Пример #1
0
def comparejson(lexicon, _id, fromdate='', todate=''):
    from src.dbhandler.dbhandler import dbselect
    import src.server.translator.jsondiff as jsondiff
    auth, permitted = validate_user()
    if lexicon not in permitted:
        raise eh.KarpAuthenticationError('You are not allowed to update')

    try:
        if not todate:
            import datetime
            todate = datetime.datetime.now()
            tojson = dbselect(lexicon, max_hits=1, to_date=todate, _id=_id)[0]
        else:
            tojson = dbselect(lexicon, exact_date=todate, _id=_id)[0]

    # TODO catch the error here and print it to the log.
    # It is probably not really sql that raises the exception
    except Exception:
        raise eh.KarpDbError('Could not find any entry from %s' % todate)

    try:
        if not fromdate:
            jsons = dbselect(lexicon, max_hits=2, to_date=todate, _id=_id)
            fromjson = {'doc': {}} if len(jsons) == 1 else jsons[1]
        else:
            fromjson = dbselect(lexicon, exact_date=fromdate, _id=_id)[0]

        fromjson = fromjson['doc']
        tojson = tojson['doc']
    # TODO catch the error here and print it to the log.
    # It is probably not really sql that raises the exception
    except Exception:
        raise eh.KarpDbError('Could not find any entry from %s' % fromdate)
    return jsonify({'diff': jsondiff.compare(fromjson, tojson)})
Пример #2
0
def printlatestversion(lexicon, debug=True, with_id=False):
    to_keep = {}
    engine, db_entry = db.get_engine(lexicon, echo=False)
    count = 0
    for entry in db.dbselect(lexicon, engine=engine, db_entry=db_entry,
                             max_hits=-1):
        _id = entry['id']
        count += 1
        if _id:  # don't add things without id, they should not be here at all
            if _id in to_keep:
                last = to_keep[_id]['date']
                if last < entry['date']:
                    to_keep[_id] = entry
            else:
                    to_keep[_id] = entry

    if debug:
        print >> sys.stderr, 'count', count
    if with_id:
        out = [{'_id': i, '_source': val['doc']} for i, val in to_keep.items()
               if val['status'] != 'removed']
        print json.dumps(out)
    else:
        print '\n'.join((json.dumps(val['doc']) for val in to_keep.values()
                         if val['status'] != 'removed'))
Пример #3
0
def recover_add(index, suffix, lexicon):
    # TODO test this
    """ Recovers the data to ES, uses SQL as the trusted base version.
        Find the last version of every SQL entry and send this to ES.
        Adds the specified lexicons to an existing index
    """
    import src.server.translator.bulkify as bulk
    es = configM.elastic(index)
    print 'Save %s to %s' % (lexicon, index)
    to_keep = {}
    for lex in lexicon:
        engine, db_entry = db.get_engine(lex, echo=False)
        for entry in db.dbselect(lex, engine=engine, db_entry=db_entry,
                                 max_hits=-1):
            _id = entry['id']
            if _id:  # don't add things without id, they are errors
                if _id in to_keep:
                    last = to_keep[_id]['date']
                    if last < entry['date']:
                        to_keep[_id] = entry
                else:
                        to_keep[_id] = entry

    print len(to_keep), 'entries to keep'
    data = bulk.bulkify_sql(to_keep, bulk_info={'index': index})
    ok, err = helpers.bulk(es, data)
    if err:
        msg = "Error during upload. %s documents successfully uploaded. \
               Message: %s.\n"
        raise Exception(msg % (ok, '\n'.join(err)))
    print 'recovery done'
Пример #4
0
def checkhistory(lexicon, lid):
    """ Shows the update log of an entry """
    from src.dbhandler.dbhandler import dbselect
    auth, permitted = validate_user(mode="read")
    settings = {'allowed': permitted}
    size = helpers.get_size(default=10, settings=settings)
    return jsonify({'updates': dbselect(lexicon, _id=lid, max_hits=size)})
Пример #5
0
def export(lexicon):
    # TODO can user with only read permissions export all the lexicon?
    # (eg saol)
    auth, permitted = validate_user(mode="read")
    if lexicon not in permitted:
        raise eh.KarpAuthenticationError('You are not allowed to search the '
                                         'lexicon %s' % lexicon)

