oj 3.16.3 → 3.17.4

data/ext/oj/dump.h

@@ -7,12 +7,17 @@
 #include <ruby.h>
 
 #include "oj.h"
+#include "simd.h"
 
 #define MAX_DEPTH 1000
 
 // Extra padding at end of buffer.
 #define BUFFER_EXTRA 64
 
+#ifdef HAVE_SIMD_NEON
+extern void initialize_neon(void);
+#endif /* HAVE_SIMD_NEON */
+
 extern void oj_dump_nil(VALUE obj, int depth, Out out, bool as_ok);
 extern void oj_dump_true(VALUE obj, int depth, Out out, bool as_ok);
 extern void oj_dump_false(VALUE obj, int depth, Out out, bool as_ok);
@@ -30,7 +35,7 @@ extern void oj_dump_xml_time(VALUE obj, Out out);
 extern void oj_dump_time(VALUE obj, Out out, int withZone);
 extern void oj_dump_obj_to_s(VALUE obj, Out out);
 
-extern const char *oj_nan_str(VALUE obj, int opt, int mode, bool plus, int *lenp);
+extern const char *oj_nan_str(VALUE obj, int opt, int mode, bool plus, size_t *lenp);
 
 // initialize an out buffer with the provided stack allocated memory
 extern void oj_out_init(Out out);
@@ -53,10 +58,12 @@ extern void oj_dump_raw_json(VALUE obj, int depth, Out out);
 extern VALUE oj_add_to_json(int argc, VALUE *argv, VALUE self);
 extern VALUE oj_remove_to_json(int argc, VALUE *argv, VALUE self);
 
-extern int oj_dump_float_printf(char *buf, size_t blen, VALUE obj, double d, const char *format);
+extern size_t oj_dump_float_printf(char *buf, size_t blen, VALUE obj, double d, const char *format);
 
 extern time_t oj_sec_from_time_hard_way(VALUE obj);
 
+extern bool oj_key_skip(VALUE key, const char *only, const char *except);
+
 inline static void assure_size(Out out, size_t len) {
     if (out->end - out->cur <= (long)len) {
         oj_grow_out(out, len);

data/ext/oj/dump_compat.c

@@ -103,7 +103,7 @@ static void dump_values_array(VALUE *values, int depth, Out out) {
 static void dump_to_json(VALUE obj, Out out) {
     volatile VALUE rs;
     const char    *s;
-    int            len;
+    size_t         len;
 
     TRACE(out->opts->trace, "to_json", obj, 0, TraceRubyIn);
     if (0 == rb_obj_method_arity(obj, oj_to_json_id)) {
@@ -115,7 +115,7 @@ static void dump_to_json(VALUE obj, Out out) {
 
     StringValue(rs);
     s   = RSTRING_PTR(rs);
-    len = (int)RSTRING_LEN(rs);
+    len = RSTRING_LEN(rs);
 
     assure_size(out, len + 1);
     APPEND_CHARS(out->cur, s, len);
@@ -124,7 +124,8 @@ static void dump_to_json(VALUE obj, Out out) {
 
 static void dump_array(VALUE a, int depth, Out out, bool as_ok) {
     size_t size;
-    int    i, cnt;
+    size_t i;
+    size_t cnt;
     int    d2 = depth + 1;
     long   id = oj_check_circular(a, out);
 
