def __init__(self, table_name, args, dest_table):
# compat for dest-table
dest_table = args.get('table', dest_table)
BaseHandler.__init__(self, table_name, args, dest_table)
# show args
self.log.debug("dispatch.init: table_name=%r, args=%r" % \
(table_name, args))
self.batch_info = None
self.dst_curs = None
self.pkeys = None
# config
self.conf = self.get_config()
hdlr_cls = ROW_HANDLERS[self.conf.row_mode]
self.row_handler = hdlr_cls(self.log)
if self.conf.encoding:
self.encoding_validator = EncodingValidator(
self.log, self.conf.encoding)
else:
self.encoding_validator = None
def __init__(self, table_name, args, dest_table):
# compat for dest-table
dest_table = args.get('table', dest_table)
BaseHandler.__init__(self, table_name, args, dest_table)
# show args
self.log.debug("dispatch.init: table_name=%r, args=%r", table_name, args)
self.batch_info = None
self.dst_curs = None
self.pkeys = None
# config
self.conf = self.get_config()
hdlr_cls = ROW_HANDLERS[self.conf.row_mode]
self.row_handler = hdlr_cls(self.log)
if self.conf.encoding:
self.encoding_validator = EncodingValidator(self.log,
self.conf.encoding)
else:
self.encoding_validator = None
class Dispatcher(BaseHandler):
"""Partitioned loader.
Splits events into partitions, if requested.
Then applies them without further processing.
"""
handler_name = 'dispatch'
def __init__(self, table_name, args, dest_table):
# compat for dest-table
dest_table = args.get('table', dest_table)
BaseHandler.__init__(self, table_name, args, dest_table)
# show args
self.log.debug("dispatch.init: table_name=%r, args=%r" % \
(table_name, args))
self.batch_info = None
self.dst_curs = None
self.pkeys = None
# config
self.conf = self.get_config()
hdlr_cls = ROW_HANDLERS[self.conf.row_mode]
self.row_handler = hdlr_cls(self.log)
if self.conf.encoding:
self.encoding_validator = EncodingValidator(
self.log, self.conf.encoding)
else:
self.encoding_validator = None
def get_config(self):
"""Processes args dict"""
conf = skytools.dbdict()
# set table mode
conf.table_mode = self.get_arg('table_mode', TABLE_MODES)
conf.analyze = self.get_arg('analyze', [0, 1])
if conf.table_mode == 'part':
conf.part_mode = self.get_arg('part_mode', PART_MODES)
conf.part_field = self.args.get('part_field')
if conf.part_mode == 'date_field' and not conf.part_field:
raise Exception('part_mode date_field requires part_field!')
conf.period = self.get_arg('period', PERIODS)
conf.part_name = self.args.get('part_name')
conf.part_template = self.args.get('part_template')
conf.pre_part = self.args.get('pre_part')
conf.post_part = self.args.get('post_part')
conf.part_func = self.args.get('part_func', PART_FUNC_NEW)
# set row mode and event types to process
conf.row_mode = self.get_arg('row_mode', ROW_MODES)
event_types = self.args.get('event_types', '*')
if event_types == '*':
event_types = EVENT_TYPES
else:
event_types = [evt.upper() for evt in event_types.split(',')]
for evt in event_types:
if evt not in EVENT_TYPES:
raise Exception('Unsupported operation: %s' % evt)
conf.event_types = event_types
# set load handler
conf.load_mode = self.get_arg('load_mode', LOAD_MODES)
conf.method = self.get_arg('method', METHODS)
# fields to skip
conf.skip_fields = [
f.strip().lower()
for f in self.args.get('skip_fields', '').split(',')
]
# get fields map (obsolete, for compatibility reasons)
fields = self.args.get('fields', '*')
if fields == "*":
conf.field_map = None
else:
conf.field_map = {}
for fval in fields.split(','):
tmp = fval.split(':')
if len(tmp) == 1:
conf.field_map[tmp[0]] = tmp[0]
else:
conf.field_map[tmp[0]] = tmp[1]
# encoding validator
conf.encoding = self.args.get('encoding')
return conf
def get_arg(self, name, value_list, default=None):
default = default or value_list[0]
val = type(default)(self.args.get(name, default))
if val not in value_list:
raise Exception('Bad argument %s value %r' % (name, val))
return val
def reset(self):
"""Called before starting to process a batch.
