Wrong blob get with charset=WIN1251

[McFris]

FB 5.0.2. firebirdsql v0.9.15.

Create test table:

CREATE TABLE TESTBLOB (
    TXT      BLOB SUB_TYPE TEXT SEGMENT SIZE 80,
    CHARTXT  CHAR(10)
);
insert into testblob(txt, chartxt) values('Проверка', 'Проверка');

Than in Go:

conn, _ := sql.Open("firebirdsql", "SYSDBA:masterkey@127.0.0.1/test?charset=WIN1251")
conn.QueryRow(`select txt, chartxt from testblob`).Scan(&bl.Txt, &bl.CharTxt)
fmt.Printf("Blob in WIN1251: '%s'\n", bl.Txt)
fmt.Printf("Char in WIN1251: '%s'\n", bl.CharTxt)
conn.Close()

conn, _ = sql.Open("firebirdsql", "SYSDBA:masterkey@127.0.0.1/test")
conn.QueryRow(`select txt, chartxt from testblob`).Scan(&bl.Txt, &bl.CharTxt)
fmt.Printf("Blob without charset (UTF8 by default): '%s'\n", bl.Txt)
fmt.Printf("Char without charset (UTF8 by default): '%s'\n", bl.CharTxt)

Result:
Blob in WIN1251: '��������'
Char in WIN1251: 'Проверка'
Blob without charset (UTF8 by default): 'Проверка'
Char without charset (UTF8 by default): 'Проверка'

Expected:
Blob in WIN1251: 'Проверка'
Char in WIN1251: 'Проверка'

[nakagami]

I think the current behavior is correct.
The blob value get as is (the same as when charset is OCTET in a char).

What do others think is right?

[McFris]

Blob has subtype 1, which is text. And it has a storage charset and collate:

This type should be handled similarly to text fields, IMO.

[McFris]

For my conditions, this trick works well. But of course you will have to check the encoding and process this subtype according to it.

Blob in WIN1251: 'Проверка'
Char in WIN1251: 'Проверка'

[nakagami]

I am also beginning to think that it would be better to convert the SUB_TYPE TEXT of the BLOB with charset.

I think I can apply this pull request #192

Could you please test this?

[McFris]

Test passed. Thank you!

Blob in WIN1251: 'Проверка'
Char in WIN1251: 'Проверка'

[McFris]

Re-tested the mater and encode_blob branches (they are identical). Conversion does not work. It's strange, but it's a fact. :(