jferard / jxbase Goto Github PK

If you try to delete *.dbf and *.dbt file, after line dbfWriter.close() in WriterWithMemoIT , you will find that you can not delete *.dbt file, "because it used by another process". As i can see it happens because MemoWriter is not closed during dbfWriter.close()

When write data to new dbf-table with memo fields, memo fields in all records have same value from last record.
And dbt file is small, and you can see in it - only one memo field value from last record.
For writer using various types: XBaseFileTypeEnum.dBASE3plusMemo, XBaseFileTypeEnum.dBASE4Memo, XBaseFileTypeEnum.dBASE7Memo
t123.zip
Original data:
Original.pdf
Test for problem:
MemoTest.zip

Before:

    final int memoRecordLength =
            BytesUtils.makeLEInt(recordHeaderBytes[4], recordHeaderBytes[5], recordHeaderBytes[6],
                    recordHeaderBytes[7]);

After:

    final int memoRecordLength = BytesUtils.extractLEInt(recordHeaderBytes, 4, 4);

I gave a try to Mockito, but I'm not convinced.

• SQLite
• MySQL/MariaDB
• PostgreSQL
• SQLServer (?)
• Oracle

And check the length in other flavours.

The current design is poor: MemoAccess holds a reference on the memo handle. Should be removed.

See: http://www.dfpug.de/loseblattsammlung%5Cmigration%5Cwhitepapers/DDIntro.htm

File extensions are DBC, DCX and DCT. Should be handled like regular files.

• ReaderWithMemoIT
• ReaderIT

In com.github.jferard.jxbase.tool.DatabaseLoader.

static final void main(...) {
     // directory or file, connection string or file (sqlite)
}

Before:

if (DB4Utils.MEMO_FIELD_RESERVED_BYTES[0]` != recordHeaderBytes[0] ||
            DB4Utils.MEMO_FIELD_RESERVED_BYTES[1] != recordHeaderBytes[1] ||
            DB4Utils.MEMO_FIELD_RESERVED_BYTES[2] != recordHeaderBytes[2] ||
            DB4Utils.MEMO_FIELD_RESERVED_BYTES[3] != recordHeaderBytes[3]) {

After:

if (BytesUtils.equals(MEMO_FIELD_RESERVED_BYTES, 0, recordHeaderBytes, 0, 4)) {