oj 3.11.5 → 3.16.5

data/ext/oj/object.c

@@ -7,7 +7,7 @@
 
 #include "encode.h"
 #include "err.h"
-#include "hash.h"
+#include "intern.h"
 #include "odd.h"
 #include "oj.h"
 #include "parse.h"
@@ -32,16 +32,17 @@ static VALUE calc_hash_key(ParseInfo pi, Val kval, char k1) {
     volatile VALUE rkey;
 
     if (':' == k1) {
-        rkey = rb_str_new(kval->key + 1, kval->klen - 1);
-        rkey = oj_encode(rkey);
-        rkey = rb_funcall(rkey, oj_to_sym_id, 0);
-    } else {
-        rkey = rb_str_new(kval->key, kval->klen);
-        rkey = oj_encode(rkey);
-        if (Yes == pi->options.sym_key) {
-            rkey = rb_str_intern(rkey);
-        }
+        return ID2SYM(rb_intern3(kval->key + 1, kval->klen - 1, oj_utf8_encoding));
+    }
+    if (Yes == pi->options.sym_key) {
+        return ID2SYM(rb_intern3(kval->key, kval->klen, oj_utf8_encoding));
     }
+#if HAVE_RB_ENC_INTERNED_STR
+    rkey = rb_enc_interned_str(kval->key, kval->klen, oj_utf8_encoding);
+#else
+    rkey = rb_utf8_str_new(kval->key, kval->klen);
+    OBJ_FREEZE(rkey);
+#endif
     return rkey;
 }
 
@@ -49,9 +50,7 @@ static VALUE str_to_value(ParseInfo pi, const char *str, size_t len, const char
     volatile VALUE rstr = Qnil;
 
     if (':' == *orig && 0 < len) {
-        rstr = rb_str_new(str + 1, len - 1);
-        rstr = oj_encode(rstr);
-        rstr = rb_funcall(rstr, oj_to_sym_id, 0);
+        rstr = ID2SYM(rb_intern3(str + 1, len - 1, oj_utf8_encoding));
     } else if (pi->circ_array && 3 <= len && '^' == *orig && 'r' == orig[1]) {
         long i = read_long(str + 2, len - 2);
 
@@ -61,17 +60,11 @@ static VALUE str_to_value(ParseInfo pi, const char *str, size_t len, const char
         }
         rstr = oj_circ_array_get(pi->circ_array, i);
     } else {
-        rstr = rb_str_new(str, len);
-        rstr = oj_encode(rstr);
+        rstr = rb_utf8_str_new(str, len);
     }
     return rstr;
 }
 
-#if (RUBY_VERSION_MAJOR == 1 && RUBY_VERSION_MINOR == 8)
-static VALUE oj_parse_xml_time(const char *str, int len) {
-    return rb_funcall(rb_cTime, oj_parse_id, 1, rb_str_new(str, len));
-}
-#else
 // The much faster approach (4x faster)
 static int parse_num(const char *str, const char *end, int cnt) {
     int  n = 0;
@@ -90,8 +83,9 @@ static int parse_num(const char *str, const char *end, int cnt) {
 
 VALUE
 oj_parse_xml_time(const char *str, int len) {
-    VALUE       args[8];
-    const char *end = str + len;
+    VALUE       args[7];
+    const char *end  = str + len;
+    const char *orig = str;
     int         n;
 
     // year
@@ -151,7 +145,9 @@ oj_parse_xml_time(const char *str, int len) {
         char c = *str++;
 
         if ('.' == c) {
-            long long nsec = 0;
+            unsigned long long       num            = 0;
+            unsigned long long       den            = 1;
+            const unsigned long long last_den_limit = ULLONG_MAX / 10;
 
             for (; str < end; str++) {
                 c = *str;
@@ -159,9 +155,14 @@ oj_parse_xml_time(const char *str, int len) {
                     str++;
                     break;
                 }
-                nsec = nsec * 10 + (c - '0');
+                if (den > last_den_limit) {
+                    // bail to Time.parse if there are more fractional digits than a ULLONG rational can hold
+                    return rb_funcall(rb_cTime, oj_parse_id, 1, rb_str_new(orig, len));
+                }
+                num = num * 10 + (c - '0');
+                den *= 10;
             }
-            args[5] = rb_float_new((double)n + ((double)nsec + 0.5) / 1000000000.0);
+            args[5] = rb_funcall(INT2NUM(n), oj_plus_id, 1, rb_rational_new(ULL2NUM(num), ULL2NUM(den)));
         } else {
             args[5] = rb_ll2inum(n);
         }
@@ -203,7 +204,6 @@ oj_parse_xml_time(const char *str, int len) {
     }
     return rb_funcall2(rb_cTime, oj_new_id, 7, args);
 }
-#endif
 
