merged with mangle-dev branch

src/Makefile.am

@@ -13,15 +13,14 @@ cc_sources = arabica.cpp \
 	Utils/utf16utf8codecvt.cpp \
 	Utils/utf8iso88591codecvt.cpp \
 	Utils/utf8ucs2codecvt.cpp \
+        Utils/uri.cpp \
 	XML/XMLCharacterClasses.cpp
 
-library_includedir=$(includedir)
-
-INCLUDES = -I$(top_srcdir)/include $(PARSER_HEADERS) $(BOOST_CPPFLAGS)
+AM_CPPFLAGS = -I$(top_srcdir)/include @PARSER_HEADERS@ $(BOOST_CPPFLAGS)
 
 lib_LTLIBRARIES = libarabica.la
 libarabica_la_SOURCES= $(cc_sources) 
-libarabica_la_LDFLAGS= $(PARSER_LIBS)
+libarabica_la_LDFLAGS= @PARSER_LIBS@
 
 
 

src/SAX/helpers/InputSourceResolver.cpp

@@ -10,6 +10,8 @@
 #include <istream>
 #include <fstream>
 #include <Utils/socket_stream.h>
+#include <Utils/uri.hpp>
+#include <cmath>
 
 InputSourceResolver::InputSourceResolver(const SAX::InputSource& inputSource) :
   deleteStream_(false),
@@ -30,11 +32,11 @@ void InputSourceResolver::open(const std::string& publicId,
       return;
     } 
 
-    // does it look like a URL?
-    std::string::size_type colonIndex = systemId.find("://");
-    if(colonIndex != std::string::npos)
+    // does it look like a URI?
+    Arabica::io::URI url(systemId);
+    if(!url.scheme().empty())
     {
-      URIResolver res = findResolver(systemId.substr(0, colonIndex));
+      URIResolver res = findResolver(url.scheme());
       if(res)
         byteStream_ = res(systemId);
       if(byteStream_)
@@ -45,7 +47,7 @@ void InputSourceResolver::open(const std::string& publicId,
     } // if ...
 
     // try and open it as a file
-    std::ifstream* ifs = new std::ifstream(systemId.c_str());
+    std::ifstream* ifs = new std::ifstream(url.path().c_str());
     if(ifs->is_open())
     {
       deleteStream_ = true;
@@ -85,10 +87,10 @@ InputSourceResolver::URIResolver InputSourceResolver::findResolver(std::string m
 
 namespace 
 {
-  std::istream* fileResolver(const std::string& fileURL)
+  std::istream* fileResolver(const std::string& fileURI)
   {
-    int colon = fileURL.find("://");
-    std::string fileName = fileURL.substr(colon+3);
+    Arabica::io::URI url(fileURI);
+    std::string fileName = url.path();
 
     std::ifstream* ifs = new std::ifstream(fileName.c_str());
     if(ifs->is_open())
@@ -113,7 +115,7 @@ namespace
 
   static bool fileReg = InputSourceResolver::registerResolver("file", fileResolver);
 
-  std::istream* httpResolver(const std::string& httpURL)
+  std::istream* httpResolver(const std::string& httpURI)
   {
 #ifdef ARABICA_USE_WINSOCK
     WORD wVersionRequested;
@@ -126,23 +128,16 @@ namespace
         return 0;
 #endif
 
-    int colon1 = httpURL.find("://");
-    colon1 += 3;
-    //int colon2 = httpURL.find("://", colon1);
-    int slash1 = httpURL.find("/", colon1);
+    Arabica::io::URI url(httpURI);
 
-    std::string hostName = httpURL.substr(colon1, slash1 - (colon1));
-    std::string path = httpURL.substr(slash1);
-
-    Arabica::socketstream* ifs = new Arabica::socketstream(hostName.c_str(), 80);
+    Arabica::socketstream* ifs = new Arabica::socketstream(url.host().c_str(), std::atoi(url.port().c_str()));
     if(!ifs->is_open())
       return 0;
-    *ifs << "GET " << path << " HTTP/1.0" << std::endl;
-    *ifs << "Host: " << hostName << std::endl;
+    *ifs << "GET " << url.path() << " HTTP/1.0" << std::endl;
+    *ifs << "Host: " << url.host() << std::endl;
     *ifs << "Connection: close" << std::endl;
     *ifs << std::endl;
 
-
     char buffer[1024];
     do
     {

src/SAX/wrappers/saxlibxml2.cpp

@@ -64,6 +64,12 @@ void lwit_processingInstruction(void *user_data, const xmlChar* target, const xm
   p->SAXprocessingInstruction(target, data);
 } // lwit_processingInstruction
 
