Changeset 12513

Timestamp:

2006-08-24T11:10:07+12:00 (18 years ago)

Author:

kjdon

Message:

Added Stefan's fileupload code to replace use of cgicc by depositor. parsing post data, generating fileupload structures and saving the file

Location:

trunk/gsdl/src/recpt

Files:

• cgiutils.cpp (modified) (3 diffs)
• cgiutils.h (modified) (2 diffs)

trunk/gsdl/src/recpt/cgiutils.cpp

@@ -25 +25 @@
 
 #include "cgiutils.h"
+#include "md5.h"
+#include "fileutil.h"
 #include "gsdlunicode.h"
 #include "fileutil.h"
 #include "unitool.h" // in mg, for output_utf8_char
+
+#if defined(GSDL_USE_OBJECTSPACE)
+#  include <ospace\std\iostream>
+#  include <ospace\std\fstream>
+#elif defined(GSDL_USE_IOS_H)
+#  include <iostream.h>
+#  include <fstream.h>
+#else
+#  include <iostream>
+#  include <fstream>
+#endif
 
 
@@ -57 +70 @@
   t.push_back(o1);
   t.push_back(o2);
+}
+
+static text_t::iterator getline (text_t::iterator first, 
+                 text_t::iterator last,
+                 bool include_crlf) {
+  while (first != last) {
+    if (((first+1) != last) && (*first == 13) && (*(first+1) == 10)) {
+      // found <CRLF>
+      if (include_crlf) first += 2;
+      break;
+    }
+    first++;
+  }
+  return first;
+}
+
+static void process_post_section (text_t &argname, text_t &argdata, text_t &filename, text_t &filedata,
+                  text_t &filetype, bool &isfile, text_t &argstr,
+                  fileupload_tmap &fileuploads, const text_t &gsdlhome) {
+
+  if (!argname.empty()) {
+
+    if (!isfile) {
+      // argdata includes a trailing <CRLF> that we must remove
+      if ((argdata.size() > 1) && (*(argdata.end()-2) == 13) && (*(argdata.end()-1) == 10)) {
+    argdata.erase(argdata.end()-2, argdata.end());
+      }
+      if (!argstr.empty()) argstr += "&";
+      argstr += argname + "=" + argdata;
+
+    } else if (!filename.empty()) {
+      // filedata includes a trailing <CRLF> that we must remove
+      if ((filedata.size() > 1) && (*(filedata.end()-2) == 13) && (*(filedata.end()-1) == 10)) {
+    filedata.erase(filedata.end()-2, filedata.end());
+      }
+
+      // create tmp_name for storing the file on disk
+      text_t tmp_name = md5data(filedata);
+      tmp_name = filename_cat(gsdlhome, "tmp", tmp_name);
+      char *tmp_name_c = tmp_name.getcstr();
+
+      // write the file data to disk
+      outconvertclass out;
+      ofstream filestream(tmp_name_c, ios::out | ios::binary);
+      filestream << out << filedata;
+      filestream.close();
+      delete tmp_name_c;
+
+      // populate the fields of a fileupload_t and put it in the
+      // fileuploads map
+      fileupload_t fu;
+      // note that filename currently may or may not include the path since
+      // some browsers (e.g. IE) include the path while others
+      // (e.g. mozilla) do not. we should probably remove the path from
+      // this field here to get a consistent value across all browsers.
+      text_t::iterator slash = findlastchar(filename.begin(), filename.end(), '\\');
+      if (slash != filename.end()) {
+    filename = substr(slash+1, filename.end());
+      }
+      fu.name = filename;
+      fu.type = filetype;
+      // size has yet to be implemented
+      fu.size = 0;
+      fu.tmp_name = tmp_name;
+      fileuploads[argname] = fu;
+    }
+  }
+  isfile = false;
+  argname.clear();
+  argdata.clear();
+  filename.clear();
+  filedata.clear();
+  filetype.clear();
+}
+
+// parse data obtained through a CGI POST request
+text_t parse_post_data (text_t &content_type, text_t &raw_post_data, 
+            fileupload_tmap &fileuploads, const text_t &gsdlhome) {
+
+  text_t argstr;
+
+  text_t::iterator content_type_begin = content_type.begin();
+  text_t::iterator content_type_end = content_type.end();
+  if (findword(content_type_begin, content_type_end, "multipart/form-data") == content_type_end) {
+    // a simple post request
+    return raw_post_data;
+
+  } else {
+    // multipart/form data - may contain one or more uploaded files
+
+    /*
+      content_type should look something like the following
+        multipart/form-data; boundary=---------------------------7d411e1a50330
+    
+      while raw_post_data will be as follows
+        -----------------------------7d43e73450330CRLF
+    Content-Disposition: form-data; name="e"<CRLF>
+    <CRLF>
+    d-0testss--1-0-00---4----0--0-110--1en-Zz-1---10-about-0--00031-001utfZz-8-0<CRLF>
+    -----------------------------7d43e73450330<CRLF>
+    Content-Disposition: form-data; name="afile"; filename="C:\somedoc.doc"<CRLF>
+    Content-Type: application/msword<CRLF>
+    <CRLF>
+    <Content of file><CRLF>
+    
+    */
+
+    // first get the boundary from content-type
+    text_t::iterator boundary_begin = findword(content_type_begin, content_type_end, "boundary=");
+    if (boundary_begin+9 < content_type_end) boundary_begin += 9;
+    else {
+      // error
+      return "";
+    }
+    text_t boundary = substr(boundary_begin, getline(boundary_begin, content_type_end, false));
+    int boundary_len = boundary.size();
+
+    text_t argname, argdata, filename, filedata, filetype;
+    bool isfile = false;
+    text_t::iterator data_here = raw_post_data.begin();
+    text_t::iterator data_end = raw_post_data.end();
+    while (data_here != data_end) {
+      
+      // get the next available line (including the trailing <CRLF>
+      text_t line = substr(data_here, getline(data_here, data_end, true));
+      data_here += line.size();
+      text_t::iterator line_begin = line.begin();
+      text_t::iterator line_end = line.end();
+      if (findword(line_begin, line_end, boundary) != line_end) {
+    // we've found a boundary
+    process_post_section(argname, argdata, filename, filedata, filetype,
+                 isfile, argstr, fileuploads, gsdlhome);
+
+      } else if (findword(line_begin, line_end, "Content-Disposition: form-data") != line_end) {
+    // we've found the the beginning of a new section
+    argname.clear();
+    argdata.clear();
+
+    // get the name of this piece of form data
+    text_t::iterator it = findword(line_begin, line_end, "name=\"");
+    if (it == line_end) break; // error - this shouldn't happen
+    it = findchar(it, line_end, '"');
+    if ((it != line_end) && (it+1 != line_end)) {
+      argname = substr(it+1, findchar(it+1, line_end, '"'));
+    }
+
+    // if this piece of form data contains filename="" it's a file
+    // upload and needs to be treated special
+    it = (findword(line_begin, line_end, "filename=\""));
+    if (it != line_end) {
+      // we've found a file upload
+      isfile = true;
+      it = findchar(it, line_end, '"');
+      if ((it != line_end) && (it+1 != line_end)) {
+        filename = substr(it+1, findchar(it+1, line_end, '"'));
+      }
+
+      // the next line is the content-type of this section
+      line = substr(data_here, getline(data_here, data_end, true));
+      data_here += line.size();
+      line_begin = line.begin();
+      line_end = line.end();
+      it = (findword(line_begin, line_end, "Content-Type: "));
+      if (it != line_end) {
+        filetype = substr(it+14, getline(it, line_end, false));
+      }
+    }
+
+    // eat up the next line as it's just a <CRLF> on it's own
+    data_here += 2;
+
+      } else {
+    if (isfile) filedata += line;
+    else argdata += line;
+      }
+    }
+
+    // process last section
+    process_post_section(argname, argdata, filename, filedata, filetype,
+             isfile, argstr, fileuploads, gsdlhome);
+
+    return argstr;
+  }
 }
 