 static int hat_cstr(ParseInfo pi, Val parent, Val kval, const char *str, size_t len) {
     const char *key  = kval->key;
@@ -228,16 +228,10 @@ static int hat_cstr(ParseInfo pi, Val parent, Val kval, const char *str, size_t
             }
             parent->val      = odd->clas;
             parent->odd_args = oj_odd_alloc_args(odd);
-        } break;
-        case 'm':
-            parent->val = rb_str_new(str + 1, len - 1);
-            parent->val = oj_encode(parent->val);
-            parent->val = rb_funcall(parent->val, oj_to_sym_id, 0);
-            break;
-        case 's':
-            parent->val = rb_str_new(str, len);
-            parent->val = oj_encode(parent->val);
             break;
+        }
+        case 'm': parent->val = ID2SYM(rb_intern3(str + 1, len - 1, oj_utf8_encoding)); break;
+        case 's': parent->val = rb_utf8_str_new(str, len); break;
         case 'c':  // class
         {
             VALUE clas = oj_name2class(pi, str, len, Yes == pi->options.auto_define, rb_eArgError);
@@ -247,7 +241,8 @@ static int hat_cstr(ParseInfo pi, Val parent, Val kval, const char *str, size_t
             } else {
                 parent->val = clas;
             }
-        } break;
+            break;
+        }
         case 't':  // time
             parent->val = oj_parse_xml_time(str, (int)len);
             break;
@@ -277,32 +272,22 @@ static int hat_num(ParseInfo pi, Val parent, Val kval, NumInfo ni) {
                 }
                 if (86400 == ni->exp) {  // UTC time
                     parent->val = rb_time_nano_new(ni->i, (long)nsec);
-                    // Since the ruby C routines alway create local time, the
+                    // Since the ruby C routines always create local time, the
                     // offset and then a conversion to UTC keeps makes the time
                     // match the expected value.
                     parent->val = rb_funcall2(parent->val, oj_utc_id, 0, 0);
                 } else if (ni->has_exp) {
-                    int64_t          t = (int64_t)(ni->i + ni->exp);
-                    struct _timeInfo ti;
-                    VALUE            args[8];
-
-                    sec_as_time(t, &ti);
-                    args[0] = LONG2NUM((long)(ti.year));
-                    args[1] = LONG2NUM(ti.mon);
-                    args[2] = LONG2NUM(ti.day);
-                    args[3] = LONG2NUM(ti.hour);
-                    args[4] = LONG2NUM(ti.min);
-                    args[5] = rb_float_new((double)ti.sec + ((double)nsec + 0.5) / 1000000000.0);
-                    args[6] = LONG2NUM(ni->exp);
-                    parent->val = rb_funcall2(rb_cTime, oj_new_id, 7, args);
+                    struct timespec ts;
+                    ts.tv_sec   = ni->i;
+                    ts.tv_nsec  = nsec;
+                    parent->val = rb_time_timespec_new(&ts, (int)ni->exp);
                 } else {
                     parent->val = rb_time_nano_new(ni->i, (long)nsec);
                 }
             }
             break;
-        case 'i':  // circular index
-            if (!ni->infinity && !ni->neg && 1 == ni->div && 0 == ni->exp &&
-                0 != pi->circ_array) {  // fixnum
+        case 'i':                                                                                    // circular index
+            if (!ni->infinity && !ni->neg && 1 == ni->div && 0 == ni->exp && 0 != pi->circ_array) {  // fixnum
                 if (Qnil == parent->val) {
                     parent->val = rb_hash_new();
                 }
@@ -331,7 +316,7 @@ static int hat_value(ParseInfo pi, Val parent, const char *key, size_t klen, vol
                 oj_set_error_at(pi, oj_parse_error_class, __FILE__, __LINE__, "Invalid struct data");
                 return 1;
             }
-            e1 = *RARRAY_PTR(value);
+            e1 = *RARRAY_CONST_PTR(value);
             // check for anonymous Struct
             if (T_ARRAY == rb_type(e1)) {
                 VALUE          args[1024];
@@ -339,46 +324,50 @@ static int hat_value(ParseInfo pi, Val parent, const char *key, size_t klen, vol
                 int            i, cnt = (int)RARRAY_LEN(e1);
 