Should clean any pending data."""
BaseHandler.reset(self)
def prepare_batch(self, batch_info, dst_curs):
"""Called on first event for this table in current batch."""
if self.conf.table_mode != 'ignore':
self.batch_info = batch_info
self.dst_curs = dst_curs
#BaseHandler.prepare_batch(self, batch_info, dst_curs)
def filter_data(self, data):
"""Process with fields skip and map"""
fskip = self.conf.skip_fields
fmap = self.conf.field_map
if fskip:
data = dict((k, v) for k, v in data.items() if k not in fskip)
if fmap:
# when field name not present in source is used then None (NULL)
# value is inserted. is it ok?
data = dict((v, data.get(k)) for k, v in fmap.items())
return data
def filter_pkeys(self, pkeys):
"""Process with fields skip and map"""
fskip = self.conf.skip_fields
fmap = self.conf.field_map
if fskip:
pkeys = [f for f in pkeys if f not in fskip]
if fmap:
pkeys = [fmap[p] for p in pkeys if p in fmap]
return pkeys
def process_event(self, ev, sql_queue_func, arg):
"""Process a event.
Event should be added to sql_queue or executed directly.
"""
if self.conf.table_mode == 'ignore':
return
# get data
data = skytools.db_urldecode(ev.data)
if self.encoding_validator:
data = self.encoding_validator.validate_dict(data, self.table_name)
if len(ev.ev_type) < 2 or ev.ev_type[1] != ':':
raise Exception('Unsupported event type: %s/extra1=%s/data=%s' %
(ev.ev_type, ev.ev_extra1, ev.ev_data))
op, pkeys = ev.type.split(':', 1)
if op not in 'IUD':
raise Exception('Unknown event type: %s' % ev.ev_type)
# process only operations specified
if not op in self.conf.event_types:
return
self.log.debug('dispatch.process_event: %s/%s' %
(ev.ev_type, ev.ev_data))
if self.pkeys is None:
self.pkeys = self.filter_pkeys(pkeys.split(','))
data = self.filter_data(data)
# prepare split table when needed
if self.conf.table_mode == 'part':
dst, part_time = self.split_format(ev, data)
if dst not in self.row_handler.table_map:
self.check_part(dst, part_time)
else:
dst = self.dest_table
if dst not in self.row_handler.table_map:
self.row_handler.add_table(dst, LOADERS[self.conf.load_mode],
self.pkeys, self.conf)
self.row_handler.process(dst, op, data)
#BaseHandler.process_event(self, ev, sql_queue_func, arg)
def finish_batch(self, batch_info, dst_curs):
"""Called when batch finishes."""
if self.conf.table_mode != 'ignore':
self.row_handler.flush(dst_curs)
#BaseHandler.finish_batch(self, batch_info, dst_curs)
def get_part_name(self):
# if custom part name template given, use it
if self.conf.part_name:
return self.conf.part_name
parts = ['year', 'month', 'day', 'hour']
name_parts = ['parent'] + parts[:parts.index(self.conf.period) + 1]
return '_'.join('%%(%s)s' % part for part in name_parts)
def split_format(self, ev, data):
"""Generates part table name from template"""
if self.conf.part_mode == 'batch_time':
dtm = self.batch_info['batch_end']
elif self.conf.part_mode == 'event_time':
dtm = ev.ev_time
elif self.conf.part_mode == 'current_time':
dtm = datetime.datetime.now()
elif self.conf.part_mode == 'date_field':
dt_str = data[self.conf.part_field]
if dt_str is None:
raise Exception('part_field(%s) is NULL: %s' %
(self.conf.part_field, ev))
dtm = datetime.datetime.strptime(dt_str[:19], "%Y-%m-%d %H:%M:%S")
else:
raise UsageError('Bad value for part_mode: %s' %\
self.conf.part_mode)
vals = {
'parent': self.dest_table,
'year': "%04d" % dtm.year,
'month': "%02d" % dtm.month,
'day': "%02d" % dtm.day,
'hour': "%02d" % dtm.hour,
}
return (self.get_part_name() % vals, dtm)
def check_part(self, dst, part_time):
"""Create part table if not exists.
