ox 1.5.1 → 1.5.2

data/README.md

@@ -30,9 +30,11 @@ A fast XML parser and Object marshaller as a Ruby gem.
 
 ## <a name="release">Release Notes</a>
 
-### Release 1.5.1
+### Release 1.5.2
 
- - Fixed XML dump character table. the '?' character was marked incorrectly.
+ - Added support for Date.
+
+ - Exception encoding and decoding support improved.
 
 ## <a name="description">Description</a>
 

data/ext/ox/cache.c

@@ -46,9 +46,7 @@ static void     slot_print(Cache cache, unsigned int depth);
 
 void
 ox_cache_new(Cache *cache) {
-    if (0 == (*cache = (Cache)malloc(sizeof(struct _Cache)))) {
-        rb_raise(rb_eNoMemError, "not enough memory\n");
-    }
+    *cache = ALLOC(struct _Cache);
     (*cache)->key = 0;
     (*cache)->value = Qundef;
     bzero((*cache)->slots, sizeof((*cache)->slots));
@@ -87,7 +85,7 @@ ox_cache_get(Cache cache, const char *key, VALUE **slot) {
                     cp2 = (*cp2)->slots + (*ck & 0x0F);
                     ox_cache_new(cp2);
                     if ('\0' == *(ck + 1)) {
-                        free(cache->key);
+                        xfree(cache->key);
                     } else {
                         (*cp2)->key = cache->key;
                     }

data/ext/ox/cache8.c

@@ -28,9 +28,7 @@ ox_cache8_new(Cache8 *cache) {
     Cache8     *cp;
     int         i;
     
-    if (0 == (*cache = (Cache8)malloc(sizeof(struct _Cache8)))) {
-        rb_raise(rb_eNoMemError, "not enough memory\n");
-    }
+    *cache = ALLOC(struct _Cache8);
     for (i = SLOT_CNT, cp = (*cache)->slots; 0 < i; i--, cp++) {
         *cp = 0;
     }
@@ -53,7 +51,7 @@ cache8_delete(Cache8 cache, int depth) {
             }
         }
     }
-    free(cache);
+    xfree(cache);
 }
 
 slot_t

data/ext/ox/dump.c

@@ -98,6 +98,7 @@ static void     dump_value(Out out, const char *value, size_t size);
 static void     dump_str_value(Out out, const char *value, size_t size);
 static int      dump_var(ID key, VALUE value, Out out);
 static void     dump_num(Out out, VALUE obj);
+static void     dump_date(Out out, VALUE obj);
 static void     dump_time_thin(Out out, VALUE obj);
 static void     dump_time_xsd(Out out, VALUE obj);
 static int      dump_hash(VALUE key, VALUE value, Out out);
@@ -145,8 +146,10 @@ dump_hex(u_char c, Out out) {
     *out->cur++ = hex_chars[d];
 }
 
-inline static Type
+static Type
 obj_class_code(VALUE obj) {
+    VALUE	clas = rb_obj_class(obj);
+
     switch (rb_type(obj)) {
     case T_NIL:            return NilClassCode;
     case T_ARRAY:          return ArrayCode;
@@ -162,9 +165,9 @@ obj_class_code(VALUE obj) {
 
         return (is_xml_friendly((u_char*)sym, (int)strlen(sym))) ? SymbolCode : Symbol64Code;
     }
-    case T_DATA:           return (rb_cTime == rb_obj_class(obj)) ? TimeCode : 0;
-    case T_STRUCT:         return (rb_cRange == rb_obj_class(obj)) ? RangeCode : StructCode;
-    case T_OBJECT:         return (ox_document_clas == rb_obj_class(obj) || ox_element_clas == rb_obj_class(obj)) ? RawCode : ObjectCode;
+    case T_DATA:           return (rb_cTime == clas) ? TimeCode : ((ox_date_class == clas) ? DateCode : 0);
+    case T_STRUCT:         return (rb_cRange == clas) ? RangeCode : StructCode;
+    case T_OBJECT:         return (ox_document_clas == clas || ox_element_clas == clas) ? RawCode : ObjectCode;
     case T_REGEXP:         return RegexpCode;
     case T_BIGNUM:         return BignumCode;
 #ifdef T_COMPLEX
@@ -256,17 +259,13 @@ static void
 grow(Out out, size_t len) {
     size_t  size = out->end - out->buf;
     long    pos = out->cur - out->buf;
-    char    *buf;
         
     size *= 2;
     if (size <= len * 2 + pos) {
         size += len;
     }
-    if (0 == (buf = (char*)realloc(out->buf, size + 10))) { // 1 extra for terminator character plus extra (paranoid)
-        rb_raise(rb_eNoMemError, "Failed to create string. [%d:%s]\n", ENOSPC, strerror(ENOSPC));
-    }
-    out->buf = buf;
-    out->end = buf + size;
+    REALLOC_N(out->buf, char, size + 10); // 10 extra for terminator character plus extra (paranoid)
+    out->end = out->buf + size;
     out->cur = out->buf + pos;
 }
 
