oj 3.16.0 → 3.17.3

data/ext/oj/string_writer.c

@@ -225,7 +225,7 @@ void oj_str_writer_pop_all(StrWriter sw) {
     }
 }
 
-static void str_writer_free(void *ptr) {
+static void string_writer_free(void *ptr) {
     StrWriter sw;
 
     if (0 == ptr) {
@@ -239,6 +239,23 @@ static void str_writer_free(void *ptr) {
     OJ_R_FREE(ptr);
 }
 
+static const rb_data_type_t oj_string_writer_type = {
+    "Oj/string_writer",
+    {
+        NULL,
+        string_writer_free,
+        NULL,
+    },
+    0,
+    0,
+};
+
+StrWriter oj_str_writer_unwrap(VALUE writer) {
+    StrWriter sw;
+    TypedData_Get_Struct(writer, struct _strWriter, &oj_string_writer_type, sw);
+    return sw;
+}
+
 /* Document-method: new
  * call-seq: new(io, options)
  *
@@ -266,7 +283,7 @@ static VALUE str_writer_new(int argc, VALUE *argv, VALUE self) {
     sw->out.argv   = argv + 1;
     sw->out.indent = sw->opts.indent;
 
-    return Data_Wrap_Struct(oj_string_writer_class, 0, str_writer_free, sw);
+    return TypedData_Wrap_Struct(oj_string_writer_class, &oj_string_writer_type, sw);
 }
 
 /* Document-method: push_key
@@ -278,7 +295,8 @@ static VALUE str_writer_new(int argc, VALUE *argv, VALUE self) {
  * - *key* [_String_] the key pending for the next push
  */
 static VALUE str_writer_push_key(VALUE self, VALUE key) {
-    StrWriter sw = (StrWriter)DATA_PTR(self);
+    StrWriter sw;
+    TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
 
     oj_str_writer_push_key(sw, StringValuePtr(key));
 
@@ -293,7 +311,8 @@ static VALUE str_writer_push_key(VALUE self, VALUE key) {
  * - *key* [_String_] the key if adding to an object in the JSON document
  */
 static VALUE str_writer_push_object(int argc, VALUE *argv, VALUE self) {
-    StrWriter sw = (StrWriter)DATA_PTR(self);
+    StrWriter sw;
+    TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
 
     switch (argc) {
     case 0: oj_str_writer_push_object(sw, 0); break;
@@ -321,7 +340,8 @@ static VALUE str_writer_push_object(int argc, VALUE *argv, VALUE self) {
  * - *key* [_String_] the key if adding to an object in the JSON document
  */
 static VALUE str_writer_push_array(int argc, VALUE *argv, VALUE self) {
-    StrWriter sw = (StrWriter)DATA_PTR(self);
+    StrWriter sw;
+    TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
 
     switch (argc) {
     case 0: oj_str_writer_push_array(sw, 0); break;
@@ -349,13 +369,16 @@ static VALUE str_writer_push_array(int argc, VALUE *argv, VALUE self) {
  * - *key* [_String_] the key if adding to an object in the JSON document
  */
 static VALUE str_writer_push_value(int argc, VALUE *argv, VALUE self) {
+    StrWriter sw;
+    TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
+
     switch (argc) {
-    case 1: oj_str_writer_push_value((StrWriter)DATA_PTR(self), *argv, 0); break;
+    case 1: oj_str_writer_push_value(sw, *argv, 0); break;
     case 2:
         if (Qnil == argv[1]) {
-            oj_str_writer_push_value((StrWriter)DATA_PTR(self), *argv, 0);
+            oj_str_writer_push_value(sw, *argv, 0);
         } else {
-            oj_str_writer_push_value((StrWriter)DATA_PTR(self), *argv, StringValuePtr(argv[1]));
+            oj_str_writer_push_value(sw, *argv, StringValuePtr(argv[1]));
         }
         break;
     default: rb_raise(rb_eArgError, "Wrong number of argument to 'push_value'."); break;
@@ -373,13 +396,16 @@ static VALUE str_writer_push_value(int argc, VALUE *argv, VALUE self) {
  * - *key* [_String_] the key if adding to an object in the JSON document
  */
 static VALUE str_writer_push_json(int argc, VALUE *argv, VALUE self) {
+    StrWriter sw;
+    TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
+
     switch (argc) {
-    case 1: oj_str_writer_push_json((StrWriter)DATA_PTR(self), StringValuePtr(*argv), 0); break;
+    case 1: oj_str_writer_push_json(sw, StringValuePtr(*argv), 0); break;
     case 2:
         if (Qnil == argv[1]) {
-            oj_str_writer_push_json((StrWriter)DATA_PTR(self), StringValuePtr(*argv), 0);
+            oj_str_writer_push_json(sw, StringValuePtr(*argv), 0);
         } else {
-            oj_str_writer_push_json((StrWriter)DATA_PTR(self), StringValuePtr(*argv), StringValuePtr(argv[1]));
+            oj_str_writer_push_json(sw, StringValuePtr(*argv), StringValuePtr(argv[1]));
         }
         break;
     default: rb_raise(rb_eArgError, "Wrong number of argument to 'push_json'."); break;
@@ -393,7 +419,10 @@ static VALUE str_writer_push_json(int argc, VALUE *argv, VALUE self) {
  * currently open.
  */
 static VALUE str_writer_pop(VALUE self) {
-    oj_str_writer_pop((StrWriter)DATA_PTR(self));
+    StrWriter sw;
+    TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
+
+    oj_str_writer_pop(sw);
     return Qnil;
 }
 
