ox 1.8.6 → 1.8.7

data/README.md

@@ -34,6 +34,14 @@ A fast XML parser and Object marshaller as a Ruby gem.
 
 ## <a name="release">Release Notes</a>
 
+### Release 1.8.7
+
+ - Added a correct check for element open and close names.
+
+ - Changed raised Exceptions to customer classes that inherit from StandardError.
+
+ - Fixed a few minor bugs.
+
 ### Release 1.8.6
 
  - Removed broken check for matching start and end element names in SAX mode. The names are still included in the

data/ext/ox/cache.c

@@ -46,8 +46,8 @@ struct _Cache {
 static void     slot_print(Cache cache, unsigned int depth);
 
 static char* dupstr(const char *s) {
-    size_t	len = strlen(s);
-    char	*d = ALLOC_N(char, len + 1);
+    size_t	len = strlen(s) + 1;
+    char	*d = ALLOC_N(char, len);
 
     memcpy(d, s, len);
 
@@ -63,7 +63,7 @@ ox_cache_new(Cache *cache) {
 }
 
 VALUE
-ox_cache_get(Cache cache, const char *key, VALUE **slot) {
+ox_cache_get(Cache cache, const char *key, VALUE **slot, char **keyp) {
     unsigned char       *k = (unsigned char*)key;
     Cache               *cp;
 
@@ -107,7 +107,9 @@ ox_cache_get(Cache cache, const char *key, VALUE **slot) {
         }
     }
     *slot = &cache->value;
-
+    if (0 != keyp) {
+	*keyp = cache->key;
+    }
     return cache->value;
 }
 

data/ext/ox/cache.h

@@ -37,7 +37,7 @@ typedef struct _Cache   *Cache;
 
 extern void     ox_cache_new(Cache *cache);
 
-extern VALUE    ox_cache_get(Cache cache, const char *key, VALUE **slot);
+extern VALUE    ox_cache_get(Cache cache, const char *key, VALUE **slot, char **keyp);
 
 extern void     ox_cache_print(Cache cache);
 

data/ext/ox/cache_test.c

@@ -49,7 +49,7 @@ ox_cache_test() {
 
     ox_cache_new(&c);
     for (d = data; 0 != *d; d++) {
-        v = ox_cache_get(c, *d, &slot);
+        v = ox_cache_get(c, *d, &slot, 0);
         if (Qundef == v) {
             if (0 == slot) {
                 /*printf("*** failed to get a slot for %s\n", *d); */

data/ext/ox/gen_load.c

@@ -326,7 +326,7 @@ add_element(PInfo pi, const char *ename, Attr attrs, int hasChildren) {
             VALUE   *slot;
 
 	    if (Yes == pi->options->sym_keys) {
-		if (Qundef == (sym = ox_cache_get(ox_symbol_cache, attrs->name, &slot))) {
+		if (Qundef == (sym = ox_cache_get(ox_symbol_cache, attrs->name, &slot, 0))) {
 #if HAS_ENCODING_SUPPORT
 		    if (0 != pi->options->rb_enc) {
 			VALUE	rstr = rb_str_new2(attrs->name);
@@ -438,7 +438,7 @@ add_instruct(PInfo pi, const char *name, Attr attrs, const char *content) {
             VALUE   *slot;
 
 	    if (Yes == pi->options->sym_keys) {
-		if (Qundef == (sym = ox_cache_get(ox_symbol_cache, attrs->name, &slot))) {
+		if (Qundef == (sym = ox_cache_get(ox_symbol_cache, attrs->name, &slot, 0))) {
 #if HAS_ENCODING_SUPPORT
 		    if (0 != pi->options->rb_enc) {
 			VALUE	rstr = rb_str_new2(attrs->name);

data/ext/ox/obj_load.c

@@ -104,7 +104,7 @@ name2var(const char *name, void *encoding) {
 
     if ('0' <= *name && *name <= '9') {
         var_id = INT2NUM(atoi(name));
-    } else if (Qundef == (var_id = ox_cache_get(ox_attr_cache, name, &slot))) {
+    } else if (Qundef == (var_id = ox_cache_get(ox_attr_cache, name, &slot, 0))) {
 #ifdef HAVE_RUBY_ENCODING_H
 	if (0 != encoding) {
 	    VALUE	rstr = rb_str_new2(name);
@@ -226,7 +226,7 @@ classname2class(const char *name, PInfo pi, VALUE base_class) {
     VALUE       *slot;
     VALUE       clas;
             