@@ -294,7 +293,7 @@ dump_start(Out out, Element e) {
     if (0 < e->attr.len) {
         fill_attr(out, 'a', e->attr.str, e->attr.len);
     }
-    if ((ObjectCode == e->type || StructCode == e->type || ClassCode == e->type) && 0 < e->clas.len) {
+    if ((ObjectCode == e->type || ExceptionCode == e->type || StructCode == e->type || ClassCode == e->type) && 0 < e->clas.len) {
         fill_attr(out, 'c', e->clas.str, e->clas.len);
     }
     if (0 < e->id) {
@@ -454,6 +453,30 @@ dump_time_thin(Out out, VALUE obj) {
     out->cur += size;
 }
 
+static void
+dump_date(Out out, VALUE obj) {
+    char        buf[64];
+    char        *b = buf + sizeof(buf) - 1;
+    long        jd = NUM2LONG(rb_funcall2(obj, ox_jd_id, 0, 0));
+    long        size;
+
+    *b-- = '\0';
+    for (; 0 < jd; b--, jd /= 10) {
+        *b = '0' + (jd % 10);
+    }
+    b++;
+    if ('\0' == *b) {
+	b--;
+	*b = '0';
+    }
+    size = sizeof(buf) - (b - buf) - 1;
+    if (out->end - out->cur <= size) {
+        grow(out, size);
+    }
+    memcpy(out->cur, b, size);
+    out->cur += size;
+}
+
 static void
 dump_time_xsd(Out out, VALUE obj) {
     struct tm   *tm;
@@ -621,24 +644,15 @@ dump_obj(ID aid, VALUE obj, unsigned int depth, Out out) {
             e.indent = -1;
             out->w_end(out, &e);
         } else {
-            char        buf64[4096];
-            char        *b64 = buf64;
             ulong       size = b64_size(cnt);
+            char        *b64 = ALLOCA_N(char, size + 1);
 
             e.type = String64Code;
-            if (sizeof(buf64) < size) {
-                if (0 == (b64 = (char*)malloc(size + 1))) {
-                    rb_raise(rb_eNoMemError, "Failed to create string. [%d:%s]\n", ENOSPC, strerror(ENOSPC));
-                }
-            }
             to_base64((u_char*)str, cnt, b64);
             out->w_start(out, &e);
             dump_value(out, b64, size);
             e.indent = -1;
             out->w_end(out, &e);
-            if (buf64 != b64) {
-                free(b64);
-            }
         }
 #else
 	e.type = StringCode;
@@ -662,24 +676,15 @@ dump_obj(ID aid, VALUE obj, unsigned int depth, Out out) {
             e.indent = -1;
             out->w_end(out, &e);
         } else {
-            char        buf64[4096];
-            char        *b64 = buf64;
             ulong       size = b64_size(cnt);
+            char        *b64 = ALLOCA_N(char, size + 1);
 
             e.type = Symbol64Code;
-            if (sizeof(buf64) < size) {
-                if (0 == (b64 = (char*)malloc(size + 1))) {
-                    rb_raise(rb_eNoMemError, "Failed to create string. [%d:%s]\n", ENOSPC, strerror(ENOSPC));
-                }
-            }
             to_base64((u_char*)sym, cnt, b64);
             out->w_start(out, &e);
             dump_value(out, b64, size);
             e.indent = -1;
             out->w_end(out, &e);
-            if (buf64 != b64) {
-                free(b64);
-            }
         }
 #else
 	e.type = SymbolCode;
@@ -701,14 +706,24 @@ dump_obj(ID aid, VALUE obj, unsigned int depth, Out out) {
             out->w_time(out, obj);
             e.indent = -1;
             out->w_end(out, &e);
-        } else {
-            if (StrictEffort == out->opts->effort) {
-                rb_raise(rb_eNotImpError, "Failed to dump T_DATA %s\n", rb_class2name(clas));
-            } else {
-                e.type = NilClassCode;
-                e.closed = 1;
-                out->w_start(out, &e);
-            }
+	} else {
+	    const char	*classname = rb_class2name(clas);
+
+	    if (0 == strcmp("Date", classname)) {
+		e.type = DateCode;
+		out->w_start(out, &e);
+		dump_date(out, obj);
+		e.indent = -1;
+		out->w_end(out, &e);
+	    } else {
+		if (StrictEffort == out->opts->effort) {
+		    rb_raise(rb_eNotImpError, "Failed to dump T_DATA %s\n", classname);
+		} else {
+		    e.type = NilClassCode;
+		    e.closed = 1;
+		    out->w_start(out, &e);
+		}
+	    }
         }
         break;
     }
@@ -779,7 +794,7 @@ dump_obj(ID aid, VALUE obj, unsigned int depth, Out out) {
         } else { // Object
 // use encoding as the indicator for Ruby 1.8.7 or 1.9.x
 #ifdef HAVE_RUBY_ENCODING_H
-            e.type = ObjectCode;
+            e.type = (Qtrue == rb_obj_is_kind_of(obj, rb_eException)) ? ExceptionCode : ObjectCode;
             cnt = (int)rb_ivar_count(obj);
             e.closed = (0 >= cnt);
             out->w_start(out, &e);
@@ -797,7 +812,7 @@ dump_obj(ID aid, VALUE obj, unsigned int depth, Out out) {
 #else
             VALUE       vars = rb_obj_instance_variables(obj);
 #endif            
-            e.type = ObjectCode;
+            e.type = (Qtrue == rb_obj_is_kind_of(obj, rb_eException)) ? ExceptionCode : ObjectCode;
             cnt = (int)RARRAY_LEN(vars);
             e.closed = (0 >= cnt);
             out->w_start(out, &e);
@@ -839,20 +854,11 @@ dump_obj(ID aid, VALUE obj, unsigned int depth, Out out) {
             //dump_value(out, "/", 1);
             dump_str_value(out, s, cnt);
         } else {
-            char        buf64[4096];
-            char        *b64 = buf64;
             ulong       size = b64_size(cnt);
+            char        *b64 = ALLOCA_N(char, size + 1);
 
