readint.c: add new function `mrb_int_read`.

Difference from `strtoul(3)`:

* reads `mrb_int` based on configuration
* specifies the end of the string
* no sign interpretation
* base 10 only

include/mruby/value.h

@@ -100,6 +100,7 @@ struct mrb_state;
 # define MRB_ENDIAN_LOHI(a,b) b a
 #endif
 
+MRB_API mrb_int mrb_int_read(const char *p, const char *e, char **endp);
 #ifndef MRB_NO_FLOAT
 MRB_API double mrb_float_read(const char*, char**);
 #ifdef MRB_USE_FLOAT32

mrbgems/mruby-pack/src/pack.c

@@ -1234,17 +1234,16 @@ alias:
 
   /* read suffix [0-9*_!<>] */
   while (tmpl->idx < tlen) {
-    ch = tptr[tmpl->idx++];
+    ch = tptr[tmpl->idx];
     if (ISDIGIT(ch)) {
-      count = ch - '0';
-      while (tmpl->idx < tlen && ISDIGIT(tptr[tmpl->idx])) {
-        if (count+10 > INT_MAX/10) {
-          mrb_raise(mrb, E_RUNTIME_ERROR, "too big template length");
-        }
-        ch = tptr[tmpl->idx++] - '0';
-        count = count * 10 + ch;
+      char *e;
+      mrb_int n = mrb_int_read(tptr+tmpl->idx, tptr+tlen, &e);
+      if (e == NULL || n > INT_MAX) {
+        mrb_raise(mrb, E_RUNTIME_ERROR, "too big template length");
       }
-      continue;  /* special case */
+      count = (int)n;
+      tmpl->idx += e - tptr;
+      continue;
     } else if (ch == '*')  {
       count = -1;
     } else if (ch == '_' || ch == '!' || ch == '<' || ch == '>') {
@@ -1258,10 +1257,11 @@ alias:
       } else if (ch == '>') {
         flags |= PACK_FLAG_GT;
       }
-    } else {
-      tmpl->idx--;
+    }
+    else {
       break;
     }
+    tmpl->idx++;
   }
 
   if ((flags & PACK_FLAG_LT) || (!(flags & PACK_FLAG_GT) && littleendian)) {

mrbgems/mruby-sprintf/src/sprintf.c

@@ -245,7 +245,6 @@ check_name_arg(mrb_state *mrb, int posarg, const char *name, size_t len)
 #define GETASTER(num) do { \
   mrb_value tmp_v; \
   t = p++; \
-  n = 0; \
   GETNUM(n, val); \
   if (*p == '$') { \
     tmp_v = GETPOSARG(n); \
@@ -260,22 +259,11 @@ check_name_arg(mrb_state *mrb, int posarg, const char *name, size_t len)
 static const char *
 get_num(mrb_state *mrb, const char *p, const char *end, int *valp)
 {
-  mrb_int next_n = (int)*valp;
-  for (; p < end && ISDIGIT(*p); p++) {
-    if (mrb_int_mul_overflow(10, next_n, &next_n)) {
-      return NULL;
-    }
-    if (MRB_INT_MAX - (*p - '0') < next_n) {
-      return NULL;
-    }
-    next_n += *p - '0';
-  }
-  if (next_n > INT_MAX || next_n < 0) return NULL;
-  if (p >= end) {
-    mrb_raise(mrb, E_ARGUMENT_ERROR, "malformed format string - %%*[0-9]");
-  }
-  *valp = (int)next_n;
-  return p;
+  char *e;
+  mrb_int n = mrb_int_read(p, end, &e);
+  if (e == NULL || n > INT_MAX) return NULL;
+  *valp = (int)n;
+  return e;
 }
 
 static void
@@ -665,7 +653,6 @@ retry:
 
       case '1': case '2': case '3': case '4':
       case '5': case '6': case '7': case '8': case '9':
-        n = 0;
         GETNUM(n, width);
         if (*p == '$') {
           if (!mrb_undef_p(nextvalue)) {
@@ -722,7 +709,6 @@ retry:
         }
         flags |= FPREC|FPREC0;
 
-        prec = 0;
         p++;
         if (*p == '*') {
           GETASTER(prec);
@@ -732,7 +718,6 @@ retry:
           p++;
           goto retry;
         }
-
         GETNUM(prec, precision);
         goto retry;
 

src/readint.c

+#include <mruby.h>
+#include <mruby/numeric.h>
+#include <errno.h>
+
+/* mrb_int_read(): read mrb_int from a string (base 10 only) */
+/* const char *p - string to read                            */
+/* const char *e - end of string                             */
+/* char **endp   - end of pased integer                      */
+
+/* if integer overflows, errno will be set to ERANGE         */
+/* also endp will be set to NULL on overflow                 */
+MRB_API mrb_int
+mrb_int_read(const char *p, const char *e, char **endp)
+{
+  mrb_int n = 0;
+  int ch;
+
+  while ((e == NULL || p < e) && ISDIGIT(*p)) {
+    ch = *p - '0';
+    if (mrb_int_mul_overflow(n, 10, &n) ||
+        mrb_int_add_overflow(n, ch, &n)) {
+      if (endp) *endp = NULL;
+      errno = ERANGE;
+      return MRB_INT_MAX;
+    }
+    p++;
+  }
+  if (endp) *endp = (char*)p;
+  return n;
+}