Mercurial > p > roundup > code
view roundup/backends/back_sqlite.py @ 6804:25d08e15e3b4
issue2551224 - Replace dbm db for sessions/otks when using sqlite
Generate new sqlite db's for storing one time keys and session and
other ephemeral data.
| author | John Rouillard <rouilj@ieee.org> |
|---|---|
| date | Mon, 25 Jul 2022 15:30:36 -0400 |
| parents | 39189dd94f2c |
| children | 3f60a71b0812 |
line wrap: on
line source
"""Implements a backend for SQLite. See https://pysqlite.sourceforge.net/ for pysqlite info NOTE: we use the rdbms_common table creation methods which define datatypes for the columns, but sqlite IGNORES these specifications. """ __docformat__ = 'restructuredtext' import os, marshal, shutil, time, logging from roundup import hyperdb, date, password from roundup.backends import rdbms_common from roundup.backends.sessions_sqlite import Sessions, OneTimeKeys from roundup.anypy.strings import uany2s sqlite_version = None try: import sqlite3 as sqlite sqlite_version = 3 except ImportError: try: from pysqlite2 import dbapi2 as sqlite if sqlite.version_info < (2, 1, 0): raise ValueError('pysqlite2 minimum version is 2.1.0+ ' '- %s found' % sqlite.version) sqlite_version = 2 except ImportError: import sqlite sqlite_version = 1 def db_exists(config): return os.path.exists(os.path.join(config.DATABASE, 'db')) def db_nuke(config): shutil.rmtree(config.DATABASE) class Database(rdbms_common.Database): """Sqlite DB backend implementation attributes: dbtype: holds the value for the type of db. It is used by indexer to identify the database type so it can import the correct indexer module when using native text search mode. """ # char to use for positional arguments if sqlite_version in (2, 3): arg = '?' else: arg = '%s' dbtype = "sqlite" # used by some code to switch styles of query implements_intersect = 1 # used in generic backend to determine if db supports # 'DOUBLE PRECISION' for floating point numbers. Note that sqlite # already has double precision as its standard 'REAL' type. So this # is set to False here. implements_double_precision = False hyperdb_to_sql_datatypes = { hyperdb.String : 'VARCHAR(255)', hyperdb.Date : 'VARCHAR(30)', hyperdb.Link : 'INTEGER', hyperdb.Interval : 'VARCHAR(255)', hyperdb.Password : 'VARCHAR(255)', hyperdb.Boolean : 'BOOLEAN', hyperdb.Number : 'REAL', hyperdb.Integer : 'INTEGER', } hyperdb_to_sql_value = { hyperdb.String : str, hyperdb.Date : lambda x: x.serialise(), hyperdb.Link : int, hyperdb.Interval : str, hyperdb.Password : str, hyperdb.Boolean : int, hyperdb.Integer : int, hyperdb.Number : lambda x: x, hyperdb.Multilink : lambda x: x, # used in journal marshalling } sql_to_hyperdb_value = { hyperdb.String : uany2s, hyperdb.Date : lambda x: date.Date(str(x)), hyperdb.Link : str, # XXX numeric ids hyperdb.Interval : date.Interval, hyperdb.Password : lambda x: password.Password(encrypted=x), hyperdb.Boolean : int, hyperdb.Integer : int, hyperdb.Number : rdbms_common._num_cvt, hyperdb.Multilink : lambda x: x, # used in journal marshalling } # We're using DBM for managing session info and one-time keys: # For SQL database storage of this info we would need two concurrent # connections to the same database which SQLite doesn't support def getSessionManager(self): if not self.Session: self.Session = Sessions(self) return self.Session def getOTKManager(self): if not self.Otk: self.Otk = OneTimeKeys(self) return self.Otk def sqlite_busy_handler(self, data, table, count): """invoked whenever SQLite tries to access a database that is locked""" now = time.time() if count == 1: # Timeout for handling locked database (default 30s) self._busy_handler_endtime = now + self.config.RDBMS_SQLITE_TIMEOUT elif now > self._busy_handler_endtime: # timeout expired - no more retries return 0 # sleep adaptively as retry count grows, # starting from about half a second time_to_sleep = 0.01 * (2 << min(5, count)) time.sleep(time_to_sleep) return 1 def sql_open_connection(self, dbname=None): """Open a standard, non-autocommitting connection. pysqlite will automatically BEGIN TRANSACTION for us. """ # make sure the database directory exists # database itself will be created by sqlite if needed if not os.path.isdir(self.config.DATABASE): os.makedirs(self.config.DATABASE) if dbname: db = os.path.join(self.config.DATABASE, 'db-' + dbname) else: db = os.path.join(self.config.DATABASE, 'db') logging.getLogger('roundup.hyperdb').info('open database %r' % db) # set timeout (30 second default is extraordinarily generous) # for handling locked database if sqlite_version == 1: conn = sqlite.connect(db=db) conn.db.sqlite_busy_handler(self.sqlite_busy_handler) else: conn = sqlite.connect(db, timeout=self.config.RDBMS_SQLITE_TIMEOUT) conn.row_factory = sqlite.Row # pysqlite2 / sqlite3 want us to store Unicode in the