Merge pull request #4532 from shuujii/fix-String-byteslice-with-MRB_UTF8_STRING-and-some-edge-cases

Fix `String#byteslice` with `MRB_UTF8_STRING` and some edge cases

include/mruby/string.h

@@ -438,6 +438,9 @@ mrb_value mrb_str_inspect(mrb_state *mrb, mrb_value str);
 #define mrb_str_buf_cat(mrb, str, ptr, len) mrb_str_cat(mrb, str, ptr, len)
 #define mrb_str_buf_append(mrb, str, str2) mrb_str_cat_str(mrb, str, str2)
 
+mrb_bool mrb_str_beg_len(mrb_int str_len, mrb_int *begp, mrb_int *lenp);
+mrb_value mrb_str_byte_subseq(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len);
+
 #ifdef MRB_UTF8_STRING
 mrb_int mrb_utf8_len(const char *str, mrb_int byte_len);
 #endif

mrbgems/mruby-string-ext/src/string.c

@@ -42,44 +42,32 @@ mrb_str_setbyte(mrb_state *mrb, mrb_value str)
 static mrb_value
 mrb_str_byteslice(mrb_state *mrb, mrb_value str)
 {
-  mrb_value a1;
-  mrb_int len;
-
-  if (mrb_get_argc(mrb) == 2) {
-    mrb_int pos;
-    mrb_get_args(mrb, "ii", &pos, &len);
-    return mrb_str_substr(mrb, str, pos, len);
+  mrb_value a1, a2;
+  mrb_int str_len = RSTRING_LEN(str), beg, len;
+  mrb_bool empty = TRUE;
+
+  if (mrb_get_args(mrb, "o|o", &a1, &a2) == 2) {
+    beg = mrb_fixnum(mrb_to_int(mrb, a1));
+    len = mrb_fixnum(mrb_to_int(mrb, a2));
+    goto subseq;
   }
-  mrb_get_args(mrb, "o|i", &a1, &len);
-  switch (mrb_type(a1)) {
-  case MRB_TT_RANGE:
-    {
-      mrb_int beg;
-
-      len = RSTRING_LEN(str);
-      switch (mrb_range_beg_len(mrb, a1, &beg, &len, len, TRUE)) {
-      case MRB_RANGE_TYPE_MISMATCH:
-        break;
-      case MRB_RANGE_OK:
-        return mrb_str_substr(mrb, str, beg, len);
-      case MRB_RANGE_OUT:
-        mrb_raisef(mrb, E_RANGE_ERROR, "%S out of range", a1);
-        break;
-      }
-      return mrb_nil_value();
+  if (mrb_type(a1) == MRB_TT_RANGE) {
+    if (mrb_range_beg_len(mrb, a1, &beg, &len, str_len, TRUE) == MRB_RANGE_OK) {
+      goto subseq;
     }
-#ifndef MRB_WITHOUT_FLOAT
-  case MRB_TT_FLOAT:
-    a1 = mrb_fixnum_value((mrb_int)mrb_float(a1));
-    /* fall through */
-#endif
-  case MRB_TT_FIXNUM:
-    return mrb_str_substr(mrb, str, mrb_fixnum(a1), 1);
-  default:
-    mrb_raise(mrb, E_TYPE_ERROR, "wrong type of argument");
+    return mrb_nil_value();
+  }
+
+  beg = mrb_fixnum(mrb_to_int(mrb, a1));
+  len = 1;
+  empty = FALSE;
+subseq:
+  if (mrb_str_beg_len(str_len, &beg, &len) && (empty || len != 0)) {
+    return mrb_str_byte_subseq(mrb, str, beg, len);
+  }
+  else {
+    return mrb_nil_value();
   }
-  /* not reached */
-  return mrb_nil_value();
 }
 
