ox 1.6.6 → 1.6.7

data/README.md

@@ -34,9 +34,9 @@ A fast XML parser and Object marshaller as a Ruby gem.
 
 ## <a name="release">Release Notes</a>
 
-### Release 1.6.6
+### Release 1.6.7
 
- - A SAX parser encoding bug discovered by bughit was fixed.
+ - Now uses the encoding of the imput XML as the default encoding for the parsed output if the default options encoding is not set and the encoding is not set in the XML file prolog.
 
 ## <a name="description">Description</a>
 

data/ext/ox/gen_load.c

@@ -116,10 +116,10 @@ create_prolog_doc(PInfo pi, const char *target, Attr attrs) {
     for (; 0 != attrs->name; attrs++) {
 	if (Yes == pi->options->sym_keys) {
 #if HAS_ENCODING_SUPPORT
-	    if (0 != pi->encoding) {
+	    if (0 != pi->options->rb_enc) {
 		VALUE	rstr = rb_str_new2(attrs->name);
 
-		rb_enc_associate(rstr, pi->encoding);
+		rb_enc_associate(rstr, pi->options->rb_enc);
 		sym = rb_funcall(rstr, ox_to_sym_id, 0);
 	    } else {
 		sym = ID2SYM(rb_intern(attrs->name));
@@ -132,15 +132,15 @@ create_prolog_doc(PInfo pi, const char *target, Attr attrs) {
 	    VALUE	rstr = rb_str_new2(attrs->name);
 
 #if HAS_ENCODING_SUPPORT
-	    if (0 != pi->encoding) {
-		rb_enc_associate(rstr, pi->encoding);
+	    if (0 != pi->options->rb_enc) {
+		rb_enc_associate(rstr, pi->options->rb_enc);
 	    }
 #endif
 	    rb_hash_aset(ah, rstr, rb_str_new2(attrs->value));
 	}
 #if HAS_ENCODING_SUPPORT
 	if (0 == strcmp("encoding", attrs->name)) {
-	    pi->encoding = rb_enc_find(attrs->value);
+	    pi->options->rb_enc = rb_enc_find(attrs->value);
 	}
 #endif
     }
@@ -210,8 +210,8 @@ add_doctype(PInfo pi, const char *docType) {
     VALUE       s = rb_str_new2(docType);
 
 #if HAS_ENCODING_SUPPORT
-    if (0 != pi->encoding) {
-        rb_enc_associate(s, pi->encoding);
+    if (0 != pi->options->rb_enc) {
+        rb_enc_associate(s, pi->options->rb_enc);
     }
 #endif
     rb_ivar_set(n, ox_at_value_id, s);
@@ -227,8 +227,8 @@ add_comment(PInfo pi, const char *comment) {
     VALUE       s = rb_str_new2(comment);
 
 #if HAS_ENCODING_SUPPORT
-    if (0 != pi->encoding) {
-        rb_enc_associate(s, pi->encoding);
+    if (0 != pi->options->rb_enc) {
+        rb_enc_associate(s, pi->options->rb_enc);
     }
 #endif
     rb_ivar_set(n, ox_at_value_id, s);
@@ -244,8 +244,8 @@ add_cdata(PInfo pi, const char *cdata, size_t len) {
     VALUE       s = rb_str_new2(cdata);
 
 #if HAS_ENCODING_SUPPORT
-    if (0 != pi->encoding) {
-        rb_enc_associate(s, pi->encoding);
+    if (0 != pi->options->rb_enc) {
+        rb_enc_associate(s, pi->options->rb_enc);
     }
 #endif
     rb_ivar_set(n, ox_at_value_id, s);
@@ -260,8 +260,8 @@ add_text(PInfo pi, char *text, int closed) {
     VALUE       s = rb_str_new2(text);
 
 #if HAS_ENCODING_SUPPORT
-    if (0 != pi->encoding) {
-        rb_enc_associate(s, pi->encoding);
+    if (0 != pi->options->rb_enc) {
+        rb_enc_associate(s, pi->options->rb_enc);
     }
 #endif
     if (0 == pi->h) { /* top level object */
@@ -276,8 +276,8 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
     VALUE       s = rb_str_new2(ename);
 
