server/sqlutils.py
author Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
Wed, 03 Dec 2014 16:29:16 +0100
changeset 10125 bc6461a7d2da
parent 10086 98bc2ca1a816
child 10200 cceb2c7c02f4
child 10375 28ec01db78b3
permissions -rw-r--r--
[server] add a db-namespace option in source definition (closes #1631339) "namespace" is preferred to the postgresql term "schema" to try and avoid confusion with cubicweb's schema. cubicweb now depends on logilab-database >= 1.13.0

# copyright 2003-2014 LOGILAB S.A. (Paris, FRANCE), all rights reserved.
# contact http://www.logilab.fr/ -- mailto:contact@logilab.fr
#
# This file is part of CubicWeb.
#
# CubicWeb is free software: you can redistribute it and/or modify it under the
# terms of the GNU Lesser General Public License as published by the Free
# Software Foundation, either version 2.1 of the License, or (at your option)
# any later version.
#
# CubicWeb is distributed in the hope that it will be useful, but WITHOUT
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
# FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License for more
# details.
#
# You should have received a copy of the GNU Lesser General Public License along
# with CubicWeb.  If not, see <http://www.gnu.org/licenses/>.
"""SQL utilities functions and classes."""

__docformat__ = "restructuredtext en"

import sys
import os
import re
import subprocess
from os.path import abspath
from itertools import ifilter
from logging import getLogger

from logilab import database as db, common as lgc
from logilab.common.shellutils import ProgressBar
from logilab.common.deprecation import deprecated
from logilab.common.logging_ext import set_log_methods
from logilab.database.sqlgen import SQLGenerator

from cubicweb import Binary, ConfigurationError
from cubicweb.uilib import remove_html_tags
from cubicweb.schema import PURE_VIRTUAL_RTYPES
from cubicweb.server import SQL_CONNECT_HOOKS
from cubicweb.server.utils import crypt_password

lgc.USE_MX_DATETIME = False
SQL_PREFIX = 'cw_'

def _run_command(cmd):
    print ' '.join(cmd)
    return subprocess.call(cmd)


def sqlexec(sqlstmts, cursor_or_execute, withpb=True,
            pbtitle='', delimiter=';', cnx=None):
    """execute sql statements ignoring DROP/ CREATE GROUP or USER statements
    error.

    :sqlstmts_as_string: a string or a list of sql statements.
    :cursor_or_execute: sql cursor or a callback used to execute statements
    :cnx: if given, commit/rollback at each statement.

    :withpb: if True, display a progresse bar
    :pbtitle: a string displayed as the progress bar title (if `withpb=True`)

    :delimiter: a string used to split sqlstmts (if it is a string)

    Return the failed statements (same type as sqlstmts)
    """
    if hasattr(cursor_or_execute, 'execute'):
        execute = cursor_or_execute.execute
    else:
        execute = cursor_or_execute
    sqlstmts_as_string = False
    if isinstance(sqlstmts, basestring):
        sqlstmts_as_string = True
        sqlstmts = sqlstmts.split(delimiter)
    if withpb:
        pb = ProgressBar(len(sqlstmts), title=pbtitle)
    failed = []
    for sql in sqlstmts:
        sql = sql.strip()
        if withpb:
            pb.update()
        if not sql:
            continue
        try:
            # some dbapi modules doesn't accept unicode for sql string
            execute(str(sql))
        except Exception, err:
            if cnx:
                cnx.rollback()
            failed.append(sql)
        else:
            if cnx:
                cnx.commit()
    if withpb:
        print
    if sqlstmts_as_string:
        failed = delimiter.join(failed)
    return failed


def sqlgrants(schema, driver, user,
              text_index=True, set_owner=True,
              skip_relations=(), skip_entities=()):
    """return sql to give all access privileges to the given user on the system
    schema
    """
    from yams.schema2sql import grant_schema
    from cubicweb.server.sources import native
    output = []
    w = output.append
    w(native.grant_schema(user, set_owner))
    w('')
    if text_index:
        dbhelper = db.get_db_helper(driver)
        w(dbhelper.sql_grant_user_on_fti(user))
        w('')
    w(grant_schema(schema, user, set_owner, skip_entities=skip_entities, prefix=SQL_PREFIX))
    return '\n'.join(output)


