/*
-Copyright (c) 2005-2009, John Hurst
+Copyright (c) 2005-2015, John Hurst
All rights reserved.
Redistribution and use in source and binary forms, with or without
#include <stack>
#include <map>
-//#undef HAVE_EXPAT
-//#define HAVE_XERCES_C
-
#ifdef HAVE_EXPAT
# ifdef HAVE_XERCES_C
# error "Both HAVE_EXPAT and HAVE_XERCES_C defined"
#include <xercesc/util/PlatformUtils.hpp>
#include <xercesc/util/XMLString.hpp>
+#include <xercesc/util/TransService.hpp>
#include <xercesc/sax/AttributeList.hpp>
#include <xercesc/sax/HandlerBase.hpp>
#include <xercesc/sax/ErrorHandler.hpp>
XERCES_CPP_NAMESPACE_USE
+
+extern "C"
+{
+ void kumu_init_xml_dom();
+ bool kumu_UTF_8_to_XercesString(const std::string& in_str, std::basic_string<XMLCh>& out_str);
+ bool kumu_UTF_8_to_XercesString_p(const char* in_str, std::basic_string<XMLCh>& out_str);
+ bool kumu_XercesString_to_UTF_8(const std::basic_string<XMLCh>& in_str, std::string& out_str);
+ bool kumu_XercesString_to_UTF_8_p(const XMLCh* in_str, std::string& out_str);
+}
+
#endif
using namespace Kumu;
//
void
-Kumu::XMLElement::Render(std::string& outbuf) const
+Kumu::XMLElement::Render(std::string& outbuf, const bool& pretty) const
{
outbuf = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
- RenderElement(outbuf, 0);
+ RenderElement(outbuf, 0, pretty);
}
//
//
void
-Kumu::XMLElement::RenderElement(std::string& outbuf, ui32_t depth) const
+Kumu::XMLElement::RenderElement(std::string& outbuf, const ui32_t& depth, const bool& pretty) const
{
- add_spacer(outbuf, depth);
+ if ( pretty )
+ {
+ add_spacer(outbuf, depth);
+ }
outbuf += "<";
outbuf += m_Name;
// render attributes
- for ( Attr_i i = m_AttrList.begin(); i != m_AttrList.end(); i++ )
+ for ( Attr_i i = m_AttrList.begin(); i != m_AttrList.end(); ++i )
{
outbuf += " ";
outbuf += (*i).name;
// render body
if ( m_Body.length() > 0 )
- outbuf += m_Body;
+ {
+ outbuf += m_Body;
+ }
- for ( Elem_i i = m_ChildList.begin(); i != m_ChildList.end(); i++ )
- (*i)->RenderElement(outbuf, depth + 1);
+ for ( Elem_i i = m_ChildList.begin(); i != m_ChildList.end(); ++i )
+ {
+ (*i)->RenderElement(outbuf, depth + 1, pretty);
+ }
- add_spacer(outbuf, depth);
+ if ( pretty )
+ {
+ add_spacer(outbuf, depth);
+ }
}
else if ( m_Body.length() > 0 )
{
}
}
+//
+bool
+Kumu::XMLElement::ParseString(const ByteString& document)
+{
+ return ParseString((const char*)document.RoData(), document.Length());
+}
+
+//
+bool
+Kumu::XMLElement::ParseString(const std::string& document)
+{
+ return ParseString(document.c_str(), document.size());
+}
+
+//
+bool
+Kumu::XMLElement::ParseFirstFromString(const ByteString& document)
+{
+ return ParseFirstFromString((const char*)document.RoData(), document.Length());
+}
+
+//
+bool
+Kumu::XMLElement::ParseFirstFromString(const std::string& document)
+{
+ return ParseFirstFromString(document.c_str(), document.size());
+}
+
//----------------------------------------------------------------------------------------------------
//
bool
-Kumu::XMLElement::ParseString(const std::string& document)
