connections.py

import json
import re
from contextlib import contextmanager
from dataclasses import dataclass
from functools import lru_cache
import agate
from requests.exceptions import ConnectionError
from typing import Optional, Any, Dict, Tuple

import google.auth
import google.auth.exceptions
import google.cloud.bigquery
import google.cloud.exceptions
from google.api_core import retry, client_info
from google.auth import impersonated_credentials
from google.oauth2 import (
    credentials as GoogleCredentials,
    service_account as GoogleServiceAccountCredentials
)

from dbt.adapters.bigquery import gcloud
from dbt.clients import agate_helper
from dbt.config.profile import INVALID_PROFILE_MESSAGE
from dbt.tracking import active_user
from dbt.contracts.connection import ConnectionState, AdapterResponse
from dbt.exceptions import (
    FailedToConnectException, RuntimeException, DatabaseException, DbtProfileError
)
from dbt.adapters.base import BaseConnectionManager, Credentials
from dbt.events import AdapterLogger
from dbt.events.functions import fire_event
from dbt.events.types import SQLQuery
from dbt.version import __version__ as dbt_version

from dbt.dataclass_schema import StrEnum

logger = AdapterLogger("BigQuery")

BQ_QUERY_JOB_SPLIT = '-----Query Job SQL Follows-----'

WRITE_TRUNCATE = google.cloud.bigquery.job.WriteDisposition.WRITE_TRUNCATE

REOPENABLE_ERRORS = (
    ConnectionResetError,
    ConnectionError,
)

RETRYABLE_ERRORS = (
    google.cloud.exceptions.ServerError,
    google.cloud.exceptions.BadRequest,
    ConnectionResetError,
    ConnectionError,
)


@lru_cache()
def get_bigquery_defaults(scopes=None) -> Tuple[Any, Optional[str]]:
    """
    Returns (credentials, project_id)

    project_id is returned available from the environment; otherwise None
    """
    # Cached, because the underlying implementation shells out, taking ~1s
    try:
        credentials, _ = google.auth.default(scopes=scopes)
        return credentials, _
    except google.auth.exceptions.DefaultCredentialsError as e:
        raise DbtProfileError(INVALID_PROFILE_MESSAGE.format(error_string=e))


class Priority(StrEnum):
    Interactive = 'interactive'
    Batch = 'batch'


class BigQueryConnectionMethod(StrEnum):
    OAUTH = 'oauth'
    SERVICE_ACCOUNT = 'service-account'
    SERVICE_ACCOUNT_JSON = 'service-account-json'
    OAUTH_SECRETS = 'oauth-secrets'


@dataclass
class BigQueryAdapterResponse(AdapterResponse):
    bytes_processed: Optional[int] = None


@dataclass
class BigQueryCredentials(Credentials):
    method: BigQueryConnectionMethod
    # BigQuery allows an empty database / project, where it defers to the
    # environment for the project
    database: Optional[str]
    execution_project: Optional[str] = None
    location: Optional[str] = None
    priority: Optional[Priority] = None
    maximum_bytes_billed: Optional[int] = None
    impersonate_service_account: Optional[str] = None

    job_retry_deadline_seconds: Optional[int] = None
    job_retries: Optional[int] = 1
    job_creation_timeout_seconds: Optional[int] = None
    job_execution_timeout_seconds: Optional[int] = 300

    # Keyfile json creds
    keyfile: Optional[str] = None
    keyfile_json: Optional[Dict[str, Any]] = None

    # oauth-secrets
    token: Optional[str] = None
    refresh_token: Optional[str] = None
    client_id: Optional[str] = None
    client_secret: Optional[str] = None
    token_uri: Optional[str] = None

    scopes: Optional[Tuple[str, ...]] = (
        'https://www.googleapis.com/auth/bigquery',
        'https://www.googleapis.com/auth/cloud-platform',
        'https://www.googleapis.com/auth/drive'
    )

    _ALIASES = {
        # 'legacy_name': 'current_name'
        'project': 'database',
        'dataset': 'schema',
        'target_project': 'target_database',
        'target_dataset': 'target_schema',
        'retries': 'job_retries',
        'timeout_seconds': 'job_execution_timeout_seconds',
    }

    @property
    def type(self):
        return 'bigquery'

    @property
    def unique_field(self):
        return self.database

    def _connection_keys(self):
        return ('method', 'database', 'schema', 'location', 'priority',
                'timeout_seconds', 'maximum_bytes_billed', 'execution_project',
                'job_retry_deadline_seconds', 'job_retries',
                'job_creation_timeout_seconds', 'job_execution_timeout_seconds')

    @classmethod
    def __pre_deserialize__(cls, d: Dict[Any, Any]) -> Dict[Any, Any]:
        # We need to inject the correct value of the database (aka project) at
        # this stage, ref
        # https://github.com/dbt-labs/dbt/pull/2908#discussion_r532927436.