@@ -404,7 +433,10 @@ static VALUE str_writer_pop(VALUE self) {
  * currently open.
  */
 static VALUE str_writer_pop_all(VALUE self) {
-    oj_str_writer_pop_all((StrWriter)DATA_PTR(self));
+    StrWriter sw;
+    TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
+
+    oj_str_writer_pop_all(sw);
 
     return Qnil;
 }
@@ -415,7 +447,8 @@ static VALUE str_writer_pop_all(VALUE self) {
  * Reset the writer back to the empty state.
  */
 static VALUE str_writer_reset(VALUE self) {
-    StrWriter sw = (StrWriter)DATA_PTR(self);
+    StrWriter sw;
+    TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
 
     sw->depth      = 0;
     *sw->types     = '\0';
@@ -434,23 +467,22 @@ static VALUE str_writer_reset(VALUE self) {
  * *return* [_String_]
  */
 static VALUE str_writer_to_s(VALUE self) {
-    StrWriter sw   = (StrWriter)DATA_PTR(self);
-    VALUE     rstr = rb_str_new(sw->out.buf, sw->out.cur - sw->out.buf);
-
-    return oj_encode(rstr);
+    StrWriter sw;
+    TypedData_Get_Struct(self, struct _strWriter, &oj_string_writer_type, sw);
+    return rb_utf8_str_new(sw->out.buf, sw->out.cur - sw->out.buf);
 }
 
 /* Document-method: as_json
- * call-seq: as_json()
+ * call-seq: as_json(*)
  *
  * Returns the contents of the writer as a JSON element. If called from inside
  * an array or hash by Oj the raw buffer will be used othersize a more
  * inefficient parse of the contents and a return of the result is
- * completed. The parse uses the strict mode.
+ * completed. The parse uses the strict mode. Optional arguments are ignored.
  *
  * *return* [_Hash_|_Array_|_String_|_Integer_|_Float_|_True_|_False_|_nil|)
  */
-static VALUE str_writer_as_json(VALUE self) {
+static VALUE str_writer_as_json(int argc, VALUE *argv, VALUE self) {
     if (string_writer_optimized) {
         return self;
     }
@@ -481,5 +513,5 @@ void oj_string_writer_init(void) {
     rb_define_method(oj_string_writer_class, "reset", str_writer_reset, 0);
     rb_define_method(oj_string_writer_class, "to_s", str_writer_to_s, 0);
     rb_define_method(oj_string_writer_class, "raw_json", str_writer_to_s, 0);
-    rb_define_method(oj_string_writer_class, "as_json", str_writer_as_json, 0);
+    rb_define_method(oj_string_writer_class, "as_json", str_writer_as_json, -1);
 }

data/ext/oj/usual.c

@@ -63,7 +63,7 @@ static VALUE form_attr(const char *str, size_t len) {
         memcpy(b + 1, str, len);
         b[len + 1] = '\0';
 