+Kumu::XMLElement::ParseString(const char* document, ui32_t doc_len)
{
+ if ( doc_len == 0 )
+ {
+ return false;
+ }
+
XML_Parser Parser = XML_ParserCreateNS("UTF-8", '|');
if ( Parser == 0 )
XML_SetCharacterDataHandler(Parser, xph_char);
XML_SetStartNamespaceDeclHandler(Parser, xph_namespace_start);
- if ( ! XML_Parse(Parser, document.c_str(), document.size(), 1) )
+ if ( ! XML_Parse(Parser, document, doc_len, 1) )
{
- XML_ParserFree(Parser);
DefaultLogSink().Error("XML Parse error on line %d: %s\n",
XML_GetCurrentLineNumber(Parser),
XML_ErrorString(XML_GetErrorCode(Parser)));
+ XML_ParserFree(Parser);
return false;
}
XML_ParserFree(Parser);
if ( ! Ctx.Namespaces->empty() )
- m_NamespaceOwner = (void*)Ctx.Namespaces;
+ {
+ m_NamespaceOwner = (void*)Ctx.Namespaces;
+ }
return true;
}
-//------------------------------------------------------------------------------------------
-
-struct xph_test_wrapper
-{
- XML_Parser Parser;
- bool Status;
-
- xph_test_wrapper(XML_Parser p) : Parser(p), Status(false) {}
-};
-
-// expat wrapper functions, map callbacks to IASAXHandler
+// expat wrapper functions
//
static void
-xph_test_start(void* p, const XML_Char* name, const XML_Char** attrs)
+xph_start_one_shot(void* p, const XML_Char* name, const XML_Char** attrs)
{
- assert(p);
- xph_test_wrapper* Wrapper = (xph_test_wrapper*)p;
-
- Wrapper->Status = true;
- XML_StopParser(Wrapper->Parser, false);
+ xph_start(p, name, attrs);
+ XML_Parser parser = (XML_Parser)p;
+ XML_StopParser(parser, false);
}
-
//
bool
-Kumu::StringIsXML(const char* document, ui32_t len)
+Kumu::XMLElement::ParseFirstFromString(const char* document, ui32_t doc_len)
{
- if ( document == 0 )
- return false;
-
- if ( len == 0 )
- len = strlen(document);
+ if ( doc_len == 0 )
+ {
+ return false;
+ }
- XML_Parser Parser = XML_ParserCreate("UTF-8");
+ XML_Parser Parser = XML_ParserCreateNS("UTF-8", '|');
if ( Parser == 0 )
{
return false;
}
- xph_test_wrapper Wrapper(Parser);
- XML_SetUserData(Parser, (void*)&Wrapper);
- XML_SetStartElementHandler(Parser, xph_test_start);
+ ExpatParseContext Ctx(this);
+ XML_SetUserData(Parser, (void*)&Ctx);
+ XML_SetElementHandler(Parser, xph_start_one_shot, xph_end);
+ XML_SetCharacterDataHandler(Parser, xph_char);
+ XML_SetStartNamespaceDeclHandler(Parser, xph_namespace_start);
+
+ if ( ! XML_Parse(Parser, document, doc_len, 1) )
+ {
+ XML_ParserFree(Parser);
+ return false;
+ }
- XML_Parse(Parser, document, len, 1);
XML_ParserFree(Parser);
- return Wrapper.Status;
+
+ if ( ! Ctx.Namespaces->empty() )
+ {
+ m_NamespaceOwner = (void*)Ctx.Namespaces;
+ }
+
+ return true;
}
+
+
#endif
//----------------------------------------------------------------------------------------------------
#ifdef HAVE_XERCES_C
-static Mutex sg_Lock;
-static bool sg_xml_init = false;
+static Mutex sg_xerces_init_lock; // protect the xerces initialized
+static bool sg_xml_init = false; // signal initialization
+static Mutex sg_coder_lock; // protect the transcoder context
+static XMLTranscoder* sg_coder = 0;