@@ -136,7 +137,7 @@ static void dump_array(VALUE a, int depth, Out out, bool as_ok) {
         dump_to_json(a, out);
         return;
     }
-    cnt         = (int)RARRAY_LEN(a);
+    cnt         = RARRAY_LEN(a);
     *out->cur++ = '[';
     assure_size(out, 2);
     if (0 == cnt) {
@@ -147,10 +148,10 @@ static void dump_array(VALUE a, int depth, Out out, bool as_ok) {
         } else {
             size = d2 * out->indent + 2;
         }
-        assure_size(out, size * cnt);
         cnt--;
         for (i = 0; i <= cnt; i++) {
             if (out->opts->dump_opts.use) {
+                assure_size(out, size);
                 if (0 < out->opts->dump_opts.array_size) {
                     APPEND_CHARS(out->cur, out->opts->dump_opts.array_nl, out->opts->dump_opts.array_size);
                 }
@@ -161,6 +162,7 @@ static void dump_array(VALUE a, int depth, Out out, bool as_ok) {
                     }
                 }
             } else {
+                assure_size(out, size);
                 fill_indent(out, d2);
             }
             oj_dump_compat_val(RARRAY_AREF(a, i), d2, out, true);
@@ -552,7 +554,7 @@ static void dump_float(VALUE obj, int depth, Out out, bool as_ok) {
     char   buf[64];
     char  *b;
     double d   = rb_num2dbl(obj);
-    int    cnt = 0;
+    size_t cnt = 0;
 
     if (0.0 == d) {
         b    = buf;
@@ -590,7 +592,7 @@ static void dump_float(VALUE obj, int depth, Out out, bool as_ok) {
         volatile VALUE rstr = oj_safe_string_convert(obj);
 
         strcpy(buf, RSTRING_PTR(rstr));
-        cnt = (int)RSTRING_LEN(rstr);
+        cnt = RSTRING_LEN(rstr);
     }
     assure_size(out, cnt);
     APPEND_CHARS(out->cur, buf, cnt);
@@ -604,6 +606,11 @@ static int hash_cb(VALUE key, VALUE value, VALUE ov) {
     if (out->omit_nil && Qnil == value) {
         return ST_CONTINUE;
     }
+    if (NULL != out->opts->dump_opts.only || NULL != out->opts->dump_opts.except) {
+        if (oj_key_skip(key, out->opts->dump_opts.only, out->opts->dump_opts.except)) {
+            return ST_CONTINUE;
+        }
+    }
     if (!out->opts->dump_opts.use) {
         assure_size(out, depth * out->indent + 1);
         fill_indent(out, depth);
@@ -800,7 +807,7 @@ static void dump_bignum(VALUE obj, int depth, Out out, bool as_ok) {
     // rb_big2str can not so unless overridden by using add_to_json(Integer)
     // this must use to_s to pass the json gem unit tests.
     volatile VALUE rs;
-    int            cnt;
+    size_t         cnt;
     bool           dump_as_string = false;
 
     if (use_bignum_alt) {
@@ -809,7 +816,7 @@ static void dump_bignum(VALUE obj, int depth, Out out, bool as_ok) {
         rs = oj_safe_string_convert(obj);
     }
     rb_check_type(rs, T_STRING);
-    cnt = (int)RSTRING_LEN(rs);
+    cnt = RSTRING_LEN(rs);
 
     if (out->opts->int_range_min != 0 || out->opts->int_range_max != 0) {
         dump_as_string = true;  // Bignum cannot be inside of Fixnum range

data/ext/oj/dump_leaf.c

@@ -17,7 +17,7 @@ inline static void dump_chars(const char *s, size_t size, Out out) {
 static void dump_leaf_str(Leaf leaf, Out out) {
     switch (leaf->value_type) {
     case STR_VAL: oj_dump_cstr(leaf->str, strlen(leaf->str), 0, 0, out); break;
-    case RUBY_VAL: oj_dump_cstr(StringValueCStr(leaf->value), (int)RSTRING_LEN(leaf->value), 0, 0, out); break;
+    case RUBY_VAL: oj_dump_cstr(StringValueCStr(leaf->value), RSTRING_LEN(leaf->value), 0, 0, out); break;
     case COL_VAL:
     default: rb_raise(rb_eTypeError, "Unexpected value type %02x.\n", leaf->value_type); break;
     }

data/ext/oj/dump_object.c

@@ -10,6 +10,24 @@ static const char hex_chars[17] = "0123456789abcdef";
 
 static void dump_obj_attrs(VALUE obj, VALUE clas, slot_t id, int depth, Out out);
 
