コード例 #1
0
def action(args):
    engine = sqlalchemy.create_engine(args.url, echo=args.verbosity > 2)
    tax = Taxonomy(engine, schema=args.schema)

    records = list(yaml.load_all(args.new_nodes))

    log.info('adding new nodes')
    retval = None
    for rec in records:
        try:
            record_type = rec.pop('type')
            if record_type not in {'node', 'name'}:
                raise ValueError
        except (KeyError, ValueError):
            log.error(
                ('Error in record for tax_id {tax_id}: "type" is '
                 'required and must be one of "node" or "name"').format(**rec))
            retval = 1
            continue

        tax_id = rec['tax_id']
        rec['source_name'] = rec.get('source_name') or args.source_name

        try:
            if record_type == 'node':
                if not rec['source_name']:
                    log.error('Error: record has no source_name:\n{}'.format(
                        pprint.pformat(rec)))
                    raise ValueError
                if tax.has_node(tax_id):
                    log.info('updating *node* "{tax_id}"'.format(**rec))
                    tax.update_node(**rec)
                else:
                    log.info('new *node* "{tax_id}"'.format(**rec))
                    tax.add_node(**rec)
            elif record_type == 'name':
                for name in rec['names']:
                    name['tax_id'] = tax_id
                    # source_name may be provided at the record or name level
                    name['source_name'] = name.get(
                        'source_name') or rec['source_name']
                    if not name['source_name']:
                        log.error(
                            'Error: record has no source_name:\n {}'.format(
                                pprint.pformat(rec)))
                        raise ValueError

                    log.info('new *name* for "{tax_id}": "{tax_name}"'.format(
                        **name))
                    tax.add_name(**name)
        except (ValueError, TypeError):
            log.error('Error in record with tax_id {}'.format(rec['tax_id']))
            log.error(''.join(traceback.format_exception(*sys.exc_info())))
            retval = 1

    engine.dispose()

    if retval:
        log.error('Error: some records were malformed')
    return retval
コード例 #2
0
def action(args):
    engine = sqlalchemy.create_engine(args.url, echo=args.verbosity > 2)
    tax = Taxonomy(engine, schema=args.schema)
    nodes = list(get_new_nodes(args.new_nodes))

    # check if there are any new ranks and exit if needed
    node_ranks = set(n['rank'] for n in nodes)
    for r in node_ranks:
        if r not in tax.ranks:
            msg = 'adding new ranks to taxonomy is not yet supported'
            raise TaxonIntegrityError(msg)

    ranksdict = tax.ranksdict()
    ranksdict.update(dict([(n['tax_id'], n['rank']) for n in nodes]))
    nodes = [verify_rank_integrity(n, ranksdict, tax.ranks) for n in nodes]
    nodes = [verify_lineage_integrity(n, ranksdict, tax.ranks, tax) for n in nodes]

    log.info('adding new nodes')
    for d in nodes:
        if args.source_name:
            d['source_name'] = args.source_name
            try:
                tax.add_node(**d)
            except sqlalchemy.exc.IntegrityError:
                if args.update:
                    tax.update_node(**d)
                else:
                    log.warn('node with tax_id %(tax_id)s already exists' % d)
            else:
                log.info('added new node with tax_id %(tax_id)s' % d)

    engine.dispose()
コード例 #3
0
ファイル: add_nodes.py プロジェクト: fhcrc/taxtastic
def action(args):
    engine = sqlalchemy.create_engine(args.url, echo=args.verbosity > 2)
    tax = Taxonomy(engine, schema=args.schema)

    records = list(yaml.load_all(args.new_nodes))

    log.info('adding new nodes')
    retval = None
    for rec in records:
        try:
            record_type = rec.pop('type')
            if record_type not in {'node', 'name'}:
                raise ValueError
        except (KeyError, ValueError):
            log.error(('Error in record for tax_id {tax_id}: "type" is '
                       'required and must be one of "node" or "name"').format(**rec))
            retval = 1
            continue

        tax_id = rec['tax_id']
        rec['source_name'] = rec.get('source_name') or args.source_name

        try:
            if record_type == 'node':
                if not rec['source_name']:
                    log.error('Error: record has no source_name:\n{}'.format(
                        pprint.pformat(rec)))
                    raise ValueError
                if tax.has_node(tax_id):
                    log.info('updating *node* "{tax_id}"'.format(**rec))
                    tax.update_node(**rec)
                else:
                    log.info('new *node* "{tax_id}"'.format(**rec))
                    tax.add_node(**rec)
            elif record_type == 'name':
                for name in rec['names']:
                    name['tax_id'] = tax_id
                    # source_name may be provided at the record or name level
                    name['source_name'] = name.get('source_name') or rec['source_name']
                    if not name['source_name']:
                        log.error(
                            'Error: record has no source_name:\n {}'.format(
                                pprint.pformat(rec)))
                        raise ValueError

                    log.info('new *name* for "{tax_id}": "{tax_name}"'.format(**name))
                    tax.add_name(**name)
        except (ValueError, TypeError):
            log.error('Error in record with tax_id {}'.format(rec['tax_id']))
            log.error(''.join(traceback.format_exception(*sys.exc_info())))
            retval = 1

    engine.dispose()

    if retval:
        log.error('Error: some records were malformed')
    return retval
コード例 #4
0
def action(args):

    dbname = args.database_file
    new_nodes = args.new_nodes
    source_name = args.source_name

    engine = create_engine('sqlite:///%s' % dbname, echo=args.verbosity > 2)
    tax = Taxonomy(engine, ncbi.RANKS)

    log.warning('adding new nodes')
    nodes = get_new_nodes(new_nodes)
    for d in nodes:
        if source_name:
            d['source_name'] = source_name
            try:
                tax.add_node(**d)
            except IntegrityError:
                log.info('node with tax_id %(tax_id)s already exists' % d)
            else:
                log.info('added new node with tax_id %(tax_id)s' % d)

    engine.dispose()
コード例 #5
0
ファイル: add_nodes.py プロジェクト: TatyanaV/taxtastic
def action(args):

    dbname = args.database_file
    new_nodes = args.new_nodes
    source_name = args.source_name

    engine = create_engine('sqlite:///%s' % dbname, echo=args.verbosity > 2)
    tax = Taxonomy(engine, ncbi.RANKS)

    log.warning('adding new nodes')
    nodes = get_new_nodes(new_nodes)
    for d in nodes:
        if source_name:
            d['source_name'] = source_name
            try:
                tax.add_node(**d)
            except IntegrityError:
                log.info('node with tax_id %(tax_id)s already exists' % d)
            else:
                log.info('added new node with tax_id %(tax_id)s' % d)

    engine.dispose()