-            if (sizeof(buf64) < size) {
-                if (0 == (b64 = (char*)malloc(size + 1))) {
-                    rb_raise(rb_eNoMemError, "Failed to create string. [%d:%s]\n", ENOSPC, strerror(ENOSPC));
-                }
-            }
             to_base64((u_char*)s, cnt, b64);
             dump_value(out, b64, size);
-            if (buf64 != b64) {
-                free(b64);
-            }
         }
 #else
 	dump_str_value(out, s, cnt);
@@ -1119,8 +1125,8 @@ dump_obj_to_xml(VALUE obj, Options copts, Out out) {
     VALUE       clas = rb_obj_class(obj);
 
     out->w_time = (Yes == copts->xsd_date) ? dump_time_xsd : dump_time_thin;
-    out->buf = (char*)malloc(65336);
-    out->end = out->buf + 65325; // 1 less than end plus extra for possible errors
+    out->buf = ALLOC_N(char, 65336);
+    out->end = out->buf + 65325; // 10 less than end plus extra for possible errors
     out->cur = out->buf;
     out->circ_cache = 0;
     out->circ_cnt = 0;
@@ -1168,6 +1174,6 @@ ox_write_obj_to_file(VALUE obj, const char *path, Options copts) {
         int err = ferror(f);
         rb_raise(rb_eIOError, "Write failed. [%d:%s]\n", err, strerror(err));
     }
-    free(out.buf);
+    xfree(out.buf);
     fclose(f);
 }

data/ext/ox/obj_load.c

@@ -50,9 +50,9 @@ static VALUE    parse_double_time(const char *text, VALUE clas);
 static VALUE    parse_regexp(const char *text);
 
 static VALUE            get_var_sym_from_attrs(Attr a, void *encoding);
-static VALUE            get_obj_from_attrs(Attr a, PInfo pi);
-static VALUE            get_class_from_attrs(Attr a, PInfo pi);
-static VALUE            classname2class(const char *name, PInfo pi);
+static VALUE            get_obj_from_attrs(Attr a, PInfo pi, VALUE base_class);
+static VALUE            get_class_from_attrs(Attr a, PInfo pi, VALUE base_class);
+static VALUE            classname2class(const char *name, PInfo pi, VALUE base_class);
 static unsigned long    get_id_from_attrs(PInfo pi, Attr a);
 static CircArray        circ_array_new(void);
 static void             circ_array_free(CircArray ca);
@@ -125,7 +125,7 @@ name2var(const char *name, void *encoding) {
 }
 
 inline static VALUE
-resolve_classname(VALUE mod, const char *class_name, Effort effort) {
+resolve_classname(VALUE mod, const char *class_name, Effort effort, VALUE base_class) {
     VALUE       clas;
     ID          ci = rb_intern(class_name);
 
@@ -141,7 +141,7 @@ resolve_classname(VALUE mod, const char *class_name, Effort effort) {
         if (rb_const_defined_at(mod, ci)) {
             clas = rb_const_get_at(mod, ci);
         } else {
-            clas = rb_define_class_under(mod, class_name, ox_bag_clas);
+            clas = rb_define_class_under(mod, class_name, base_class);
         }
         break;
     case StrictEffort:
@@ -154,8 +154,8 @@ resolve_classname(VALUE mod, const char *class_name, Effort effort) {
 }
 
 inline static VALUE
-classname2obj(const char *name, PInfo pi) {
-    VALUE   clas = classname2class(name, pi);
+classname2obj(const char *name, PInfo pi, VALUE base_class) {
+    VALUE   clas = classname2class(name, pi, base_class);
     
     if (Qundef == clas) {
         return Qnil;
@@ -189,6 +189,20 @@ structname2obj(const char *name) {
 }
 #endif
 
+inline static VALUE
+parse_ulong(const char *s, PInfo pi) {
+    unsigned long	n = 0;
+
+    for (; '\0' != *s; s++) {
+	if ('0' <= *s && *s <= '9') {
+	    n = n * 10 + (*s - '0');
+	} else {
+	    raise_error("Invalid number for a julian day", pi->str, pi->s);
+	}
+    }
+    return ULONG2NUM(n);
+}
+
 // 2010-07-09T10:47:45.895826162+09:00
 inline static VALUE
 parse_time(const char *text, VALUE clas) {
@@ -206,7 +220,7 @@ parse_time(const char *text, VALUE clas) {
 }
 
 static VALUE
-classname2class(const char *name, PInfo pi) {
+classname2class(const char *name, PInfo pi, VALUE base_class) {
     VALUE       *slot;
     VALUE       clas;
             