-    if (Qundef == (clas = ox_cache_get(ox_class_cache, name, &slot))) {
+    if (Qundef == (clas = ox_cache_get(ox_class_cache, name, &slot, 0))) {
         char            class_name[1024];
         char            *s;
         const char      *n = name;
@@ -483,7 +483,7 @@ add_text(PInfo pi, char *text, int closed) {
         VALUE   sym;
         VALUE   *slot;
 
-        if (Qundef == (sym = ox_cache_get(ox_symbol_cache, text, &slot))) {
+        if (Qundef == (sym = ox_cache_get(ox_symbol_cache, text, &slot, 0))) {
 	    sym = str2sym(text, (void*)pi->options->rb_enc);
             *slot = sym;
         }
@@ -532,7 +532,7 @@ add_text(PInfo pi, char *text, int closed) {
         char            *str = ALLOCA_N(char, str_size + 1);
         
         from_base64(text, (uchar*)str);
-        if (Qundef == (sym = ox_cache_get(ox_symbol_cache, str, &slot))) {
+        if (Qundef == (sym = ox_cache_get(ox_symbol_cache, str, &slot, 0))) {
 	    sym = str2sym(str, (void*)pi->options->rb_enc);
             *slot = sym;
         }

data/ext/ox/ox.c

@@ -95,6 +95,7 @@ VALUE	ox_encoding_sym;
 VALUE	ox_empty_string;
 VALUE	ox_zero_fixnum;
 
+VALUE	ox_arg_error_class;
 VALUE	ox_bag_clas;
 VALUE	ox_cdata_clas;
 VALUE	ox_comment_clas;
@@ -103,6 +104,7 @@ VALUE	ox_doctype_clas;
 VALUE	ox_document_clas;
 VALUE	ox_element_clas;
 VALUE	ox_instruct_clas;
+VALUE	ox_parse_error_class;
 VALUE	ox_stringio_class;
 VALUE	ox_struct_class;
 VALUE	ox_time_class;
@@ -175,7 +177,7 @@ defuse_bom(char *xml, Options options) {
 	    options->rb_enc = ox_utf8_encoding;
 	    xml += 3;
 	} else {
-	    rb_raise(rb_eArgError, "Invalid BOM in XML string.\n");
+	    rb_raise(ox_parse_error_class, "Invalid BOM in XML string.\n");
 	}
 	break;
 #if 0
@@ -184,7 +186,7 @@ defuse_bom(char *xml, Options options) {
 	    options->rb_enc = ox_utf16be_encoding;
 	    xml += 2;
 	} else {
-	    rb_raise(rb_eArgError, "Invalid BOM in XML string.\n");
+	    rb_raise(ox_parse_error_class, "Invalid BOM in XML string.\n");
 	}
 	break;
     case 0xFF: /* UTF-16LE or UTF-32LE */
@@ -197,7 +199,7 @@ defuse_bom(char *xml, Options options) {
 		xml += 2;
 	    }
 	} else {
-	    rb_raise(rb_eArgError, "Invalid BOM in XML string.\n");
+	    rb_raise(ox_parse_error_class, "Invalid BOM in XML string.\n");
 	}
 	break;
     case 0x00: /* UTF-32BE */
@@ -205,7 +207,7 @@ defuse_bom(char *xml, Options options) {
 	    options->rb_enc = ox_utf32be_encoding;
 	    xml += 4;
 	} else {
-	    rb_raise(rb_eArgError, "Invalid BOM in XML string.\n");
+	    rb_raise(ox_parse_error_class, "Invalid BOM in XML string.\n");
 	}
 	break;
 #endif
@@ -332,7 +334,7 @@ set_def_opts(VALUE self, VALUE opts) {
     } else if (limited_sym == v) {
 	ox_default_options.mode = LimMode;
     } else {
-	rb_raise(rb_eArgError, ":mode must be :object, :generic, :limited, or nil.\n");
+	rb_raise(ox_parse_error_class, ":mode must be :object, :generic, :limited, or nil.\n");
     }
 