def sqlschema(schema, driver, text_index=True,
              user=None, set_owner=False,
              skip_relations=PURE_VIRTUAL_RTYPES, skip_entities=()):
    """return the system sql schema, according to the given parameters"""
    from yams.schema2sql import schema2sql
    from cubicweb.server.sources import native
    if set_owner:
        assert user, 'user is argument required when set_owner is true'
    output = []
    w = output.append
    w(native.sql_schema(driver))
    w('')
    dbhelper = db.get_db_helper(driver)
    if text_index:
        w(dbhelper.sql_init_fti().replace(';', ';;'))
        w('')
    w(schema2sql(dbhelper, schema, prefix=SQL_PREFIX,
                 skip_entities=skip_entities,
                 skip_relations=skip_relations).replace(';', ';;'))
    if dbhelper.users_support and user:
        w('')
        w(sqlgrants(schema, driver, user, text_index, set_owner,
                    skip_relations, skip_entities).replace(';', ';;'))
    return '\n'.join(output)


def sqldropschema(schema, driver, text_index=True,
                  skip_relations=PURE_VIRTUAL_RTYPES, skip_entities=()):
    """return the sql to drop the schema, according to the given parameters"""
    from yams.schema2sql import dropschema2sql
    from cubicweb.server.sources import native
    output = []
    w = output.append
    if text_index:
        dbhelper = db.get_db_helper(driver)
        w(dbhelper.sql_drop_fti())
        w('')
    w(dropschema2sql(dbhelper, schema, prefix=SQL_PREFIX,
                     skip_entities=skip_entities,
                     skip_relations=skip_relations))
    w('')
    w(native.sql_drop_schema(driver))
    return '\n'.join(output)


_SQL_DROP_ALL_USER_TABLES_FILTER_FUNCTION = re.compile('^(?!(sql|pg)_)').match
def sql_drop_all_user_tables(driver_or_helper, sqlcursor):
    """Return ths sql to drop all tables found in the database system."""
    if not getattr(driver_or_helper, 'list_tables', None):
        dbhelper = db.get_db_helper(driver_or_helper)
    else:
        dbhelper = driver_or_helper

    cmds = [dbhelper.sql_drop_sequence('entities_id_seq')]
    # for mssql, we need to drop views before tables
    if hasattr(dbhelper, 'list_views'):
        cmds += ['DROP VIEW %s;' % name
                 for name in ifilter(_SQL_DROP_ALL_USER_TABLES_FILTER_FUNCTION, dbhelper.list_views(sqlcursor))]
    cmds += ['DROP TABLE %s;' % name
             for name in ifilter(_SQL_DROP_ALL_USER_TABLES_FILTER_FUNCTION, dbhelper.list_tables(sqlcursor))]
    return '\n'.join(cmds)


class ConnectionWrapper(object):
    """handle connection to the system source, at some point associated to a
    :class:`Session`
    """

    # since 3.19, we only have to manage the system source connection
    def __init__(self, system_source):
        # dictionary of (source, connection), indexed by sources'uri
        self._source = system_source
        self.cnx = system_source.get_connection()
        self.cu = self.cnx.cursor()

    def commit(self):
        """commit the current transaction for this user"""
        # let exception propagates
        self.cnx.commit()

    def rollback(self):
        """rollback the current transaction for this user"""
        # catch exceptions, rollback other sources anyway
        try:
            self.cnx.rollback()
        except Exception:
            self._source.critical('rollback error', exc_info=sys.exc_info())
            # error on rollback, the connection is much probably in a really
            # bad state. Replace it by a new one.
            self.reconnect()

    def close(self, i_know_what_i_do=False):
        """close all connections in the set"""
        if i_know_what_i_do is not True: # unexpected closing safety belt
            raise RuntimeError('connections set shouldn\'t be closed')
        try:
            self.cu.close()
            self.cu = None
        except Exception:
            pass
        try:
            self.cnx.close()
            self.cnx = None
        except Exception:
            pass

    # internals ###############################################################

    def cnxset_freed(self):
        """connections set is being freed from a session"""
        pass # no nothing by default

    def reconnect(self):
        """reopen a connection for this source or all sources if none specified
        """
        try:
            # properly close existing connection if any
            self.cnx.close()
        except Exception:
            pass
        self._source.info('trying to reconnect')
        self.cnx = self._source.get_connection()
        self.cu = self.cnx.cursor()

    @deprecated('[3.19] use .cu instead')
    def __getitem__(self, uri):
        assert uri == 'system'
        return self.cu

    @deprecated('[3.19] use repo.system_source instead')
    def source(self, uid):
        assert uid == 'system'
        return self._source