It part_template present, execute it
else if part function present in db, call it
else clone master table"""
curs = self.dst_curs
if skytools.exists_table(curs, dst):
return
dst = quote_fqident(dst)
vals = {
'dest': dst,
'part': dst,
'parent': self.fq_dest_table,
'pkeys': ",".join(self.pkeys), # quoting?
# we do this to make sure that constraints for
# tables who contain a schema will still work
'schema_table': dst.replace(".", "__"),
'part_field': self.conf.part_field,
'part_time': part_time,
'period': self.conf.period,
}
def exec_with_vals(tmpl):
if tmpl:
sql = tmpl % vals
curs.execute(sql)
return True
return False
exec_with_vals(self.conf.pre_part)
if not exec_with_vals(self.conf.part_template):
self.log.debug('part_template not provided, using part func')
# if part func exists call it with val arguments
pfargs = ', '.join('%%(%s)s' % arg for arg in PART_FUNC_ARGS)
# set up configured function
pfcall = 'select %s(%s)' % (self.conf.part_func, pfargs)
have_func = skytools.exists_function(curs, self.conf.part_func,
len(PART_FUNC_ARGS))
# backwards compat
if not have_func and self.conf.part_func == PART_FUNC_NEW:
pfcall = 'select %s(%s)' % (PART_FUNC_OLD, pfargs)
have_func = skytools.exists_function(curs, PART_FUNC_OLD,
len(PART_FUNC_ARGS))
if have_func:
self.log.debug('check_part.exec: func:%s, args: %s' %
(pfcall, vals))
curs.execute(pfcall, vals)
else:
#
# Otherwise crete simple clone.
#
# FixMe: differences from create_partitions():
# - check constraints
# - inheritance
#
self.log.debug('part func %s not found, cloning table' %
self.conf.part_func)
struct = TableStruct(curs, self.dest_table)
struct.create(curs, T_ALL, dst)
exec_with_vals(self.conf.post_part)
self.log.info("Created table: %s" % dst)
def real_copy(self, tablename, src_curs, dst_curs, column_list):
"""do actual table copy and return tuple with number of bytes and rows
copyed
"""
_src_cols = _dst_cols = column_list
condition = ''
if self.conf.skip_fields:
_src_cols = [
col for col in column_list if col not in self.conf.skip_fields
]
_dst_cols = _src_cols
if self.conf.field_map:
_src_cols = [
col for col in _src_cols if col in self.conf.field_map
]
_dst_cols = [self.conf.field_map[col] for col in _src_cols]
if self.encoding_validator:
def _write_hook(obj, data):
return self.encoding_validator.validate_copy(
data, _src_cols, tablename)
else:
_write_hook = None
return skytools.full_copy(tablename,
src_curs,
dst_curs,
_src_cols,
condition,
dst_tablename=self.dest_table,
dst_column_list=_dst_cols,
write_hook=_write_hook)
class Dispatcher(BaseHandler):
"""Partitioned loader.
Splits events into partitions, if requested.
Then applies them without further processing.