@@ -223,7 +237,7 @@ classname2class(const char *name, PInfo pi) {
 		if (':' != *n) {
                     raise_error("Invalid classname, expected another ':'", pi->str, pi->s);
 		}
-                if (Qundef == (clas = resolve_classname(clas, class_name, pi->effort))) {
+                if (Qundef == (clas = resolve_classname(clas, class_name, pi->effort, base_class))) {
                     return Qundef;
                 }
                 s = class_name;
@@ -232,7 +246,7 @@ classname2class(const char *name, PInfo pi) {
             }
         }
         *s = '\0';
-        if (Qundef != (clas = resolve_classname(clas, class_name, pi->effort))) {
+        if (Qundef != (clas = resolve_classname(clas, class_name, pi->effort, base_class))) {
             *slot = clas;
         }
     }
@@ -250,10 +264,10 @@ get_var_sym_from_attrs(Attr a, void *encoding) {
 }
 
 static VALUE
-get_obj_from_attrs(Attr a, PInfo pi) {
+get_obj_from_attrs(Attr a, PInfo pi, VALUE base_class) {
     for (; 0 != a->name; a++) {
         if ('c' == *a->name && '\0' == *(a->name + 1)) {
-            return classname2obj(a->value, pi);
+            return classname2obj(a->value, pi, base_class);
         }
     }
     return Qundef;
@@ -272,10 +286,10 @@ get_struct_from_attrs(Attr a) {
 #endif
 
 static VALUE
-get_class_from_attrs(Attr a, PInfo pi) {
+get_class_from_attrs(Attr a, PInfo pi, VALUE base_class) {
     for (; 0 != a->name; a++) {
         if ('c' == *a->name && '\0' == *(a->name + 1)) {
-            return classname2class(a->value, pi);
+            return classname2class(a->value, pi, base_class);
         }
     }
     return Qundef;
@@ -307,9 +321,7 @@ static CircArray
 circ_array_new() {
     CircArray   ca;
     
-    if (0 == (ca = (CircArray)malloc(sizeof(struct _CircArray)))) {
-        rb_raise(rb_eNoMemError, "not enough memory\n");
-    }
+    ca = ALLOC(struct _CircArray);
     ca->objs = ca->obj_array;
     ca->size = sizeof(ca->obj_array) / sizeof(VALUE);
     ca->cnt = 0;
@@ -320,9 +332,9 @@ circ_array_new() {
 static void
 circ_array_free(CircArray ca) {
     if (ca->objs != ca->obj_array) {
-        free(ca->objs);
+        xfree(ca->objs);
     }
-    free(ca);
+    xfree(ca);
 }
 
 static void
@@ -334,14 +346,10 @@ circ_array_set(CircArray ca, VALUE obj, unsigned long id) {
             unsigned long       cnt = id + 512;
 
             if (ca->objs == ca->obj_array) {
-                if (0 == (ca->objs = (VALUE*)malloc(sizeof(VALUE) * cnt))) {
-                    rb_raise(rb_eNoMemError, "not enough memory\n");
-                }
+                ca->objs = ALLOC_N(VALUE, cnt);
                 memcpy(ca->objs, ca->obj_array, sizeof(VALUE) * ca->cnt);
-            } else { 
-                if (0 == (ca->objs = (VALUE*)realloc(ca->objs, sizeof(VALUE) * cnt))) {
-                    rb_raise(rb_eNoMemError, "not enough memory\n");
-                }
+            } else {
+		REALLOC_N(ca->objs, VALUE, cnt);
             }
             ca->size = cnt;
         }
@@ -461,21 +469,23 @@ add_text(PInfo pi, char *text, int closed) {
         pi->h->obj = sym;
         break;
     }
+    case DateCode:
+    {
+	VALUE	args[1];
+
+	*args = parse_ulong(text, pi);
+        pi->h->obj = rb_funcall2(ox_date_class, ox_jd_id, 1, args);
+        break;
+    }
     case TimeCode:
         pi->h->obj = parse_time(text, ox_time_class);
         break;
     case String64Code:
     {
-        char            buf[1024];
-        char            *str = buf;
         unsigned long   str_size = b64_orig_size(text);
         VALUE           v;
+        char            *str = ALLOCA_N(char, str_size + 1);
         
-        if (sizeof(buf) <= str_size) {
-            if (0 == (str = (char*)malloc(str_size + 1))) {
-                rb_raise(rb_eNoMemError, "not enough memory\n");
-            }
-        }
         from_base64(text, (u_char*)str);
         v = rb_str_new(str, str_size);
 #ifdef HAVE_RUBY_ENCODING_H
@@ -487,53 +497,32 @@ add_text(PInfo pi, char *text, int closed) {
             circ_array_set(pi->circ_array, v, (unsigned long)pi->h->obj);
         }
         pi->h->obj = v;
-        if (buf != str) {
-            free(str);
-        }
         break;
     }
     case Symbol64Code:
     {
         VALUE           sym;
         VALUE           *slot;
-        char            buf[1024];
-        char            *str = buf;
         unsigned long   str_size = b64_orig_size(text);
+        char            *str = ALLOCA_N(char, str_size + 1);
         