     v = rb_hash_aref(opts, effort_sym);
@@ -345,7 +347,7 @@ set_def_opts(VALUE self, VALUE opts) {
     } else if (auto_define_sym == v) {
 	ox_default_options.effort = AutoEffort;
     } else {
-	rb_raise(rb_eArgError, ":effort must be :strict, :tolerant, :auto_define, or nil.\n");
+	rb_raise(ox_parse_error_class, ":effort must be :strict, :tolerant, :auto_define, or nil.\n");
     }
     for (o = ynos; 0 != o->attr; o++) {
 	v = rb_hash_lookup(opts, o->sym);
@@ -356,7 +358,7 @@ set_def_opts(VALUE self, VALUE opts) {
 	} else if (Qfalse == v) {
 	    *o->attr = No;
 	} else {
-	    rb_raise(rb_eArgError, "%s must be true or false.\n", rb_id2name(SYM2ID(o->sym)));
+	    rb_raise(ox_parse_error_class, "%s must be true or false.\n", rb_id2name(SYM2ID(o->sym)));
 	}
     }
     return Qnil;
@@ -446,7 +448,7 @@ load(char *xml, int argc, VALUE *argv, VALUE self, VALUE encoding) {
 	    } else if (limited_sym == v) {
 		options.mode = LimMode;
 	    } else {
-		rb_raise(rb_eArgError, ":mode must be :generic, :object, or :limited.\n");
+		rb_raise(ox_parse_error_class, ":mode must be :generic, :object, or :limited.\n");
 	    }
 	}
 	if (Qnil != (v = rb_hash_lookup(h, effort_sym))) {
@@ -457,7 +459,7 @@ load(char *xml, int argc, VALUE *argv, VALUE self, VALUE encoding) {
 	    } else if (strict_sym == v) {
 		options.effort = StrictEffort;
 	    } else {
-		rb_raise(rb_eArgError, ":effort must be :strict, :tolerant, or :auto_define.\n");
+		rb_raise(ox_parse_error_class, ":effort must be :strict, :tolerant, or :auto_define.\n");
 	    }
 	}
 	if (Qnil != (v = rb_hash_lookup(h, trace_sym))) {
@@ -625,7 +627,7 @@ sax_parse(int argc, VALUE *argv, VALUE self) {
     int convert = 0;
 
     if (argc < 2) {
-	rb_raise(rb_eArgError, "Wrong number of arguments to sax_parse.\n");
+	rb_raise(ox_parse_error_class, "Wrong number of arguments to sax_parse.\n");
     }
     if (3 <= argc && rb_cHash == rb_obj_class(argv[2])) {
 	VALUE	h = argv[2];
@@ -657,19 +659,19 @@ parse_dump_options(VALUE ropts, Options copts) {
 	
 	if (Qnil != (v = rb_hash_lookup(ropts, indent_sym))) {
 	    if (rb_cFixnum != rb_obj_class(v)) {
-		rb_raise(rb_eArgError, ":indent must be a Fixnum.\n");
+		rb_raise(ox_parse_error_class, ":indent must be a Fixnum.\n");
 	    }
 	    copts->indent = NUM2INT(v);
 	}
 	if (Qnil != (v = rb_hash_lookup(ropts, trace_sym))) {
 	    if (rb_cFixnum != rb_obj_class(v)) {
-		rb_raise(rb_eArgError, ":trace must be a Fixnum.\n");
+		rb_raise(ox_parse_error_class, ":trace must be a Fixnum.\n");
 	    }
 	    copts->trace = NUM2INT(v);
 	}
 	if (Qnil != (v = rb_hash_lookup(ropts, ox_encoding_sym))) {
 	    if (rb_cString != rb_obj_class(v)) {
-		rb_raise(rb_eArgError, ":encoding must be a String.\n");
+		rb_raise(ox_parse_error_class, ":encoding must be a String.\n");
 	    }
 	    strncpy(copts->encoding, StringValuePtr(v), sizeof(copts->encoding) - 1);
 	}
@@ -681,7 +683,7 @@ parse_dump_options(VALUE ropts, Options copts) {
 	    } else if (strict_sym == v) {
 		copts->effort = StrictEffort;
 	    } else {
-		rb_raise(rb_eArgError, ":effort must be :strict, :tolerant, or :auto_define.\n");
+		rb_raise(ox_parse_error_class, ":effort must be :strict, :tolerant, or :auto_define.\n");
 	    }
 	}
 	for (o = ynos; 0 != o->attr; o++) {
@@ -693,7 +695,7 @@ parse_dump_options(VALUE ropts, Options copts) {
 		} else if (rb_cFalseClass == c) {
 		    *o->attr = No;
 		} else {
-		    rb_raise(rb_eArgError, "%s must be true or false.\n", rb_id2name(SYM2ID(o->sym)));
+		    rb_raise(ox_parse_error_class, "%s must be true or false.\n", rb_id2name(SYM2ID(o->sym)));
 		}
 	    }
 	}
@@ -846,6 +848,8 @@ void Init_ox() {
 