-        id = rb_intern3(buf, len + 1, oj_utf8_encoding);
+        id = rb_intern3(b, len + 1, oj_utf8_encoding);
         OJ_R_FREE(b);
         return id;
     }
@@ -200,7 +200,10 @@ static void push_key(ojParser p) {
         d->ktail = d->khead + pos;
         d->kend  = d->khead + cap;
     }
-    d->ktail->len = klen;
+    if (32000 < klen) {
+        rb_raise(oj_json_parser_error_class, "Key too long. Keys are limited to 32,000 bytes.");
+    }
+    d->ktail->len = (int16_t)klen;
     if (klen < sizeof(d->ktail->buf)) {
         memcpy(d->ktail->buf, key, klen);
         d->ktail->buf[klen] = '\0';
@@ -281,7 +284,6 @@ static void close_object(ojParser p) {
     VALUE         *head = d->vhead + c->vi + 1;
     volatile VALUE obj  = rb_hash_new();
 
-#if HAVE_RB_HASH_BULK_INSERT
     for (vp = head; kp < d->ktail; kp++, vp += 2) {
         *vp = d->get_key(p, kp);
         if (sizeof(kp->buf) <= (size_t)kp->len) {
@@ -289,18 +291,15 @@ static void close_object(ojParser p) {
         }
     }
     rb_hash_bulk_insert(d->vtail - head, head, obj);
-#else
-    for (vp = head; kp < d->ktail; kp++, vp += 2) {
-        rb_hash_aset(obj, d->get_key(p, kp), *(vp + 1));
-        if (sizeof(kp->buf) <= (size_t)kp->len) {
-            OJ_R_FREE(kp->key);
-        }
-    }
-#endif
     d->ktail = d->khead + c->ki;
+
     d->vtail = head;
     head--;
     *head = obj;
+    if (1 == d->vtail - d->vhead && rb_block_given_p()) {
+        d->vtail = d->vhead;
+        rb_yield(obj);
+    }
 }
 
 static void close_object_class(ojParser p) {
@@ -341,7 +340,6 @@ static void close_object_create(ojParser p) {
         head++;
         if (Qnil == d->hash_class) {
             obj = rb_hash_new();
-#if HAVE_RB_HASH_BULK_INSERT
             for (vp = head; kp < d->ktail; kp++, vp += 2) {
                 *vp = d->get_key(p, kp);
                 if (sizeof(kp->buf) <= (size_t)kp->len) {
@@ -349,14 +347,6 @@ static void close_object_create(ojParser p) {
                 }
             }
             rb_hash_bulk_insert(d->vtail - head, head, obj);
-#else
-            for (vp = head; kp < d->ktail; kp++, vp += 2) {
-                rb_hash_aset(obj, d->get_key(p, kp), *(vp + 1));
-                if (sizeof(kp->buf) <= (size_t)kp->len) {
-                    OJ_R_FREE(kp->key);
-                }
-            }
-#endif
         } else {
             obj = rb_class_new_instance(0, NULL, d->hash_class);
             for (vp = head; kp < d->ktail; kp++, vp += 2) {
@@ -373,7 +363,6 @@ static void close_object_create(ojParser p) {
         if (!d->ignore_json_create && rb_respond_to(clas, oj_json_create_id)) {
             volatile VALUE arg = rb_hash_new();
 
-#if HAVE_RB_HASH_BULK_INSERT
             for (vp = head; kp < d->ktail; kp++, vp += 2) {
                 *vp = d->get_key(p, kp);
                 if (sizeof(kp->buf) <= (size_t)kp->len) {
@@ -381,14 +370,6 @@ static void close_object_create(ojParser p) {
                 }
             }
             rb_hash_bulk_insert(d->vtail - head, head, arg);
-#else
-            for (vp = head; kp < d->ktail; kp++, vp += 2) {
-                rb_hash_aset(arg, d->get_key(p, kp), *(vp + 1));
-                if (sizeof(kp->buf) <= (size_t)kp->len) {
-                    OJ_R_FREE(kp->key);
-                }
-            }
-#endif
             obj = rb_funcall(clas, oj_json_create_id, 1, arg);
         } else {
             obj = rb_class_new_instance(0, NULL, clas);
@@ -599,7 +580,21 @@ static VALUE result(ojParser p) {
     Usual d = (Usual)p->ctx;
 
