Python sqlaload.upsert函数代码示例

def load_transcript(engine, wp, session, incremental=True):
    url = URL % (wp, session)
    Speech = sl.get_table(engine, 'speech')
    if incremental and sl.find_one(engine, Speech,
        source_url=url, matched=True):
        return True
    if '404 Seite nicht gefunden' in fetch(url):
        return False
    sio = fetch_stream(url)
    if sio is None:
        return False
    log.info("Loading transcript: %s/%s" % (wp, session))
    seq = 0
    parser = SpeechParser(engine, sio)
    for contrib in parser:
        if not len(contrib['text'].strip()):
            continue
        contrib['sitzung'] = session
        contrib['sequence'] = seq
        contrib['wahlperiode'] = wp
        contrib['source_url'] = url
        contrib['matched'] = True
        sl.upsert(engine, Speech, contrib, 
                  unique=['sequence', 'sitzung', 'wahlperiode'])
        seq += 1
    if parser.missing_recon:
        sl.upsert(engine, Speech, {
                    'matched': False,
                    'sitzung': session,
                    'wahlperiode': wp
            }, unique=['sitzung', 'wahlperiode'])

    return True

def update_network_entities(engine, file_name):
    log.info("Updating network entities reference sheet: %s", file_name)
    network_entities = set()
    table = sl.get_table(engine, 'network_entity')
    if os.path.exists(file_name):
        fh = open(file_name, 'rb')
        reader = csv.DictReader(fh)
        for d in reader:
            e = [(k, v.decode('utf-8')) for (k, v) in d.items()]
            e = dict(e)
            network_entities.add((e['representativeEtlId'], e['etlFingerPrint']))
            sl.upsert(engine, table, e, ['representativeEtlId', 'etlFingerPrint'])
        fh.close()
        reps = set([ne[0] for ne in network_entities])
        rep_table = sl.get_table(engine, 'representative')
        for rep in reps:
            sl.update(engine, rep_table, {'etlId': rep}, {'network_extracted': True})

    for row in sl.all(engine, table):
        network_entities.add((row['representativeEtlId'], row['etlFingerPrint']))

    fh = open(file_name, 'wb')
    writer = None
    table = sl.get_table(engine, 'network_entity')
    for ic, fp in network_entities:
        row = {
            'representativeEtlId': ic,
            'etlFingerPrint': fp
        }
        if writer is None:
            writer = csv.DictWriter(fh, row.keys())
            writer.writerow(dict(zip(row.keys(), row.keys())))
        r = [(k, unicode(v).encode('utf-8')) for k, v in row.items()]
        writer.writerow(dict(r))
    fh.close()

def generate_person_long_names(engine):
    log.info("Generating person fingerprints and slugs...")
    from offenesparlament.transform.namematch import match_speaker
    nkp = nk_persons()
    Person = sl.get_table(engine, 'person')
    for person in sl.find(engine, Person):
        long_name = make_long_name(person)
        try:
            long_name = match_speaker(long_name)
        except NKNoMatch:
            pass
        log.info(" -> %s" % long_name.strip())
        slug = url_slug(long_name)
        sl.upsert(engine, Person, {
                         'fingerprint': long_name,
                         'slug': slug,
                         'id': person['id']},
                         unique=['id'])
        tries = 0
        while True:
            try:
                nkp.ensure_value(long_name, data=person)
            except ValueError, E:
                log.warn('Exception: %s' % str(E))
                tries = tries + 1
                if tries > 5:
                    raise
            else:
                break

def merge_speeches(engine):
    # desired result: (position_id, debatte_id)
    referenzen = referenzen_index(engine)
    items = item_index(engine)
    
    log.info("Finding best matches.... ")
    matches = {}
    for (ablauf_id, rwp, rsession), rdrs in referenzen.items():
        for (iwp, isession, item_id), idrs in items.items():
            if iwp != rwp or rsession != isession:
                continue
            ints = len(idrs.intersection(rdrs))
            if ints == 0:
                continue
            k = (ablauf_id, rwp, rsession)
            if k in matches and matches[k][1] > ints:
                continue
            matches[k] = (item_id, ints)

    log.info("Saving position associations....")
    pos_tbl = sl.get_table(engine, 'position')
    for (ablauf_id, wp, session), (item_id, n) in matches.items():
        for pos in sl.find(engine, pos_tbl, ablauf_id="%s/%s" % (wp, ablauf_id)):
            if not pos['fundstelle_url']:
                continue
            if 'btp/%s/%s%03d.pdf' % (wp, wp, int(session)) in pos['fundstelle_url']:
                d = {'ablauf_id': pos['ablauf_id'], 
                     'hash': pos['hash'],
                     'debatte_wp': wp,
                     'debatte_session': session,
                     'debatte_item_id': item_id}
                sl.upsert(engine, pos_tbl, d, unique=['ablauf_id', 'hash'])