        # `database` is an alias of `project` in BigQuery
        if 'database' not in d:
            _, database = get_bigquery_defaults()
            d['database'] = database
        # `execution_project` default to dataset/project
        if 'execution_project' not in d:
            d['execution_project'] = d['database']
        return d


class BigQueryConnectionManager(BaseConnectionManager):
    TYPE = 'bigquery'

    QUERY_TIMEOUT = 300
    RETRIES = 1
    DEFAULT_INITIAL_DELAY = 1.0  # Seconds
    DEFAULT_MAXIMUM_DELAY = 1.0  # Seconds

    @classmethod
    def handle_error(cls, error, message):
        error_msg = "\n".join([item['message'] for item in error.errors])
        raise DatabaseException(error_msg)

    def clear_transaction(self):
        pass

    @contextmanager
    def exception_handler(self, sql):
        try:
            yield

        except google.cloud.exceptions.BadRequest as e:
            message = "Bad request while running query"
            self.handle_error(e, message)

        except google.cloud.exceptions.Forbidden as e:
            message = "Access denied while running query"
            self.handle_error(e, message)

        except google.auth.exceptions.RefreshError as e:
            message = "Unable to generate access token, if you're using " \
                      "impersonate_service_account, make sure your " \
                      'initial account has the "roles/' \
                      'iam.serviceAccountTokenCreator" role on the ' \
                      'account you are trying to impersonate.\n\n' \
                      f'{str(e)}'
            raise RuntimeException(message)

        except Exception as e:
            logger.debug("Unhandled error while running:\n{}".format(sql))
            logger.debug(e)
            if isinstance(e, RuntimeException):
                # during a sql query, an internal to dbt exception was raised.
                # this sounds a lot like a signal handler and probably has
                # useful information, so raise it without modification.
                raise
            exc_message = str(e)
            # the google bigquery library likes to add the query log, which we
            # don't want to log. Hopefully they never change this!
            if BQ_QUERY_JOB_SPLIT in exc_message:
                exc_message = exc_message.split(BQ_QUERY_JOB_SPLIT)[0].strip()
            raise RuntimeException(exc_message)

    def cancel_open(self) -> None:
        pass

    @classmethod
    def close(cls, connection):
        connection.state = ConnectionState.CLOSED

        return connection

    def begin(self):
        pass

    def commit(self):
        pass

    def format_bytes(self, num_bytes):
        if num_bytes:
            for unit in ['Bytes', 'KB', 'MB', 'GB', 'TB', 'PB']:
                if abs(num_bytes) < 1024.0:
                    return f"{num_bytes:3.1f} {unit}"
                num_bytes /= 1024.0

            num_bytes *= 1024.0
            return f"{num_bytes:3.1f} {unit}"

        else:
            return num_bytes

    def format_rows_number(self, rows_number):
        for unit in ['', 'k', 'm', 'b', 't']:
            if abs(rows_number) < 1000.0:
                return f"{rows_number:3.1f}{unit}".strip()
            rows_number /= 1000.0

        rows_number *= 1000.0
        return f"{rows_number:3.1f}{unit}".strip()

    @classmethod
    def get_bigquery_credentials(cls, profile_credentials):
        method = profile_credentials.method
        creds = GoogleServiceAccountCredentials.Credentials

        if method == BigQueryConnectionMethod.OAUTH:
            credentials, _ = get_bigquery_defaults(scopes=profile_credentials.scopes)
            return credentials

        elif method == BigQueryConnectionMethod.SERVICE_ACCOUNT:
            keyfile = profile_credentials.keyfile
            return creds.from_service_account_file(keyfile, scopes=profile_credentials.scopes)

        elif method == BigQueryConnectionMethod.SERVICE_ACCOUNT_JSON:
            details = profile_credentials.keyfile_json
            return creds.from_service_account_info(details, scopes=profile_credentials.scopes)

        elif method == BigQueryConnectionMethod.OAUTH_SECRETS:
            return GoogleCredentials.Credentials(
                token=profile_credentials.token,
                refresh_token=profile_credentials.refresh_token,
                client_id=profile_credentials.client_id,
                client_secret=profile_credentials.client_secret,
                token_uri=profile_credentials.token_uri,
                scopes=profile_credentials.scopes
            )

        error = ('Invalid `method` in profile: "{}"'.format(method))
        raise FailedToConnectException(error)