@@ -465 +661 @@
 }
 
+void add_fileupload_args (const cgiargsinfoclass &argsinfo,
+              cgiargsclass &args,
+              fileupload_tmap &fileuploads,
+              ostream &logout) {
+ 
+  const cgiarginfo *info = argsinfo.getarginfo("a");
+  fileupload_tmap::const_iterator this_file = fileuploads.begin();
+  fileupload_tmap::const_iterator end_file = fileuploads.end();
+  while (this_file != end_file) {
+    const cgiarginfo *info = argsinfo.getarginfo((*this_file).first);
+    if (info != NULL) {
+      if ((*info).fileupload && (file_exists((*this_file).second.tmp_name))) {
+    args.setargfile((*this_file).first, (*this_file).second);
+      }
+    }
+    this_file++;
+  }
+}
 
 // compress_save_args will compress the arguments and return

trunk/gsdl/src/recpt/cgiutils.h

@@ -32 +32 @@
 #include "cgiargs.h"
 
+// parse data obtained through a CGI POST request
+text_t parse_post_data (text_t &content_type, text_t &raw_post_data, 
+            fileupload_tmap &fileuploads, const text_t &gsdlhome);
 
 // convert %xx and + to their appropriate equivalents
@@ -82 +85 @@
                ostream &logout);
 
+void add_fileupload_args (const cgiargsinfoclass &argsinfo,
+              cgiargsclass &args,
+              fileupload_tmap &fileuploads,
+              ostream &logout);
+
 // compress_save_args will compress the arguments and return
 // them in compressed_args. If an error was encountered