1. awkward yet working support for POST file upload (requires mimetic)

2. support for NOT handling/parsing POSTed data (including application/x-www-form-urlencoded)

4 files changed, 190 insertions, 19 deletions

diff --git a/configure.ac b/configure.ac
index b3141f1..b51d64d 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1,27 +1,27 @@
 AC_INIT([kingate], [0.0.1], [kingate-bugs@klever.net])
 AC_CONFIG_SRCDIR([include/kingate/cgi_gateway.h])
-AC_CONFIG_HEADER([config.h])
+AC_CONFIG_HEADERS([config.h])
 AM_INIT_AUTOMAKE([dist-bzip2])
 
 AC_PROG_INSTALL
 AC_PROG_AWK
 AC_PROG_CXX
 AC_PROG_CC
 AC_PROG_LIBTOOL
 
 AC_HEADER_STDC
 AC_CHECK_HEADERS([sys/types.h sys/stat.h])
 AC_CHECK_DECLS([environ],,,[
  #include <unistd.h>
 ])
 
 HAVE_FCGI=false
 AC_LANG_PUSH(C++)
 AC_CHECK_HEADERS([fcgio.h],[
  AC_CHECK_LIB(fcgi,FCGX_Init,[
   LIBS="-lfcgi ${LIBS}"
   AC_CHECK_LIB([fcgi++],[main],[
    LIBS="-lfcgi++ ${LIBS}"
    HAVE_FCGI=true
   ])
  ])
@@ -39,32 +39,46 @@ AC_FUNC_REALLOC
 
 AC_PATH_PROG([XSLTPROC],[xsltproc],[true])
 
 AC_WITH_PKGCONFIG
 
 PKG_CHECK_MODULES([KONFORKA],[konforka],,[
  AC_MSG_ERROR([no konforka library found. get one from http://kin.klever.net/konforka/])
 ])
 
 WANT_DOXYGEN="yes"
 AC_ARG_ENABLE([doxygen],
  AC_HELP_STRING([--disable-doxygen],[do not generate documentation]),
  [
   test "${enableval}" = "no" && WANT_DOXYGEN="no"
  ]
 )
 if test "${WANT_DOXYGEN}" = "yes" ; then
  AC_WITH_DOXYGEN
  AC_WITH_DOT
 else
  AM_CONDITIONAL([HAVE_DOXYGEN],[false])
  AM_CONDITIONAL([HAVE_DOT],[false])
 fi
 
+HAVE_MIMETIC=false
+AC_LANG_PUSH(C++)
+  AC_CHECK_LIB([mimetic],[main],[
+   MIMETIC_LIBS=-lmimetic
+   HAVE_MIMETIC=true
+  ]
+ )
+AC_LANG_POP(C++)
+AC_SUBST([MIMETIC_LIBS])
+AC_SUBST([MIMETIC_CFLAGS])
+if ${HAVE_MIMETIC} ; then
+ AC_DEFINE([HAVE_MIMETIC],,[defined in presence of mimetic])
+fi
+
 AC_CONFIG_FILES([
  Makefile
   kingate.pc kingate-fcgi.pc kingate-plaincgi.pc
   Doxyfile
  include/Makefile
  src/Makefile
 ])
 AC_OUTPUT
diff --git a/include/kingate/cgi_gateway.h b/include/kingate/cgi_gateway.h
index f20d72b..a26b0ae 100644
--- a/include/kingate/cgi_gateway.h
+++ b/include/kingate/cgi_gateway.h
@@ -22,60 +22,97 @@ namespace kingate {
     using namespace std;
 
     /**
      * The main class interfacing with the CGI environment.
      */
     class cgi_gateway {
         public:
             /**
              * The interface to CGI environment (e.g. fastcgi).
              */
             cgi_interface& iface;
             /**
              * The type describing map holding parameters parsed from query string or input.
              */
             typedef multimap<string,string> params_t;
             /**
              * The GET-passed parameters.
              */
             params_t get;
             /**
              * The POST-passed parameters.
              */
             params_t post;
             /**
+             * Abstract base class for retrieving posted files.
+             */
+            class basic_file_t {
+                public:
+                    /**
+                     * Retrieve file name.
+                     * @return filename
+                     */
+                    virtual const string& filename() const = 0;
+                    /**
+                     * Retrieve file content type.
+                     * @return content type
+                     */
+                    virtual const string& content_type() const = 0;
+                    /**
+                     * Retrieve file contents.
+                     * @return reference to the stream for accessing file content.
+                     */
+                    virtual istream& content() = 0;
+                    virtual ~basic_file_t();
+            };
+            typedef basic_file_t *file_t;
+            /**
+             * The map holding information pertaining to files uploaded via post.
+             */
+            typedef multimap<string,file_t> files_t;
+            /**
+             * Files uploaded via post
+             */
+            files_t files;
+            /**
              * Cookies passed.
              */
             cookies_t cookies;
             /**
              * Was the stdin content parsed?
              */
             bool b_parsed_content;
 