 #if HAS_ENCODING_SUPPORT
-    if (0 != pi->encoding) {
-        rb_enc_associate(s, pi->encoding);
+    if (0 != pi->options->rb_enc) {
+        rb_enc_associate(s, pi->options->rb_enc);
     }
 #endif
     e = rb_obj_alloc(ox_element_clas);
@@ -292,10 +292,10 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
 	    if (Yes == pi->options->sym_keys) {
 		if (Qundef == (sym = ox_cache_get(ox_symbol_cache, attrs->name, &slot))) {
 #if HAS_ENCODING_SUPPORT
-		    if (0 != pi->encoding) {
+		    if (0 != pi->options->rb_enc) {
 			VALUE	rstr = rb_str_new2(attrs->name);
 
-			rb_enc_associate(rstr, pi->encoding);
+			rb_enc_associate(rstr, pi->options->rb_enc);
 			sym = rb_funcall(rstr, ox_to_sym_id, 0);
 		    } else {
 			sym = ID2SYM(rb_intern(attrs->name));
@@ -308,15 +308,15 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
 	    } else {
 		sym = rb_str_new2(attrs->name);
 #if HAS_ENCODING_SUPPORT
-		if (0 != pi->encoding) {
-		    rb_enc_associate(sym, pi->encoding);
+		if (0 != pi->options->rb_enc) {
+		    rb_enc_associate(sym, pi->options->rb_enc);
 		}
 #endif
 	    }
             s = rb_str_new2(attrs->value);
 #if HAS_ENCODING_SUPPORT
-            if (0 != pi->encoding) {
-                rb_enc_associate(s, pi->encoding);
+            if (0 != pi->options->rb_enc) {
+                rb_enc_associate(s, pi->options->rb_enc);
             }
 #endif
             rb_hash_aset(ah, sym, s);

data/ext/ox/obj_load.c

@@ -399,7 +399,7 @@ instruct(PInfo pi, const char *target, Attr attrs) {
 #if HAS_ENCODING_SUPPORT
         for (; 0 != attrs->name; attrs++) {
             if (0 == strcmp("encoding", attrs->name)) {
-                pi->encoding = rb_enc_find(attrs->value);
+                pi->options->rb_enc = rb_enc_find(attrs->value);
             }
         }
 #endif
@@ -422,8 +422,8 @@ add_text(PInfo pi, char *text, int closed) {
     case StringCode:
         pi->h->obj = rb_str_new2(text);
 #if HAS_ENCODING_SUPPORT
-        if (0 != pi->encoding) {
-            rb_enc_associate(pi->h->obj, pi->encoding);
+        if (0 != pi->options->rb_enc) {
+            rb_enc_associate(pi->h->obj, pi->options->rb_enc);
         }
 #endif
         if (0 != pi->circ_array) {
@@ -463,7 +463,7 @@ add_text(PInfo pi, char *text, int closed) {
         VALUE   *slot;
 
         if (Qundef == (sym = ox_cache_get(ox_symbol_cache, text, &slot))) {
-	    sym = str2sym(text, pi->encoding);
+	    sym = str2sym(text, pi->options->rb_enc);
             *slot = sym;
         }
         pi->h->obj = sym;
@@ -489,8 +489,8 @@ add_text(PInfo pi, char *text, int closed) {
         from_base64(text, (uchar*)str);
         v = rb_str_new(str, str_size);
 #if HAS_ENCODING_SUPPORT
-        if (0 != pi->encoding) {
-            rb_enc_associate(v, pi->encoding);
+        if (0 != pi->options->rb_enc) {
+            rb_enc_associate(v, pi->options->rb_enc);
         }
 #endif
         if (0 != pi->circ_array) {
@@ -508,7 +508,7 @@ add_text(PInfo pi, char *text, int closed) {
         
         from_base64(text, (uchar*)str);
         if (Qundef == (sym = ox_cache_get(ox_symbol_cache, str, &slot))) {
-	    sym = str2sym(str, pi->encoding);
+	    sym = str2sym(str, pi->options->rb_enc);
             *slot = sym;
         }
         pi->h->obj = sym;
@@ -572,7 +572,7 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
     }
     h = pi->h;
     h->type = *ename;
-    h->var = get_var_sym_from_attrs(attrs, pi->encoding);
+    h->var = get_var_sym_from_attrs(attrs, pi->options->rb_enc);
     switch (h->type) {
     case NilClassCode:
         h->obj = Qnil;