                 for (i = 0; i < cnt; i++) {
-                    rstr    = rb_ary_entry(e1, i);
+                    rstr    = RARRAY_AREF(e1, i);
                     args[i] = rb_funcall(rstr, oj_to_sym_id, 0);
                 }
                 sc = rb_funcall2(rb_cStruct, oj_new_id, cnt, args);
             } else {
                 // If struct is not defined then we let this fail and raise an exception.
-                sc = oj_name2struct(pi, *RARRAY_PTR(value), rb_eArgError);
+                sc = oj_name2struct(pi, *RARRAY_CONST_PTR(value), rb_eArgError);
             }
-            // Create a properly initialized struct instance without calling the initialize method.
-            parent->val = rb_obj_alloc(sc);
-            // If the JSON array has more entries than the struct class allows, we record an error.
+            if (sc == rb_cRange) {
+                parent->val = rb_class_new_instance(len - 1, RARRAY_CONST_PTR(value) + 1, rb_cRange);
+            } else {
+                // Create a properly initialized struct instance without calling the initialize method.
+                parent->val = rb_obj_alloc(sc);
+                // If the JSON array has more entries than the struct class allows, we record an error.
 #ifdef RSTRUCT_LEN
 #if RSTRUCT_LEN_RETURNS_INTEGER_OBJECT
-            slen = (int)NUM2LONG(RSTRUCT_LEN(parent->val));
+                slen = (int)NUM2LONG(RSTRUCT_LEN(parent->val));
 #else   // RSTRUCT_LEN_RETURNS_INTEGER_OBJECT
-            slen = (int)RSTRUCT_LEN(parent->val);
+                slen = (int)RSTRUCT_LEN(parent->val);
 #endif  // RSTRUCT_LEN_RETURNS_INTEGER_OBJECT
 #else
-            slen = FIX2INT(rb_funcall2(parent->val, oj_length_id, 0, 0));
+                slen = FIX2INT(rb_funcall2(parent->val, oj_length_id, 0, 0));
 #endif
-            // MRI >= 1.9
-            if (len - 1 > slen) {
-                oj_set_error_at(pi, oj_parse_error_class, __FILE__, __LINE__, "Invalid struct data");
-            } else {
-                int i;
+                // MRI >= 1.9
+                if (len - 1 > slen) {
+                    oj_set_error_at(pi, oj_parse_error_class, __FILE__, __LINE__, "Invalid struct data");
+                } else {
+                    int i;
 
-                for (i = 0; i < len - 1; i++) {
-                    rb_struct_aset(parent->val, INT2FIX(i), RARRAY_PTR(value)[i + 1]);
+                    for (i = 0; i < len - 1; i++) {
+                        rb_struct_aset(parent->val, INT2FIX(i), RARRAY_CONST_PTR(value)[i + 1]);
+                    }
                 }
             }
             return 1;
         } else if (3 <= klen && '#' == key[1]) {
-            volatile VALUE *a;
+            volatile const VALUE *a;
 
             if (2 != len) {
                 oj_set_error_at(pi, oj_parse_error_class, __FILE__, __LINE__, "invalid hash pair");
                 return 1;
             }
             parent->val = rb_hash_new();
-            a           = RARRAY_PTR(value);
+            a           = RARRAY_CONST_PTR(value);
             rb_hash_aset(parent->val, *a, a[1]);
 
             return 1;
@@ -388,55 +377,17 @@ static int hat_value(ParseInfo pi, Val parent, const char *key, size_t klen, vol
 }
 