 /*

mrbgems/mruby-string-ext/test/string.rb

@@ -26,10 +26,61 @@ end
 
 assert('String#byteslice') do
   str1 = "hello"
+  str2 = "\u3042ab"  # "\xE3\x81\x82ab"
+
+  assert_equal("h", str1.byteslice(0))
   assert_equal("e", str1.byteslice(1))
+  assert_equal(nil, str1.byteslice(5))
   assert_equal("o", str1.byteslice(-1))
+  assert_equal(nil, str1.byteslice(-6))
+  assert_equal("\xE3", str2.byteslice(0))
+  assert_equal("\x81", str2.byteslice(1))
+  assert_equal(nil, str2.byteslice(5))
+  assert_equal("b", str2.byteslice(-1))
+  assert_equal(nil, str2.byteslice(-6))
+
+  assert_equal("", str1.byteslice(0, 0))
+  assert_equal(str1, str1.byteslice(0, 6))
+  assert_equal("el", str1.byteslice(1, 2))
+  assert_equal("", str1.byteslice(5, 1))
+  assert_equal("o", str1.byteslice(-1, 6))
+  assert_equal(nil, str1.byteslice(-6, 1))
+  assert_equal(nil, str1.byteslice(0, -1))
+  assert_equal("", str2.byteslice(0, 0))
+  assert_equal(str2, str2.byteslice(0, 6))
+  assert_equal("\x81\x82", str2.byteslice(1, 2))
+  assert_equal("", str2.byteslice(5, 1))
+  assert_equal("b", str2.byteslice(-1, 6))
+  assert_equal(nil, str2.byteslice(-6, 1))
+  assert_equal(nil, str2.byteslice(0, -1))
+
   assert_equal("ell", str1.byteslice(1..3))
   assert_equal("el", str1.byteslice(1...3))
+  assert_equal("h", str1.byteslice(0..0))
+  assert_equal("", str1.byteslice(5..0))
+  assert_equal("o", str1.byteslice(4..5))
+  assert_equal(nil, str1.byteslice(6..0))
+  assert_equal("", str1.byteslice(-1..0))
+  assert_equal("llo", str1.byteslice(-3..5))
+  assert_equal("\x81\x82a", str2.byteslice(1..3))
+  assert_equal("\x81\x82", str2.byteslice(1...3))
+  assert_equal("\xE3", str2.byteslice(0..0))
+  assert_equal("", str2.byteslice(5..0))
+  assert_equal("b", str2.byteslice(4..5))
+  assert_equal(nil, str2.byteslice(6..0))
+  assert_equal("", str2.byteslice(-1..0))
+  assert_equal("\x82ab", str2.byteslice(-3..5))
+
+  assert_raise(ArgumentError) { str1.byteslice }
+  assert_raise(ArgumentError) { str1.byteslice(1, 2, 3) }
+  assert_raise(TypeError) { str1.byteslice("1") }
+  assert_raise(TypeError) { str1.byteslice("1", 2) }
+  assert_raise(TypeError) { str1.byteslice(1, "2") }
+  assert_raise(TypeError) { str1.byteslice(1..2, 3) }
+
+  skip unless Object.const_defined?(:Float)
+  assert_equal("o", str1.byteslice(4.0))
+  assert_equal("\x82ab", str2.byteslice(2.0, 3.0))
 end
 
 assert('String#dump') do

src/string.c

@@ -410,8 +410,8 @@ str_make_shared(mrb_state *mrb, struct RString *orig, struct RString *s)
   }
 }
 
-static mrb_value
-byte_subseq(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len)
+mrb_value
+mrb_str_byte_subseq(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len)
 {
   struct RString *orig, *s;
 
@@ -434,32 +434,33 @@ str_subseq(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len)
   beg = chars2bytes(str, 0, beg);
   len = chars2bytes(str, beg, len);
 
-  return byte_subseq(mrb, str, beg, len);
+  return mrb_str_byte_subseq(mrb, str, beg, len);
 }
 #else
-#define str_subseq(mrb, str, beg, len) byte_subseq(mrb, str, beg, len)
+#define str_subseq(mrb, str, beg, len) mrb_str_byte_subseq(mrb, str, beg, len)
 #endif
 
-static mrb_value
-str_substr(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len)
+mrb_bool
+mrb_str_beg_len(mrb_int str_len, mrb_int *begp, mrb_int *lenp)
 {
-  mrb_int clen = RSTRING_CHAR_LEN(str);
-
-  if (len < 0) return mrb_nil_value();
-  if (clen == 0) {
-    len = 0;
+  if (str_len < *begp || *lenp < 0) return FALSE;
+  if (*begp < 0) {
+    *begp += str_len;
+    if (*begp < 0) return FALSE;
   }
-  if (beg > clen) return mrb_nil_value();
-  if (beg < 0) {
-    beg += clen;
-    if (beg < 0) return mrb_nil_value();
+  if (*lenp > str_len - *begp)
+    *lenp = str_len - *begp;
+  if (*lenp <= 0) {
+    *lenp = 0;
   }
-  if (len > clen - beg)
-    len = clen - beg;
-  if (len <= 0) {
-    len = 0;
-  }
-  return str_subseq(mrb, str, beg, len);
+  return TRUE;
+}
+
+static mrb_value
+str_substr(mrb_state *mrb, mrb_value str, mrb_int beg, mrb_int len)
+{
+  return mrb_str_beg_len(RSTRING_CHAR_LEN(str), &beg, &len) ?
+    str_subseq(mrb, str, beg, len) : mrb_nil_value();
 }
 
 MRB_API mrb_int
@@ -1919,7 +1920,7 @@ mrb_str_split_m(mrb_state *mrb, mrb_value str)
         }
       }
       else if (ISSPACE(c)) {
-        mrb_ary_push(mrb, result, byte_subseq(mrb, str, beg, end-beg));
+        mrb_ary_push(mrb, result, mrb_str_byte_subseq(mrb, str, beg, end-beg));
         mrb_gc_arena_restore(mrb, ai);
         skip = TRUE;
         beg = idx;
@@ -1944,7 +1945,7 @@ mrb_str_split_m(mrb_state *mrb, mrb_value str)
       else {
         end = chars2bytes(str, idx, 1);
       }
-      mrb_ary_push(mrb, result, byte_subseq(mrb, str, idx, end));
+      mrb_ary_push(mrb, result, mrb_str_byte_subseq(mrb, str, idx, end));
       mrb_gc_arena_restore(mrb, ai);
       idx += end + pat_len;
       if (lim_p && lim <= ++i) break;
@@ -1956,7 +1957,7 @@ mrb_str_split_m(mrb_state *mrb, mrb_value str)
       tmp = mrb_str_new_empty(mrb, str);
     }
     else {
-      tmp = byte_subseq(mrb, str, beg, RSTRING_LEN(str)-beg);
+      tmp = mrb_str_byte_subseq(mrb, str, beg, RSTRING_LEN(str)-beg);
     }
     mrb_ary_push(mrb, result, tmp);
   }