private void readRecord(LittleEndianInputStream is, int recordSize, Table table)
throws IOException {
int deletedFlag = is.readUnsignedByte();
byte[] data = new byte[recordSize];
Vector rowData = new Vector();
java.util.Iterator iterator = this.fields.iterator();
while (iterator.hasNext()) {
DBFField field = (DBFField) iterator.next();
is.read(data, 0, field.length);
switch (field.type) {
case 'C': // character string
String string = bytesToString(data, field.length, charsetName);
rowData.add(string.trim());
break;
case 'F': // floating number
try {
rowData.add(new Double(new String(data, 0, field.length)));
} catch (NumberFormatException exc) {
rowData.add(this.DEFAULT_NUMBER);
}
break;
case 'N': // number
try {
rowData.add(new Double(new String(data, 0, field.length)));
} catch (NumberFormatException exc) {
rowData.add(this.DEFAULT_NUMBER);
}
break;
case '8':
case 'O':
// little endian 8 byte double. Not tested !!! ???
long byte1 = data[0];
long byte2 = data[1];
long byte3 = data[2];
long byte4 = data[3];
long byte5 = data[4];
long byte6 = data[5];
long byte7 = data[6];
long byte8 = data[7];
long l =
(byte8 << 56)
+ (byte7 << 48)
+ (byte6 << 40)
+ (byte5 << 32)
+ (byte4 << 24)
+ (byte3 << 16)
+ (byte2 << 8)
+ byte1;
rowData.add(new Double(Double.longBitsToDouble(l)));
break;
case '4':
case 'I':
// little endian 4 byte integer. Not tested !!! ???
int i = (data[3] << 24) + (data[2] << 16) + (data[1] << 8) + data[0];
rowData.add(new Double(i));
break;
case '2':
// little endian 2 byte integer. Not tested !!! ???
rowData.add(new Double((data[1] << 8) + data[0]));
break;
/*
case 'D': // date
System.out.println ("Date objects in DBF files not tested."); // !!! ???
// Date in format YYYYMMDD.
if (field.length != 8)
throw new Exception("Date object has non-standard length in DBF file.");
int year = Integer.parseInt(new String(data, 0, 4));
int month = Integer.parseInt(new String(data, 4, 2));
int day = Integer.parseInt(new String(data, 6, 2));
Date date = new Date(year, month, day);
rowData.add(date.toString());
break;
case 'L': // logical
// ? Not initialised (default)
// Y,y Yes
// N,n No
// F,f False
// T,t True
final byte b = data[0];
if (b == 'Y' || b == 'y' || b == 'T' || b == 't')
rowData.add("True");
else if (b == 'N' || b == 'n' || b == 'F' || b == 'f')
rowData.add("False");
else
rowData.add("?");
break;
*/
default:
// add the raw bytes as String
rowData.add(new String(data, 0, field.length));
;
}
}
table.addRow(rowData);
}
public Table read(String filePath) throws IOException {
LittleEndianInputStream is = null;
if (filePath == null) return null;
try {
// read all data into a data buffer.
is = new LittleEndianInputStream(new BufferedInputStream(new FileInputStream(filePath)));
int fileCode = unsignedByteToInt(is.readByte());
int year = unsignedByteToInt(is.readByte());
int month = unsignedByteToInt(is.readByte());
int day = unsignedByteToInt(is.readByte());
long nbrRecords = is.readInt(); // & 0xffffffffL;// unsigned int
int headerSize = is.readShort();
int recordSize = is.readShort();
is.readShort(); // overread reserved value
is.readByte(); // transaction byte
int encrypted = is.readUnsignedByte(); // encription byte
is.skipBytes(13);
int codepage = is.readUnsignedByte();
is.skipBytes(2);
// map the codepage to a string. Based on:
// http://www.clicketyclick.dk/databases/xbase/format/dbf.html#DBF_STRUCT
switch (codepage) {
case 0x01: // DOS USA code page 437
this.charsetName = "IBM437";
break;
case 0x02: // DOS Multilingual code page 850
this.charsetName = "IBM850";
break;
case 0x03: // Windows ANSI code page 1252
this.charsetName = "windows-1252";
break;
case 0x04: // Standard Macintosh
this.charsetName = "MacRoman";
break;
// ESRI shape files use code 0x57 to indicate that
// data is written in ANSI (whatever that means).
