src/tim/prune/load/FileSplitter.java

   1 package tim.prune.load;
   2
   3 /**
   4  * Class responsible for splitting the file contents into an array
   5  * based on the selected delimiter character
   6  */
   7 public class FileSplitter
   8 {
   9         private ContentCacher _cacher = null;
  10         private int _numRows = 0;
  11         private int _numColumns = 0;
  12         private boolean[] _columnStates = null;
  13         private String[] _firstFullRow = null;
  14
  15
  16         /**
  17          * Constructor
  18          * @param inCacher cacher object holding file contents
  19          */
  20         public FileSplitter(ContentCacher inCacher)
  21         {
  22                 _cacher = inCacher;
  23         }
  24
  25         /**
  26          * Split the FileCacher's contents into a 2d array
  27          * @param inDelim delimiter character
  28          * @return 2d Object array
  29          */
  30         public String[][] splitFieldData(char inDelim)
  31         {
  32                 _firstFullRow = null;
  33                 if (_cacher == null) return null;
  34                 String[] contents = _cacher.getContents();
  35                 if (contents == null || contents.length == 0) return null;
  36                 String delimStr = checkDelimiter(inDelim);
  37                 // Count non-blank rows and max field count
  38                 _numRows = 0;
  39                 int maxFields = 0;
  40                 for (int i=0; i<contents.length; i++)
  41                 {
  42                         if (contents[i] != null && !contents[i].trim().equals(""))
  43                         {
  44                                 _numRows++;
  45                                 String[] splitLine = contents[i].split(delimStr);
  46                                 if (splitLine != null && splitLine.length > maxFields)
  47                                 {
  48                                         maxFields = splitLine.length;
  49                                         _firstFullRow = splitLine;
  50                                 }
  51                         }
  52                 }
  53                 _numColumns = maxFields;
  54                 _columnStates = new boolean[maxFields];
  55
  56                 // Create array and populate it
  57                 // Note that array will be rectangular even if data is ragged
  58                 String[][] result = new String[_numRows][];
  59                 for (int i=0; i<contents.length; i++)
  60                 {
  61                         result[i] = new String[maxFields];
  62                         if (contents[i] != null)
  63                         {
  64                                 String wholeLine = contents[i];
  65                                 if (!wholeLine.trim().equals(""))
  66                                 {
  67                                         // Don't use trimmed string here because you'll lose empty fields at beginning
  68                                         // if separated by spaces or tabs
  69                                         String[] splitLine = wholeLine.split(delimStr);
  70                                         if (splitLine != null)
  71                                         {
  72                                                 System.arraycopy(splitLine, 0, result[i], 0, splitLine.length);
  73                                                 // Check if columns are blank or not
  74                                                 for (int j=0; j<splitLine.length; j++)
  75                                                 {
  76                                                         if (!_columnStates[j] && splitLine[j].trim().length() > 0)
  77                                                         {
  78                                                                 _columnStates[j] = true;
  79                                                         }
  80                                                 }
  81                                         }
  82                                 }
  83                         }
  84                 }
  85                 return result;
  86         }
  87
  88
  89         /**
  90          * @return the number of rows in the data
  91          */
  92         public int getNumRows()
  93         {
  94                 return _numRows;
  95         }
  96
  97
  98         /**
  99          * @return the number of columns in the data
 100          */
 101         public int getNumColumns()
 102         {
 103                 return _numColumns;
 104         }
 105
 106         /**
 107          * @return the fields in the first full row
 108          */
 109         public String[] getFirstFullRow()
 110         {
 111                 return _firstFullRow;
 112         }
 113
 114
 115         /**
 116          * Check if the specified column of the data is blank
 117          * @param inColumnNum number of column, starting with 0
 118          * @return true if no data exists in this column
 119          */
 120         public boolean isColumnBlank(int inColumnNum)
 121         {
 122                 // Should probably trap out of range values
 123                 return !_columnStates[inColumnNum];
 124         }
 125
 126         /**
 127          * Check the delimiter for proper regular expression matching
 128          * @param inDelim character selected as delimiter
 129          * @return regular expression for splitting
 130          */
 131         private static String checkDelimiter(char inDelim)
 132         {
 133                 String result = "" + inDelim;
 134                 // Don't pass asterisks or dots without escaping them for RE
 135                 if (inDelim == '*' || inDelim == '.') {
 136                         result = "\\" + result;
 137                 }
 138                 return result;
 139         }
 140 }