db but # that's not what's been done historically and it's definitely # not what the other backends do, so we'll stick with UTF-8 if sqlite_version in (2, 3): conn.text_factory = str cursor = conn.cursor() return (conn, cursor) def open_connection(self): # ensure files are group readable and writable os.umask(self.config.UMASK) (self.conn, self.cursor) = self.sql_open_connection() try: self.load_dbschema() except sqlite.DatabaseError as error: if str(error) != 'no such table: schema': raise self.init_dbschema() self.sql('create table schema (schema varchar)') self.sql('create table ids (name varchar, num integer)') self.sql('create index ids_name_idx on ids(name)') self.create_version_2_tables() self._add_fts5_table() def create_version_2_tables(self): self.sql('create table otks (otk_key varchar, ' 'otk_value varchar, otk_time integer)') self.sql('create index otks_key_idx on otks(otk_key)') self.sql('create table sessions (session_key varchar, ' 'session_time integer, session_value varchar)') self.sql('create index sessions_key_idx on ' 'sessions(session_key)') # full-text indexing store self.sql('CREATE TABLE __textids (_class varchar, ' '_itemid varchar, _prop varchar, _textid' ' integer primary key) ') self.sql('CREATE TABLE __words (_word varchar, ' '_textid integer)') self.sql('CREATE INDEX words_word_ids ON __words(_word)') self.sql('CREATE INDEX words_by_id ON __words (_textid)') self.sql('CREATE UNIQUE INDEX __textids_by_props ON ' '__textids (_class, _itemid, _prop)') sql = 'insert into ids (name, num) values (%s,%s)' % ( self.arg, self.arg) self.sql(sql, ('__textids', 1)) def add_new_columns_v2(self): # update existing tables to have the new actor column tables = self.database_schema['tables'] for classname, spec in self.classes.items(): if classname in tables: dbspec = tables[classname] self.update_class(spec, dbspec, force=1, adding_v2=1) # we've updated - don't try again tables[classname] = spec.schema() def fix_version_3_tables(self): # NOOP - no restriction on column length here pass def _add_fts5_table(self): self.sql('CREATE virtual TABLE __fts USING fts5(_class, ' '_itemid, _prop, _textblob)' ) def fix_version_6_tables(self): # note sqlite has no limit on column size so v6 fixes # to __words._word length are not needed. # Add native full-text indexing table self._add_fts5_table() def update_class(self, spec, old_spec, force=0, adding_v2=0): """ Determine the differences between the current spec and the database version of the spec, and update where necessary. If 'force' is true, update the database anyway. SQLite doesn't have ALTER TABLE, so we have to copy and regenerate the tables with the new schema. """ new_spec = spec.schema() new_spec[1].sort() old_spec[1].sort() if not force and new_spec == old_spec: # no changes return 0 logging.getLogger('roundup.hyperdb').info( 'update_class %s' % spec.classname) # detect multilinks that have been removed, and drop their table old_has = {} for name, prop in old_spec[1]: old_has[name] = 1 if name in spec.properties or not \ isinstance(prop, hyperdb.Multilink): continue # it's a multilink, and it's been removed - drop the old # table. First drop indexes. self.drop_multilink_table_indexes(spec.classname, name) sql = 'drop table %s_%s' % (spec.classname, prop) self.sql(sql) # now figure how we populate the new table if adding_v2: fetch = ['_activity', '_creation', '_creator'] else: fetch = ['_actor', '_activity', '_creation', '_creator'] properties = spec.getprops() for propname, _x in new_spec[1]: prop = properties[propname] if isinstance(prop, hyperdb.Multilink): if propname not in old_has: # we need to create the new table self.create_multilink_table(spec, propname) elif force: tn = '%s_%s' % (spec.classname, propname) # grabe the current values sql = 'select linkid, nodeid from %s' % tn self.sql(sql) rows = self.cursor.fetchall() # drop the old table self.drop_multilink_table_indexes(spec.classname, propname) sql = 'drop table %s' % tn self.sql(sql) # re-create and populate the new table self.create_multilink_table(spec, propname) sql = """insert into %s (linkid, nodeid) values (%s, %s)""" % (tn, self.arg, self.arg) for linkid, nodeid in rows: self.sql(sql, (int(linkid), int(nodeid))) elif propname in old_has: # we copy this col over from the old table fetch.append('_'+propname) # select the data out of the old table fetch.append('id') fetch.append('__retired__') fetchcols = ','.join(fetch) cn = spec.classname sql = 'select %s from _%s' % (fetchcols, cn) self.sql(sql) olddata = self.cursor.fetchall() # TODO: update all the other index dropping code self.drop_class_table_indexes(cn, old_spec[0]) # drop the old table self.sql('drop table _%s' % cn) # create the new table self.create_class_table(spec) if olddata: inscols = ['id', '_actor', '_activity', '_creation', '_creator', '__retired__'] for propname, _x in new_spec[1]: prop = properties[propname] if isinstance(prop, hyperdb.Multilink): continue elif isinstance(prop, hyperdb.Interval): inscols.append('_'+propname) inscols.append('__'+propname+'_int__') elif propname in old_has: # we copy this col over from the old table inscols.append('_'+propname) # do the insert of the old data - the new columns will have # NULL values args = ','.join([self.arg for x in inscols]) cols = ','.join(inscols) sql = 'insert into _%s (%s) values (%s)' % (cn, cols, args) for entry in olddata: d = [] retired_id = None for name in inscols: # generate the new value for the Interval int column if name.endswith('_int__'): name = name[2:-6] if sqlite_version in (2, 3): try: v = hyperdb.Interval(entry[name]).as_seconds() except IndexError: v = None elif name in entry: v = hyperdb.Interval(entry[name]).as_seconds() else: v = None elif sqlite_version in (2, 3): try: v = entry[name] except IndexError: v = None elif (sqlite_version == 1 and name in entry): v = entry[name] else: v = None if name == 'id': retired_id = v elif name == '__retired__' and retired_id and \ v not in ['0', 0]: v = retired_id d.append(v) self.sql(sql, tuple(d)) return 1 def sql_close(self): """ Squash any error caused by us already having closed the connection. """ try: self.conn.close() except sqlite.ProgrammingError as value: if str(value) != 'close failed - Connection is closed.': raise def sql_rollback(self): """ Squash any error caused by us having closed the connection (and therefore not having anything to roll back) """ try: self.conn.rollback() except sqlite.ProgrammingError as value: if str(value) != 'rollback failed - Connection is closed.': raise def __repr__(self): return '<roundlite 0x%x>' % id(self) def sql_commit(self): """ Actually commit to the database. Ignore errors if there's nothing to commit. """ try: self.conn.commit() except sqlite.DatabaseError as error: if str(error) != 'cannot commit - no transaction is active': raise # open a new cursor for subsequent work self.cursor = self.conn.cursor() def sql_index_exists(self, table_name, index_name): self.sql('pragma index_list(%s)' % table_name) for entry in self.cursor.fetchall(): if entry[1] == index_name: return 1 return 0 # old-skool id generation def newid(self, classname): """ Generate a new id for the given class """ # Prevent other processes from reading while we increment. # Otherwise multiple processes can end up with the same # new id and hilarity results. # # Defeat pysqlite's attempts to do locking by setting # isolation_level to None. Pysqlite can commit # on it's own even if we don't want it to end the transaction. # If we rewrite to use another sqlite library like apsw we # don't have to deal with this autocommit/autotransact foolishness. self.conn.isolation_level = None # Manage the transaction locks manually. self.sql("BEGIN IMMEDIATE") # get the next ID sql = 'select num from ids where name=%s' % self.arg self.sql(sql, (classname, )) newid = int(self.cursor.fetchone()[0]) # leave the next larger number as the next newid sql = 'update ids set num=num+1 where name=%s' % self.arg vals = (classname,) self.sql(sql, vals) # reset pysqlite's auto transact stuff to default since the # rest of the code expects it. self.conn.isolation_level = '' # commit writing the data, clearing locks for other processes # and create a new cursor to the database. self.sql_commit() # return as string return str(newid) def setid(self, classname, setid): """ Set the id counter: used during import of database We add one to make it behave like the sequences in postgres. """ sql = 'update ids set num=%s where name=%s' % (self.arg, self.arg) vals = (int(setid)+1, classname) self.sql(sql, vals) def clear(self): rdbms_common.Database.clear(self) # set the id counters to 0 (setid adds one) so we start at 1 for cn in self.classes.keys(): self.setid(cn, 0) def create_class(self, spec): rdbms_common.Database.create_class(self, spec) sql = 'insert into ids (name, num) values (%s, %s)' %( self.arg, self.arg) vals = (spec.classname, 1) self.sql(sql, vals) if sqlite_version in (2, 3): def load_journal(self, classname, cols, nodeid): """We need to turn the sqlite3.Row into a tuple so it can be unpacked""" l = rdbms_common.Database.load_journal(self, classname, cols, nodeid) cols = range(5) return [[row[col] for col in cols] for row in l] class sqliteClass: def filter(self, *args, **kw): """ If there's NO matches to a fetch, sqlite returns NULL instead of nothing """ return [f for f in rdbms_common.Class.filter(self, *args, **kw) if f] class Class(sqliteClass, rdbms_common.Class): pass class IssueClass(sqliteClass, rdbms_common.IssueClass): pass class FileClass(sqliteClass, rdbms_common.FileClass): pass # vim: set et sts=4 sw=4 :
