) || 
-           (*pStart >  EndBuffer)) {
-               return (-1);
+       if (*pStart == NULL)
+       {
+               *pStart = Source->buf; /* we're starting to examine this buffer. */
+       }
+       else if ((*pStart < Source->buf) || 
+                (*pStart > EndBuffer  )   ) 
+       {
+               return -1; /* no more tokens to find. */
        }
-
 
        s = *pStart;
-
-       //cit_backtrace();
-       //lprintf (CTDL_DEBUG, "test >: n: %d sep: %c source: %s \n willi \n", parmnum, separator, source);
-
-       while ((s<EndBuffer) && !IsEmptyStr(s)) {
-               if (*s == separator) {
+       /* start to find the next token */
+       while ((s <= EndBuffer)      && 
+              (current_token == 0) ) 
+       {
+               if (*s == separator) 
+               {
+                       /* we found the next token */
                        ++current_token;
                }
-               if (len >= dest->BufSize) {
+
+               if (len >= dest->BufSize) 
+               {
+                       /* our Dest-buffer isn't big enough, increase it. */
                        dest->BufUsed = len;
-                       if (!IncreaseBuf(dest, 1, -1)) {
-                               *pStart = EndBuffer + 1;
+
+                       if (IncreaseBuf(dest, 1, -1) < 0) {
+                               /* WHUT? no more mem? bail out. */
+                               s = EndBuffer;
+                               dest->BufUsed --;
                                break;
                        }
                }
-               if ( (current_token == 0) && 
-                    (*s != separator)) {
-                       dest->buf[len] = *s;
-                       ++len;
-               }
-               else if (current_token > 0) {
-                       *pStart = s;
-                       break;
+
+               if ( (current_token == 0 ) &&   /* are we in our target token? */
+                    (!IsEmptyStr(s)     ) &&
+                    (separator     != *s)    ) /* don't copy the token itself */
+               {
+                       dest->buf[len] = *s;    /* Copy the payload */
+                       ++len;                  /* remember the bigger size. */
                }
+
                ++s;
        }
-       *pStart = s;
-       (*pStart) ++;
 
+       /* did we reach the end? */
+       if ((s > EndBuffer)) {
+               EndBuffer = StrBufNOTNULL;
+               *pStart = EndBuffer;
+       }
+       else {
+               *pStart = s;  /* remember the position for the next run */
+       }
+
+       /* sanitize our extracted token */
        dest->buf[len] = '\0';
-       dest->BufUsed = len;
-               //lprintf (CTDL_DEBUG,"test <!: %s\n", dest);
-       //lprintf (CTDL_DEBUG,"test <: %d; %s\n", len, dest);
-       return(len);
+       dest->BufUsed  = len;
+
+       return (len);
 }
 
 
@@ -1548,6 +1746,9 @@ int StrBufTCP_read_buffered_line(StrBuf *Line,
 
 }
 
+static const char *ErrRBLF_WrongFDFlags="StrBufTCP_read_buffered_line_fast: don't work with fdflags & O_NONBLOCK";
+static const char *ErrRBLF_SelectFailed="StrBufTCP_read_buffered_line_fast: Select failed without reason";
+static const char *ErrRBLF_NotEnoughSentFromServer="StrBufTCP_read_buffered_line_fast: No complete line was sent from peer";
 /**
  * \brief Read a line from socket
  * flushes and closes the FD on error
@@ -1612,8 +1813,10 @@ int StrBufTCP_read_buffered_line_fast(StrBuf *Line,
        }
 
        fdflags = fcntl(*fd, F_GETFL);
-       if ((fdflags & O_NONBLOCK) == O_NONBLOCK)
+       if ((fdflags & O_NONBLOCK) == O_NONBLOCK) {
+               *Error = ErrRBLF_WrongFDFlags;
                return -1;
+       }
 
        pch = NULL;
        while ((nSuccessLess < timeout) && (pch == NULL)) {
@@ -1626,6 +1829,8 @@ int StrBufTCP_read_buffered_line_fast(StrBuf *Line,
                        *Error = strerror(errno);
                        close (*fd);
                        *fd = -1;
+                       if (*Error == NULL)
+                               *Error = ErrRBLF_SelectFailed;
                        return -1;
                }               
                if (FD_ISSET(*fd, &rfds) != 0) {
@@ -1667,6 +1872,7 @@ int StrBufTCP_read_buffered_line_fast(StrBuf *Line,
                *Pos = pos + len + 1;
                return len - rlen;
        }
+       *Error = ErrRBLF_NotEnoughSentFromServer;
        return -1;
 