 void oj_set_obj_ivar(Val parent, Val kval, VALUE value) {
-    const char *key  = kval->key;
-    int         klen = kval->klen;
-    ID          var_id;
-    ID *        slot;
-
-#ifdef HAVE_PTHREAD_MUTEX_INIT
-    pthread_mutex_lock(&oj_cache_mutex);
-#else
-    rb_mutex_lock(oj_cache_mutex);
-#endif
-    if (0 == (var_id = oj_attr_hash_get(key, klen, &slot))) {
-        char attr[256];
-
-        if ((int)sizeof(attr) <= klen + 2) {
-            char *buf = ALLOC_N(char, klen + 2);
-
-            if ('~' == *key) {
-                strncpy(buf, key + 1, klen - 1);
-                buf[klen - 1] = '\0';
-            } else {
-                *buf = '@';
-                strncpy(buf + 1, key, klen);
-                buf[klen + 1] = '\0';
-            }
-            var_id = rb_intern(buf);
-            xfree(buf);
-        } else {
-            if ('~' == *key) {
-                strncpy(attr, key + 1, klen - 1);
-                attr[klen - 1] = '\0';
-            } else {
-                *attr = '@';
-                strncpy(attr + 1, key, klen);
-                attr[klen + 1] = '\0';
-            }
-            var_id = rb_intern(attr);
-        }
-        *slot = var_id;
+    if (kval->klen == 5 && strncmp("~mesg", kval->key, 5) == 0 && rb_obj_is_kind_of(parent->val, rb_eException)) {
+        parent->val = rb_funcall(parent->val, rb_intern("exception"), 1, value);
+    } else if (kval->klen == 3 && strncmp("~bt", kval->key, 3) == 0 && rb_obj_is_kind_of(parent->val, rb_eException)) {
+        rb_funcall(parent->val, rb_intern("set_backtrace"), 1, value);
+    } else {
+        rb_ivar_set(parent->val, oj_attr_intern(kval->key, kval->klen), value);
     }
-#ifdef HAVE_PTHREAD_MUTEX_INIT
-    pthread_mutex_unlock(&oj_cache_mutex);
-#else
-    rb_mutex_unlock(oj_cache_mutex);
-#endif
-    rb_ivar_set(parent->val, var_id, value);
 }
 
 static void hash_set_cstr(ParseInfo pi, Val kval, const char *str, size_t len, const char *orig) {
-    const char *   key    = kval->key;
+    const char    *key    = kval->key;
     int            klen   = kval->klen;
     Val            parent = stack_peek(&pi->stack);
     volatile VALUE rval   = Qnil;
@@ -451,9 +402,7 @@ WHICH_TYPE:
         }
         break;
     case T_HASH:
-        rb_hash_aset(parent->val,
-                     calc_hash_key(pi, kval, parent->k1),
-                     str_to_value(pi, str, len, orig));
+        rb_hash_aset(parent->val, calc_hash_key(pi, kval, parent->k1), str_to_value(pi, str, len, orig));
         break;
     case T_STRING:
         rval = str_to_value(pi, str, len, orig);
@@ -505,13 +454,11 @@ WHICH_TYPE:
                         rb_class2name(rb_obj_class(parent->val)));
         return;
     }
-    if (Yes == pi->options.trace) {
-        oj_trace_parse_call("set_string", pi, __FILE__, __LINE__, rval);
-    }
+    TRACE_PARSE_CALL(pi->options.trace, "set_string", pi, rval);
 }
 
 static void hash_set_num(ParseInfo pi, Val kval, NumInfo ni) {
-    const char *   key    = kval->key;
+    const char    *key    = kval->key;
     int            klen   = kval->klen;
     Val            parent = stack_peek(&pi->stack);
     volatile VALUE rval   = Qnil;
@@ -530,8 +477,8 @@ WHICH_TYPE:
         rb_hash_aset(parent->val, calc_hash_key(pi, kval, parent->k1), rval);
         break;
     case T_OBJECT:
-        if (2 == klen && '^' == *key && 'i' == key[1] && !ni->infinity && !ni->neg &&
-            1 == ni->div && 0 == ni->exp && 0 != pi->circ_array) {  // fixnum
+        if (2 == klen && '^' == *key && 'i' == key[1] && !ni->infinity && !ni->neg && 1 == ni->div && 0 == ni->exp &&
+            0 != pi->circ_array) {  // fixnum
             oj_circ_array_set(pi->circ_array, parent->val, ni->i);
         } else {
             rval = oj_num_as_value(ni);
@@ -576,9 +523,7 @@ WHICH_TYPE:
                         rb_class2name(rb_obj_class(parent->val)));
         return;
     }
-    if (Yes == pi->options.trace) {
-        oj_trace_parse_call("add_number", pi, __FILE__, __LINE__, rval);
-    }
+    TRACE_PARSE_CALL(pi->options.trace, "add_number", pi, rval);
 }
 
 static void hash_set_value(ParseInfo pi, Val kval, VALUE value) {
@@ -604,15 +549,11 @@ WHICH_TYPE:
             }
         } else {
             if (3 <= klen && '^' == *key && '#' == key[1] && T_ARRAY == rb_type(value)) {
-                long            len = RARRAY_LEN(value);
-                volatile VALUE *a   = RARRAY_PTR(value);
+                long                  len = RARRAY_LEN(value);
+                volatile const VALUE *a   = RARRAY_CONST_PTR(value);
 