+void lwit_comment(void *user_data, const xmlChar* comment)
+{
+  libxml2_base* p = reinterpret_cast<libxml2_base*>(user_data);
+  p->SAXcomment(comment);
+} // lwit_comment
+
 void lwit_warning(void *user_data, const char* fmt, ...)
 {
   va_list arg;
@@ -168,7 +174,7 @@ static xmlSAXHandler saxHandler = {
 		lwit_characters,		// charactersSAXFunc characters;
 		lwit_ignorableWhitespace,		// ignorableWhitespaceSAXFunc ignorableWhitespace;
 		lwit_processingInstruction,		// processingInstructionSAXFunc processingInstruction;
-		0,		// commentSAXFunc comment;
+		lwit_comment,		// commentSAXFunc comment;
 		lwit_warning,		// warningSAXFunc warning;
 		lwit_error,		// errorSAXFunc error;
 		lwit_fatalError,		// fatalErrorSAXFunc fatalError;

src/Utils/base64codecvt.cpp

@@ -82,7 +82,7 @@ std::codecvt_base::result base64codecvt::do_in(std::mbstate_t& state,
   from_next = from;
   to_next = to;
 
-  while((from_next != from_end) && (to != to_limit))
+  while((from_next != from_end) && (to_next != to_limit))
   {
     char b = *from_next++;
     size_t i = base64_charset.find(b);

src/Utils/uri.cpp

@@ -0,0 +1,141 @@
+
+#include <algorithm>
+#include <Utils/uri.hpp>
+
+using namespace Arabica::io;
+
+namespace {
+  const std::string ZERO = "0";
+  const std::string PORT_EIGHTY = "80";
+  const std::string PORT_443 = "443";
+  
+  const std::string& wellKnownPort(const std::string& scheme)
+  {
+    if(scheme.empty())
+      return ZERO;
+
+    if(scheme == "http")
+      return PORT_EIGHTY;
+    if(scheme == "https")
+      return PORT_443;
+
+    return ZERO;
+  } // wellKnownPort
+} // namespace 
+
+URI::URI(const std::string& uri) 
+{
+  parse(uri);
+} // URI
+
+URI::URI(const URI& base, const std::string& relativeUrl) :
+  scheme_(base.scheme_),
+  host_(base.host_),
+  path_(base.path_),
+  port_(base.port_)
+{
+  URI relUrl(relativeUrl);
+  absolutise(relUrl);
+} // URI         
+
+const std::string& URI::port() const
+{
+  if(port_.empty())
+    return wellKnownPort(scheme_);
+  return port_;
+} // port()
+
+std::string URI::as_string() const
+{
+  std::string str;
+  if(!scheme_.empty())
+    str.append(scheme_).append("://");
+  if(!host_.empty())
+  {
+    str.append(host_);
+    if(!port_.empty())
+      str.append(":").append(port_);
+  }
+  str.append(path_);
+  return str;
+} // as_string
+
+
+void URI::parse(const std::string& uri)
+{
+  // I'd like to use something a bit stronger - http://code.google.com/p/uri-grammar/
+  // but that would put a Boost Spirit dependence right in the core, which I'm not prepared to do at the moment
+
+  int d = uri.find_first_of(":");
+  if(d == std::string::npos)
+  {
+    path_ = uri;
+    return;
+  } // if ...
+
+  scheme_ = uri.substr(0, d);
+
+  std::string::const_iterator u = uri.begin() + d;
+  std::string::const_iterator ue = uri.end();
+
+  ++u;
+  if(*u == '/' && *(u+1) == '/')
+  {
+    u += 2;
+    parseAuthority(u, ue);
+  } // if ...
+
+  path_.append(u, ue);
+} // parse
+
+void URI::parseAuthority(std::string::const_iterator& u, std::string::const_iterator& ue)
+{
+  std::string::const_iterator slash = std::find(u, ue, '/');
+  if(slash == ue)
+    return;
+
+  std::string::const_iterator colon = std::find(u, slash, ':');
+  host_.append(u, colon);
+
+  if(colon != slash)
+    port_.append(colon+1, slash);
+
+  u = slash;
+} // parseAuthority
+
+void URI::absolutise(URI& relative)
+{
+  if(!relative.scheme().empty())
+  {
+    swap(relative);
+    return;
+  }
+
+  if(relative.path_[0] == '/')
+    path_ = relative.path_;
+  else 
+    combinePath(relative.path_);
+} // absolutise
+
+void URI::combinePath(const std::string& relPath)
+{
+  if(*(path_.rbegin()) != '/')
+    path_.erase(path_.rfind('/')+1);
+
+  path_.append(relPath);
+
+  int dots = path_.find("/../");
+  while(dots != std::string::npos)
+  {
+    int preceding_slash = (dots > 0) ? path_.rfind('/', dots-1) : 0;
+    path_.erase(preceding_slash, dots+3-preceding_slash);
+    dots = path_.find("/../");
+  } // while
+
+  int dot = path_.find("/./");
+  while(dot != std::string::npos)
+  {
+    path_.erase(dot, 2);
+    dot = path_.find("/./", dot);
+  }
+} // combinePath