metadata_reset.py

MAX_META_SIZE = 2000000
ids = sa_session.execute("select id from history_dataset_association where length(metadata)  > %s" % MAX_META_SIZE)
ids = [id for id in ids]
print len(ids), " large metadata blobs detected"
for id in ids:
    id = id['id']
    # Let's see what keys we have.
    data = sa_session.query(galaxy.model.HistoryDatasetAssociation).get(id)
    for name, spec in data.metadata.spec.items():
        # We need to be careful about the attributes we are resetting!
        if name in ['bam_header', 'read_groups', 'reference_lengths', 'reference_names']:
            print "Stripped %s from hda %s" % (name, id)
            del data._metadata[name]
    sa_session.flush()