+// An odd class is dumped as {"^O":class,...} and is rebuilt with a create
+// function when loaded so there is no place to put a "^i" circular id on it.
+// Registering it in the circular cache would consume an id that is never
+// written and any later occurrence would then be dumped as a "^r" that refers
+// to nothing. Instead dump odd classes in full each time they are encountered
+// just like Time, Rational, and Complex are.
+static void dump_circular_obj(VALUE obj, VALUE clas, int depth, Out out) {
+    if (NULL != oj_get_odd(clas)) {
+        dump_obj_attrs(obj, clas, 0, depth, out);
+    } else {
+        long id = oj_check_circular(obj, out);
+
+        if (0 <= id) {
+            dump_obj_attrs(obj, clas, id, depth, out);
+        }
+    }
+}
+
 static void dump_time(VALUE obj, Out out) {
     switch (out->opts->time_format) {
     case RubyTime:
@@ -33,7 +51,7 @@ static void dump_data(VALUE obj, int depth, Out out, bool as_ok) {
         if (oj_bigdecimal_class == clas) {
             volatile VALUE rstr = oj_safe_string_convert(obj);
             const char    *str  = RSTRING_PTR(rstr);
-            int            len  = (int)RSTRING_LEN(rstr);
+            size_t         len  = RSTRING_LEN(rstr);
 
             if (No != out->opts->bigdec_as_num) {
                 oj_dump_raw(str, len, out);
@@ -47,11 +65,7 @@ static void dump_data(VALUE obj, int depth, Out out, bool as_ok) {
                 oj_dump_cstr(str, len, 0, 0, out);
             }
         } else {
-            long id = oj_check_circular(obj, out);
-
-            if (0 <= id) {
-                dump_obj_attrs(obj, clas, id, depth, out);
-            }
+            dump_circular_obj(obj, clas, depth, out);
         }
     }
 }
@@ -62,7 +76,7 @@ static void dump_obj(VALUE obj, int depth, Out out, bool as_ok) {
     if (oj_bigdecimal_class == clas) {
         volatile VALUE rstr = oj_safe_string_convert(obj);
         const char    *str  = RSTRING_PTR(rstr);
-        int            len  = (int)RSTRING_LEN(rstr);
+        size_t         len  = RSTRING_LEN(rstr);
 
         if (0 == strcasecmp("Infinity", str)) {
             str = oj_nan_str(obj, out->opts->dump_opts.nan_dump, out->opts->mode, true, &len);
@@ -74,11 +88,7 @@ static void dump_obj(VALUE obj, int depth, Out out, bool as_ok) {
             oj_dump_raw(str, len, out);
         }
     } else {
-        long id = oj_check_circular(obj, out);
-
-        if (0 <= id) {
-            dump_obj_attrs(obj, clas, id, depth, out);
-        }
+        dump_circular_obj(obj, clas, depth, out);
     }
 }
 
@@ -95,7 +105,8 @@ static void dump_class(VALUE obj, int depth, Out out, bool as_ok) {
 
 static void dump_array_class(VALUE a, VALUE clas, int depth, Out out) {
     size_t size;
-    int    i, cnt;
+    size_t i;
+    size_t cnt;
     int    d2 = depth + 1;
     long   id = oj_check_circular(a, out);
 
@@ -106,7 +117,7 @@ static void dump_array_class(VALUE a, VALUE clas, int depth, Out out) {
         dump_obj_attrs(a, clas, 0, depth, out);
         return;
     }
-    cnt         = (int)RARRAY_LEN(a);
+    cnt         = RARRAY_LEN(a);
     *out->cur++ = '[';
     if (0 < id) {
         assure_size(out, d2 * out->indent + 16);
@@ -181,7 +192,7 @@ static void dump_str_class(VALUE obj, VALUE clas, int depth, Out out) {
         dump_obj_attrs(obj, clas, 0, depth, out);
     } else {
         const char *s   = RSTRING_PTR(obj);
-        size_t      len = (int)RSTRING_LEN(obj);
+        size_t      len = RSTRING_LEN(obj);
         char        s1  = s[1];
 
         oj_dump_cstr(s, len, 0, (':' == *s || ('^' == *s && ('r' == s1 || 'i' == s1))), out);
@@ -195,7 +206,7 @@ static void dump_str(VALUE obj, int depth, Out out, bool as_ok) {
 static void dump_sym(VALUE obj, int depth, Out out, bool as_ok) {
     volatile VALUE s = rb_sym2str(obj);
 