     ox_time_class = rb_const_get(rb_cObject, rb_intern("Time"));
     ox_date_class = rb_const_get(rb_cObject, rb_intern("Date"));
+    ox_parse_error_class = rb_const_get_at(Ox, rb_intern("ParseError"));
+    ox_arg_error_class = rb_const_get_at(Ox, rb_intern("ArgError"));
     ox_struct_class = rb_const_get(rb_cObject, rb_intern("Struct"));
     ox_stringio_class = rb_const_get(rb_cObject, rb_intern("StringIO"));
 
@@ -910,5 +914,5 @@ _ox_raise_error(const char *msg, const char *xml, const char *current, const cha
 	    xline++;
 	}
     }
-    rb_raise(rb_eSyntaxError, "%s at line %d, column %d [%s:%d]\n", msg, xline, col, file, line);
+    rb_raise(ox_parse_error_class, "%s at line %d, column %d [%s:%d]\n", msg, xline, col, file, line);
 }

data/ext/ox/ox.h

@@ -272,12 +272,14 @@ extern VALUE		ox_utf8_encoding;
 extern void		*ox_utf8_encoding;
 #endif
 
+extern VALUE	ox_arg_error_class;
 extern VALUE	ox_date_class;
 extern VALUE	ox_empty_string;
 extern VALUE	ox_encoding_sym;
+extern VALUE	ox_parse_error_class;
+extern VALUE	ox_stringio_class;
 extern VALUE	ox_struct_class;
 extern VALUE	ox_time_class;
-extern VALUE	ox_stringio_class;
 extern VALUE	ox_zero_fixnum;
 
 extern VALUE	ox_document_clas;

data/ext/ox/sax.c

@@ -202,11 +202,11 @@ is_white(char c) {
 }
 
 inline static VALUE
-str2sym(const char *str, SaxDrive dr) {
+str2sym(const char *str, SaxDrive dr, char **strp) {
     VALUE       *slot;
     VALUE       sym;
 
-    if (Qundef == (sym = ox_cache_get(ox_symbol_cache, str, &slot))) {
+    if (Qundef == (sym = ox_cache_get(ox_symbol_cache, str, &slot, strp))) {
 #if HAS_ENCODING_SUPPORT
         if (0 != dr->encoding) {
 	    VALUE	rstr = rb_str_new2(str);
@@ -312,7 +312,7 @@ sax_drive_init(SaxDrive dr, VALUE handler, VALUE io, int convert) {
         }
 #endif
     } else {
-        rb_raise(rb_eArgError, "sax_parser io argument must respond to readpartial() or read().\n");
+        rb_raise(ox_arg_error_class, "sax_parser io argument must respond to readpartial() or read().\n");
     }
     dr->buf = dr->base_buf;
     *dr->buf = '\0';
@@ -427,7 +427,7 @@ sax_drive_error(SaxDrive dr, const char *msg, int critical) {
         rb_funcall2(dr->handler, ox_error_id, 3, args);
     } else if (critical) {
         sax_drive_cleanup(dr);
-        rb_raise(rb_eSyntaxError, "%s at line %d, column %d\n", msg, dr->line, dr->col);
+        rb_raise(ox_parse_error_class, "%s at line %d, column %d\n", msg, dr->line, dr->col);
     }
 }
 