     if (d->vhead < d->vtail) {
-        return *d->vhead;
+        long            cnt = d->vtail - d->vhead;
+        volatile VALUE  ary;
+        volatile VALUE *vp;
+
+        if (1 == cnt) {
+            return *d->vhead;
+        }
+        ary = rb_ary_new();
+        for (vp = d->vhead; vp < d->vtail; vp++) {
+            rb_ary_push(ary, *vp);
+        }
+        return ary;
+    }
+    if (d->raise_on_empty) {
+        rb_raise(oj_parse_error_class, "empty string");
     }
     return Qnil;
 }
@@ -616,12 +611,16 @@ static void dfree(ojParser p) {
     Usual d = (Usual)p->ctx;
 
     cache_free(d->str_cache);
+    d->str_cache = NULL;
     cache_free(d->attr_cache);
+    d->attr_cache = NULL;
     if (NULL != d->sym_cache) {
         cache_free(d->sym_cache);
+        d->sym_cache = NULL;
     }
     if (NULL != d->class_cache) {
         cache_free(d->class_cache);
+        d->class_cache = NULL;
     }
     OJ_R_FREE(d->vhead);
     OJ_R_FREE(d->chead);
@@ -648,6 +647,12 @@ static void mark(ojParser p) {
     if (NULL != d->class_cache) {
         cache_mark(d->class_cache);
     }
+    if (Qnil != d->hash_class) {
+        rb_gc_mark(d->hash_class);
+    }
+    if (Qnil != d->array_class) {
+        rb_gc_mark(d->array_class);
+    }
     for (vp = d->vhead; vp < d->vtail; vp++) {
         if (Qundef != *vp) {
             rb_gc_mark(*vp);
@@ -842,8 +847,8 @@ static VALUE opt_create_id_set(ojParser p, VALUE value) {
         rb_check_type(value, T_STRING);
         size_t len = RSTRING_LEN(value);
 
-        if (1 << sizeof(d->create_id_len) <= len) {
-            rb_raise(rb_eArgError, "The create_id values is limited to %d bytes.", 1 << sizeof(d->create_id_len));
+        if (1 << (8 * sizeof(d->create_id_len)) <= len) {
+            rb_raise(rb_eArgError, "The create_id values is limited to %d bytes.", 1 << (8 * sizeof(d->create_id_len)));
         }
         d->create_id_len                  = (uint8_t)len;
         d->create_id                      = str_dup(RSTRING_PTR(value), len);
@@ -1058,14 +1063,28 @@ static VALUE opt_symbol_keys_set(ojParser p, VALUE value) {
         if (NULL != d->sym_cache) {
             cache_free(d->sym_cache);
             d->sym_cache = NULL;
+            d->key_cache = NULL;
         }
-        if (!d->cache_keys) {
-            d->get_key = str_key;
-        }
+        d->cache_keys = false;
+        d->get_key    = str_key;
     }
     return (NULL != d->sym_cache) ? Qtrue : Qfalse;
 }
 
+static VALUE opt_raise_on_empty(ojParser p, VALUE value) {
+    Usual d = (Usual)p->ctx;
+
+    return d->raise_on_empty ? Qtrue : Qfalse;
+}
+
+static VALUE opt_raise_on_empty_set(ojParser p, VALUE value) {
+    Usual d = (Usual)p->ctx;
+
+    d->raise_on_empty = (Qtrue == value);
+
+    return d->raise_on_empty ? Qtrue : Qfalse;
+}
+
 static VALUE option(ojParser p, const char *key, VALUE value) {
     struct opt *op;
     struct opt  opts[] = {
@@ -1095,6 +1114,8 @@ static VALUE option(ojParser p, const char *key, VALUE value) {
         {.name = "omit_null=", .func = opt_omit_null_set},
         {.name = "symbol_keys", .func = opt_symbol_keys},
         {.name = "symbol_keys=", .func = opt_symbol_keys_set},
+        {.name = "raise_on_empty", .func = opt_raise_on_empty},
+        {.name = "raise_on_empty=", .func = opt_raise_on_empty_set},
         {.name = NULL},
     };
 