 }
@@ -1728,6 +1934,7 @@ int StrBufReadBLOB(StrBuf *Buf, int *fd, int append, long nBytes, const char **E
        return nRead;
 }
 
+const char *ErrRBB_too_many_selects = "StrBufReadBLOBBuffered: to many selects; aborting.";
 /**
  * \brief Input binary data from socket
  * flushes and closes the FD on error
@@ -1847,6 +2054,7 @@ int StrBufReadBLOBBuffered(StrBuf *Blob,
                        }
                        if (nSelects > 10) {
                                FlushStrBuf(IOBuf);
+                               *Error = ErrRBB_too_many_selects;
                                return -1;
                        }
                }
@@ -1962,6 +2170,36 @@ void StrBufLowerCase(StrBuf *Buf)
        }
 }
 
+/**
+ * \Brief removes double slashes from pathnames
+ * \param Dir directory string to filter
+ * \param RemoveTrailingSlash allows / disallows trailing slashes
+ */
+void StrBufStripSlashes(StrBuf *Dir, int RemoveTrailingSlash)
+{
+       char *a, *b;
+
+       a = b = Dir->buf;
+
+       while (!IsEmptyStr(a)) {
+               if (*a == '/') {
+                       while (*a == '/')
+                               a++;
+                       *b = '/';
+                       b++;
+               }
+               else {
+                       *b = *a;
+                       b++; a++;
+               }
+       }
+       if ((RemoveTrailingSlash) && (*(b - 1) != '/')){
+               *b = '/';
+               b++;
+       }
+       *b = '\0';
+       Dir->BufUsed = b - Dir->buf;
+}
 
 /**
  * \brief unhide special chars hidden to the HTML escaper
@@ -2115,10 +2353,12 @@ int CompressBuffer(StrBuf *Buf)
        char *compressed_data = NULL;
        size_t compressed_len, bufsize;
        int i = 0;
-       
-       bufsize = compressed_len = ((Buf->BufUsed * 101) / 100) + 100;
+
+       bufsize = compressed_len = Buf->BufUsed +  (Buf->BufUsed / 100) + 100;
        compressed_data = malloc(compressed_len);
        
+       if (compressed_data == NULL)
+               return -1;
        /* Flush some space after the used payload so valgrind shuts up... */
         while ((i < 10) && (Buf->BufUsed + i < Buf->BufSize))
                Buf->buf[Buf->BufUsed + i++] = '\0';
@@ -2417,8 +2657,9 @@ void StrBufConvert(StrBuf *ConvertBuf, StrBuf *TmpBuf, void *pic)
        size_t obuflen;                 /**< Length of output buffer */
 
 
-       if (ConvertBuf->BufUsed >= TmpBuf->BufSize)
-               IncreaseBuf(TmpBuf, 0, ConvertBuf->BufUsed);
+       /* since we're converting to utf-8, one glyph may take up to 6 bytes */
+       if (ConvertBuf->BufUsed * 6 >= TmpBuf->BufSize)
+               IncreaseBuf(TmpBuf, 0, ConvertBuf->BufUsed * 6);
 TRYAGAIN:
        ic = *(iconv_t*)pic;
        ibuf = ConvertBuf->buf;
@@ -2587,8 +2828,13 @@ void StrBuf_RFC822_to_Utf8(StrBuf *Target, const StrBuf *DecodeMe, const StrBuf*
 
        ConvertBuf2 = NewStrBufPlain(NULL, StrLength(DecodeMe));
 
-       if (start != DecodeMe->buf)
-               StrBufAppendBufPlain(Target, DecodeMe->buf, start - DecodeMe->buf, 0);
+       if (start != DecodeMe->buf) {
+               long nFront;
+               
+               nFront = start - DecodeMe->buf;
+               StrBufAppendBufPlain(Target, DecodeMe->buf, nFront, 0);
+               len -= nFront;
+       }
        /*
         * Since spammers will go to all sorts of absurd lengths to get their
         * messages through, there are LOTS of corrupt headers out there.