ResultsStorage.py

# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements.  See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You under the Apache License, Version 2.0
# (the "License"); you may not use this file except in compliance with
# the License.  You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import configparser
import json
import logging
import uuid
from datetime import datetime
from time import sleep

from pathlib import Path
from cassandra.auth import PlainTextAuthProvider
from cassandra.cluster import Cluster
from cassandra.concurrent import execute_concurrent_with_args
from cassandra.policies import TokenAwarePolicy, DCAwareRoundRobinPolicy
from pytz import UTC
from webservice.algorithms.doms.BaseDomsHandler import DomsEncoder
from webservice.webmodel import NexusProcessingException

BATCH_SIZE = 1024

logger = logging.getLogger(__name__)


class ResultInsertException(IOError):
    pass


class AbstractResultsContainer:
    def __init__(self, config=None):
        self._log = logging.getLogger(__name__)
        self._log.info("Creating DOMS Results Storage Instance")

        self._session = None
        self._config = configparser.RawConfigParser()
        self._config.read(AbstractResultsContainer._get_config_files('domsconfig.ini'))

        if config:
            self.override_config(config)
        else:
            print('Config NOT provided from params...')

    def __enter__(self):
        cassHost = self._config.get("cassandra", "host")
        cassKeyspace = self._config.get("cassandra", "keyspace")
        cassDatacenter = self._config.get("cassandra", "local_datacenter")
        cassVersion = int(self._config.get("cassandra", "protocol_version"))
        cassUsername = self._config.get("cassandra", "username")
        cassPassword = self._config.get("cassandra", "password")

        auth_provider = PlainTextAuthProvider(username=cassUsername, password=cassPassword)

        dc_policy = DCAwareRoundRobinPolicy(cassDatacenter)
        token_policy = TokenAwarePolicy(dc_policy)

        logger.info(f'Connecting to Cassandra cluster @ {[host for host in cassHost.split(",")]}; datacenter: '
                    f'{cassDatacenter}; protocol version: {cassVersion}')

        self._cluster = Cluster([host for host in cassHost.split(',')], load_balancing_policy=token_policy,
                                protocol_version=cassVersion, auth_provider=auth_provider)

        self._session = self._cluster.connect(cassKeyspace)
        return self

    def __exit__(self, exc_type, exc_val, exc_tb):
        self._cluster.shutdown()

    def _parseDatetime(self, dtString):
        dt = datetime.strptime(dtString, "%Y-%m-%dT%H:%M:%SZ")
        epoch = datetime.utcfromtimestamp(0)
        time = (dt - epoch).total_seconds() * 1000.0
        return int(time)

    def override_config(self, config):
        for section in config.sections():
            if self._config.has_section(section):
                for option in config.options(section):
                    if config.get(section, option) is not None:
                        self._config.set(section, option, config.get(section, option))

    @staticmethod
    def _get_config_files(filename):
        log = logging.getLogger(__name__)
        candidates = []
        extensions = ['.default', '']
        for extension in extensions:
            try:
                candidate = str(Path(__file__).parent / (filename + extension))
                log.info('use config file {}'.format(filename + extension))
                candidates.append(candidate)
            except KeyError as ke:
                log.warning('configuration file {} not found'.format(filename + extension))
        return candidates


class ResultsStorage(AbstractResultsContainer):
    def __init__(self, config=None):
        AbstractResultsContainer.__init__(self, config)

    def insertInitialExecution(self, params, startTime, status, userEmail='', execution_id=None):
        """
        Initial insert into database for CDMS matchup request. This
        populates the execution and params table.
        """
        if isinstance(execution_id, str):
            execution_id = uuid.UUID(execution_id)

        execution_id = self.__insertExecution(execution_id, startTime, None, userEmail, status)
        self.__insertParams(execution_id, params)
        return execution_id

    def updateExecution(self, execution_id, completeTime, status, message, stats, results):
        if stats:
            self.__insertStats(execution_id, stats)
        if results:
            self.__insertResults(execution_id, results)
        self.__updateExecution(execution_id, completeTime, status, message)

    def __insertExecution(self, execution_id, startTime, completeTime, userEmail, status):
        """
        Insert new entry into execution table
        """
        if execution_id is None:
            execution_id = uuid.uuid4()

        cql = "INSERT INTO doms_executions (id, time_started, time_completed, user_email, status) VALUES (%s, %s, %s, %s, %s)"
        self._session.execute(cql, (execution_id, startTime, completeTime, userEmail, status))
        return execution_id

    def __updateExecution(self, execution_id, complete_time, status, message=None):
        # Only update the status if it's "running". Any other state is immutable.
        cql = "UPDATE doms_executions SET time_completed = %s, status = %s, message = %s WHERE id=%s IF status = 'running'"
        self._session.execute(cql, (complete_time, status, message, execution_id))
        return execution_id