@@ -1129,6 +1150,7 @@ void oj_init_usual(ojParser p, Usual d) {
     d->get_key            = cache_key;
     d->cache_keys         = true;
     d->ignore_json_create = false;
+    d->raise_on_empty     = false;
     d->cache_str          = 6;
     d->array_class        = Qnil;
     d->hash_class         = Qnil;

data/ext/oj/usual.h

@@ -60,6 +60,7 @@ typedef struct _usual {
     uint8_t miss_class;
     bool    cache_keys;
     bool    ignore_json_create;
+    bool    raise_on_empty;
 } *Usual;
 
 // Initialize the parser with the usual delegate. If the usual delegate is

data/ext/oj/val_stack.c

@@ -8,7 +8,7 @@
 #include "odd.h"
 #include "oj.h"
 
-static void mark(void *ptr) {
+static void stack_mark(void *ptr) {
     ValStack stack = (ValStack)ptr;
     Val      v;
 
@@ -46,6 +46,17 @@ static void mark(void *ptr) {
 #endif
 }
 
+static const rb_data_type_t oj_stack_type = {
+    "Oj/stack",
+    {
+        stack_mark,
+        NULL,
+        NULL,
+    },
+    0,
+    0,
+};
+
 VALUE
 oj_stack_init(ValStack stack) {
 #ifdef HAVE_PTHREAD_MUTEX_INIT
@@ -70,7 +81,7 @@ oj_stack_init(ValStack stack) {
     stack->head->clen      = 0;
     stack->head->next      = NEXT_NONE;
 
-    return Data_Wrap_Struct(oj_cstack_class, mark, 0, stack);
+    return TypedData_Wrap_Struct(oj_cstack_class, &oj_stack_type, stack);
 }
 
 const char *oj_stack_next_string(ValNext n) {

data/ext/oj/wab.c

@@ -19,7 +19,7 @@
 // Workaround in case INFINITY is not defined in math.h or if the OS is CentOS
 #define OJ_INFINITY (1.0 / 0.0)
 
-static char hex_chars[256] = "\
+static char hex_chars[257] = "\
 ................................\
 ................xxxxxxxxxx......\
 .xxxxxx.........................\
@@ -111,10 +111,11 @@ 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;
 
-    cnt         = (int)RARRAY_LEN(a);
+    cnt         = RARRAY_LEN(a);
     *out->cur++ = '[';
     size        = 2;
     assure_size(out, size);
@@ -226,7 +227,7 @@ static void dump_obj(VALUE obj, int depth, Out out, bool as_ok) {
     } else 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 if (resolve_wab_uuid_class() == clas) {
         oj_dump_str(oj_safe_string_convert(obj), depth, out, false);
     } else if (resolve_uri_http_class() == clas) {

data/lib/oj/mimic.rb

@@ -1,11 +1,7 @@
 # frozen_string_literal: false
 
 require 'bigdecimal'
-begin
-  require 'ostruct'
-rescue Exception
-  # ignore
-end
+require 'ostruct'
 
 module Oj
 

data/lib/oj/schandler.rb

@@ -64,13 +64,14 @@ module Oj
   #
   #    hash_end
   #
-  # When a hash key is encountered the hash_key method is called with the parsed
-  # hash value key. The return value from the call is then used as the key in
-  # the key-value pair that follows.
+  #  At the end of a JSON object element the hash_end() callback is called if
+  #  public.
   #
   #    hash_key
   #
-  # At the end of a JSON object element the hash_end() callback is called if public.
+  # When a hash key is encountered the hash_key() method is called with the
+  # parsed hash value key. The return value from the call is then used as the
+  # key in the key-value pair that follows.
   #
   #    hash_set
   #

data/lib/oj/version.rb

@@ -1,4 +1,4 @@
 module Oj
   # Current version of the module.
-  VERSION = '3.16.0'
+  VERSION = '3.17.3'
 end

data/pages/Encoding.md

@@ -15,7 +15,7 @@ in a JSON document. The formatting follows these rules.
  * The `'^'` character denotes a special key value when in a JSON Object sequence.
 
  * A Ruby String that starts with `':'`or the sequence `'^i'` or `'^r'` are
-   encoded by excaping the first character so that it appears as `'\u005e'` or
+   encoded by escaping the first character so that it appears as `'\u005e'` or
    `'\u003a'` instead of `':'` or `'^'`.
 
  * A `"^c"` JSON Object key indicates the value should be converted to a Ruby