-    oj_dump_cstr(RSTRING_PTR(s), (int)RSTRING_LEN(s), 1, 0, out);
+    oj_dump_cstr(RSTRING_PTR(s), RSTRING_LEN(s), 1, 0, out);
 }
 
 static int hash_cb(VALUE key, VALUE value, VALUE ov) {
@@ -209,6 +220,11 @@ static int hash_cb(VALUE key, VALUE value, VALUE ov) {
     if (out->omit_nil && Qnil == value) {
         return ST_CONTINUE;
     }
+    if (NULL != out->opts->dump_opts.only || NULL != out->opts->dump_opts.except) {
+        if (oj_key_skip(key, out->opts->dump_opts.only, out->opts->dump_opts.except)) {
+            return ST_CONTINUE;
+        }
+    }
     assure_size(out, size);
     fill_indent(out, depth);
     switch (rb_type(key)) {
@@ -389,7 +405,7 @@ static void dump_odd(VALUE obj, Odd odd, VALUE clas, int depth, Out out) {
             rb_raise(rb_eEncodingError, "Invalid type for raw JSON.");
         } else {
             const char *s    = RSTRING_PTR(v);
-            int         len  = (int)RSTRING_LEN(v);
+            size_t      len  = RSTRING_LEN(v);
             const char *name = rb_id2name(*odd->attrs);
             size_t      nlen = strlen(name);
 
@@ -489,7 +505,7 @@ static void dump_obj_attrs(VALUE obj, VALUE clas, slot_t id, int depth, Out out)
         *out->cur++ = ',';
         fill_indent(out, d2);
         APPEND_CHARS(out->cur, "\"self\":", 7);
-        oj_dump_cstr(RSTRING_PTR(obj), (int)RSTRING_LEN(obj), 0, 0, out);
+        oj_dump_cstr(RSTRING_PTR(obj), RSTRING_LEN(obj), 0, 0, out);
         break;
     case T_ARRAY:
         assure_size(out, d2 * out->indent + 14);
@@ -564,11 +580,12 @@ static void dump_regexp(VALUE obj, int depth, Out out, bool as_ok) {
 static void dump_struct(VALUE obj, int depth, Out out, bool as_ok) {
     VALUE       clas       = rb_obj_class(obj);
     const char *class_name = rb_class2name(clas);
-    int         i;
-    int         d2   = depth + 1;
-    int         d3   = d2 + 1;
-    size_t      len  = strlen(class_name);
-    size_t      size = d2 * out->indent + d3 * out->indent + 10 + len;
+    size_t      i;
+    int         d2       = depth + 1;
+    int         d3       = d2 + 1;
+    size_t      len      = strlen(class_name);
+    size_t      size     = d2 * out->indent + d3 * out->indent + 10 + len;
+    char        circular = out->opts->circular;
 
     assure_size(out, size);
     *out->cur++ = '{';
@@ -577,14 +594,14 @@ static void dump_struct(VALUE obj, int depth, Out out, bool as_ok) {
     if ('#' == *class_name) {
         VALUE       ma = rb_struct_s_members(clas);
         const char *name;
-        int         cnt = (int)RARRAY_LEN(ma);
+        size_t      cnt = RARRAY_LEN(ma);
 