    @deprecated('[3.19] use .cnx instead')
    def connection(self, uid):
        assert uid == 'system'
        return self.cnx


class SqliteConnectionWrapper(ConnectionWrapper):
    """Sqlite specific connection wrapper: close the connection each time it's
    freed (and reopen it later when needed)
    """
    def __init__(self, system_source):
        # don't call parent's __init__, we don't want to initiate the connection
        self._source = system_source

    _cnx = None

    def cnxset_freed(self):
        self.cu.close()
        self.cnx.close()
        self.cnx = self.cu = None

    @property
    def cnx(self):
        if self._cnx is None:
            self._cnx = self._source.get_connection()
            self._cu = self._cnx.cursor()
        return self._cnx
    @cnx.setter
    def cnx(self, value):
        self._cnx = value

    @property
    def cu(self):
        if self._cnx is None:
            self._cnx = self._source.get_connection()
            self._cu = self._cnx.cursor()
        return self._cu
    @cu.setter
    def cu(self, value):
        self._cu = value


class SQLAdapterMixIn(object):
    """Mixin for SQL data sources, getting a connection from a configuration
    dictionary and handling connection locking
    """
    cnx_wrap = ConnectionWrapper

    def __init__(self, source_config, repairing=False):
        try:
            self.dbdriver = source_config['db-driver'].lower()
            dbname = source_config['db-name']
        except KeyError:
            raise ConfigurationError('missing some expected entries in sources file')
        dbhost = source_config.get('db-host')
        port = source_config.get('db-port')
        dbport = port and int(port) or None
        dbuser = source_config.get('db-user')
        dbpassword = source_config.get('db-password')
        dbencoding = source_config.get('db-encoding', 'UTF-8')
        dbextraargs = source_config.get('db-extra-arguments')
        dbnamespace = source_config.get('db-namespace')
        self.dbhelper = db.get_db_helper(self.dbdriver)
        self.dbhelper.record_connection_info(dbname, dbhost, dbport, dbuser,
                                             dbpassword, dbextraargs,
                                             dbencoding, dbnamespace)
        self.sqlgen = SQLGenerator()
        # copy back some commonly accessed attributes
        dbapi_module = self.dbhelper.dbapi_module
        self.OperationalError = dbapi_module.OperationalError
        self.InterfaceError = dbapi_module.InterfaceError
        self.DbapiError = dbapi_module.Error
        self._binary = self.dbhelper.binary_value
        self._process_value = dbapi_module.process_value
        self._dbencoding = dbencoding
        if self.dbdriver == 'sqlite':
            self.cnx_wrap = SqliteConnectionWrapper
            self.dbhelper.dbname = abspath(self.dbhelper.dbname)
        if not repairing:
            statement_timeout = int(source_config.get('db-statement-timeout', 0))
            if statement_timeout > 0:
                def set_postgres_timeout(cnx):
                    cnx.cursor().execute('SET statement_timeout to %d' % statement_timeout)
                    cnx.commit()
                postgres_hooks = SQL_CONNECT_HOOKS['postgres']
                postgres_hooks.append(set_postgres_timeout)

    def wrapped_connection(self):
        """open and return a connection to the database, wrapped into a class
        handling reconnection and all
        """
        return self.cnx_wrap(self)

    def get_connection(self):
        """open and return a connection to the database"""
        return self.dbhelper.get_connection()

    def backup_to_file(self, backupfile, confirm):
        for cmd in self.dbhelper.backup_commands(backupfile,
                                                 keepownership=False):
            if _run_command(cmd):
                if not confirm('   [Failed] Continue anyway?', default='n'):
                    raise Exception('Failed command: %s' % cmd)

    def restore_from_file(self, backupfile, confirm, drop=True):
        for cmd in self.dbhelper.restore_commands(backupfile,
                                                  keepownership=False,
                                                  drop=drop):
            if _run_command(cmd):
                if not confirm('   [Failed] Continue anyway?', default='n'):
                    raise Exception('Failed command: %s' % cmd)

    def merge_args(self, args, query_args):
        if args is not None:
            newargs = {}
            for key, val in args.iteritems():
                # convert cubicweb binary into db binary
                if isinstance(val, Binary):
                    val = self._binary(val.getvalue())
                newargs[key] = val
            # should not collide
            newargs.update(query_args)
            return newargs
        return query_args

    def process_result(self, cursor, cnx=None, column_callbacks=None):
        """return a list of CubicWeb compliant values from data in the given cursor
        """
        return list(self.iter_process_result(cursor, cnx, column_callbacks))

