// dataset_text.cxx

// Main program to list and extract text objects in a TextDataset or
// compound datset holding a TextDataset.

#include <string>
#include <iostream>
#include "dataset_util/FileStatus.h"
#include "dataset_xml/XmlParser.h"
#include "dataset_base/TextDataset.h"
#include "dataset_base/DatasetCreator.h"
#include "dataset_base/DatasetRepository.h"

using std::string;
using std::cout;
using std::cerr;
using std::endl;
using dset::Dataset;
using dset::TextDataset;
using dset::DatasetCreator;
using dset::DatasetRepository;

typedef TextDataset::TextList TextList;

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

  string exename = argv[0];
  string xfile;
  string sid;
  string action;
  string textname;
  int error = 0;
  int iarg = 1;
  while ( iarg < argc ) {
    // Action defined--this must be the text name.
    string flag = argv[iarg++];
    if ( action.size() ) {
      textname = flag;
      break;
    }
    // Not a flag--this must be the action.
    if ( flag[0] != '-' ) {
      action = flag;
      continue;
    }
    // Otherwise this is a flag.
    string arg = "";
    if ( iarg < argc ) {
      arg = argv[iarg];
      if ( arg.size() && arg[0] == '-' ) {
        arg = "";
      } else {
        ++iarg;
      }
    }
    // Input file
    if ( flag == "-h" ) {
      action = "help";
      break;
    } else if ( flag == "-f" ) {
      if ( arg == "" ) {
        error = 13;
        break;
      }
      xfile = arg;
      sid = "";
    } else if ( flag == "-i" ) {
      if ( arg == "" ) {
        error = 14;
        break;
      }
      xfile = "";
      sid = arg;
    } else {
      cerr << "Invalid flag: " << flag << endl;
      error = 15;
    }
  }

  if ( error ) {
    cerr << "Unable to parse command line." << endl;
    return error;
  }
      
  if ( action == "help" ) {
    cout << "Usage: " << exename 
         << "  [-h] (-i iii.jjj | -f filename) action textname" << endl;
    cout << "  action = help, list, extract" << endl;
    cout << "  name is from the list or blank for all" << endl;
    cout << "  -f DSFILE - input dataset XML file" << endl;
    cout << "  -i ID - ID of the dataset in format iii-jjj" << endl;
    return 0;
  }

  // Open the repository.
  DatasetRepository::set_default_instance();
  DatasetRepository& rep = DatasetRepository::default_instance();
  if ( ! rep.is_valid() ) {
    cerr << "Unable to access default dataset repository";
  }

  // Open the dataset.
  XmlParser parser;
  const Dataset* pdst = 0;
  if ( xfile.size() ) {
    FileStatus fstat(xfile);
    if ( ! fstat.exists() ) {
      cerr << "File does not exist:" << endl;
      cerr << "  " << xfile << endl;
      return 31;
    }
    const XmlElement* pele = parser.parse(xfile);
    if ( pele == 0 ) {
      cerr << "Unable to parse dataset XML file" << endl;
      return 32;
    }
    pdst = DatasetCreator::create(*pele, &rep);
    delete pele;
    if ( pdst == 0 ) {
      int ecode = rep.error();
      string msg = rep.error_message();
      cerr << "Error retrieving input dataset: " << ecode << endl;
      cerr << "  \"" << msg << "\"" << endl;
      return 39;
    }
  } else if ( sid.size() ) {
    DatasetId did;
    did = DatasetId(sid);
    if ( ! did.is_valid() ) {
      cerr << "Unable to parse id" << endl;
      return 34;
    }
    pdst = rep.extract(did);
    if ( pdst == 0 ) {
      int ecode = rep.error();
      string msg = rep.error_message();
      cerr << "Error retrieving input dataset: " << ecode << endl;
      cerr << "  \"" << msg << "\"" << endl;
       return 39;
    }
    assert( pdst != 0 );
  } else {
   cerr << "either filename or ID must be provided" << endl;
    return 30;
  }

  // Check type.
  if ( ! pdst->is<TextDataset>() ) {
    cerr << "Dataset is not a TextDataset" << endl;
    cerr << "  Full type is " << pdst->fulltype() << endl;
    return 41;
  }

  // Extract the text objects.
  const TextDataset* ptdst = dynamic_cast<const TextDataset*>(pdst);
  TextList texts0 = ptdst->texts();
  // Find texts matching name.
  TextList texts;
  for ( TextList::const_iterator itxt=texts0.begin();
        itxt!=texts0.end(); ++itxt ) {
    if ( textname.size() == 0 || itxt->name() == textname ) {
      texts.push_back(*itxt);
    }
  }
  if ( texts.size() == 0 ) {
    cerr << "No matching texts" << endl;
    return 71;
  }
  // Handle action.
  if ( action == "list" || action == "extract" ) {
    string slist;
    for ( TextList::const_iterator itxt=texts.begin();
          itxt!=texts.end(); ++itxt ) {
      if ( action == "extract" ) {
        int wstat = itxt->write();
        if ( wstat != 0 ) {
          cerr << "Error " << wstat << " writing text "
               << itxt->name() << endl;
          if ( slist.size() ) cout << slist << endl;
          return 72;
        }
      }
      if ( itxt != texts.begin() ) slist += " ";
      slist += itxt->name();
    }
    cout << slist << endl;
  } else {
    cerr << "Unknown action: " << action << endl;
    return 81;
  }


  return 0;
}