// make_text_dataset.cxx
//
// David Adams
// June 2003
//
// Create a text dataset from a file.

#include <string>
#include <iostream>
#include <dlfcn.h>
#include "dataset_util/FileStatus.h"
#include "dataset_util/FileName.h"
#include "dataset_util/Environment.h"
#include "dataset_util/XmlElement.h"
#include "dataset_util/DtdRegistry.h"
#include "dataset_xml/XmlParser.h"
#include "dataset_id/UniqueIdGenerator.h"
#include "dataset_file/FileCatalog.h"
#include "dataset_credential/GssCredentialManager.h"
#include "dataset_base/TextDataset.h"

using std::string;
using std::cout;
using std::cerr;
using std::endl;
using dset::FileCatalog;
using dset::GssCredentialManager;
using dset::TextDataset;

//**********************************************************************

int main(int argc, char* argv[]) {

  // Fetch arguments.
  string arg0 = argv[0];

  // First argument.
  string snames = "";
  string id_name = "";
  string content_label = "";
  bool use_uid = true;
  string uid_name = "";
  int stat = 0;
  // Print help and exit.
  bool help = false;
  // ID for dataset.
  DatasetId dstid;
  assert( ! dstid.is_valid() );
  // Name for dataset XML file.
  string xfile = "dataset.xml";
  // Flag to insert file into file catalog.
  for ( int iarg=1; iarg<argc; ++iarg ) {
    string arg = argv[iarg];
    // help
    if ( arg == "-h" ) {
      help = true;
    // specify ID
    } else if ( arg == "-i" ) {
      if ( iarg<argc && argv[iarg+1][0]!='-' ) {
        id_name = argv[++iarg];
	use_uid = false;
      } else {
	cerr << "Value missing for option -i" << endl;
	stat = 4;
      }
    // specify unique ID generator
    } else if ( arg == "-u" ) {
      use_uid = true;
      id_name = "";
      if ( iarg<argc && argv[iarg+1][0]!='-' ) {
        uid_name = argv[++iarg];
      }
    // specify content label
    } else if ( arg == "-c" ) {
      if ( iarg < argc ) {
        arg = argv[++iarg];
	content_label = arg;
      }
    // specify XML file name
    } else if ( arg == "-f" ) {
      if ( iarg < argc ) {
        arg = argv[++iarg];
	xfile = arg;
      }
    // Put file in catalog
    } else if ( snames == "" ) {
      snames = arg;
    } else {
      snames += " ";
      snames += arg;
    }
  }

  if ( help || stat!=0 ) {
    cout << "Usage: " << arg0 << " [-i iii.jjj | -u [UID_DIR]] "
         << "-c label [-f outfile] file1 [file2 ...]" << endl;
    cout << "       -i iii-jjj sets ID to iii-jjj" << endl;
    cout << "                  (normally better to use option -u)" << endl;
    cout << "       -u [CONN] to specify unique ID connection string" << endl;
    cout << "                 default is $DIAL_UIDS" << endl;
    cout << "       -c label sets the content label" << endl;
    cout << "       -f fname writes XML to file fname" << endl;
    return stat;
  }

  // Check content label.
  if ( content_label == "" ) {
    cerr << "Content label must be provided" << endl;
    return 10;
  }

  // Check that unique ID generator and explicit ID are not both
  // specified.
  bool use_id = id_name.size();
  if ( use_id && use_uid ) {
    cerr << "Use only one of -i and -u" << endl;
    return 11;
  }

  // Initialize the unique ID generator.
  if ( use_uid ) {
    if ( ! uid_name.size() ) {
      uid_name = Environment::initial().value("DIAL_UIDS");
      if ( uid_name == "" ) {
        cerr << "Unable to locate $DIAL_UIDS" << endl;
        return 12;
      }
    }
    int gstat = UniqueIdGenerator::set_generator(uid_name);
    if ( gstat != 0 ) {
      cerr << "Unable to set unique ID generator" << endl;
      cerr << "  " << uid_name << endl;
      cerr << "  Error code is " << gstat << endl;
      return 17;
    }
  }

  // Set the credential for this thread.
  // This is needed to find owner for task.
  int cstat = GssCredentialManager::set_default();
  if ( cstat != 0 ) {
    cerr << "Unable to find credential" << endl;
    return 5;
  }

  // Create dataset.
  Text::set_warn();
  TextDataset dst(content_label, snames);
  if ( ! dst.is_valid() ) {
    cerr << "Unable to create dataset" << endl;
    int err = dst.error();
    cerr << "TextDataset reports error " << err << endl;
    string msg = "Unknown error";
    if ( err == 10 ) {
      msg = "Invalid content label";
    } else if ( err = 11 ) {
      msg = "Duplicate file name";
    } else if ( err = 12 ) {
      msg = "File name is not relative";
    } else if ( err = 13 ) {
      msg = "Invalid text object";
    }
    cerr << "  " << msg << endl;
    return 15;
  }

  // Set ID.
  if ( use_id ) {
    DatasetId newid(id_name);
    int idstat = dst.set_id(newid);
    if ( idstat != 0 ) {
      cerr << "Error " << idstat << " setting dataset ID " << dstid << endl;
      return 16;
    }
  }
  dst.lock();

  // Create XML.
  const XmlElement* pele = dst.xml();
  if ( pele == 0 ) {
    cerr << "Unable to create XML" << endl;
    return 17;
  }

  // Write XML description.
  if ( xfile.size() ) {
    XmlParser parser;
    int wstat = parser.write(xfile, *pele);
    if ( wstat ) {
      cerr << "Error writing XML: " << wstat << endl;
      cerr << *pele << endl;
    } else {
      cerr << "Output in " << xfile << endl;
    }
    delete pele;
  }

}