-        if (sizeof(buf) <= str_size) {
-            if (0 == (str = (char*)malloc(str_size + 1))) {
-                rb_raise(rb_eNoMemError, "not enough memory\n");
-            }
-        }
         from_base64(text, (u_char*)str);
         if (Qundef == (sym = ox_cache_get(ox_symbol_cache, str, &slot))) {
 	    sym = str2sym(str, pi->encoding);
             *slot = sym;
         }
         pi->h->obj = sym;
-        if (buf != str) {
-            free(str);
-        }
         break;
     }
     case RegexpCode:
         if ('/' == *text) {
             pi->h->obj = parse_regexp(text);
         } else {
-            char                buf[1024];
-            char                *str = buf;
             unsigned long       str_size = b64_orig_size(text);
+	    char		*str = ALLOCA_N(char, str_size + 1);
         
-            if (sizeof(buf) <= str_size) {
-                if (0 == (str = (char*)malloc(str_size + 1))) {
-                    rb_raise(rb_eNoMemError, "not enough memory\n");
-                }
-            }
             from_base64(text, (u_char*)str);
             pi->h->obj = parse_regexp(str);
-            if (sizeof(buf) <= str_size) {
-                free(str);
-            }
         }
         break;
     case BignumCode:
@@ -609,6 +598,7 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
     case RegexpCode:
     case BignumCode:
     case ComplexCode:
+    case DateCode:
     case TimeCode:
     case RationalCode: // sub elements read next
         // value will be read in the following add_text
@@ -645,8 +635,14 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
             h->obj = Qnil;
         }
         break;
+    case ExceptionCode:
+        h->obj = get_obj_from_attrs(attrs, pi, rb_eException);
+        if (0 != pi->circ_array && Qnil != h->obj) {
+            circ_array_set(pi->circ_array, h->obj, get_id_from_attrs(pi, attrs));
+        }
+        break;
     case ObjectCode:
-        h->obj = get_obj_from_attrs(attrs, pi);
+        h->obj = get_obj_from_attrs(attrs, pi, ox_bag_clas);
         if (0 != pi->circ_array && Qnil != h->obj) {
             circ_array_set(pi->circ_array, h->obj, get_id_from_attrs(pi, attrs));
         }
@@ -662,7 +658,7 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
 #endif
         break;
     case ClassCode:
-        h->obj = get_class_from_attrs(attrs, pi);
+        h->obj = get_class_from_attrs(attrs, pi, ox_bag_clas);
         break;
     case RefCode:
         h->obj = Qundef;
@@ -711,6 +707,7 @@ end_element(PInfo pi, const char *ename) {
             case ArrayCode:
                 rb_ary_push(pi->h->obj, h->obj);
                 break;
+            case ExceptionCode:
             case ObjectCode:
                 if (Qnil != pi->h->obj) {
                     rb_ivar_set(pi->h->obj, h->var, h->obj);
@@ -910,7 +907,7 @@ debug_stack(PInfo pi, const char *comment) {
                     
                     v = rb_funcall2(h->var, rb_intern("to_s"), 0, 0);
                     key = StringValuePtr(v);
-                } else if (ObjectCode == (h - 1)->type || RangeCode == (h - 1)->type || StructCode == (h - 1)->type) {
+                } else if (ObjectCode == (h - 1)->type || ExceptionCode == (h - 1)->type || RangeCode == (h - 1)->type || StructCode == (h - 1)->type) {
                     key = rb_id2name(h->var);
                 } else {
                     printf("%s*** corrupt stack ***\n", indent);

data/ext/ox/ox.c

@@ -60,6 +60,7 @@ ID	ox_excl_id;
 ID	ox_fileno_id;
 ID	ox_inspect_id;
 ID	ox_instruct_id;
+ID	ox_jd_id;
 ID	ox_keys_id;
 ID	ox_local_id;
 ID	ox_mesg_id;
@@ -91,6 +92,7 @@ VALUE   ox_element_clas;
 VALUE   ox_bag_clas;
 VALUE   ox_struct_class;
 VALUE   ox_time_class;
+VALUE   ox_date_class;
 
 Cache   ox_symbol_cache = 0;
 Cache   ox_class_cache = 0;
@@ -285,15 +287,16 @@ set_def_opts(VALUE self, VALUE opts) {
  */
 static VALUE
 to_obj(VALUE self, VALUE ruby_xml) {
-    VALUE       obj;
     char        *xml;
+    size_t	len;
 
     Check_Type(ruby_xml, T_STRING);
     // the xml string gets modified so make a copy of it
-    xml = strdup(StringValuePtr(ruby_xml));
-    obj = ox_parse(xml, ox_obj_callbacks, 0, 0, StrictEffort);
-    free(xml);
-    return obj;
+    len = RSTRING_LEN(ruby_xml) + 1;
+    xml = ALLOCA_N(char, len);
+    strcpy(xml, StringValuePtr(ruby_xml));
+
+    return ox_parse(xml, ox_obj_callbacks, 0, 0, StrictEffort);
 }
 
 /* call-seq: parse(xml) => Ox::Document or Ox::Element
@@ -305,15 +308,16 @@ to_obj(VALUE self, VALUE ruby_xml) {
  */
 static VALUE
 to_gen(VALUE self, VALUE ruby_xml) {
-    VALUE       obj;
     char        *xml;
+    size_t	len;
 