             /**
              * @param ci the interface to use.
              */
-            cgi_gateway(cgi_interface& ci);
+            cgi_gateway(cgi_interface& ci,bool parsebody = true);
+            virtual ~cgi_gateway() throw();
+
+            /**
+             * Parse request body.
+             */
+            void parse_request_body();
 
             /**
              * Check whether there is an 'environment' meta-variable with specific name
              * passed to CGI.
              * @param n variable name.
              * @return true if yes.
              * @see cgi_interface::has_meta()
              * @see get_meta()
              */
             bool has_meta(const string& n) const { return iface.has_meta(n); }
             /**
              * Retrieve the 'environment' meta-variable value.
              * @param n variable name.
              * @return variable contents.
              * @see exception_notfound
              * @see cgi_interface::get_meta()
              */
             const string& get_meta(const string& n) const { return iface.get_meta(n); }
 
             /**
              * fetch reference to the 'stdin' stream.
              * @return the reference to the corresponding istream object.
              * @see cgi_interface::in()
              */
@@ -115,90 +152,104 @@ namespace kingate {
             /**
              * Check to see whether the parameter was passed via POST.
              * @param n the parameter name.
              * @return true if yes.
              */
             bool has_POST(const string& n) const;
             /**
              * Retrieve the POST-parameter.
              * @param n the parameter name.
              * @return the parameter contents.
              * @see exception_notfound
              */
             const string& get_POST(const string& n) const;
             /**
              * Check to see whether the parameter was passed either via POST or
              * GET.
              * @param n the parameter name.
              * @return true if yes.
              */
             bool has_param(const string& n) const;
             /**
              * Retrieve the parameter passed either via POST or GET
              * (GET-parameter takes precedence).
              * @param n the parameter name.
-             * @return true if yes.
+             * @return the parameter contents.
              * @see exception_notfound.
              */
             const string& get_param(const string& n) const;
+            /**
+             * Check to see whether the file was uploaded in the request body.
+             * @param n the parameter name.
+             * @return true if yes.
+             */
+            bool has_file(const string& n) const;
+            /**
+             * Retrieve the file uploaded in the request body.
+             * @param n the parameter name.
+             * @return the file.
+             * @see exception_notfound.
+             */
+            const file_t get_file(const string& n) const;
+            file_t get_file(const string& n);
 
             /**
              * Retrieve the POST content-type (as passed via CONTENT_TYPE
              * environment variable).
              * @return the content type.
              */
             const string& __deprecated get_content_type() const;
             /**
              * Retrieve the POST content length (as passed via the
              * CONTENT_LENGTH environment variable).
              * @return the content length.
              */
             unsigned long __deprecated get_content_length() const;
 
             /**
              * Check to see whether the content from stdin stream was parsed.
              * @return true if yes.
              */
             bool is_content_parsed() const { return b_parsed_content; }
 
             /**
              * Retrieve the HTTP header value from the HTTP_ meta-variable.
              * (see RFC3875)
              * @param hn header field name.
              * @return the HTTP header value.
              */
             const string& http_request_header(const string& hn) const;
 
             /**
              * Retrieve the AUTH_TYPE meta-variable (see RFC3875)
              * @return authentication type.
              */
             const string& auth_type() const;
             /**
              * Retrieve the CONTENT_LENGTH meta-variable (see RFC3875)
              * @return size of the request message body.
              */
-            unsigned long cgi_gateway::content_length() const;
+            unsigned long content_length() const;
             /**
              * Retrieve the CONTENT_TYPE meta-variable (see RFC3875)
              * @return media type of the request message body.
              */
             const string& content_type() const;
             /**
              * Retrieve the GATEWAY_INTERFACE meta-variable (see RFC3875)
              * @return the gateway interface dialect.
              */
             const string& gateway_interface() const;
             /**
              * Retrieve the PATH_INFO meta-variable (see RFC3875)
              * @return path to be interpreted by the script.
              */
             const string& path_info() const;
             /**
              * Retrieve the PATH_TRANSLATED meta-variable (see RFC3875)
              * @return the translated path to the document.
              */
             const string& path_translated() const;
             /**
              * Retrieve the QUERY_STRING meta-variable (see RFC3875)
              * @return the query string.
              */
diff --git a/kingate.pc.in b/kingate.pc.in
index 671faac..05cfe1d 100644
--- a/kingate.pc.in
+++ b/kingate.pc.in
@@ -1,11 +1,11 @@
 prefix=@prefix@
 exec_prefix=@exec_prefix@
 libdir=@libdir@
 includedir=@includedir@
 
