xmlhash 1.2.2 → 1.2.4

data/History.txt

@@ -1,3 +1,11 @@
+=== 1.2.4 / 2012-09-27
+
+* set the encoding of the result to the encoding of the input
+
+=== 1.2.3 / 2012-06-22
+
+* (pulled)
+
 === 1.2.2 / 2012-06-20
 
 * use pkg-config for libxml

data/ext/xmlhash/xmlhash.c

@@ -1,6 +1,16 @@
 #include <assert.h>
 #include <ruby.h>
-#include <st.h>
+#ifdef HAVE_RUBY_ST_H
+# include <ruby/st.h>
+#else
+# include <st.h>
+#endif
+
+/* API_VERSION_CODE is only defined in those we want */
+#ifdef HAVE_RUBY_ENCODING_H
+# include <ruby/encoding.h>
+#endif
+
 #include <libxml/parser.h>
 #include <libxml/xmlreader.h>
 
@@ -8,6 +18,9 @@ static VALUE m_current = Qnil;
 static VALUE m_stack = Qnil;
 static VALUE m_cstring = Qnil;
 static VALUE m_result = Qnil;
+#ifdef HAVE_RUBY_ENCODING_H
+static rb_encoding *m_current_encoding = NULL;
+#endif
 
 void init_XmlhashParserData()
 {
@@ -18,9 +31,10 @@ void init_XmlhashParserData()
 
 void xml_hash_start_element(const xmlChar *name)
 {
-  // needed for further attributes
+  VALUE pair;
+  /* needed for further attributes */
   m_current = rb_hash_new();
-  VALUE pair = rb_ary_new();
+  pair = rb_ary_new();
   rb_ary_push(pair, rb_str_new2((const char*)name));
   rb_ary_push(pair, m_current);
   rb_ary_push(m_stack, pair);
@@ -29,18 +43,24 @@ void xml_hash_start_element(const xmlChar *name)
 
 void xml_hash_end_element(const xmlChar *name)
 {
+  VALUE pair, cname, chash, phash, obj;
+
   assert(m_stack != Qnil);
-  VALUE pair = rb_ary_pop(m_stack);
+  pair = rb_ary_pop(m_stack);
   assert(pair != Qnil);
-  VALUE cname = rb_ary_entry(pair, 0);
-  VALUE chash = rb_ary_entry(pair, 1);
+  cname = rb_ary_entry(pair, 0);
+  chash = rb_ary_entry(pair, 1);
   assert(!strcmp((const char*)name, RSTRING_PTR(cname)));
 
   if (rb_obj_is_kind_of(chash, rb_cHash) && RHASH_SIZE(chash) == 0) {
-    // now check if the cstring array contains non-empty string
-    VALUE string = rb_ary_join(m_cstring, Qnil);
-    const char *string_ptr = RSTRING_PTR(string);
-    long string_len = RSTRING_LEN(string);
+    VALUE string;
+    const char *string_ptr;
+    long string_len;
+
+    /* now check if the cstring array contains non-empty string */
+    string = rb_ary_join(m_cstring, Qnil);
+    string_ptr = RSTRING_PTR(string);
+    string_len = RSTRING_LEN(string);
     while (string_len > 0 && (string_ptr[0] == ' ' || string_ptr[0] == '\t' || string_ptr[0] == '\n')) {
       string_ptr++;
       string_len--;
@@ -48,7 +68,7 @@ void xml_hash_end_element(const xmlChar *name)
     while (string_len > 0 && (string_ptr[string_len-1] == ' ' || string_ptr[string_len-1] == '\t' || string_ptr[string_len-1] == '\n')) {
       string_len--;
     }
-    // avoid overwriting empty hash with empty string
+    /* avoid overwriting empty hash with empty string */
     if (string_len > 0)
       chash = string;
   }
@@ -58,10 +78,9 @@ void xml_hash_end_element(const xmlChar *name)
   }
 
   pair = rb_ary_entry(m_stack, RARRAY_LEN(m_stack)-1);
-  //VALUE pname = rb_ary_entry(pair, 0);
-  VALUE phash = rb_ary_entry(pair, 1);
+  phash = rb_ary_entry(pair, 1);
 