    @classmethod
    def get_impersonated_bigquery_credentials(cls, profile_credentials):
        source_credentials = cls.get_bigquery_credentials(profile_credentials)
        return impersonated_credentials.Credentials(
            source_credentials=source_credentials,
            target_principal=profile_credentials.impersonate_service_account,
            target_scopes=list(profile_credentials.scopes),
            lifetime=profile_credentials.job_execution_timeout_seconds,
        )

    @classmethod
    def get_bigquery_client(cls, profile_credentials):
        if profile_credentials.impersonate_service_account:
            creds =\
                cls.get_impersonated_bigquery_credentials(profile_credentials)
        else:
            creds = cls.get_bigquery_credentials(profile_credentials)
        execution_project = profile_credentials.execution_project
        location = getattr(profile_credentials, 'location', None)

        info = client_info.ClientInfo(user_agent=f'dbt-{dbt_version}')
        return google.cloud.bigquery.Client(
            execution_project,
            creds,
            location=location,
            client_info=info,
        )

    @classmethod
    def open(cls, connection):
        if connection.state == 'open':
            logger.debug('Connection is already open, skipping open.')
            return connection

        try:
            handle = cls.get_bigquery_client(connection.credentials)

        except google.auth.exceptions.DefaultCredentialsError:
            logger.info("Please log into GCP to continue")
            gcloud.setup_default_credentials()

            handle = cls.get_bigquery_client(connection.credentials)

        except Exception as e:
            logger.debug("Got an error when attempting to create a bigquery "
                         "client: '{}'".format(e))

            connection.handle = None
            connection.state = 'fail'

            raise FailedToConnectException(str(e))

        connection.handle = handle
        connection.state = 'open'
        return connection

    @classmethod
    def get_job_execution_timeout_seconds(cls, conn):
        credentials = conn.credentials
        return credentials.job_execution_timeout_seconds

    @classmethod
    def get_job_retries(cls, conn) -> int:
        credentials = conn.credentials
        return credentials.job_retries

    @classmethod
    def get_job_creation_timeout_seconds(cls, conn):
        credentials = conn.credentials
        return credentials.job_creation_timeout_seconds

    @classmethod
    def get_job_retry_deadline_seconds(cls, conn):
        credentials = conn.credentials
        return credentials.job_retry_deadline_seconds

    @classmethod
    def get_table_from_response(cls, resp):
        column_names = [field.name for field in resp.schema]
        return agate_helper.table_from_data_flat(resp, column_names)

    def raw_execute(self, sql, fetch=False, *, use_legacy_sql=False):
        conn = self.get_thread_connection()
        client = conn.handle

        fire_event(SQLQuery(conn_name=conn.name, sql=sql))

        if self.profile.query_comment and self.profile.query_comment.job_label:
            query_comment = self.query_header.comment.query_comment
            labels = self._labels_from_query_comment(query_comment)
        else:
            labels = {}

        if active_user:
            labels['dbt_invocation_id'] = active_user.invocation_id

        job_params = {'use_legacy_sql': use_legacy_sql, 'labels': labels}

        priority = conn.credentials.priority
        if priority == Priority.Batch:
            job_params['priority'] = google.cloud.bigquery.QueryPriority.BATCH
        else:
            job_params[
                'priority'] = google.cloud.bigquery.QueryPriority.INTERACTIVE

        maximum_bytes_billed = conn.credentials.maximum_bytes_billed
        if maximum_bytes_billed is not None and maximum_bytes_billed != 0:
            job_params['maximum_bytes_billed'] = maximum_bytes_billed

        job_creation_timeout = self.get_job_creation_timeout_seconds(conn)
        job_execution_timeout = self.get_job_execution_timeout_seconds(conn)

        def fn():
            return self._query_and_results(
                client, sql, job_params,
                job_creation_timeout=job_creation_timeout,
                job_execution_timeout=job_execution_timeout
            )

        query_job, iterator = self._retry_and_handle(msg=sql, conn=conn, fn=fn)

        return query_job, iterator

    def execute(
        self, sql, auto_begin=False, fetch=None
    ) -> Tuple[BigQueryAdapterResponse, agate.Table]:
        sql = self._add_query_comment(sql)
        # auto_begin is ignored on bigquery, and only included for consistency
        query_job, iterator = self.raw_execute(sql, fetch=fetch)