"""
handler_name = 'dispatch'
def __init__(self, table_name, args, dest_table):
# compat for dest-table
dest_table = args.get('table', dest_table)
BaseHandler.__init__(self, table_name, args, dest_table)
# show args
self.log.debug("dispatch.init: table_name=%r, args=%r" % \
(table_name, args))
self.batch_info = None
self.dst_curs = None
self.pkeys = None
# config
self.conf = self.get_config()
hdlr_cls = ROW_HANDLERS[self.conf.row_mode]
self.row_handler = hdlr_cls(self.log)
if self.conf.encoding:
self.encoding_validator = EncodingValidator(self.log,
self.conf.encoding)
else:
self.encoding_validator = None
def _parse_args_from_doc (self):
doc = __doc__
params_descr = []
params_found = False
for line in doc.splitlines():
ln = line.strip()
if params_found:
if ln.startswith("=="):
break
m = re.match ("^(\w+):$", ln)
if m:
name = m.group(1)
expr = text = ""
elif not params_descr:
continue
else:
name, expr, text = params_descr.pop()
text += ln + "\n"
params_descr.append ((name, expr, text))
elif ln == "== HANDLER ARGUMENTS ==":
params_found = True
return params_descr
def get_config(self):
"""Processes args dict"""
conf = skytools.dbdict()
# set table mode
conf.table_mode = self.get_arg('table_mode', TABLE_MODES)
conf.analyze = self.get_arg('analyze', [0, 1])
if conf.table_mode == 'part':
conf.part_mode = self.get_arg('part_mode', PART_MODES)
conf.part_field = self.args.get('part_field')
if conf.part_mode == 'date_field' and not conf.part_field :
raise Exception('part_mode date_field requires part_field!')
conf.period = self.get_arg('period', PERIODS)
conf.part_name = self.args.get('part_name')
conf.part_template = self.args.get('part_template')
conf.pre_part = self.args.get('pre_part')
conf.post_part = self.args.get('post_part')
conf.part_func = self.args.get('part_func', PART_FUNC_NEW)
# set row mode and event types to process
conf.row_mode = self.get_arg('row_mode', ROW_MODES)
event_types = self.args.get('event_types', '*')
if event_types == '*':
event_types = EVENT_TYPES
else:
event_types = [evt.upper() for evt in event_types.split(',')]
for evt in event_types:
if evt not in EVENT_TYPES:
raise Exception('Unsupported operation: %s' % evt)
conf.event_types = event_types
# set load handler
conf.load_mode = self.get_arg('load_mode', LOAD_MODES)
conf.method = self.get_arg('method', METHODS)
# fields to skip
conf.skip_fields = [f.strip().lower()
for f in self.args.get('skip_fields','').split(',')]
# get fields map (obsolete, for compatibility reasons)
fields = self.args.get('fields', '*')
if fields == "*":
conf.field_map = None
else:
conf.field_map = {}
for fval in fields.split(','):
tmp = fval.split(':')
if len(tmp) == 1:
conf.field_map[tmp[0]] = tmp[0]
else:
conf.field_map[tmp[0]] = tmp[1]
# encoding validator
conf.encoding = self.args.get('encoding')
return conf
def get_arg(self, name, value_list, default = None):
default = default or value_list[0]
val = type(default)(self.args.get(name, default))
if val not in value_list:
raise Exception('Bad argument %s value %r' % (name, val))
return val
def reset(self):
"""Called before starting to process a batch.
Should clean any pending data."""
BaseHandler.reset(self)
def prepare_batch(self, batch_info, dst_curs):
"""Called on first event for this table in current batch."""
if self.conf.table_mode != 'ignore':
self.batch_info = batch_info
self.dst_curs = dst_curs
#BaseHandler.prepare_batch(self, batch_info, dst_curs)
def filter_data(self, data):
"""Process with fields skip and map"""
fskip = self.conf.skip_fields
fmap = self.conf.field_map
if fskip:
data = dict((k, v) for k, v in data.items()
if k not in fskip)
if fmap:
# when field name not present in source is used then None (NULL)
# value is inserted. is it ok?
data = dict( (v, data.get(k)) for k, v in fmap.items())
return data
def filter_pkeys(self, pkeys):
"""Process with fields skip and map"""
fskip = self.conf.skip_fields
fmap = self.conf.field_map
if fskip:
pkeys = [f for f in pkeys if f not in fskip]
if fmap:
pkeys = [fmap[p] for p in pkeys if p in fmap]
return pkeys
def process_event(self, ev, sql_queue_func, arg):
"""Process a event.
