On Mon, 25 Aug 2008 10:21:54 -0400 "John T. Dow" <john@xxxxxxxxxxxx> wrote: > By "bad data", I mean a character that's not UTF8, such as hex 98. > > As far as I can tell, pg_dump is the tool to use. But it has > serious drawbacks. > > If you dump in the custom format, the data is compressed (nice) and > includes large objects (very nice). But, from my tests and the > postings of others, if there is invalid data in a table, although > PostgreSQL won't complain and pg_dump won't complain, pg_restore will > strenuously object, rejecting all rows for that particular table (not > nice at all). You can use the TOC feature of -Fc to remove restoring of that single table. You can then convert that single table to a plain text dump and clean the data. Then restore it separately. If you have foregin keys and indexes on the bad data table, don't restore the keys until *after* you have done the above. Sincerely, Joshua D. Drake -- The PostgreSQL Company since 1997: http://www.commandprompt.com/ PostgreSQL Community Conference: http://www.postgresqlconference.org/ United States PostgreSQL Association: http://www.postgresql.us/ Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