        if fetch:
            table = self.get_table_from_response(iterator)
        else:
            table = agate_helper.empty_table()

        message = 'OK'
        code = None
        num_rows = None
        bytes_processed = None

        if query_job.statement_type == 'CREATE_VIEW':
            code = 'CREATE VIEW'

        elif query_job.statement_type == 'CREATE_TABLE_AS_SELECT':
            conn = self.get_thread_connection()
            client = conn.handle
            query_table = client.get_table(query_job.destination)
            code = 'CREATE TABLE'
            num_rows = query_table.num_rows
            num_rows_formated = self.format_rows_number(num_rows)
            bytes_processed = query_job.total_bytes_processed
            processed_bytes = self.format_bytes(bytes_processed)
            message = f'{code} ({num_rows_formated} rows, {processed_bytes} processed)'

        elif query_job.statement_type == 'SCRIPT':
            code = 'SCRIPT'
            bytes_processed = query_job.total_bytes_processed
            message = f'{code} ({self.format_bytes(bytes_processed)} processed)'

        elif query_job.statement_type in ['INSERT', 'DELETE', 'MERGE', 'UPDATE']:
            code = query_job.statement_type
            num_rows = query_job.num_dml_affected_rows
            num_rows_formated = self.format_rows_number(num_rows)
            bytes_processed = query_job.total_bytes_processed
            processed_bytes = self.format_bytes(bytes_processed)
            message = f'{code} ({num_rows_formated} rows, {processed_bytes} processed)'

        elif query_job.statement_type == 'SELECT':
            conn = self.get_thread_connection()
            client = conn.handle
            # use anonymous table for num_rows
            query_table = client.get_table(query_job.destination)
            code = 'SELECT'
            num_rows = query_table.num_rows
            num_rows_formated = self.format_rows_number(num_rows)
            bytes_processed = query_job.total_bytes_processed
            processed_bytes = self.format_bytes(bytes_processed)
            message = f'{code} ({num_rows_formated} rows, {processed_bytes} processed)'

        response = BigQueryAdapterResponse(
            _message=message,
            rows_affected=num_rows,
            code=code,
            bytes_processed=bytes_processed
        )

        return response, table

    def get_partitions_metadata(self, table):
        def standard_to_legacy(table):
            return table.project + ':' + table.dataset + '.' + table.identifier

        legacy_sql = 'SELECT * FROM ['\
            + standard_to_legacy(table) + '$__PARTITIONS_SUMMARY__]'

        sql = self._add_query_comment(legacy_sql)
        # auto_begin is ignored on bigquery, and only included for consistency
        _, iterator =\
            self.raw_execute(sql, fetch='fetch_result', use_legacy_sql=True)
        return self.get_table_from_response(iterator)

    def copy_bq_table(self, source, destination, write_disposition):
        conn = self.get_thread_connection()
        client = conn.handle

# -------------------------------------------------------------------------------
#  BigQuery allows to use copy API using two different formats:
#  1. client.copy_table(source_table_id, destination_table_id)
#     where source_table_id = "your-project.source_dataset.source_table"
#  2. client.copy_table(source_table_ids, destination_table_id)
#     where source_table_ids = ["your-project.your_dataset.your_table_name", ...]
#  Let's use uniform function call and always pass list there
# -------------------------------------------------------------------------------
        if type(source) is not list:
            source = [source]

        source_ref_array = [
            self.table_ref(src_table.database, src_table.schema, src_table.table)
            for src_table in source
        ]
        destination_ref = self.table_ref(
            destination.database, destination.schema, destination.table
        )

        logger.debug(
            'Copying table(s) "{}" to "{}" with disposition: "{}"',
            ', '.join(source_ref.path for source_ref in source_ref_array),
            destination_ref.path, write_disposition)

        def copy_and_results():
            job_config = google.cloud.bigquery.CopyJobConfig(
                write_disposition=write_disposition)
            copy_job = client.copy_table(
                source_ref_array, destination_ref, job_config=job_config)
            iterator = copy_job.result(timeout=self.get_job_execution_timeout_seconds(conn))
            return copy_job, iterator

        self._retry_and_handle(
            msg='copy table "{}" to "{}"'.format(
                ', '.join(source_ref.path for source_ref in source_ref_array),
                destination_ref.path),
            conn=conn, fn=copy_and_results)

    @staticmethod
    def dataset_ref(database, schema):
        return google.cloud.bigquery.DatasetReference(project=database, dataset_id=schema)