@@ -748,17 +748,16 @@ read_comment(SaxDrive dr) {
  */
 static int
 read_element(SaxDrive dr) {
+    char	*ename = 0;
     VALUE       name = Qnil;
     const char	*err;
     char        c;
-    char	*ename = 0;
     int         closed;
 
     if ('\0' == (c = read_name_token(dr))) {
         return -1;
     }
-    ename = dr->str;
-    name = str2sym(dr->str, dr);
+    name = str2sym(dr->str, dr, &ename);
     if (dr->has_start_element) {
         VALUE       args[1];
 
@@ -794,7 +793,11 @@ read_element(SaxDrive dr) {
         if (0 != read_children(dr, 0)) {
             return -1;
         }
-	/* no check of matching start and end as it would require coping the name */
+	if (0 != ename && 0 != strcmp(ename, dr->str)) {
+	    //printf("*** ename: %s  close: %s\n", ename, dr->str);
+            sax_drive_error(dr, "invalid format, element start and end names do not match", 1);
+	    return -1;
+	}
         if (0 != dr->has_end_element) {
             VALUE       args[1];
 
@@ -872,7 +875,7 @@ read_attrs(SaxDrive dr, char c, char termc, char term2, int is_xml) {
         }
 	/* TBD use symbol cache */
         if (dr->has_attr || dr->has_attr_value) {
-            name = str2sym(dr->str, dr);
+            name = str2sym(dr->str, dr, 0);
         }
         if (is_white(c)) {
             c = next_non_white(dr);
@@ -1275,7 +1278,7 @@ sax_value_as_sym(VALUE self) {
     if ('\0' == *dr->str) {
 	return Qnil;
     }
-    return str2sym(dr->str, dr);
+    return str2sym(dr->str, dr, 0);
 }
 
 static VALUE
@@ -1308,7 +1311,7 @@ sax_value_as_i(VALUE self) {
 	if ('0' <= *s && *s <= '9') {
 	    n = n * 10 + (*s - '0');
 	} else {
-	    rb_raise(rb_eArgError, "Not a valid Fixnum.\n");
+	    rb_raise(ox_arg_error_class, "Not a valid Fixnum.\n");
 	}
     }
     if (neg) {

data/lib/ox.rb

@@ -87,7 +87,7 @@ module Ox
 end
 
 require 'ox/version'
-require 'ox/invalidpath'
+require 'ox/error'
 require 'ox/hasattrs'
 require 'ox/node'
 require 'ox/comment'

data/lib/ox/error.rb

@@ -0,0 +1,22 @@
+
+module Ox
+
+  class Error < StandardError
+  end # Error
+
+  # An Exception that is raised as a result of a parse error while parsing a XML document.
+  class ParseError < Error
+  end # ParseError
+
+  # An Exception that is raised as a result of an invalid argument.
+  class ArgError < Error
+  end # ArgError
+
+  # An Exception raised if a path is not valid.
+  class InvalidPath < Error
+    def initialize(path)
+      super("#{path.join('/')} is not a valid location.")
+    end
+  end # InvalidPath
+
+end # Ox

data/lib/ox/version.rb

@@ -1,5 +1,5 @@
 
 module Ox
   # Current version of the module. 
-  VERSION = '1.8.6'
+  VERSION = '1.8.7'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ox
 version: !ruby/object:Gem::Version
-  version: 1.8.6
+  version: 1.8.7
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-02-07 00:00:00.000000000 Z
+date: 2013-02-08 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
@@ -29,9 +29,9 @@ files:
 - lib/ox/doctype.rb
 - lib/ox/document.rb
 - lib/ox/element.rb
+- lib/ox/error.rb
 - lib/ox/hasattrs.rb
 - lib/ox/instruct.rb
-- lib/ox/invalidpath.rb
 - lib/ox/node.rb
 - lib/ox/sax.rb
 - lib/ox/version.rb

data/lib/ox/invalidpath.rb

@@ -1,10 +0,0 @@
-module Ox
-
-  # 
-  class InvalidPath < Exception
-    def initialize(path)
-      super("#{path.join('/')} is not a valid location.")
-    end
-
-  end # InvalidPath
-end # Ox