     Check_Type(ruby_xml, T_STRING);
     // the xml string gets modified so make a copy of it
-    xml = strdup(StringValuePtr(ruby_xml));
-    obj = ox_parse(xml, ox_gen_callbacks, 0, 0, StrictEffort);
-    free(xml);
-    return obj;
+    len = RSTRING_LEN(ruby_xml) + 1;
+    xml = ALLOCA_N(char, len);
+    strcpy(xml, StringValuePtr(ruby_xml));
+
+    return ox_parse(xml, ox_gen_callbacks, 0, 0, StrictEffort);
 }
 
 static VALUE
@@ -371,8 +375,6 @@ load(char *xml, int argc, VALUE *argv, VALUE self) {
         obj = ox_parse(xml, ox_gen_callbacks, 0, options.trace, options.effort);
         break;
     }
-    free(xml);
-
     return obj;
 }
 
@@ -396,10 +398,13 @@ load(char *xml, int argc, VALUE *argv, VALUE self) {
 static VALUE
 load_str(int argc, VALUE *argv, VALUE self) {
     char        *xml;
+    size_t	len;
     
     Check_Type(*argv, T_STRING);
     // the xml string gets modified so make a copy of it
-    xml = strdup(StringValuePtr(*argv));
+    len = RSTRING_LEN(*argv) + 1;
+    xml = ALLOCA_N(char, len);
+    strcpy(xml, StringValuePtr(*argv));
 
     return load(xml, argc - 1, argv + 1, self);
 }
@@ -435,10 +440,7 @@ load_file(int argc, VALUE *argv, VALUE self) {
     }
     fseek(f, 0, SEEK_END);
     len = ftell(f);
-    if (0 == (xml = malloc(len + 1))) {
-        fclose(f);
-        rb_raise(rb_eNoMemError, "Could not allocate memory for %ld byte file.\n", len);
-    }
+    xml = ALLOCA_N(char, len + 1);
     fseek(f, 0, SEEK_SET);
     if (len != fread(xml, 1, len, f)) {
         fclose(f);
@@ -569,7 +571,7 @@ dump(int argc, VALUE *argv, VALUE self) {
         rb_enc_associate(rstr, rb_enc_find(copts.encoding));
     }
 #endif
-    free(xml);
+    xfree(xml);
 
     return rstr;
 }
@@ -653,6 +655,7 @@ void Init_ox() {
     ox_fileno_id = rb_intern("fileno");
     ox_inspect_id = rb_intern("inspect");
     ox_instruct_id = rb_intern("instruct");
+    ox_jd_id = rb_intern("jd");
     ox_keys_id = rb_intern("keys");
     ox_local_id = rb_intern("local");
     ox_mesg_id = rb_intern("mesg");
@@ -672,6 +675,7 @@ void Init_ox() {
     ox_value_id = rb_intern("@value");
 
     ox_time_class = rb_const_get(rb_cObject, rb_intern("Time"));
+    ox_date_class = rb_const_get(rb_cObject, rb_intern("Date"));
     ox_struct_class = rb_const_get(rb_cObject, rb_intern("Struct"));
 
     ox_encoding_sym = ID2SYM(rb_intern("encoding"));		rb_ary_push(keep, ox_encoding_sym);

data/ext/ox/ox.h

@@ -121,6 +121,8 @@ typedef enum {
     String64Code   = 'b', // base64 encoded String
     ClassCode      = 'c',
     Symbol64Code   = 'd', // base64 encoded Symbol
+    DateCode       = 'D',
+    ExceptionCode  = 'e',
     FloatCode      = 'f',
     RegexpCode     = 'g',
     HashCode       = 'h',
@@ -231,6 +233,7 @@ extern ID	ox_excl_id;
 extern ID	ox_fileno_id;
 extern ID	ox_inspect_id;
 extern ID	ox_instruct_id;
+extern ID	ox_jd_id;
 extern ID	ox_keys_id;
 extern ID	ox_local_id;
 extern ID	ox_mesg_id;
@@ -249,6 +252,7 @@ extern ID	ox_tv_sec_id;
 extern ID	ox_tv_usec_id;
 extern ID	ox_value_id;
 
+extern VALUE    ox_date_class;
 extern VALUE    ox_empty_string;
 extern VALUE    ox_encoding_sym;
 extern VALUE    ox_struct_class;

data/ext/ox/parse.c

@@ -469,18 +469,14 @@ read_text(PInfo pi) {
 		
 		if (0 == alloc_buf) {
 		    size = sizeof(buf) * 2;
-		    if (0 == (alloc_buf = (char*)malloc(size))) {
-			raise_error("text too long", pi->str, pi->s);
-		    }
+		    alloc_buf = ALLOC_N(char, size);
 		    memcpy(alloc_buf, buf, b - buf);
 		    b = alloc_buf + (b - buf);
 		} else {
 		    unsigned long	pos = b - alloc_buf;
 
