> On 25/05/2023 13:26 CEST Peter J. Holzer <hjp-pgsql@xxxxxx> wrote: > > On 2023-05-25 07:14:40 +0000, Laura Smith wrote: > > I'm currently doing a CSV export using COPY: > > > > COPY (select * from foo where bar='foo') TO '/tmp/bar.csv' DELIMITER ',' CSV HEADER; > > > > This works great apart from accents are not preserved in the output, > > for example é gets converted to random characters, e.g. √© or similar. > > How do you check the output? > > If a single character is turned into 2 or 3 characters the issue is > usually that the program which produces the output (in the case of COPY > I think that would be the PostgreSQL server, not the client) produces > UTF-8, but the program consuming it expects an 8-bit character set > (typically windows-1252). See if oyu can tell that program that the file > is in UTF-8. > > > How can I preserve accents ? > > They probably already are preserved. You're right. The bytes are probably interpreted as Mac OS Roman: $ echo é | iconv -f macintosh √© $ echo -n é | xxd 00000000: c3a9 -- Erik