                 if (2 != len) {
-                    oj_set_error_at(pi,
-                                    oj_parse_error_class,
-                                    __FILE__,
-                                    __LINE__,
-                                    "invalid hash pair");
+                    oj_set_error_at(pi, oj_parse_error_class, __FILE__, __LINE__, "invalid hash pair");
                     return;
                 }
                 rb_hash_aset(parent->val, *a, a[1]);
@@ -666,15 +607,11 @@ WHICH_TYPE:
                         rb_class2name(rb_obj_class(parent->val)));
         return;
     }
-    if (Yes == pi->options.trace) {
-        oj_trace_parse_call("add_value", pi, __FILE__, __LINE__, value);
-    }
+    TRACE_PARSE_CALL(pi->options.trace, "add_value", pi, value);
 }
 
 static VALUE start_hash(ParseInfo pi) {
-    if (Yes == pi->options.trace) {
-        oj_trace_parse_in("start_hash", pi, __FILE__, __LINE__);
-    }
+    TRACE_PARSE_IN(pi->options.trace, "start_hash", pi);
     return Qnil;
 }
 
@@ -686,24 +623,18 @@ static void end_hash(ParseInfo pi) {
     } else if (NULL != parent->odd_args) {
         OddArgs oa = parent->odd_args;
 
-        parent->val = rb_funcall2(oa->odd->create_obj,
-                                  oa->odd->create_op,
-                                  oa->odd->attr_cnt,
-                                  oa->args);
+        parent->val = rb_funcall2(oa->odd->create_obj, oa->odd->create_op, oa->odd->attr_cnt, oa->args);
         oj_odd_free(oa);
         parent->odd_args = NULL;
     }
-    if (Yes == pi->options.trace) {
-        oj_trace_parse_hash_end(pi, __FILE__, __LINE__);
-    }
+    TRACE_PARSE_HASH_END(pi->options.trace, pi);
 }
 
 static void array_append_cstr(ParseInfo pi, const char *str, size_t len, const char *orig) {
     volatile VALUE rval = Qnil;
 
     // orig lets us know whether the string was ^r1 or \u005er1
-    if (3 <= len && 0 != pi->circ_array && '^' == orig[0] &&
-        0 == rb_array_len(stack_peek(&pi->stack)->val)) {
+    if (3 <= len && 0 != pi->circ_array && '^' == orig[0] && 0 == rb_array_len(stack_peek(&pi->stack)->val)) {
         if ('i' == str[1]) {
             long i = read_long(str + 2, len - 2);
 
@@ -722,32 +653,24 @@ static void array_append_cstr(ParseInfo pi, const char *str, size_t len, const c
     }
     rval = str_to_value(pi, str, len, orig);
     rb_ary_push(stack_peek(&pi->stack)->val, rval);
-    if (Yes == pi->options.trace) {
-        oj_trace_parse_call("append_string", pi, __FILE__, __LINE__, rval);
-    }
+    TRACE_PARSE_CALL(pi->options.trace, "append_string", pi, rval);
 }
 
 static void array_append_num(ParseInfo pi, NumInfo ni) {
     volatile VALUE rval = oj_num_as_value(ni);
 
     rb_ary_push(stack_peek(&pi->stack)->val, rval);
-    if (Yes == pi->options.trace) {
-        oj_trace_parse_call("append_number", pi, __FILE__, __LINE__, rval);
-    }
+    TRACE_PARSE_CALL(pi->options.trace, "append_number", pi, rval);
 }
 
 static void add_cstr(ParseInfo pi, const char *str, size_t len, const char *orig) {
     pi->stack.head->val = str_to_value(pi, str, len, orig);
-    if (Yes == pi->options.trace) {
-        oj_trace_parse_call("add_string", pi, __FILE__, __LINE__, pi->stack.head->val);
-    }
+    TRACE_PARSE_CALL(pi->options.trace, "add_string", pi, pi->stack.head->val);
 }
 
 static void add_num(ParseInfo pi, NumInfo ni) {
     pi->stack.head->val = oj_num_as_value(ni);
-    if (Yes == pi->options.trace) {
-        oj_trace_parse_call("add_num", pi, __FILE__, __LINE__, pi->stack.head->val);
-    }
+    TRACE_PARSE_CALL(pi->options.trace, "add_num", pi, pi->stack.head->val);
 }
 
 void oj_set_object_callbacks(ParseInfo pi) {