+static const int sg_coder_buf_len = 128 * 1024;
+static char sg_coder_buf[sg_coder_buf_len + 8];
+static unsigned char sg_coder_counts[sg_coder_buf_len / sizeof(XMLCh)]; // see XMLTranscoder::transcodeFrom
+static const XMLCh sg_LS[] = { chLatin_L, chLatin_S, chNull };
+static const XMLCh sg_label_UTF_8[] = { chLatin_U, chLatin_T, chLatin_F,
+ chDash, chDigit_8, chNull};
//
void
-asdcp_init_xml_dom()
+kumu_init_xml_dom()
{
if ( ! sg_xml_init )
{
- AutoMutex AL(sg_Lock);
+ AutoMutex AL(sg_xerces_init_lock);
if ( ! sg_xml_init )
{
{
XMLPlatformUtils::Initialize();
sg_xml_init = true;
+
+ XMLTransService::Codes ret;
+ sg_coder = XMLPlatformUtils::fgTransService->makeNewTranscoderFor(sg_label_UTF_8, ret, sg_coder_buf_len);
+
+ if ( ret != XMLTransService::Ok )
+ {
+ const char* message = "Undefined Error";
+
+ switch ( ret )
+ {
+ case XMLTransService::UnsupportedEncoding: message = "Unsupported encoding"; break;
+ case XMLTransService::InternalFailure: message = "Internal failure"; break;
+ case XMLTransService::SupportFilesNotFound: message = "Support files not found"; break;
+ }
+
+ DefaultLogSink().Error("Xerces transform initialization error: %s\n", message);
+ }
}
catch (const XMLException &e)
{
}
}
+//
+bool
+kumu_XercesString_to_UTF_8(const std::basic_string<XMLCh>& in_str, std::string& out_str) {
+ return kumu_XercesString_to_UTF_8_p(in_str.c_str(), out_str);
+}
+
+//
+bool
+kumu_XercesString_to_UTF_8_p(const XMLCh* in_str, std::string& out_str)
+{
+ assert(in_str);
+ assert(sg_xml_init);
+ AutoMutex AL(sg_coder_lock);
+ ui32_t str_len = XMLString::stringLen(in_str);
+ ui32_t read_total = 0;
+
+ try
+ {
+ while ( str_len > 0 )
+ {
+#if XERCES_VERSION_MAJOR < 3
+ ui32_t read_count = 0;
+#else
+ XMLSize_t read_count = 0;
+#endif
+ ui32_t write_count = sg_coder->transcodeTo(in_str + read_total, str_len,
+ (XMLByte*)sg_coder_buf, sg_coder_buf_len,
+ read_count, XMLTranscoder::UnRep_Throw);
+
+ out_str.append(sg_coder_buf, write_count);
+ str_len -= read_count;
+ read_total += read_count;
+ assert(str_len >= 0);
+ }
+ }
+ catch (...)
+ {
+ return false;
+ }
+
+ return true;
+}
+
+//
+bool
+kumu_UTF_8_to_XercesString(const std::string& in_str, std::basic_string<XMLCh>& out_str) {
+ return kumu_UTF_8_to_XercesString_p(in_str.c_str(), out_str);
+}
+
+//
+bool
+kumu_UTF_8_to_XercesString_p(const char* in_str, std::basic_string<XMLCh>& out_str)
+{
+ assert(in_str);
+ assert(sg_xml_init);
+ AutoMutex AL(sg_coder_lock);
+ ui32_t str_len = strlen(in_str);
+ ui32_t read_total = 0;
+
+ try
+ {
+ while ( str_len > 0 )
+ {
+#if XERCES_VERSION_MAJOR < 3
+ ui32_t read_count = 0;
+#else
+ XMLSize_t read_count = 0;
+#endif
+ ui32_t write_count = sg_coder->transcodeFrom((const XMLByte*)(in_str + read_total), str_len,
+ (XMLCh*)sg_coder_buf, sg_coder_buf_len / sizeof(XMLCh),
+ read_count, sg_coder_counts);
+
+ out_str.append((XMLCh*)sg_coder_buf, write_count * sizeof(XMLCh));
+ str_len -= read_count;
+ read_total += read_count;
+ assert(str_len >= 0);
+ }
+ }
+ catch (...)
