add scan_netstring

scan/scan_8long.c

@@ -1,13 +1,5 @@
 #include "scan.h"
 
 size_t scan_8long(const char *src,unsigned long *dest) {
-  register const char *tmp=src;
-  register unsigned long l=0;
-  register unsigned char c;
-  while ((c=*tmp-'0')<8) {
-    l=l*8+c;
-    ++tmp;
-  }
-  *dest=l;
-  return tmp-src;
+  return scan_8longn(src,-1,dest);
 }

scan/scan_long.c

@@ -1,21 +1,5 @@
 #include "scan.h"
 
 size_t scan_long(const char *src,long *dest) {
-  register const char *tmp;
-  register long int l;
-  register unsigned char c;
-  int neg;
-  int ok;
-  tmp=src; l=0; ok=neg=0;
-  switch (*tmp) {
-  case '-': neg=1;
-  case '+': ++tmp;
-  }
-  while ((c=*tmp-'0')<10) {
-    l=l*10+c;
-    ++tmp;
-    ok=1;
-  }
-  if (ok) *dest=(neg?-l:l);
-  return tmp-src;
+  return scan_longn(src,-1,dest);
 }

scan/scan_netstring.c

@@ -0,0 +1,25 @@
+#include "scan.h"
+
+/* parse a netstring, input buffer is in (len bytes).
+ * if parsing is successful:
+ *   *dest points to string and *slen is size of string
+ *   return number of bytes parsed
+ * else
+ *   return 0
+ * Note: *dest will point inside the input buffer!
+ */
+size_t scan_netstring(const char* in,size_t len,char** dest,size_t* slen) {
+  // [len]":"[string]","
+  // 3:foo,3:bar,4:fini,
+  unsigned long l;
+  size_t n=scan_ulongn(in,len,&l);
+  if (!n || /* did not start with a number */
+      n+2+l<l || /* overflow */
+      n+2+l>len || /* longer than we have input data */
+      in[n]!=':' || /* syntax error */
+      in[n+l+1]!=',')
+    return 0;
+  *dest=(char*)in+n+1;
+  *slen=n;
+  return n+2+l;
+}

scan/scan_ulong.c

@@ -1,19 +1,5 @@
 #include "scan.h"
 
 size_t scan_ulong(const char* src,unsigned long int* dest) {
-  register const char *tmp=src;
-  register unsigned long int l=0;
-  register unsigned char c;
-  while ((c=*tmp-'0')<10) {
-    unsigned long int n;
-    /* division is very slow on most architectures */
-    n=l<<3; if ((n>>3)!=l) break;
-    if (n+(l<<1) < n) break;
-    n+=l<<1;
-    if (n+c < n) break;
-    l=n+c;
-    ++tmp;
-  }
-  if (tmp-src) *dest=l;
-  return tmp-src;
+  return scan_ulongn(src,(size_t)-1,dest);
 }

scan/scan_xlong.c

@@ -1,13 +1,5 @@
 #include "scan.h"
 
 size_t scan_xlong(const char *src,unsigned long *dest) {
-  register const char *tmp=src;
-  register unsigned long l=0;
-  register unsigned char c;
-  while ((c=scan_fromhex(*tmp))<16) {
-    l=(l<<4)+c;
-    ++tmp;
-  }
-  *dest=l;
-  return tmp-src;
+  return scan_xlongn(src,(size_t)-1,dest);
 }

test/scan_long.c

@@ -6,6 +6,14 @@
 int main() {
   char buf[1024];
   unsigned long long int i;
+  unsigned long l;
+  if (sizeof(unsigned long)==4) {
+    assert(scan_ulong("4294967295",&l) == 10 && l==4294967295ul);
+    assert(scan_ulong("4294967296",&l) == 9 && l==429496729);
+  } else {
+    assert(scan_ulong("18446744073709551615",&l) == 20 && l==18446744073709551615ull);
+    assert(scan_ulong("18446744073709551616",&l) == 19 && l==1844674407370955161ull);
+  }
   if (sizeof(unsigned long) != 4)
     return 0;
   for (i=1; i<0xfffffffffull; i+=i+1) {

test/scan_netstring.c

@@ -0,0 +1,11 @@
+#include "scan.h"
+#include <assert.h>
+
+int main() {
+  char* s;
+  size_t l;
+  const char* orig;
+  orig="3:foo,"; assert(scan_netstring(orig,6,&s,&l)==6); assert(s==orig+2);
+  orig="4294967295:foo,"; assert(scan_netstring(orig,15,&s,&l)==0);
+  orig="18446744073709551615:foo,"; assert(scan_netstring(orig,25,&s,&l)==0);
+}