    @staticmethod
    def table_ref(database, schema, table_name):
        dataset_ref = google.cloud.bigquery.DatasetReference(database, schema)
        return google.cloud.bigquery.TableReference(dataset_ref, table_name)

    def get_bq_table(self, database, schema, identifier):
        """Get a bigquery table for a schema/model."""
        conn = self.get_thread_connection()
        table_ref = self.table_ref(database, schema, identifier)
        return conn.handle.get_table(table_ref)

    def drop_dataset(self, database, schema):
        conn = self.get_thread_connection()
        dataset_ref = self.dataset_ref(database, schema)
        client = conn.handle

        def fn():
            return client.delete_dataset(dataset_ref, delete_contents=True, not_found_ok=True)

        self._retry_and_handle(
            msg='drop dataset', conn=conn, fn=fn)

    def create_dataset(self, database, schema):
        conn = self.get_thread_connection()
        client = conn.handle
        dataset_ref = self.dataset_ref(database, schema)

        def fn():
            return client.create_dataset(dataset_ref, exists_ok=True)
        self._retry_and_handle(msg='create dataset', conn=conn, fn=fn)

    def _query_and_results(
        self, client, sql, job_params,
        job_creation_timeout=None,
        job_execution_timeout=None
    ):
        """Query the client and wait for results."""
        # Cannot reuse job_config if destination is set and ddl is used
        job_config = google.cloud.bigquery.QueryJobConfig(**job_params)
        query_job = client.query(
            query=sql,
            job_config=job_config,
            timeout=job_creation_timeout
        )
        iterator = query_job.result(timeout=job_execution_timeout)

        return query_job, iterator

    def _retry_and_handle(self, msg, conn, fn):
        """retry a function call within the context of exception_handler."""
        def reopen_conn_on_error(error):
            if isinstance(error, REOPENABLE_ERRORS):
                logger.warning('Reopening connection after {!r}'.format(error))
                self.close(conn)
                self.open(conn)
                return

        with self.exception_handler(msg):
            return retry.retry_target(
                target=fn,
                predicate=_ErrorCounter(self.get_job_retries(conn)).count_error,
                sleep_generator=self._retry_generator(),
                deadline=self.get_job_retry_deadline_seconds(conn),
                on_error=reopen_conn_on_error)

    def _retry_generator(self):
        """Generates retry intervals that exponentially back off."""
        return retry.exponential_sleep_generator(
            initial=self.DEFAULT_INITIAL_DELAY,
            maximum=self.DEFAULT_MAXIMUM_DELAY)

    def _labels_from_query_comment(self, comment: str) -> Dict:
        try:
            comment_labels = json.loads(comment)
        except (TypeError, ValueError):
            return {'query_comment': _sanitize_label(comment)}
        return {
            _sanitize_label(key): _sanitize_label(str(value))
            for key, value in comment_labels.items()
        }


class _ErrorCounter(object):
    """Counts errors seen up to a threshold then raises the next error."""

    def __init__(self, retries):
        self.retries = retries
        self.error_count = 0

    def count_error(self, error):
        if self.retries == 0:
            return False  # Don't log
        self.error_count += 1
        if _is_retryable(error) and self.error_count <= self.retries:
            logger.debug(
                'Retry attempt {} of {} after error: {}'.format(
                    self.error_count, self.retries, repr(error)
                ))
            return True
        else:
            return False


def _is_retryable(error):
    """Return true for errors that are unlikely to occur again if retried."""
    if isinstance(error, RETRYABLE_ERRORS):
        return True
    elif isinstance(error, google.api_core.exceptions.Forbidden) and any(
            e['reason'] == 'rateLimitExceeded' for e in error.errors):
        return True
    return False


_SANITIZE_LABEL_PATTERN = re.compile(r"[^a-z0-9_-]")

_VALIDATE_LABEL_LENGTH_LIMIT = 63


def _sanitize_label(value: str) -> str:
    """Return a legal value for a BigQuery label."""
    value = value.strip().lower()
    value = _SANITIZE_LABEL_PATTERN.sub("_", value)
    value_length = len(value)
    if value_length > _VALIDATE_LABEL_LENGTH_LIMIT:
        error_msg = (
            f"Job label length {value_length} is greater than length limit: "
            f"{_VALIDATE_LABEL_LENGTH_LIMIT}\n"
            f"Current sanitized label: {value}"
        )
        raise RuntimeException(error_msg)
    else:
        return value