+ {
+ return false;
+ }
+
+ return true;
+}
//
class MyTreeHandler : public HandlerBase
ns_map* m_Namespaces;
std::stack<XMLElement*> m_Scope;
XMLElement* m_Root;
+ bool m_HasEncodeErrors;
public:
- MyTreeHandler(XMLElement* root) : m_Namespaces(0), m_Root(root) {
+ MyTreeHandler(XMLElement* root) : m_Namespaces(0), m_Root(root), m_HasEncodeErrors(false)
+ {
assert(m_Root);
m_Namespaces = new ns_map;
}
delete m_Namespaces;
}
- ns_map* TakeNamespaceMap() {
+ bool HasEncodeErrors() const { return m_HasEncodeErrors; }
+
+ ns_map* TakeNamespaceMap()
+ {
if ( m_Namespaces == 0 || m_Namespaces->empty() )
return 0;
ns_prefix = "";
}
- ns_map::iterator ni = m_Namespaces->find(ns_name);
+ ns_map::iterator ni = m_Namespaces->find(ns_prefix);
if ( ni != m_Namespaces->end() )
{
XERCES_CPP_NAMESPACE::AttributeList& attributes)
{
assert(x_name);
+ std::string tx_name;
+
+ if ( ! kumu_XercesString_to_UTF_8(x_name, tx_name) )
+ m_HasEncodeErrors = true;
- const char* tx_name = XMLString::transcode(x_name);
- const char* name = tx_name;
+ const char* name = tx_name.c_str();
XMLElement* Element;
const char* ns_root = name;
const char* local_name = strchr(name, ':');
for ( ui32_t i = 0; i < a_len; i++)
{
- const XMLCh* aname = attributes.getName(i);
- const XMLCh* value = attributes.getValue(i);
- assert(aname);
- assert(value);
+ std::string aname, value;
+ if ( ! kumu_XercesString_to_UTF_8(attributes.getName(i), aname) )
+ m_HasEncodeErrors = true;
- char* x_aname = XMLString::transcode(aname);
- char* x_value = XMLString::transcode(value);
+ if ( ! kumu_XercesString_to_UTF_8(attributes.getValue(i), value) )
+ m_HasEncodeErrors = true;
+
+ const char* x_aname = aname.c_str();
+ const char* x_value = value.c_str();
if ( strncmp(x_aname, "xmlns", 5) == 0 )
AddNamespace(x_aname+5, x_value);
local_name++;
Element->SetAttr(local_name, x_value);
-
- XMLString::release(&x_aname);
- XMLString::release(&x_value);
}
// map the namespace
ns_map::iterator ni = m_Namespaces->find(key);
if ( ni != m_Namespaces->end() )
Element->SetNamespace(ni->second);
-
- XMLString::release((char**)&tx_name);
}
void endElement(const XMLCh *const name) {
m_Scope.pop();
}
+#if XERCES_VERSION_MAJOR < 3
void characters(const XMLCh *const chars, const unsigned int length)
+#else
+ void characters(const XMLCh* const chars, const XMLSize_t length)
+#endif
{
if ( length > 0 )
{
- char* text = XMLString::transcode(chars);
- m_Scope.top()->AppendBody(text);
- XMLString::release(&text);
+ std::string tmp;
+ if ( ! kumu_XercesString_to_UTF_8(chars, tmp) )
+ m_HasEncodeErrors = true;
+
+ m_Scope.top()->AppendBody(tmp);
}
}
};
//
bool
-Kumu::XMLElement::ParseString(const std::string& document)
+Kumu::XMLElement::ParseString(const char* document, ui32_t doc_len)
{
- if ( document.empty() )
- return false;
+ if ( doc_len == 0 )
+ {
+ return false;
+ }
- asdcp_init_xml_dom();
+ kumu_init_xml_dom();
int errorCount = 0;
SAXParser* parser = new SAXParser();
-// #if XERCES_VERSION_MAJOR < 3
-// parser->setDoValidation(true);
-// #else
parser->setValidationScheme(SAXParser::Val_Always);
-// #endif
-
parser->setDoNamespaces(true); // optional
MyTreeHandler* docHandler = new MyTreeHandler(this);
try
{
- MemBufInputSource xmlSource(reinterpret_cast<const XMLByte*>(document.c_str()),