 		    size = (end - alloc_buf) * 2;
-		    if (0 == (alloc_buf = (char*)realloc(alloc_buf, size))) {
-			raise_error("text too long", pi->str, pi->s);
-		    }
+		    REALLOC_N(alloc_buf, char, size);
 		    b = alloc_buf + pos;
 		}
 		end = alloc_buf + size - 2;
@@ -492,7 +488,7 @@ read_text(PInfo pi) {
     *b = '\0';
     if (0 != alloc_buf) {
 	pi->pcb->add_text(pi, alloc_buf, ('/' == *(pi->s + 1)));
-	free(alloc_buf);
+	xfree(alloc_buf);
     } else {
 	pi->pcb->add_text(pi, buf, ('/' == *(pi->s + 1)));
     }
@@ -534,18 +530,14 @@ read_reduced_text(PInfo pi) {
 		
 		if (0 == alloc_buf) {
 		    size = sizeof(buf) * 2;
-		    if (0 == (alloc_buf = (char*)malloc(size))) {
-			raise_error("text too long", pi->str, pi->s);
-		    }
+		    alloc_buf = ALLOC_N(char, size);
 		    memcpy(alloc_buf, buf, b - buf);
 		    b = alloc_buf + (b - buf);
 		} else {
 		    unsigned long	pos = b - alloc_buf;
 
 		    size = (end - alloc_buf) * 2;
-		    if (0 == (alloc_buf = (char*)realloc(alloc_buf, size))) {
-			raise_error("text too long", pi->str, pi->s);
-		    }
+		    REALLOC(alloc_buf, char, size);
 		    b = alloc_buf + pos;
 		}
 		end = alloc_buf + size - 2;
@@ -561,7 +553,7 @@ read_reduced_text(PInfo pi) {
     *b = '\0';
     if (0 != alloc_buf) {
 	pi->pcb->add_text(pi, alloc_buf, ('/' == *(pi->s + 1)));
-	free(alloc_buf);
+	xfree(alloc_buf);
     } else {
 	pi->pcb->add_text(pi, buf, ('/' == *(pi->s + 1)));
     }

data/ext/ox/sax.c

@@ -270,7 +270,7 @@ sax_drive_init(SaxDrive dr, VALUE handler, VALUE io, int convert) {
 static void
 sax_drive_cleanup(SaxDrive dr) {
     if (dr->base_buf != dr->buf) {
-        free(dr->buf);
+        xfree(dr->buf);
     }
 }
 
@@ -291,14 +291,10 @@ sax_drive_read(SaxDrive dr) {
             size_t      size = dr->buf_end - dr->buf;
         
             if (dr->buf == dr->base_buf) {
-                if (0 == (dr->buf = (char*)malloc(size * 2))) {
-                    rb_raise(rb_eNoMemError, "Could not allocate memory for large element.\n");
-                }
+                dr->buf = ALLOC_N(char, size * 2);
                 memcpy(dr->buf, old, size);
             } else {
-                if (0 == (dr->buf = (char*)realloc(dr->buf, size * 2))) {
-                    rb_raise(rb_eNoMemError, "Could not allocate memory for large element.\n");
-                }
+		REALLOC_N(dr->buf, char, size * 2);
             }
             dr->buf_end = dr->buf + size * 2;
             dr->cur = dr->buf + (dr->cur - old);

data/lib/ox/version.rb

@@ -1,5 +1,5 @@
 
 module Ox
   # Current version of the module. 
-  VERSION = '1.5.1'
+  VERSION = '1.5.2'
 end

data/test/bug3.rb

@@ -0,0 +1,21 @@
+#!/usr/bin/env ruby -wW1
+
+$: << '../lib'
+$: << '../ext'
+
+require 'ox'
+
+def name_it()
+  begin
+    "x".foo
+  rescue Exception => e
+    #puts e.message
+    xml = Ox.dump(e, effort: :tolerant)
+    puts xml
+    o = Ox.load(xml, mode: :object)
+    puts o.message
+    puts Ox.dump(e)
+  end
+end
+
+name_it()

data/test/bug4.rb

@@ -0,0 +1,60 @@
+#!/usr/bin/env ruby -wW1
+# encoding: UTF-8
+
+$: << File.join(File.dirname(__FILE__), "../lib")
+$: << File.join(File.dirname(__FILE__), "../ext")
+
+require 'stringio'
+require 'ox'
+
+x1 = %(<?xml version="1.0" encoding="ISO-8859-1" ?><tag key="value">Français</tag>).encode("ISO-8859-1")
+# => "<?xml version=\"1.0\" encoding=\"ISO-8859-1\" ?><tag key=\"value\">Fran\xE7ais</tag>" 
+x1.encoding
+# => #<Encoding:ISO-8859-1> 
+
+x2 = %(<?xml version="1.0" encoding="UTF-8" ?><tag key="value">Français</tag>)
+# => "<?xml version=\"1.0\" encoding=\"UTF-8\" ?><tag key=\"value\">Fran\xC3\xA7ais</tag>"
+x2.encoding
+# => #<Encoding:UTF-8> 
+
+class OH < ::Ox::Sax
+  def start_element(name)
+    puts "EL: #{name} (#{name.encoding})"
+  end
+
+  def end_element(name)
+  end
+
+  def attr(key, value)
+    puts "AT: #{key} => #{value} (#{key.encoding} => #{value.encoding})"
+  end
+
+  def text(value)
+    puts "TX: #{value} (#{value.encoding})"
+  end
+end
+
+::Ox.sax_parse OH.new, StringIO.new(x1)
+   # => AT: version => 1.0 (US-ASCII => ASCII-8BIT)
+   #    AT: encoding => ISO-8859-1 (US-ASCII => ISO-8859-1)
+   #    EL: tag (US-ASCII)
+   #    AT: key => value (US-ASCII => ISO-8859-1)
+   #    TX: Fran�ais (ISO-8859-1)
+
+puts
+
+::Ox.sax_parse OH.new, StringIO.new(x2)
+   # => AT: version => 1.0 (US-ASCII => ASCII-8BIT)
+   #    AT: encoding => UTF-8 (US-ASCII => UTF-8)
+   #    EL: tag (US-ASCII)
+   #    AT: key => value (US-ASCII => UTF-8)
+   #    TX: Français (UTF-8)
+
+puts
+x3 = %(<?xml version="1.0" encoding="ISO-8859-1" ?><tag Português="Español">Français</tag>).encode("ISO-8859-1")
+::Ox.sax_parse OH.new, StringIO.new(x3)
+
+puts
+x4 = %(<?xml version="1.0" encoding="UTF-8" ?><tag Português="Español">Français</tag>)
+::Ox.sax_parse OH.new, StringIO.new(x4)
+