    def __insertParams(self, execution_id, params):
        cql = """INSERT INTO doms_params
                    (execution_id, primary_dataset, matchup_datasets, depth_min, depth_max, time_tolerance, radius_tolerance, start_time, end_time, platforms, bounding_box, parameter)
                 VALUES
                    (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
        """
        self._session.execute(cql, (execution_id,
                                    params["primary"],
                                    ",".join(params["matchup"]) if type(params["matchup"]) == list else params[
                                        "matchup"],
                                    params["depthMin"] if "depthMin" in list(params.keys()) else None,
                                    params["depthMax"] if "depthMax" in list(params.keys()) else None,
                                    int(params["timeTolerance"]),
                                    params["radiusTolerance"],
                                    params["startTime"],
                                    params["endTime"],
                                    params["platforms"],
                                    params["bbox"],
                                    params["parameter"]
                                    ))

    def __insertStats(self, execution_id, stats):
        cql = """
           INSERT INTO doms_execution_stats
                (execution_id, num_gridded_matched, num_gridded_checked, num_insitu_matched, num_insitu_checked, time_to_complete, num_unique_secondaries)
           VALUES
                (%s, %s, %s, %s, %s, %s, %s)
        """
        self._session.execute(cql, (
            execution_id,
            stats['numPrimaryMatched'],
            None,
            stats['numSecondaryMatched'],
            None,
            stats['timeToComplete'],
            stats['numUniqueSecondaries']
        ))

    def __insertResults(self, execution_id, results):

        cql = """
           INSERT INTO doms_data
                (id, execution_id, value_id, primary_value_id, x, y, source_dataset, measurement_time, platform, device, measurement_values_json, is_primary, depth, file_url)
           VALUES
                (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
        """
        insertStatement = self._session.prepare(cql)

        inserts = []

        for result in results:
            # 'PRIMARY' arg since primary values cannot have primary_value_id be null anymore
            # Secondary matches are prepped recursively from this call
            inserts.extend(self.__prepare_result(execution_id, 'PRIMARY', result, insertStatement))

        for i in range(5):
            success, failed_entries = self.__insert_result_batches(inserts, insertStatement)

            if not success:
                if i < 4:
                    self._log.warning('Some write attempts failed; retrying')
                    inserts = failed_entries
                    sleep(10)
                else:
                    self._log.error('Some write attempts failed; max retries exceeded')
                    raise ResultInsertException('Some result inserts failed')
            else:
                break


    def __insert_result_batches(self, insert_params, insertStatement):
        query_batches = [insert_params[i:i + BATCH_SIZE] for i in range(0, len(insert_params), BATCH_SIZE)]
        move_successful = True

        n_inserts = len(insert_params)
        writing = 0

        failed = []

        self._log.info(f'Inserting {n_inserts} matchup entries in JSON format')

        for batch in query_batches:
            futures = []
            writing += len(batch)
            self._log.info(
                f'Writing batch of {len(batch)} matchup entries | ({writing}/{n_inserts}) [{writing / n_inserts * 100:7.3f}%]')

            for entry in batch:
                futures.append((entry, self._session.execute_async(insertStatement, entry)))

            for entry, future in futures:
                try:
                    future.result()
                except Exception:
                    move_successful = False
                    failed.append(entry)

        self._log.info('Result data write attempt completed')
        return move_successful, failed

    def __prepare_result(self, execution_id, primaryId, result, insertStatement):
        if 'primary' in result:
            data = result['primary']
        elif 'secondary' in result:
            data = result['secondary']
        else:
            data = []

        result_id = uuid.uuid4()

        insert_params = (
            result_id,
            execution_id,
            result["id"],
            primaryId,
            result["lon"],
            result["lat"],
            result["source"],
            result["time"],
            result["platform"] if "platform" in result else None,
            result["device"] if "device" in result else None,
            json.dumps(data, cls=DomsEncoder),
            1 if primaryId == 'PRIMARY' else 0,
            result["depth"],
            result['fileurl']
        )

        params_list = [insert_params]

        if "matches" in result:
            for match in result["matches"]:
                params_list.extend(self.__prepare_result(execution_id, result["id"], match, insertStatement))

        return params_list


class ResultsRetrieval(AbstractResultsContainer):
    def __init__(self, config=None):
        AbstractResultsContainer.__init__(self, config)

    def retrieveResults(self, execution_id, trim_data=False, page_num=1, page_size=1000):
        if isinstance(execution_id, str):
            execution_id = uuid.UUID(execution_id)

        params = self.retrieveParams(execution_id)
        stats = self.retrieveStats(execution_id)
        data = self.__retrieveData(execution_id, trim_data=trim_data, page_num=page_num, page_size=page_size)
        return params, stats, data