- static_cast<const unsigned int>(document.size()),
+ MemBufInputSource xmlSource(reinterpret_cast<const XMLByte*>(document),
+ static_cast<const unsigned int>(doc_len),
"pidc_rules_file");
parser->parse(xmlSource);
//
bool
-Kumu::StringIsXML(const char* document, ui32_t len)
+Kumu::XMLElement::ParseFirstFromString(const char* document, ui32_t doc_len)
{
- if ( document == 0 || *document == 0 )
- return false;
+ if ( doc_len == 0 )
+ {
+ return false;
+ }
- asdcp_init_xml_dom();
+ kumu_init_xml_dom();
+
+ int errorCount = 0;
+ SAXParser* parser = new SAXParser();
- if ( len == 0 )
- len = strlen(document);
+ parser->setValidationScheme(SAXParser::Val_Always);
+ parser->setDoNamespaces(true); // optional
- SAXParser parser;
+ MyTreeHandler* docHandler = new MyTreeHandler(this);
+ parser->setDocumentHandler(docHandler);
+ parser->setErrorHandler(docHandler);
XMLPScanToken token;
- bool status = false;
try
{
MemBufInputSource xmlSource(reinterpret_cast<const XMLByte*>(document),
- static_cast<const unsigned int>(len),
+ static_cast<const unsigned int>(doc_len),
"pidc_rules_file");
- if ( parser.parseFirst(xmlSource, token) )
+ if ( ! parser->parseFirst(xmlSource, token) )
{
- if ( parser.parseNext(token) )
- status = true;
+ ++errorCount;
+ }
+
+ if ( ! parser->parseNext(token) )
+ {
+ ++errorCount;
}
}
catch (...)
{
+ errorCount++;
}
- return status;
+ if ( errorCount == 0 )
+ {
+ m_NamespaceOwner = (void*)docHandler->TakeNamespaceMap();
+ }
+
+ delete parser;
+ delete docHandler;
+
+ return errorCount > 0 ? false : true;
}
//
bool
-Kumu::XMLElement::ParseString(const std::string& document)
+Kumu::XMLElement::ParseString(const char* document, ui32_t doc_len)
{
DefaultLogSink().Error("Kumu compiled without XML parser support.\n");
return false;
}
-//
bool
-Kumu::StringIsXML(const char* document, ui32_t len)
+Kumu::XMLElement::ParseFirstFromString(const char* document, ui32_t doc_len)
{
DefaultLogSink().Error("Kumu compiled without XML parser support.\n");
return false;
#endif
+//----------------------------------------------------------------------------------------------------
+
+//
+bool
+Kumu::GetXMLDocType(const ByteString& buf, std::string& ns_prefix, std::string& type_name, std::string& namespace_name,
+ AttributeList& doc_attr_list)
+{
+ return GetXMLDocType(buf.RoData(), buf.Length(), ns_prefix, type_name, namespace_name, doc_attr_list);
+}
+
+//
+bool
+Kumu::GetXMLDocType(const std::string& buf, std::string& ns_prefix, std::string& type_name, std::string& namespace_name,
+ AttributeList& doc_attr_list)
+{
+ return GetXMLDocType((const byte_t*)buf.c_str(), buf.size(), ns_prefix, type_name, namespace_name, doc_attr_list);
+}
+
+//
+bool
+Kumu::GetXMLDocType(const byte_t* buf, ui32_t buf_len, std::string& ns_prefix, std::string& type_name, std::string& namespace_name,
+ AttributeList& doc_attr_list)
+{
+ XMLElement tmp_element("tmp");
+
+ if ( ! tmp_element.ParseFirstFromString((const char*)buf, buf_len) )
+ {
+ return false;
+ }
+
+ const XMLNamespace* ns = tmp_element.Namespace();
+
+ if ( ns != 0 )
+ {
+ ns_prefix = ns->Prefix();
+ namespace_name = ns->Name();
+ }
+
+ type_name = tmp_element.GetName();
+ doc_attr_list = tmp_element.GetAttributes();
+ return true;
+}
+
+
//
// end KM_xml.cpp
//