// http://www.esricanada.com/english/support/faqs/arcview/avfaq21.asp
case 0x57:
this.charsetName = "windows-1252";
break;
case 0x64: // EE MS-DOS code page 852
this.charsetName = "IBM852";
break;
case 0x65: // Nordic MS-DOS code page 865
this.charsetName = "IBM865";
break;
case 0x66: // Russian MS-DOS code page 866
this.charsetName = "IBM866";
break;
case 0x67: // Icelandic MS-DOS
this.charsetName = "IBM861";
break;
/*
case 0x68: // Kamenicky (Czech) MS-DOS
// ?
break;
case 0x69: // Mazovia (Polish) MS-DOS
// ?
break;
*/
case 0x6A: // Greek MS-DOS (437G) [?]
this.charsetName = "x-IBM737";
break;
case 0x6B: // Turkish MS-DOS
this.charsetName = "IBM857";
break;
case 0x96: // Russian Macintosh
this.charsetName = "x-MacCyrillic";
break;
case 0x97: // Eastern European Macintosh
this.charsetName = "x-MacCentralEurope";
break;
case 0x98: // Greek Macintosh
this.charsetName = "x-MacGreek";
break;
case 0xC8: // Windows EE (=Eastern Europe?) code page 1250
this.charsetName = "windows-1250";
break;
case 0xC9: // Russian Windows
this.charsetName = "windows-1251";
break;
case 0xCA: // Turkish Windows
this.charsetName = "windows-1254";
break;
case 0xCB: // Greek Windows
this.charsetName = "windows-1253";
break;
default:
this.charsetName = "IBM437";
}
this.printInfo("File Code: " + fileCode);
this.printInfo("Year: " + year);
this.printInfo("Month: " + month);
this.printInfo("Day: " + day);
this.printInfo("Nbr Records: " + nbrRecords);
this.printInfo("Header Size: " + headerSize);
this.printInfo("Record Size: " + recordSize);
this.printInfo("Ecrypted: " + encrypted);
this.printInfo("Codepage: " + codepage);
if (encrypted != 0) throw new IOException("Encrypted DBF not supported.");
int nFields = (headerSize - 32) / 32;
this.readFieldDescriptors(is, nFields);
// create an new table
Table table = this.initTable("ShapeAttributes");
// read the records and fill the table
for (int i = 0; i < nbrRecords; i++) {
this.printInfo("Reading Record " + i);
this.readRecord(is, recordSize, table);
}
return table;
} finally {
if (is != null) is.close();
}
}
private void readFieldDescriptors(LittleEndianInputStream is, int nFields) throws IOException {
// read description of each field
byte[] asciiFieldName = new byte[11];
for (int i = 0; i < nFields; i++) {
DBFField field = new DBFField();
is.read(asciiFieldName);
field.name = bytesToString(asciiFieldName, asciiFieldName.length, this.charsetName);
field.type = is.readUnsignedByte();
field.address = is.readInt();
field.length = is.readUnsignedByte();
field.decimalCount = is.readUnsignedByte();
is.readShort(); // overread reserved value
field.workAreaID = is.readUnsignedByte();
field.multiUserDBase = is.readShort();
field.setFields = is.readUnsignedByte();
is.skipBytes(7); // overread 7 reserved bytes
field.fieldInMDXIndex = is.readUnsignedByte();
this.fields.add(field);
this.printInfo("\n" + field.toString());
}
// overread the Header Record Terminator, which should be 0x0D
byte terminator = is.readByte();
if (terminator != 0x0D) {
throw new IOException("DBF file is corrupt.");
}
}