    settings = parser.make_settings(permitted, {
        "size": -1,
        "resource": lexicon
    })
    query = request.query_string
    parsed = parser.parse_qs(query)
    parser.parse_extra(parsed, settings)
    date = settings.get('date', '')
    mode = settings.get('mode', '')
    if date:
        from dateutil.parser import parserinfo, parse
        from datetime import datetime
        # parse the date as inclusive (including the whole selected day)
        date = parse(date,
                     parserinfo(yearfirst=True),
                     default=datetime(1999, 01, 01, 23, 59))

    to_keep = {}
    engine, db_entry = db.get_engine(lexicon, echo=False)
    logging.debug('exporting entries from %s ' % lexicon)
    for entry in db.dbselect(lexicon,
                             engine=engine,
                             db_entry=db_entry,
                             max_hits=-1,
                             to_date=date):
        _id = entry['id']
        if _id in to_keep:
            last = to_keep[_id]['date']
            if last < entry['date']:
                to_keep[_id] = entry
        else:
            to_keep[_id] = entry

    ans = [
        val['doc'] for val in to_keep.values() if val['status'] != 'removed'
    ]
    ans = ans[:settings['size']]

    logging.debug('exporting %s entries' % len(ans))
    if settings.get('format', ''):
        toformat = settings.get('format')
        index, typ = configM.get_mode_index(mode)
        msg = 'Unkown %s %s for mode %s' % ('format', toformat, mode)
        format_posts = configM.extra_src(mode, 'exportformat',
                                         helpers.notdefined(msg))
        lmf, err = format_posts(ans, lexicon, mode, toformat)
        return Response(lmf, mimetype='text/xml')

    else:
        return jsonify({lexicon: ans})
Пример #6
0
def recover(alias, suffix, lexicon, create_new=True):
    # TODO test this
    """ Recovers the data to ES, uses SQL as the trusted base version.
        Find the last version of every SQL entry and send this to ES.
    """
    import src.server.translator.bulkify as bulk
    to_keep = {}
    # if not lexicon:
    #     lexicon = conf.keys()
    mapping = get_mapping(alias)
    index = make_indexname(alias, suffix)
    typ = configM.get_mode_type(alias)
    print 'Save %s to %s' % (lexicon or 'all', index)

    es = configM.elastic(alias)
    if create_new:
        # Create the index
        ans = es.indices.create(index=index, body=mapping)
        print ans

    for lex in lexicon:
        engine, db_entry = db.get_engine(lex, echo=False)
        for entry in db.dbselect(lex, engine=engine, db_entry=db_entry, max_hits=-1):
            _id = entry['id']
            if _id:  # don't add things without id, they are errors
                if _id in to_keep:
                    last = to_keep[_id]['date']
                    if last < entry['date']:
                        to_keep[_id] = entry
                else:
                        to_keep[_id] = entry
    print len(to_keep), 'entries to keep'
    data = bulk.bulkify_sql(to_keep, bulk_info={'index': index, 'type': typ})
    try:
        ok, err = helpers.bulk(es, data)
    except:
        print data
    if err:
        msg = "Error during upload. %s documents successfully uploaded. \
               Message: %s.\n"
        raise Exception(msg % (ok, '\n'.join(err)))
    print 'recovery done'
Пример #7
0
def checkuserhistory():
    """ Shows the updates a user has made """
    try:
        auth, permitted = validate_user()
        user = helpers.get_user()
    except AttributeError:
        raise eh.KarpGeneralError('No user name provided', 'checkuserhistory')
    try:
        size = helpers.get_size(default=10, settings={'allowed': permitted})
        from src.dbhandler.dbhandler import dbselect
        updates = []
        for lexicon in permitted:
            # add updates from lexicons that are kept in sql
            if configM.get_lexicon_sql(lexicon):
                updates.extend(dbselect(lexicon, user=user, max_hits=size))

        return jsonify({'updates': updates})
    except Exception as e:
        logging.exception(e)
        raise eh.KarpGeneralError(str(e))
Пример #8
0
def checklexiconhistory(lexicon, date):
    """ Shows the updates on one lexicon """
    try:
        auth, permitted = validate_user()
        if lexicon not in permitted:
            raise eh.KarpAuthenticationError(
                'You are not allowed to update lexicon %s' % lexicon)
        settings = {"allowed": permitted}
        helpers.get_querysettings(settings)
        size = settings.get('size', 10)
        status = settings.get('status', ['added', 'changed', 'removed'])

        from src.dbhandler.dbhandler import dbselect
        return jsonify({
            'resource':
            lexicon,
            'updates':
            dbselect(lexicon, status=status, from_date=date, max_hits=size)
        })
    except Exception as e:
        raise eh.KarpGeneralError(str(e))