data/ext/ox/ox.c

@@ -62,6 +62,7 @@ ID	ox_end_element_id;
 ID	ox_end_id;
 ID	ox_error_id;
 ID	ox_excl_id;
+ID	ox_external_encoding_id;
 ID	ox_fileno_id;
 ID	ox_inspect_id;
 ID	ox_instruct_id;
@@ -145,7 +146,8 @@ struct _Options	 ox_default_options = {
     No,			/* xsd_date */
     NoMode,		/* mode */
     StrictEffort,	/* effort */
-    Yes			/* sym_keys */
+    Yes,		/* sym_keys */
+    0			/* rb_enc */
 };
 
 extern ParseCallbacks	ox_obj_callbacks;
@@ -241,6 +243,9 @@ set_def_opts(VALUE self, VALUE opts) {
     } else {
 	Check_Type(v, T_STRING);
 	strncpy(ox_default_options.encoding, StringValuePtr(v), sizeof(ox_default_options.encoding) - 1);
+#ifdef HAVE_RUBY_ENCODING_H
+	ox_default_options.rb_enc = rb_enc_find(ox_default_options.encoding);
+#endif
     }
 
     v = rb_hash_aref(opts, indent_sym);
@@ -357,7 +362,7 @@ to_gen(VALUE self, VALUE ruby_xml) {
 }
 
 static VALUE
-load(char *xml, int argc, VALUE *argv, VALUE self) {
+load(char *xml, int argc, VALUE *argv, VALUE self, VALUE encoding) {
     VALUE		obj;
     struct _Options	options = ox_default_options;
     
@@ -397,6 +402,17 @@ load(char *xml, int argc, VALUE *argv, VALUE self) {
 	    options.sym_keys = (Qfalse == v) ? No : Yes;
 	}
     }
+#ifdef HAVE_RUBY_ENCODING_H
+    if ('\0' == *options.encoding) {
+	if (Qnil != encoding) {
+	    options.rb_enc = rb_enc_from_index(rb_enc_get_index(encoding));
+	} else {
+	    options.rb_enc = 0;
+	}
+    } else if (0 == options.rb_enc) {
+	options.rb_enc = rb_enc_find(options.encoding);
+    }
+#endif
     switch (options.mode) {
     case ObjMode:
 	obj = ox_parse(xml, ox_obj_callbacks, 0, &options);
@@ -440,6 +456,7 @@ load_str(int argc, VALUE *argv, VALUE self) {
     char	*xml;
     size_t	len;
     VALUE	obj;
+    VALUE	encoding;
     
     Check_Type(*argv, T_STRING);
     /* the xml string gets modified so make a copy of it */
@@ -449,8 +466,13 @@ load_str(int argc, VALUE *argv, VALUE self) {
     } else {
 	xml = ALLOCA_N(char, len);
     }
+#ifdef HAVE_RUBY_ENCODING_H
+    encoding = rb_obj_encoding(*argv);
+#else
+    encoding = Qnil;
+#endif
     strcpy(xml, StringValuePtr(*argv));
-    obj = load(xml, argc - 1, argv + 1, self);
+    obj = load(xml, argc - 1, argv + 1, self, encoding);
     if (SMALL_XML < len) {
 	xfree(xml);
     }
@@ -502,7 +524,7 @@ load_file(int argc, VALUE *argv, VALUE self) {
     }
     fclose(f);
     xml[len] = '\0';
-    obj = load(xml, argc - 1, argv + 1, self);
+    obj = load(xml, argc - 1, argv + 1, self, Qnil);
     if (SMALL_XML < len) {
 	xfree(xml);
     }
@@ -706,6 +728,7 @@ void Init_ox() {
     ox_comment_id = rb_intern("comment");
     ox_den_id = rb_intern("@den");
     ox_doctype_id = rb_intern("doctype");
+    ox_external_encoding_id = rb_intern("external_encoding");
     ox_end_element_id = rb_intern("end_element");
     ox_end_id = rb_intern("@end");
     ox_error_id = rb_intern("error");

data/ext/ox/ox.h

@@ -186,6 +186,11 @@ typedef struct _Options {
     char	mode;		/* LoadMode */
     char	effort;		/* Effort */
     char	sym_keys;	/* symbolize keys */
+#ifdef HAVE_RUBY_ENCODING_H
+    rb_encoding	*rb_enc;
+#else
+    void	*rb_enc;
+#endif
 } *Options;
 