         *out->cur++ = '[';
         for (i = 0; i < cnt; i++) {
             volatile VALUE s = rb_sym2str(RARRAY_AREF(ma, i));
 
             name = RSTRING_PTR(s);
-            len  = (int)RSTRING_LEN(s);
+            len  = RSTRING_LEN(s);
             size = len + 3;
             assure_size(out, size);
             if (0 < i) {
@@ -607,14 +624,18 @@ static void dump_struct(VALUE obj, int depth, Out out, bool as_ok) {
     {
         VALUE v;
         int   cnt;
+
 #if RSTRUCT_LEN_RETURNS_INTEGER_OBJECT
         cnt = (int)NUM2LONG(RSTRUCT_LEN(obj));
 #else   // RSTRUCT_LEN_RETURNS_INTEGER_OBJECT
         cnt = (int)RSTRUCT_LEN(obj);
 #endif  // RSTRUCT_LEN_RETURNS_INTEGER_OBJECT
 
-        for (i = 0; i < cnt; i++) {
-            v = RSTRUCT_GET(obj, i);
+        if (0 == strcmp(class_name, "Range")) {
+            out->opts->circular = 'n';
+        }
+        for (i = 0; i < (size_t)cnt; i++) {
+            v = RSTRUCT_GET(obj, (int)i);
             if (dump_ignore(out->opts, v)) {
                 v = Qnil;
             }
@@ -630,6 +651,9 @@ static void dump_struct(VALUE obj, int depth, Out out, bool as_ok) {
         // class in interpreted Ruby so length() may not be defined.
         int slen = FIX2INT(rb_funcall2(obj, oj_length_id, 0, 0));
 
+        if (0 == strcmp(class_name, "Range")) {
+            out->opts->circular = 'n';
+        }
         for (i = 0; i < slen; i++) {
             assure_size(out, size);
             fill_indent(out, d3);
@@ -641,6 +665,7 @@ static void dump_struct(VALUE obj, int depth, Out out, bool as_ok) {
         }
     }
 #endif
+    out->opts->circular = circular;
     out->cur--;
     APPEND_CHARS(out->cur, "]}", 2);
     *out->cur = '\0';

data/ext/oj/dump_strict.c

@@ -30,7 +30,7 @@ static void dump_float(VALUE obj, int depth, Out out, bool as_ok) {
     char   buf[64];
     char*  b;
     double d   = rb_num2dbl(obj);
-    int    cnt = 0;
+    size_t cnt = 0;
 
     if (0.0 == d) {
         b    = buf;
@@ -92,7 +92,7 @@ static void dump_float(VALUE obj, int depth, Out out, bool as_ok) {
         } else if (0 == out->opts->float_prec) {
             volatile VALUE rstr = oj_safe_string_convert(obj);
 
-            cnt = (int)RSTRING_LEN(rstr);
+            cnt = RSTRING_LEN(rstr);
             if ((int)sizeof(buf) <= cnt) {
                 cnt = sizeof(buf) - 1;
             }
@@ -109,7 +109,8 @@ static void dump_float(VALUE obj, int depth, Out out, bool as_ok) {
 
 static void dump_array(VALUE a, int depth, Out out, bool as_ok) {
     size_t size;
-    int    i, cnt;
+    size_t i;
+    size_t cnt;
     int    d2 = depth + 1;
 