def extend_position(engine, table, data):
    dt, rest = data['fundstelle'].split("-", 1)
    data['date'] = datetime.strptime(dt.strip(), "%d.%m.%Y").isoformat()
    if ',' in data['urheber']:
        typ, quelle = data['urheber'].split(',', 1)
        data['quelle'] = re.sub("^.*Urheber.*:", "", quelle).strip()
        data['typ'] = typ.strip()
    else:
        data['typ'] = data['urheber']

    br = 'Bundesregierung, '
    if data['urheber'].startswith(br):
        data['urheber'] = data['urheber'][len(br):]

    data['fundstelle_doc'] = None
    if data['fundstelle_url'] and \
            'btp' in data['fundstelle_url']:
        data['fundstelle_doc'] = data['fundstelle_url']\
                .rsplit('#',1)[0]

    hash = sha1(data['fundstelle'].encode('utf-8') \
            + data['urheber'].encode('utf-8') + \
            data['source_url'].encode('utf-8')).hexdigest()
    data['hash'] = hash[:10]
    sl.upsert(engine, table, data, unique=['id'])

def extend_positions(engine):
    log.info("Amending positions ...")
    Position = sl.get_table(engine, 'position')
    for i, data in enumerate(sl.find(engine, Position)):
        if i % 1000 == 0:
            sys.stdout.write('.')
            sys.stdout.flush()
        dt, rest = data['fundstelle'].split("-", 1)
        data['date'] = datetime.strptime(dt.strip(), "%d.%m.%Y").isoformat()
        if ',' in data['urheber']:
            typ, quelle = data['urheber'].split(',', 1)
            data['quelle'] = re.sub("^.*Urheber.*:", "", quelle).strip()
            data['typ'] = typ.strip()
        else:
            data['typ'] = data['urheber']

        br = 'Bundesregierung, '
        if data['urheber'].startswith(br):
            data['urheber'] = data['urheber'][len(br):]

        data['fundstelle_doc'] = None
        if data['fundstelle_url'] and \
                'btp' in data['fundstelle_url']:
            data['fundstelle_doc'] = data['fundstelle_url']\
                    .rsplit('#',1)[0]

        hash = sha1(data['fundstelle'].encode('utf-8') \
                + data['urheber'].encode('utf-8') + \
                data['ablauf_id'].encode('utf-8')).hexdigest()
        data['hash'] = hash[:10]
        sl.upsert(engine, Position, data, unique=UNIQUE)

def cleanup_resource(engine, source_table, row, force):
    if not row["combine_status"]:
        return

    # Skip over tables we have already cleaned up
    if (
        not force
        and sl.find_one(
            engine, source_table, resource_id=row["resource_id"], cleanup_status=True, cleanup_hash=row["combine_hash"]
        )
        is not None
    ):
        return

    log.info("Cleanup: %s, Resource %s", row["package_name"], row["resource_id"])

    status = True
    for sheet_id in range(0, row["sheets"]):
        sheet_status = cleanup_sheet(engine, row, sheet_id)
        if status and not sheet_status:
            status = False
    sl.upsert(
        engine,
        source_table,
        {"resource_id": row["resource_id"], "cleanup_hash": row["combine_hash"], "cleanup_status": status},
        unique=["resource_id"],
    )

def map_columns():
    engine, columns_table = connect()

    q = select([columns_table.c.normalised, columns_table.c.count, columns_table.c.valid], order_by=[columns_table.c.count.desc().nullslast()])

    for normalised, count, valid in engine.execute(q):
        if valid is not None:
            continue
        try:
            columns = map_column(engine, columns_table, normalised, count)
            if columns is not None:
                sl.upsert(engine, columns_table, 
                          {'normalised': normalised,
                           'valid': True,
                           'column_map': json.dumps(columns)},
                          ['normalised'])
            else:
                sl.upsert(engine, columns_table, 
                          {'normalised': normalised,
                           'valid': False}, 
                          ['normalised'])
        except SystemExit:
            raise
        except:
            traceback.print_exc()

def scrape_transcript(engine, url, force=False):
    wp, session = url_metadata(url)
    table = sl.get_table(engine, 'speech')
    sample = sl.find_one(engine, table, source_url=url, matched=True)
    response, sio = fetch_stream(url)
    sample = check_tags(sample or {}, response, force)
    base_data = {'source_url': url, 
                 'sitzung': session,
                 'wahlperiode': wp,
                 'matched': False,
                 'loaded': False,
                 'source_etag': sample['source_etag']}
    log.info("Loading transcript: %s/%s, from %s" , wp, session, url)
    seq = 0
    parser = SpeechParser(sio)
    for contrib in parser:
        if not len(contrib['text'].strip()):
            continue
        contrib.update(base_data)
        contrib['sequence'] = seq
        sl.upsert(engine, table, contrib, 
                  unique=['source_url', 'sequence'])
        seq += 1
    if not parser.missing_recon:
        sl.upsert(engine, table, {
                    'matched': True,
                    'source_url': url,
            }, unique=['source_url'])
    else:
        raise InvalidReference()
    return base_data

