finalize merge pr #4

Author: Arzaroth Lekva

rapidxml/c_ext/src/document_object.cpp

@@ -25,7 +25,8 @@ static void rapidxml_DocumentObject_dealloc(rapidxml_DocumentObject* self) {
 
 static int _parse(rapidxml_DocumentObject* self,
                   Py_buffer* text_buff,
-                  bool from_file, bool parse_cdata=false) {
+                  bool from_file,
+                  bool parse_cdata) {
   const char* text;
   std::vector<char> text_vector;
 
@@ -43,17 +44,14 @@ static int _parse(rapidxml_DocumentObject* self,
   }
   try {
     self->base.base.document->clear();
-	if (!parse_cdata){
-		(self->base.base.document
-			->parse<rapidxml::parse_no_utf8 | rapidxml::parse_no_data_nodes>)
-				(self->base.base.document->allocate_string(text));
-		
-	}else{
-		(self->base.base.document
-			->parse<rapidxml::parse_declaration_node>)
-				(self->base.base.document->allocate_string(text));
-	}
-    
+    char* data = self->base.base.document->allocate_string(text);
+    if (!parse_cdata) {
+      (self->base.base.document
+       ->parse<rapidxml::parse_no_utf8 | rapidxml::parse_no_data_nodes>)(data);
+    } else {
+      (self->base.base.document
+       ->parse<rapidxml::parse_declaration_node>)(data);
+    }
   } catch (rapidxml::parse_error &e) {
     PyErr_SetString(rapidxml_RapidXmlError, e.what());
     return 0;
@@ -78,7 +76,8 @@ static PyObject* rapidxml_DocumentObject_parse(rapidxml_DocumentObject* self,
   }
 
   if (!_parse(self, &text_buff,
-              (from_file_obj != NULL) && PyObject_IsTrue(from_file_obj), (read_cdata != NULL) && PyObject_IsTrue(read_cdata))) {
+              (from_file_obj != NULL) && PyObject_IsTrue(from_file_obj),
+              (read_cdata != NULL) && PyObject_IsTrue(read_cdata))) {
     return NULL;
   }
   Py_INCREF(Py_None);
@@ -106,7 +105,10 @@ static int rapidxml_DocumentObject_init(rapidxml_DocumentObject* self,
   self->base.base.underlying_obj = new rapidxml::xml_document<>();
   self->base.base.document = static_cast<rapidxml::xml_document<>*>(self->base.base.underlying_obj);
   if (text_buff.buf) {
-    return _parse(self, &text_buff, (from_file_obj != NULL) && PyObject_IsTrue(from_file_obj), (read_cdata != NULL) && PyObject_IsTrue(read_cdata)) - 1;
+    return (_parse(self, &text_buff,
+                   (from_file_obj != NULL) && PyObject_IsTrue(from_file_obj),
+                   (read_cdata != NULL) && PyObject_IsTrue(read_cdata))
+            - 1);
   }
   return 0;
 }

rapidxml/rapidxml.py

@@ -101,12 +101,11 @@ def __init__(self,
                  attribute_prefix='@',
                  cdata_key='#text',
                  always_aslist=False,
-				 parse_cdata=False):
+                 parse_cdata=False):
         DictNode.__init__(self, attribute_prefix, cdata_key, always_aslist)
         rapidxml.c_ext.Document.__init__(self, text, from_file, parse_cdata)
 
     def allocate_node(self, *args):
         return DictNode(self.attribute_prefix,
                         self.cdata_key,
                         self.always_aslist).copy(super(RapidXml, self).allocate_node(*args))
-