 /* parse information structure */
@@ -197,11 +202,6 @@ struct _PInfo {
     VALUE		obj;
     ParseCallbacks	pcb;
     CircArray		circ_array;
-#ifdef HAVE_RUBY_ENCODING_H
-    rb_encoding		*encoding;
-#else
-    void		*encoding;
-#endif
     unsigned long	id;		/* set for text types when cirs_array is set */
     Options		options;
 };
@@ -234,6 +234,7 @@ extern ID	ox_end_element_id;
 extern ID	ox_end_id;
 extern ID	ox_error_id;
 extern ID	ox_excl_id;
+extern ID	ox_external_encoding_id;
 extern ID	ox_fileno_id;
 extern ID	ox_inspect_id;
 extern ID	ox_instruct_id;

data/ext/ox/parse.c

@@ -116,7 +116,6 @@ ox_parse(char *xml, ParseCallbacks pcb, char **endp, Options options) {
     pi.pcb = pcb;
     pi.obj = Qnil;
     pi.circ_array = 0;
-    pi.encoding = 0;
     pi.options = options;
     while (1) {
 	next_non_white(&pi);	/* skip white space */
@@ -797,10 +796,10 @@ read_coded_chars(PInfo pi, char *text) {
 	    pi->s = s;
 	    if (u <= 0x000000000000007FULL) {
 		*text++ = (char)u;
-	    } else if (ox_utf8_encoding == pi->encoding) {
+	    } else if (ox_utf8_encoding == pi->options->rb_enc) {
 		text = ucs_to_utf8_chars(text, u);
-	    } else if (0 == pi->encoding) {
-		pi->encoding = ox_utf8_encoding;
+	    } else if (0 == pi->options->rb_enc) {
+		pi->options->rb_enc = ox_utf8_encoding;
 		text = ucs_to_utf8_chars(text, u);
 	    } else {
 		/*raise_error("Invalid encoding, need UTF-8 or UTF-16 encoding to parse &#nnnn; character sequences.", pi->str, pi->s); */
@@ -857,11 +856,11 @@ collapse_special(PInfo pi, char *str) {
 		}
 		if (u <= 0x000000000000007FULL) {
 		    *b++ = (char)u;
-		} else if (ox_utf8_encoding == pi->encoding) {
+		} else if (ox_utf8_encoding == pi->options->rb_enc) {
 		    b = ucs_to_utf8_chars(b, u);
 		    /* TBD support UTF-16 */
-		} else if (0 == pi->encoding) {
-		    pi->encoding = ox_utf8_encoding;
+		} else if (0 == pi->options->rb_enc) {
+		    pi->options->rb_enc = ox_utf8_encoding;
 		    b = ucs_to_utf8_chars(b, u);
 		} else {
 		    /* raise_error("Invalid encoding, need UTF-8 or UTF-16 encoding to parse &#nnnn; character sequences.", pi->str, pi->s);*/

data/ext/ox/sax.c

@@ -321,7 +321,13 @@ sax_drive_init(SaxDrive dr, VALUE handler, VALUE io, int convert) {
     dr->has_error = respond_to(handler, ox_error_id);
 #if HAS_ENCODING_SUPPORT
     if ('\0' == *ox_default_options.encoding) {
-        dr->encoding = 0;
+	VALUE	encoding;
+
+	if (rb_respond_to(io, ox_external_encoding_id) && Qnil != (encoding = rb_funcall(io, ox_external_encoding_id, 0))) {
+	    dr->encoding = rb_enc_from_index(rb_enc_get_index(encoding));
+	} else {
+	    dr->encoding = 0;
+	}
     } else {
         dr->encoding = rb_enc_find(ox_default_options.encoding);
     }

data/lib/ox/version.rb

@@ -1,5 +1,5 @@
 
 module Ox
   # Current version of the module. 
-  VERSION = '1.6.6'
+  VERSION = '1.6.7'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ox
 version: !ruby/object:Gem::Version
-  version: 1.6.6
+  version: 1.6.7
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-11-02 00:00:00.000000000 Z
+date: 2012-11-15 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