    def __retrieveData(self, id, trim_data=False, page_num=1, page_size=1000):
        dataMap = self.__retrievePrimaryData(id, trim_data=trim_data, page_num=page_num, page_size=page_size)
        self.__enrichPrimaryDataWithMatches(id, dataMap, trim_data=trim_data)
        data = [dataMap[name] for name in dataMap]
        return data

    def __enrichPrimaryDataWithMatches(self, id, dataMap, trim_data=False):
        cql = f"SELECT * FROM doms_data where execution_id = {str(id)} and is_primary = false and primary_value_id = ?"
        statement = self._session.prepare(cql)

        primary_ids = list(dataMap.keys())

        logger.info(f'Getting secondary data for {len(primary_ids)} primaries of {str(id)}')

        for (success, rows) in execute_concurrent_with_args(
            self._session, statement, [(i,) for i in primary_ids], concurrency=50, results_generator=True
        ):
            for row in rows:
                entry = self.__rowToDataEntry(row, trim_data=trim_data)
                if row.primary_value_id in dataMap:
                    if not "matches" in dataMap[row.primary_value_id]:
                        dataMap[row.primary_value_id]["matches"] = []
                    dataMap[row.primary_value_id]["matches"].append(entry)

    def __retrievePrimaryData(self, id, trim_data=False, page_num=2, page_size=10):
        cql = "SELECT * FROM doms_data where execution_id = %s and is_primary = true limit %s"
        rows = self._session.execute(cql, [id, page_num * page_size])

        dataMap = {}
        for row in rows[(page_num-1)*page_size:page_num*page_size]:
            entry = self.__rowToDataEntry(row, trim_data=trim_data)
            dataMap[row.value_id] = entry
        return dataMap

    def __rowToDataEntry(self, row, trim_data=False):
        if trim_data:
            entry = {
                "lon": float(row.x),
                "lat": float(row.y),
                "source": row.source_dataset,
                "time": row.measurement_time.replace(tzinfo=UTC)
            }
        else:
            entry = {
                "platform": row.platform,
                "device": row.device,
                "lon": str(row.x),
                "lat": str(row.y),
                "point": f"Point({float(row.x):.3f} {float(row.y):.3f})",
                "time": row.measurement_time.replace(tzinfo=UTC),
                "depth": float(row.depth) if row.depth is not None else None,
                "fileurl": row.file_url if hasattr(row, 'file_url') else None,
                "id": row.value_id,
                "source": row.source_dataset,
            }

        # If doms_data uses the old schema, default to original behavior

        try:
            entry['primary' if row.is_primary else 'secondary'] = json.loads(row.measurement_values_json)
        except AttributeError:
            for key in row.measurement_values:
                value = float(row.measurement_values[key])
                entry[key] = value

        return entry

    def retrieveStats(self, id):
        cql = "SELECT num_gridded_matched, num_insitu_matched, time_to_complete, num_unique_secondaries FROM doms_execution_stats where execution_id = %s limit 1"
        rows = self._session.execute(cql, (id,))
        for row in rows:
            stats = {
                'timeToComplete': row.time_to_complete,
                'numSecondaryMatched': row.num_insitu_matched,
                'numPrimaryMatched': row.num_gridded_matched,
                'numUniqueSecondaries': row.num_unique_secondaries
            }
            return stats

    def retrieveParams(self, id):
        cql = "SELECT * FROM doms_params where execution_id = %s limit 1"
        rows = self._session.execute(cql, (id,))
        for row in rows:
            matchup = row.matchup_datasets.split(",")

            if len(matchup) == 1:
                matchup = matchup[0]

            params = {
                "primary": row.primary_dataset,
                "matchup": matchup,
                "startTime": row.start_time.replace(tzinfo=UTC),
                "endTime": row.end_time.replace(tzinfo=UTC),
                "bbox": row.bounding_box,
                "timeTolerance": int(row.time_tolerance) if row.time_tolerance is not None else None,
                "radiusTolerance": float(row.radius_tolerance) if row.radius_tolerance is not None else None,
                "platforms": row.platforms,
                "parameter": row.parameter,
                "depthMin": float(row.depth_min) if row.depth_min is not None else None,
                "depthMax": float(row.depth_max) if row.depth_max is not None else None,
            }
            return params

        raise NexusProcessingException(reason=f'No params found for id {str(id)}', code=404)

    def retrieveExecution(self, execution_id):
        """
        Retrieve execution details from database.

        :param execution_id: Execution ID
        :return: execution status dictionary
        """

        cql = "SELECT * FROM doms_executions where id = %s limit 1"
        rows = self._session.execute(cql, (execution_id,))
        for row in rows:
            return {
                'status': row.status,
                'message': row.message,
                'timeCompleted': row.time_completed,
                'timeStarted': row.time_started
            }

        raise NexusProcessingException(reason=f'Execution not found with id {str(execution_id)}', code=404)