def load_budget(base_url, year, engine, table):
    context = {'data_year': year}
    print "\nHaushalt: %s" % year
    i = 0
    for row in load_einzelplaene(base_url % year, context):
        row['titel_id'] = row['id']
        del row['id']
        row['remarks'] = "\n\n".join(row['remarks'])
        commitment_appropriations = row['commitment_appropriations'].copy()
        del row['commitment_appropriations']
        #if len(commitment_appropriations):
        #    #print len(commitment_appropriations)
        
        row['commitment_year'] = None
        row['source_id'] = str(year) + "." + str(i)
        sl.upsert(engine, table, row, UNIQUE_COLUMNS)
        i += 1

        for year, amount in commitment_appropriations.items():
            ca = row.copy()
            ca['commitment_year'] = context['data_year']
            ca['year'] = year
            ca['amount'] = amount
            ca['financial_type'] = 'VE'
            ca['source_id'] = str(year) + "." + str(i)
            sl.upsert(engine, table, ca, UNIQUE_COLUMNS)
            i += 1

def make_fingerprint(engine, person):
    try:
        long_name = make_long_name(person)
        try:
            long_name = resolve_person(long_name)
            log.info(" -> %s" % long_name.strip())
        except:
            log.error("Resolve did not work")
            pass

        Person = sl.get_table(engine, 'person')
        sl.upsert(engine, Person, {
            'fingerprint': long_name,
            'slug': url_slug(long_name),
            'mdb_id': person['mdb_id']
            }, unique=['mdb_id'])
        Rolle = sl.get_table(engine, 'rolle')
        sl.upsert(engine, Rolle, {
            'mdb_id': person['mdb_id'],
            'fingerprint': long_name
            }, unique=['mdb_id'])
        person['fingerprint'] = long_name
    except BadReference:
        log.error("Bad Reference %s", person)
        pass

def process_rows(handlefunc, engine=None):
    if engine is None:
        engine = make_engine()
    table = sl.get_table(engine, 'fts')
    for row in sl.all(engine, table):
        out = handlefunc(row)
        sl.upsert(engine, table, out, ['id'])
    return table

def condense(engine, resource_id, table_id, force):
    table_suffix = '%s_table%s' % (resource_id, table_id)

    if not engine.has_table('raw_%s' % table_suffix):
        return

    condensed_table = sl.get_table(engine, 'condensed')

    # Skip over tables we have already extracted
    if not force and sl.find_one(engine, condensed_table, resource_id=resource_id, table_id=table_id) is not None:
        return

    connection = engine.connect()
    trans = connection.begin()

    start = time.time()

    try:
        raw_table = sl.get_table(connection, 'raw_%s' % table_suffix)
        sl.drop_table(connection, 'spending_%s' % table_suffix)
        spending_table = sl.get_table(connection, 'spending_%s' % table_suffix)
        columns_table = sl.get_table(connection, 'column_sets')

        normalise_map = normalised_columns_map(raw_table)
        normalised_headers = ','.join(sorted(normalise_map.values()))
        mapping_row = sl.find_one(connection, columns_table, normalised=normalised_headers)

        if mapping_row is None or not mapping_row.get('valid'):
            # This table is unmapped, cannot be condensed
            return

        column_mapping = json.loads(mapping_row['column_map'])

        # Build the final mapping from input column to output column
        mapping = {}
        for k,n in normalise_map.iteritems():
            if n in column_mapping and column_mapping[n] is not None and len(column_mapping[n]) > 0:
                mapping[k] = column_mapping[n]
        
        for row in sl.all(connection, raw_table):
            spending_row = {}
            for key, value in row.items():
                if key not in mapping:
                    continue
                if not value or not len(value.strip()):
                    continue
                if mapping[key] in spending_row:
                    continue
                spending_row[mapping[key]] = value.strip()
            #print spending_row
            sl.add_row(connection, spending_table, spending_row)
        sl.upsert(connection, condensed_table, {'resource_id': resource_id,
                                                'table_id': table_id,
                                                'condense_time': time.time() - start,
                                                }, ['resource_id', 'table_id'])
        trans.commit()
    finally:
        connection.close()

def speechmatcher_alignment_post(wp, session):
    engine = etl_engine()
    table = sl.get_table(engine, 'alignments')
    data = dict(request.form.items())
    data['sequence'] = int(data['sequence'])
    data['wp'] = wp
    data['session'] = session
    sl.upsert(engine, table, data, ['wp', 'session', 'sequence'])
    return speechmatcher_alignment_get(wp, session)

def merge():
    engine = util.make_engine()
    table = sl.get_table(engine, 'fts')
    for row in sl.distinct(engine, table, 'beneficiary', 'country_code'):
        canonical, uri, score = lookup(row.get('beneficiary'), row.get('country_code'), engine)
        row['beneficiary_canonical'] = canonical
        row['beneficiary_uri'] = uri
        row['beneficiary_score'] = score
        sl.upsert(engine, table, row, ['beneficiary', 'country'])