add mkCharLen for strings with known length and pssobly embedded nuls

git-svn-id: https://svn.r-project.org/R/trunk@43074 00db46b3-68df-0310-9c12-caf00c1e9a41

src/include/Rinternals.h

@@ -604,6 +604,7 @@ SEXP Rf_match(SEXP, SEXP, int);
 SEXP Rf_namesgets(SEXP, SEXP);
 SEXP Rf_mkChar(const char *);
 SEXP Rf_mkCharEnc(const char *, int);
+SEXP Rf_mkCharLen(const char *, int);
 Rboolean Rf_NonNullStringMatch(SEXP, SEXP);
 int Rf_ncols(SEXP);
 int Rf_nrows(SEXP);
@@ -906,6 +907,7 @@ int R_system(const char *);
 #define match			Rf_match
 #define mkChar			Rf_mkChar
 #define mkCharEnc		Rf_mkCharEnc
+#define mkCharLen		Rf_mkCharLen
 #define mkString		Rf_mkString
 #define namesgets		Rf_namesgets
 #define ncols			Rf_ncols

src/main/character.c

@@ -2459,7 +2459,7 @@ SEXP attribute_hidden do_charToRaw(SEXP call, SEXP op, SEXP args, SEXP env)
 /* <UTF8>  rawToChar should work at byte level */
 SEXP attribute_hidden do_rawToChar(SEXP call, SEXP op, SEXP args, SEXP env)
 {
-    SEXP ans, c, x = CAR(args);
+    SEXP ans, x = CAR(args);
     int i, nc = LENGTH(x), multiple, len;
     char buf[2];
 
@@ -2482,9 +2482,7 @@ SEXP attribute_hidden do_rawToChar(SEXP call, SEXP op, SEXP args, SEXP env)
         PROTECT(ans = allocVector(STRSXP, 1));
         /* String is not necessarily 0-terminated and may contain nuls
            so don't use mkChar */
-        c = allocString(len); /* adds zero terminator */
-        memcpy(CHAR_RW(c), RAW(x), len);
-        SET_STRING_ELT(ans, 0, c);
+        SET_STRING_ELT(ans, 0, mkCharLen((const char *)RAW(x), len));
     }
     UNPROTECT(1);
     return ans;

src/main/connections.c

@@ -3217,7 +3217,6 @@ static SEXP readFixedString(Rconnection con, int len)
 {
     char *buf;
     int  pos, m;
-    SEXP ans;
 
 #ifdef SUPPORT_UTF8
     if(utf8locale) {
@@ -3249,9 +3248,7 @@ static SEXP readFixedString(Rconnection con, int len)
 	pos = m;
     }
     /* String may contain nuls so don't use mkChar */
-    ans = allocString(pos);
-    memcpy(CHAR_RW(ans), buf, pos);
-    return ans;
+    return mkCharLen(buf, pos);
 }
 
 

src/main/envir.c

@@ -3212,6 +3212,18 @@ SEXP attribute_hidden do_envprofile(SEXP call, SEXP op, SEXP args, SEXP rho)
     return ans;
 }
 
+/* 
+   Version for strings with embedded nuls:
+   these do not currently go in the cache,
+   and do not have an encoding. 
+*/
+SEXP mkCharLen(const char *name, int len)
+{
+    SEXP c = allocString(len);
+    memcpy(CHAR_RW(c), name, len);
+    return c;
+}
+
 
 #ifndef USE_CHAR_HASHING
 SEXP mkChar(const char *name)
@@ -3244,8 +3256,8 @@ SEXP mkCharEnc(const char *name, int enc)
 /* We can reuse the hash structure, but need separate code for get/set
    of values since our keys are char* and not SEXP symbol types. 
 
-   Experience has shown that it is better to use a different hash function:
-   this is the simplest known good one and is likely to be replaced.
+   Experience has shown that it is better to use a different hash function,
+   and a power of 2 for the hash size.
 */
 
 /* char_hash_size MUST be a power of 2 and char_hash_mask ==
@@ -3257,8 +3269,9 @@ static unsigned int char_hash_mask = 65535;
 
 static unsigned int char_hash(const char *s)
 {
+    /* djb2 as from http://www.cse.yorku.ca/~oz/hash.html */
     char *p;
-    unsigned int h = 0;
+    unsigned int h = 5381;
     for (p = (char *) s; *p; p++)
 	h = ((h << 5) + h) + (*p);
     return h;
@@ -3373,7 +3386,7 @@ SEXP mkCharEnc(const char *name, int enc)
     }
     if (cval == R_NilValue) {
         /* no cached value; need to allocate one and add to the cache */
-        PROTECT(cval = allocString(strlen(name)));
+        PROTECT(cval = allocString(len));
         strcpy(CHAR_RW(cval), name);
         switch(enc) {
         case 0:

src/main/sysutils.c

@@ -506,10 +506,7 @@ SEXP attribute_hidden do_iconv(SEXP call, SEXP op, SEXP args, SEXP env)
 		nout = cbuff.bufsize - 1 - outb;
 		for(j = 0; j < nout; j++) if(!*p++) {has_nul = TRUE; break;}
 		if(has_nul) {
-		    si = allocString(nout);
-		    memcpy(CHAR_RW(si), cbuff.data, nout);
-		    if(isLatin1) SET_LATIN1(si);
-		    if(isUTF8) SET_UTF8(si);
+		    si = mkCharLen(cbuff.data, nout);
 		} else {
 		    if(isLatin1) si = mkCharEnc(cbuff.data, LATIN1_MASK);
 		    else if(isUTF8) si = mkCharEnc(cbuff.data, UTF8_MASK);