-  VALUE obj = rb_hash_aref(phash, cname);
+  obj = rb_hash_aref(phash, cname);
   if (obj != Qnil) {
     if (rb_obj_is_kind_of(obj, rb_cArray)) {
       rb_ary_push(obj, chash);
@@ -72,20 +91,36 @@ void xml_hash_end_element(const xmlChar *name)
       rb_hash_aset(phash, cname, nobj);
     }
   } else {
-    // implement force_array
+    /* implement force_array */
     rb_hash_aset(phash, cname, chash);
   }
 }
 
 void xml_hash_add_attribute(const xmlChar *name, const xmlChar *value)
 {
+#ifdef HAVE_RUBY_ENCODING_H
+  VALUE v_name, v_value;
+#endif
+
   assert(m_current != Qnil);
+#ifdef HAVE_RUBY_ENCODING_H
+  v_name = rb_external_str_new_with_enc((const char*)name, xmlStrlen(name), m_current_encoding);
+  v_value = rb_external_str_new_with_enc((const char*)value, xmlStrlen(value), m_current_encoding);
+  rb_hash_aset(m_current, v_name, v_value);
+#else
   rb_hash_aset(m_current, rb_str_new2((const char*)name), rb_str_new2((const char*)value));
+#endif
 }
 
 void xml_hash_add_text(const xmlChar *text)
 {
+#ifdef HAVE_RUBY_ENCODING_H
+  VALUE str;
+  str = rb_external_str_new_with_enc((const char*)text, xmlStrlen(text), m_current_encoding);
+  rb_ary_push(m_cstring, str);
+#else
   rb_ary_push(m_cstring, rb_str_new2((const char*)text));
+#endif
 }
 
 void processAttribute(xmlTextReaderPtr reader) 
@@ -153,7 +188,10 @@ static VALUE parse_xml_hash(VALUE self, VALUE rb_xml)
   int ret;
 
   Check_Type(rb_xml, T_STRING);
- 
+#ifdef HAVE_RUBY_ENCODING_H
+  m_current_encoding = rb_enc_get(rb_xml);
+#endif
+
   data = (char*)calloc(RSTRING_LEN(rb_xml), sizeof(char));
   memcpy(data, StringValuePtr(rb_xml), RSTRING_LEN(rb_xml));
 
@@ -173,13 +211,18 @@ static VALUE parse_xml_hash(VALUE self, VALUE rb_xml)
   }
 
   free(data);
+#ifdef HAVE_RUBY_ENCODING_H
+  m_current_encoding = 0;
+#endif
   return m_result;
 }
 
 void Init_xmlhash()
 {
+  VALUE mXmlhash;
+
   LIBXML_TEST_VERSION
-  VALUE mXmlhash = rb_define_module("Xmlhash");
+  mXmlhash = rb_define_module("Xmlhash");
   rb_define_singleton_method(mXmlhash, "parse", &parse_xml_hash, 1);
   m_stack = rb_ary_new();
   rb_global_variable(&m_stack);

data/lib/xmlhash.rb

@@ -1,5 +1,5 @@
 require 'xmlhash/xmlhash'
 
 module Xmlhash
-  VERSION = '1.2.2'
+  VERSION = '1.2.4'
 end 

data/test/test_xmlhash.rb

@@ -1,3 +1,5 @@
+# encoding: UTF-8
+
 require "test/unit"
 require "xmlhash"
 require 'json'
@@ -89,4 +91,15 @@ eos
       assert_equal ret, rubyoutput
   end
 
+  def test_encoding
+     xml = "<?xml version='1.0' encoding='UTF-8'?><name>Adrian Schröter</name>"
+
+     ret = Xmlhash.parse(xml)
+     assert_equal ret, "Adrian Schröter"
+
+     xml = "<?xml version='1.0' encoding='UTF-8'?><name value='Adrian Schröter'/>"
+     ret = Xmlhash.parse(xml)
+     assert_equal ret, {"value"=>"Adrian Schröter"}
+  end
+
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: xmlhash
 version: !ruby/object:Gem::Version
-  version: 1.2.2
+  version: 1.2.4
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-06-20 00:00:00.000000000 Z
+date: 2012-09-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: pkg-config
@@ -66,7 +66,7 @@ dependencies:
     requirements:
     - - ~>
       - !ruby/object:Gem::Version
-        version: '3.0'
+        version: '3.1'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
@@ -74,7 +74,7 @@ dependencies:
     requirements:
     - - ~>
       - !ruby/object:Gem::Version
-        version: '3.0'
+        version: '3.1'
 description: ! 'A small C module that wraps libxml2''s xmlreader to parse a XML
 
   string into a ruby hash'