 Name: kingate
 Description: C++ CGI support library
 Version: @VERSION@
 Requires: konforka
-Libs: -L${libdir} -lkingate
-Cflags: -I${includedir}
+Libs: -L${libdir} -lkingate @MIMETIC_LIBS@
+Cflags: -I${includedir} @MIMETIC_CFLAGS@
diff --git a/src/cgi_gateway.cc b/src/cgi_gateway.cc
index ab48f78..1706679 100644
--- a/src/cgi_gateway.cc
+++ b/src/cgi_gateway.cc
@@ -1,89 +1,195 @@
 #include <errno.h>
 #include <ctype.h>
+#include <sstream>
 #include "kingate/cgi_gateway.h"
 #include "kingate/util.h"
 #include "kingate/exception.h"
+#include "config.h"
+#ifdef HAVE_MIMETIC
+# include <mimetic/mimeentity.h>
+# include <mimetic/parser/itparser.h>
+#endif /* HAVE_MIMETIC */
 
 namespace kingate {
+    using mimetic::MimeEntity;
+
+#ifdef HAVE_MIMETIC
+    struct TornMimeEntity : public MimeEntity {
+        typedef istreambuf_iterator<char> it_type;
+        typedef it_type::iterator_category it_cat;
+        struct IParser : public mimetic::IteratorParser<it_type,it_cat> {
+            typedef mimetic::IteratorParser<it_type,it_cat> BT;
+            IParser(MimeEntity& me)
+                : BT::IteratorParser<it_type,it_cat>(me) { }
+            void loadHeader(it_type bit,it_type eit) {
+                m_bit = bit; m_eit = eit;
+                BT::loadHeader();
+            }
+            void loadBody(it_type bit,it_type eit) {
+                m_bit = bit; m_eit = eit;
+                BT::loadBody();
+            }
+        };
+        void load(istream& hs,istream& bs,int mask=0) {
+            IParser prs(*this);
+            prs.iMask(mask);
+            prs.loadHeader(it_type(hs),it_type());
+            prs.loadBody(it_type(bs),it_type());
+        }
+    };
+#endif /* HAVE_MIMETIC */
 
     static string empty_string;
 
-    cgi_gateway::cgi_gateway(cgi_interface& ci)
+    cgi_gateway::basic_file_t::~basic_file_t() { }
+
+    class string_file_t : public cgi_gateway::basic_file_t {
+        public:
+            string _file_name;
+            string _content_type;
+            stringstream _content;
+
+            string_file_t(const string& fn,const string& ct,const string& s)
+                : _file_name(fn), _content_type(ct), _content(s,ios::in) { }
+            const string& filename() const { return _file_name; }
+            const string& content_type() const { return _content_type; }
+            istream& content() { return _content; }
+    };
+
+    cgi_gateway::cgi_gateway(cgi_interface& ci,bool parsebody)
         : iface(ci), b_parsed_content(false) {
             // Fetch GET content
             try {
                 string qs = get_meta("QUERY_STRING");
                 parse_query(qs,get);
             }catch(exception_notfound& enf) { }
-            // Fetch POST content
-            if(!strcasecmp(content_type().c_str(),"application/x-www-form-urlencoded")) {
-                unsigned long cl = content_length();
-                if(cl) {
-                    char * tmp = new char[cl];
-                    iface.in().read(tmp,cl);
-                    string qs(tmp,cl);
-                    delete tmp;
-                    parse_query(qs,post);
-                }
-                b_parsed_content = true;
-            }
+            if(parsebody)
+                parse_request_body();
             // Parse cookies
             try {
                 cookies.parse_cookies(get_meta("HTTP_COOKIE"));
             }catch(exception_notfound& enf) { }
         }
 
