On Mon, 29 Nov 2010, Cornelia Plott wrote: >> $ python -c "from invenio.search_engine import search_pattern; \ >> print search_pattern(p='POETRY', f='980__a')" > > This return results for a demo site collection as well as for our > defined collection! (intbitset([105, 106, 107, 108, 109, ..., 5440390, > 5440391, 5440392, 5440393, 5440394]))
OK, so the problem seems to be with the `collection' index definition. You can double check in the BibIndex Admin interface to see how it is linked, or you can use the following set of SQL statements: $ /opt/invenio/bin/dbexec -i mysql> SELECT id_field FROM idxINDEX_field, idxINDEX WHERE id_idxINDEX=id AND name='collection'; +----------+ | id_field | +----------+ | 10 | +----------+ 1 row in set (0.00 sec) mysql> SELECT * FROM field_tag, tag WHERE id_field=10 AND id_tag=tag.id; +----------+--------+-------+----+-----------------------+--------+ | id_field | id_tag | score | id | name | value | +----------+--------+-------+----+-----------------------+--------+ | 10 | 11 | 100 | 11 | collection identifier | 980__% | +----------+--------+-------+----+-----------------------+--------+ 1 row in set (0.02 sec) This shows that `collection' index is related to field ID 10 that is related to any 980__* tag, which is what we want. Is everything alright from this point of view? If yes, you can try to list what terms the `collection' word and phrase indexes contain: $ echo "SELECT term FROM idxWORD10F" | /opt/invenio/bin/dbexec $ echo "SELECT term FROM idxPHRASE10F" | /opt/invenio/bin/dbexec > 2010-11-26 12:24:12 --> Input file > '/cdsware/home/cdsware/scopus_simple_test.xml', input mode > 'replace_or_insert'. > 2010-11-26 12:24:12 --> Error during the creation_new_record function : > (1062, "Duplicate entry '0' for key 1") How does `scopus_simple_test.xml' look like? You can use `bibupload -v9 -ir scopus_simple_test.xml' to see more verbose messages to debug this issue. Best regards -- Tibor Simko