     if (Yes == out->opts->circular) {
@@ -118,7 +119,7 @@ static void dump_array(VALUE a, int depth, Out out, bool as_ok) {
             return;
         }
     }
-    cnt         = (int)RARRAY_LEN(a);
+    cnt         = RARRAY_LEN(a);
     *out->cur++ = '[';
     size        = 2;
     assure_size(out, size);
@@ -192,14 +193,19 @@ static int hash_cb(VALUE key, VALUE value, VALUE ov) {
     if (out->omit_nil && Qnil == value) {
         return ST_CONTINUE;
     }
+    if (NULL != out->opts->dump_opts.only || NULL != out->opts->dump_opts.except) {
+        if (oj_key_skip(key, out->opts->dump_opts.only, out->opts->dump_opts.except)) {
+            return ST_CONTINUE;
+        }
+    }
     if (!out->opts->dump_opts.use) {
         size = depth * out->indent + 1;
         assure_size(out, size);
         fill_indent(out, depth);
-        if (rtype == T_STRING) {
-            oj_dump_str(key, 0, out, false);
-        } else {
-            oj_dump_sym(key, 0, out, false);
+        switch (rtype) {
+        case T_STRING: oj_dump_str(key, 0, out, false); break;
+        case T_SYMBOL: oj_dump_sym(key, 0, out, false); break;
+        default: oj_dump_str(oj_safe_string_convert(key), 0, out, false); break;
         }
         *out->cur++ = ':';
     } else {
@@ -214,10 +220,10 @@ static int hash_cb(VALUE key, VALUE value, VALUE ov) {
                 APPEND_CHARS(out->cur, out->opts->dump_opts.indent_str, out->opts->dump_opts.indent_size);
             }
         }
-        if (rtype == T_STRING) {
-            oj_dump_str(key, 0, out, false);
-        } else {
-            oj_dump_sym(key, 0, out, false);
+        switch (rtype) {
+        case T_STRING: oj_dump_str(key, 0, out, false); break;
+        case T_SYMBOL: oj_dump_sym(key, 0, out, false); break;
+        default: oj_dump_str(oj_safe_string_convert(key), 0, out, false); break;
         }
         size = out->opts->dump_opts.before_size + out->opts->dump_opts.after_size + 2;
         assure_size(out, size);
@@ -290,7 +296,7 @@ static void dump_data_strict(VALUE obj, int depth, Out out, bool as_ok) {
     if (oj_bigdecimal_class == clas) {
         volatile VALUE rstr = oj_safe_string_convert(obj);
 
-        oj_dump_raw(RSTRING_PTR(rstr), (int)RSTRING_LEN(rstr), out);
+        oj_dump_raw(RSTRING_PTR(rstr), RSTRING_LEN(rstr), out);
     } else {
         raise_strict(obj);
     }
@@ -302,7 +308,7 @@ static void dump_data_null(VALUE obj, int depth, Out out, bool as_ok) {
     if (oj_bigdecimal_class == clas) {
         volatile VALUE rstr = oj_safe_string_convert(obj);
 
-        oj_dump_raw(RSTRING_PTR(rstr), (int)RSTRING_LEN(rstr), out);
+        oj_dump_raw(RSTRING_PTR(rstr), RSTRING_LEN(rstr), out);
     } else {
         oj_dump_nil(Qnil, depth, out, false);
     }

data/ext/oj/extconf.rb

@@ -29,20 +29,22 @@ dflags = {
 have_func('rb_gc_mark_movable')
 have_func('stpcpy')
 have_func('pthread_mutex_init')
+have_func('getrlimit', 'sys/resource.h')
 have_func('rb_enc_interned_str')
 have_func('rb_ext_ractor_safe', 'ruby.h')
-# rb_hash_bulk_insert is deep down in a header not included in normal build and that seems to fool have_func.
-have_func('rb_hash_bulk_insert', 'ruby.h') unless '2' == version[0] && '6' == version[1]
 
 dflags['OJ_DEBUG'] = true unless ENV['OJ_DEBUG'].nil?
 
-if with_config('--with-sse42')
-  if try_cflags('-msse4.2')
-    $CPPFLAGS += ' -msse4.2'
-    dflags['OJ_USE_SSE4_2'] = 1
-  else
-    warn 'SSE 4.2 is not supported on this platform.'
-  end
+# SIMD optimizations use runtime CPU detection and function-level target attributes
+# We do NOT add global -msse4.2/-msse2 flags here because:
+# 1. It would cause illegal instruction errors on CPUs without SSE4.2
+# 2. The code uses __attribute__((target("sse4.2"))) for SSE4.2 functions
+# 3. Runtime detection in oj_get_simd_implementation() selects the right path
+#
+# We only add -msse2 if available, since SSE2 is baseline for all x86_64 CPUs
+# and needed for compiling the SSE2 fallback code on 32-bit x86
+if try_cflags('-msse2')
+  $CPPFLAGS += ' -msse2'
 end
 
 if enable_config('trace-log', false)