+    cgi_gateway::~cgi_gateway() throw() {
+        for(files_t::iterator i=files.begin();i!=files.end();++i)
+            delete i->second;
+        files.clear();
+    }
+
+    void cgi_gateway::parse_request_body() {
+        if(b_parsed_content)
+            throw konforka::exception(CODEPOINT,"request body is already parsed");
+        // Fetch POST content
+        if(!strncasecmp(
+                    content_type().c_str(),
+                    "application/x-www-form-urlencoded",
+                    sizeof("application/x-www-form-urlencoded")-1) ) {
+            unsigned long cl = content_length();
+            if(cl) {
+                char * tmp = new char[cl];
+                iface.in().read(tmp,cl);
+                string qs(tmp,cl);
+                delete tmp;
+                parse_query(qs,post);
+            }
+            b_parsed_content = true;
+        }
+#ifdef HAVE_MIMETIC
+         else if(!strncasecmp(
+                    content_type().c_str(),
+                    "multipart/form-data",
+                    sizeof("multipart/form-data")-1) ) {
+            stringstream h;
+            h
+                << "Content-Type: " << content_type() << "\r\n"
+                << "Content-Length: " << content_length() << "\r\n\n";
+            TornMimeEntity me;
+            me.load(h,iface.in(),0);
+            mimetic::MimeEntityList& parts = me.body().parts();
+            for(mimetic::MimeEntityList::iterator i=parts.begin();i!=parts.end();++i) {
+                MimeEntity *p = *i;
+                const mimetic::ContentDisposition& cd = p->header().contentDisposition();
+                string n = cd.param("name");
+                string fn = cd.param("filename");
+                if(fn.empty()) {
+                    post.insert(params_t::value_type(n,p->body()));
+                }else{
+                    const mimetic::ContentType& ct = p->header().contentType();
+                    files.insert(files_t::value_type(n,new string_file_t(fn,ct.str(),p->body())));
+                }
+            }
+            b_parsed_content = true;
+        }
+#endif /* HAVE_MIMETIC */
+    }
+
     bool cgi_gateway::has_GET(const string& n) const {
         return get.find(n) != get.end();
     }
     const string& cgi_gateway::get_GET(const string& n) const {
         params_t::const_iterator i = get.find(n);
         if(i==get.end())
             throw exception_notfound(CODEPOINT,"no such parameter");
         return i->second;
     }
     bool cgi_gateway::has_POST(const string& n) const {
         return post.find(n) != post.end();
     }
     const string& cgi_gateway::get_POST(const string& n) const {
         params_t::const_iterator i = post.find(n);
         if(i==post.end())
             throw exception_notfound(CODEPOINT,"no such parameter");
         return i->second;
     }
     bool cgi_gateway::has_param(const string& n) const {
         return has_GET(n) || has_POST(n);
     }
     const string& cgi_gateway::get_param(const string& n) const {
         params_t::const_iterator i = get.find(n);
         if(i!=get.end())
             return i->second;
         i = post.find(n);
         if(i!=post.end())
             return i->second;
         throw exception_notfound(CODEPOINT,"no such parameter");
     }
+    bool cgi_gateway::has_file(const string& n) const {
+        return files.find(n) != files.end();
+    }
+    const cgi_gateway::file_t cgi_gateway::get_file(const string& n) const {
+        files_t::const_iterator i = files.find(n);
+        if(i==files.end())
+            throw exception_notfound(CODEPOINT,"no such parameter");
+        return i->second;
+    }
+    cgi_gateway::file_t cgi_gateway::get_file(const string& n) {
+        files_t::const_iterator i = files.find(n);
+        if(i==files.end())
+            throw exception_notfound(CODEPOINT,"no such parameter");
+        return i->second;
+    }
 
     /*
      * deprecated stuff.
      */
     const string& cgi_gateway::get_content_type() const {
         if(!has_meta("CONTENT_TYPE"))
             return empty_string;
         return get_meta("CONTENT_TYPE");
     }
     unsigned long cgi_gateway::get_content_length() const {
         if(!has_meta("CONTENT_LENGTH"))
             return 0;
         string cl = get_meta("CONTENT_LENGTH");
         return strtol(cl.c_str(),NULL,10);
     }
     /*
      *
      */
 
     const string& cgi_gateway::http_request_header(const string& hn) const {
         string mvn = "HTTP_";
         for(const char* p=hn.c_str();*p;p++) {
             if(*p=='-')
                 mvn += '_';