Rural Hunter wrote:
> My db is in utf-8, I have a row in my table say tmp_article and I
wanted
> to generate ts_vector from the article content:
> select to_tsvector(content) from tmp_article;
> But I got this error:
> ERROR: invalid byte sequence for encoding "UTF8": 0xf481
>
> I am wondering how this could happen. I think if there was invalid
UTF8
> bytes in the content, it shouldn't have been able to inserted into the
> tmp_article table as I sometimes see similar errors when inserting
> records to tmp_article. Am I right?
You are right in theory. A lot depends on your PostgreSQL version,
because
the efforts to prevent invalid strings from entering the database have
led to changes over the versions. Older versions are more permissive.
To test the theory that the contents of the table are bad, you can
test if the same happens if you
SELECT convert_to(content, 'UTF8') FROM tmp_article;
Yours,
Laurenz Albe