    def iter_process_result(self, cursor, cnx, column_callbacks=None):
        """return a iterator on tuples of CubicWeb compliant values from data
        in the given cursor
        """
        # use two different implementations to avoid paying the price of
        # callback lookup for each *cell* in results when there is nothing to
        # lookup
        if not column_callbacks:
            return self.dbhelper.dbapi_module.process_cursor(cursor, self._dbencoding,
                                                             Binary)
        assert cnx
        return self._cb_process_result(cursor, column_callbacks, cnx)

    def _cb_process_result(self, cursor, column_callbacks, cnx):
        # begin bind to locals for optimization
        descr = cursor.description
        encoding = self._dbencoding
        process_value = self._process_value
        binary = Binary
        # /end
        cursor.arraysize = 100
        while True:
            results = cursor.fetchmany()
            if not results:
                break
            for line in results:
                result = []
                for col, value in enumerate(line):
                    if value is None:
                        result.append(value)
                        continue
                    cbstack = column_callbacks.get(col, None)
                    if cbstack is None:
                        value = process_value(value, descr[col], encoding, binary)
                    else:
                        for cb in cbstack:
                            value = cb(self, cnx, value)
                    result.append(value)
                yield result

    def preprocess_entity(self, entity):
        """return a dictionary to use as extra argument to cursor.execute
        to insert/update an entity into a SQL database
        """
        attrs = {}
        eschema = entity.e_schema
        converters = getattr(self.dbhelper, 'TYPE_CONVERTERS', {})
        for attr, value in entity.cw_edited.iteritems():
            if value is not None and eschema.subjrels[attr].final:
                atype = str(entity.e_schema.destination(attr))
                if atype in converters:
                    # It is easier to modify preprocess_entity rather
                    # than add_entity (native) as this behavior
                    # may also be used for update.
                    value = converters[atype](value)
                elif atype == 'Password': # XXX could be done using a TYPE_CONVERTERS callback
                    # if value is a Binary instance, this mean we got it
                    # from a query result and so it is already encrypted
                    if isinstance(value, Binary):
                        value = value.getvalue()
                    else:
                        value = crypt_password(value)
                    value = self._binary(value)
                elif isinstance(value, Binary):
                    value = self._binary(value.getvalue())
            attrs[SQL_PREFIX+str(attr)] = value
        attrs[SQL_PREFIX+'eid'] = entity.eid
        return attrs

    # these are overridden by set_log_methods below
    # only defining here to prevent pylint from complaining
    info = warning = error = critical = exception = debug = lambda msg,*a,**kw: None

set_log_methods(SQLAdapterMixIn, getLogger('cubicweb.sqladapter'))


# connection initialization functions ##########################################

def init_sqlite_connexion(cnx):

    class group_concat(object):
        def __init__(self):
            self.values = set()
        def step(self, value):
            if value is not None:
                self.values.add(value)
        def finalize(self):
            return ', '.join(unicode(v) for v in self.values)

    cnx.create_aggregate("GROUP_CONCAT", 1, group_concat)

    def _limit_size(text, maxsize, format='text/plain'):
        if len(text) < maxsize:
            return text
        if format in ('text/html', 'text/xhtml', 'text/xml'):
            text = remove_html_tags(text)
        if len(text) > maxsize:
            text = text[:maxsize] + '...'
        return text

    def limit_size3(text, format, maxsize):
        return _limit_size(text, maxsize, format)
    cnx.create_function("LIMIT_SIZE", 3, limit_size3)

    def limit_size2(text, maxsize):
        return _limit_size(text, maxsize)
    cnx.create_function("TEXT_LIMIT_SIZE", 2, limit_size2)

    from logilab.common.date import strptime
    def weekday(ustr):
        try:
            dt = strptime(ustr, '%Y-%m-%d %H:%M:%S')
        except:
            dt =  strptime(ustr, '%Y-%m-%d')
        # expect sunday to be 1, saturday 7 while weekday method return 0 for
        # monday
        return (dt.weekday() + 1) % 7
    cnx.create_function("WEEKDAY", 1, weekday)

    import yams.constraints
    yams.constraints.patch_sqlite_decimal()

sqlite_hooks = SQL_CONNECT_HOOKS.setdefault('sqlite', [])
sqlite_hooks.append(init_sqlite_connexion)


def init_postgres_connexion(cnx):
    cnx.cursor().execute('SET TIME ZONE UTC')
    # commit is needed, else setting are lost if the connection is first
    # rolled back
    cnx.commit()

postgres_hooks = SQL_CONNECT_HOOKS.setdefault('postgres', [])
postgres_hooks.append(init_postgres_connexion)