示例#1
0
bool lemur::file::SortMergeTextFiles::_readLine(FILE *_in, char*& beginLine, size_t& lineLength, indri::utility::Buffer &_buffer ) {
  lineLength = 0;
  size_t actual;

  // make a buffer of a reasonable size so we're not always allocating
  if( _buffer.size() < 1024*1024 )
    _buffer.grow( 1024*1024 );
  // if we're running out of room, add 25MB
  if( (_buffer.size() -  _buffer.position()) < 512*1024 ) {
    _buffer.grow( _buffer.size() + 1024*1024*25 );
  }
  if( _buffer.position() ) {
    // get rid of null terminator from previous call
    _buffer.unwrite(1);
  }

  size_t readAmount = _buffer.size() - _buffer.position() - 2;

  // fetch next document line
  char* buffer = _buffer.write( readAmount );
  char* result = fgets( buffer, (int)readAmount, _in );

  if(!result) {
    return false;
  }

  actual = strlen(buffer);
  lineLength += actual;
  _buffer.unwrite( readAmount - actual );

  // all finished reading
  *_buffer.write(1) = 0;
  beginLine = _buffer.front() + _buffer.position() - lineLength - 1;

  // strip the \n off the end
  if (beginLine[lineLength-1]=='\n') {
    beginLine[lineLength-1]=0;
  }

  return true;
}
static void zlib_read_document( z_stream_s& stream, indri::file::File& infile, UINT64 offset, indri::utility::Buffer& outputBuffer ) {
  // read in data from the file until the stream ends
  // split up the data as necessary
  // decompress positional info

  // read some data
  char inputBuffer[INPUT_BUFFER_SIZE];
  outputBuffer.grow( INPUT_BUFFER_SIZE );
  outputBuffer.write( sizeof(indri::api::ParsedDocument) );

  stream.avail_in = 0;
  stream.avail_out = 0;
  
  while(true) {
    if( !stream.avail_in ) {
      UINT64 readSize = infile.read( inputBuffer, offset, sizeof inputBuffer );
      offset += readSize; 
      
      stream.avail_in = readSize;
      stream.next_in = (Bytef*) inputBuffer;
    }

    stream.avail_out = outputBuffer.size() - outputBuffer.position();
    stream.next_out = (Bytef*) outputBuffer.write( outputBuffer.size() - outputBuffer.position() );

    int result = inflate( &stream, Z_NO_FLUSH );
    outputBuffer.unwrite( stream.avail_out );

    if( result == Z_STREAM_END ) {
      result = inflate( &stream, Z_FINISH );
      
      if( result < 0 )
        LEMUR_THROW( result, "Something bad happened while trying to finish decompressing a document." );

      inflateEnd( &stream );
      break;
    }

    if( result < 0 ) {
      LEMUR_THROW( result, "Something bad happened while trying to decompress a document." );
    }

    if( stream.avail_out == 0 ) {
      outputBuffer.grow();
    }
  }
}
      void _fetchText( indri::utility::greedy_vector<TagExtent *>& tags, indri::utility::greedy_vector<char*>& terms ) {
        // now, fetch the additional terms
        char line[65536];
        _buffer.clear();

        for( int i=0; i<_count; i++ ) {
          // LINK
          _in.getline( line, sizeof line-1 );

          // LINKDOCNO 
          _in.getline( line, sizeof line-1 );
          
          // TEXT=
          _in.getline( line, sizeof line-1 );
          int textLen = strlen(line+6);
          strcpy( _buffer.write(textLen+1), line+6 );
          _buffer.unwrite(1);
          
          assert( *(_buffer.front()+_buffer.position()-1) == '\"' && "Last character should be a quote" );
        }
        *(_buffer.write(1)) = 0;

        // now there's a bunch of text in _buffer, space separated, with each
        // link separated by a " symbol

        char* beginWord = 0;
        int beginIndex = 0;
        char* buffer = _buffer.front();

        for( unsigned int i=0; i<_buffer.position(); i++ ) {
          if( isalnum(buffer[i]) && !beginWord ) {
            beginWord = buffer+i;

            if(!beginIndex)
              beginIndex = terms.size();
          } else if( isspace(buffer[i]) ) {
            buffer[i] = 0;
            if( beginWord )
              terms.push_back( beginWord );
            beginWord = 0;
          } else if( buffer[i] == '\"' ) {
            buffer[i] = 0;
            if( beginWord )
              terms.push_back( beginWord );
            beginWord = 0;
        
            TagExtent * extent = new TagExtent;
            extent->name = "inlink";
            extent->begin = beginIndex;
            extent->end = terms.size();
            extent->number = 0;
            extent->parent = 0;

            assert( extent->begin <= extent->end );

            if( beginIndex ) {
              tags.push_back(extent);
              if( terms.size() > 125000 )
                break;
            }


            beginIndex = 0;
          }

        }

      }