tests/conftest.py

@@ -22,14 +22,24 @@ def init_rapidxml():
   <test>some text</test>
 </root>""")
     return r
-	
+
 @pytest.fixture(scope="function")
 def init_rapidxml_with_CDADA():
-    datra_str = "<root><test attr1=\"one\" attr2=\"two\" attr3=\"three\"/><test2><node id=\"1\"/><node id=\"2\"/><node id=\"3\"/></test2><test>some text</test><ns2:AdditionalData><ns2:Data TID=\"AD_1\"><![CDATA[{\"Cart\":{\"expirationTime\":\"2017-04-22T09:40\",\"id\":\"b469df3b-f626-4fe3-898c-825373e546a2\",\"products\":[\"1223\"],\"creationTime\":\"2017-04-21T09:40\",\"totalPrice\":{\"currencyCode\":\"EUR\",\"amount\":\"138.000\"}}}]]></ns2:Data></ns2:AdditionalData></root>"
-    r = rapidxml.RapidXml(datra_str,
-                          from_file=False,
-                          attribute_prefix='@',
-                          cdata_key='#text',
-                          always_aslist=False,
-                          parse_cdata=True)
+    data = b"<root><test attr1=\"one\" attr2=\"two\" attr3=\"three\"/><test2><node id=\"1\"/><node id=\"2\"/><node id=\"3\"/></test2><test>some text</test><ns2:AdditionalData><ns2:Data TID=\"AD_1\"><![CDATA[{\"Cart\":{\"expirationTime\":\"2017-04-22T09:40\",\"id\":\"b469df3b-f626-4fe3-898c-825373e546a2\",\"products\":[\"1223\"],\"creationTime\":\"2017-04-21T09:40\",\"totalPrice\":{\"currencyCode\":\"EUR\",\"amount\":\"138.000\"}}}]]></ns2:Data></ns2:AdditionalData></root>"
+    r = rapidxml.RapidXml(data, parse_cdata=True)
     return r
+
+@pytest.fixture(scope="module")
+def cdata_obj():
+    return {
+        "Cart": {
+            "expirationTime": "2017-04-22T09:40",
+            "id": "b469df3b-f626-4fe3-898c-825373e546a2",
+            "products": ["1223"],
+            "creationTime":"2017-04-21T09:40",
+            "totalPrice": {
+                "currencyCode": "EUR",
+                "amount": "138.000",
+            },
+        },
+    }

tests/test_cdata.py

@@ -6,7 +6,7 @@
 # arzaroth@arzaroth.com
 #
 
-import ast
+import json
 
 def test_first_node(init_rapidxml):
     root = init_rapidxml.first_node()
@@ -38,26 +38,11 @@ def test_first_node_cdata(init_rapidxml_with_CDADA):
     assert root.first_node() == root.first_node("test")
     assert root.first_node() != root.first_node("test2")
 
-def test_last_node_cdata(init_rapidxml_with_CDADA):
+def test_last_node_cdata(init_rapidxml_with_CDADA, cdata_obj):
     root = init_rapidxml_with_CDADA.first_node()
     assert root.first_node() != root.last_node()
     assert root.first_node("test2") == root.last_node("test2")
-    object = \
-        {
-            "Cart":
-                {
-                    "expirationTime":"2017-04-22T09:40",
-                    "id":"b469df3b-f626-4fe3-898c-825373e546a2",
-                    "products":["1223"],
-                    "creationTime":"2017-04-21T09:40",
-                    "totalPrice":
-                        {
-                            "currencyCode":"EUR",
-                            "amount":"138.000"
-                        }
-                }
-        }
-    assert ast.literal_eval(root.last_node().first_node().first_node().value) == object
+    assert json.loads(root.last_node().first_node().first_node().value) == cdata_obj
 
 def test_nested_node_cdata(init_rapidxml_with_CDADA):
     test = init_rapidxml_with_CDADA.first_node().first_node("test")

tests/test_nodes.py

@@ -19,3 +19,14 @@ def test_utf8():
     assert data == r.unparse(raw=True)
     r.first_node().value = b'\x85'
     assert r.unparse(raw=True).decode('cp1252').encode('utf-8') == b"<root>\xe2\x80\xa6</root>"
+
+def test_utf8_cdata():
+    data = "<root><![CDATA[\"Càrt\"]]></root>"
+    if sys.version_info[0] >= 3:
+        data = data.encode('utf-8')
+    r = rapidxml.RapidXml(data, parse_cdata=True)
+    assert str(r) == r.unparse(pretty=True)
+    assert repr(r) == r.unparse()
+    assert data == r.unparse(raw=True)
+    r.first_node().first_node().value = b'\x85'
+    assert r.unparse(raw=True).decode('cp1252').encode('utf-8') == b"<root><![CDATA[\xe2\x80\xa6]]></root>"