data/test/func.rb

@@ -6,6 +6,7 @@ $: << File.join(File.dirname(__FILE__), "../ext")
 
 require 'test/unit'
 require 'optparse'
+require 'date'
 require 'ox'
 
 $indent = 2
@@ -107,6 +108,10 @@ class Func < ::Test::Unit::TestCase
     dump_and_load(Time.now, false)
   end
 
+  def test_date
+    dump_and_load(Date.new(2011, 1, 5), false)
+  end
+
   def test_array
     dump_and_load([], false)
     dump_and_load([1, 'a'], false)
@@ -262,10 +267,30 @@ class Func < ::Test::Unit::TestCase
       e = StandardError.new("Some Error")
       e.set_backtrace(["./func.rb:119: in test_exception",
                        "./fake.rb:57: in fake_func"])
+      dump_and_load(e, false)
+    end
+  end
+
+  def test_exception_bag
+    if RUBY_VERSION.start_with?('1.8')
+      assert(true)
+    else
+      xml = %{
+<e c="FakeError">
+  <s a="mesg">Some Error</s>
+  <a a="bt">
+    <s>./func.rb:119: in test_exception</s>
+    <s>./fake.rb:57: in fake_func</s>
+  </a>
+</e>
+}
+      x = Ox.load(xml, :mode => :object, :effort => :auto_define)
+      assert_equal('Some Error', x.message())
+      assert(x.is_a?(Exception))
     end
-    dump_and_load(e, false)
   end
 
+
   def test_struct
     s = Struct.new('Box', :x, :y, :w, :h)
     dump_and_load(s.new(2, 4, 10, 20), false)

data/test/perf.rb

@@ -45,7 +45,7 @@ class Perf
       end
     end
     puts
-    puts "Comparison Matrix\n(performance factor, 2.0 row is means twice as fast as column)"
+    puts "Comparison Matrix\n(performance factor, 2.0 means row is twice as fast as column)"
     puts ([' ' * width] + iva.map { |i| "%*s" % [width, i.title] }).join('  ')
     puts (['-' * width] + iva.map { |i| '-' * width }).join('  ')
     iva.each do |i|

data/test/sax_example.rb

@@ -0,0 +1,37 @@
+#!/usr/bin/env ruby -wW1
+
+$: << '../lib'
+$: << '../ext'
+
+require 'stringio'
+require 'ox'
+
+class Sample < ::Ox::Sax
+  def start_element(name); puts "start: #{name}"; end
+  def end_element(name); puts "end: #{name}"; end
+  def attr(name, value); puts "  #{name} => #{value}"; end
+  def text(value); puts "text #{value}"; end
+end
+
+io = StringIO.new(%{
+<top name="sample">
+  <middle name="second">
+    <bottom name="third"/>
+  </middle>
+</top>
+})
+
+handler = Sample.new()
+Ox.sax_parse(handler, io)
+
+# outputs
+# start: top
+#   name => sample
+# start: middle
+#   name => second
+# start: bottom
+#   name => third
+# end: bottom
+# end: middle
+# end: top
+

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ox
 version: !ruby/object:Gem::Version
-  version: 1.5.1
+  version: 1.5.2
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-03-07 00:00:00.000000000 Z
+date: 2012-03-15 00:00:00.000000000 Z
 dependencies: []
 description: ! "A fast XML parser and object serializer that uses only standard C
   lib.\n            \nOptimized XML (Ox), as the name implies was written to provide
@@ -53,6 +53,8 @@ files:
 - test/bench.rb
 - test/bug1.rb
 - test/bug2.rb
+- test/bug3.rb
+- test/bug4.rb
 - test/cache16_test.rb
 - test/cache8_test.rb
 - test/cache_test.rb
@@ -93,6 +95,7 @@ files:
 - test/sample/shape.rb
 - test/sample/text.rb
 - test/sample.rb
+- test/sax_example.rb
 - test/sax_test.rb
 - test/test.rb
 - test/Sample.graffle
@@ -126,4 +129,3 @@ signing_key:
 specification_version: 3
 summary: A fast XML parser and object serializer.
 test_files: []
-has_rdoc: true