Event should be added to sql_queue or executed directly.
"""
if self.conf.table_mode == 'ignore':
return
# get data
data = skytools.db_urldecode(ev.data)
if self.encoding_validator:
data = self.encoding_validator.validate_dict(data, self.table_name)
if len(ev.ev_type) < 2 or ev.ev_type[1] != ':':
raise Exception('Unsupported event type: %s/extra1=%s/data=%s' % (
ev.ev_type, ev.ev_extra1, ev.ev_data))
op, pkeys = ev.type.split(':', 1)
if op not in 'IUD':
raise Exception('Unknown event type: %s' % ev.ev_type)
# process only operations specified
if not op in self.conf.event_types:
return
self.log.debug('dispatch.process_event: %s/%s' % (
ev.ev_type, ev.ev_data))
if self.pkeys is None:
self.pkeys = self.filter_pkeys(pkeys.split(','))
data = self.filter_data(data)
# prepare split table when needed
if self.conf.table_mode == 'part':
dst, part_time = self.split_format(ev, data)
if dst not in self.row_handler.table_map:
self.check_part(dst, part_time)
else:
dst = self.dest_table
if dst not in self.row_handler.table_map:
self.row_handler.add_table(dst, LOADERS[self.conf.load_mode],
self.pkeys, self.conf)
self.row_handler.process(dst, op, data)
#BaseHandler.process_event(self, ev, sql_queue_func, arg)
def finish_batch(self, batch_info, dst_curs):
"""Called when batch finishes."""
if self.conf.table_mode != 'ignore':
self.row_handler.flush(dst_curs)
#BaseHandler.finish_batch(self, batch_info, dst_curs)
def get_part_name(self):
# if custom part name template given, use it
if self.conf.part_name:
return self.conf.part_name
parts = ['year', 'month', 'day', 'hour']
name_parts = ['parent'] + parts[:parts.index(self.conf.period)+1]
return '_'.join('%%(%s)s' % part for part in name_parts)
def split_format(self, ev, data):
"""Generates part table name from template"""
if self.conf.part_mode == 'batch_time':
dtm = self.batch_info['batch_end']
elif self.conf.part_mode == 'event_time':
dtm = ev.ev_time
elif self.conf.part_mode == 'current_time':
dtm = datetime.datetime.now()
elif self.conf.part_mode == 'date_field':
dt_str = data[self.conf.part_field]
if dt_str is None:
raise Exception('part_field(%s) is NULL: %s' % (self.conf.part_field, ev))
dtm = datetime.datetime.strptime(dt_str[:19], "%Y-%m-%d %H:%M:%S")
else:
raise UsageError('Bad value for part_mode: %s' %\
self.conf.part_mode)
vals = {'parent': self.dest_table,
'year': "%04d" % dtm.year,
'month': "%02d" % dtm.month,
'day': "%02d" % dtm.day,
'hour': "%02d" % dtm.hour,
}
return (self.get_part_name() % vals, dtm)
def check_part(self, dst, part_time):
"""Create part table if not exists.
It part_template present, execute it
else if part function present in db, call it
else clone master table"""
curs = self.dst_curs
if skytools.exists_table(curs, dst):
return
dst = quote_fqident(dst)
vals = {'dest': dst,
'part': dst,
'parent': self.fq_dest_table,
'pkeys': ",".join(self.pkeys), # quoting?
# we do this to make sure that constraints for
# tables who contain a schema will still work
'schema_table': dst.replace(".", "__"),
'part_field': self.conf.part_field,
'part_time': part_time,
'period': self.conf.period,
}
def exec_with_vals(tmpl):
if tmpl:
sql = tmpl % vals
curs.execute(sql)
return True
return False
exec_with_vals(self.conf.pre_part)
if not exec_with_vals(self.conf.part_template):
self.log.debug('part_template not provided, using part func')
# if part func exists call it with val arguments
pfargs = ', '.join('%%(%s)s' % arg for arg in PART_FUNC_ARGS)
# set up configured function
pfcall = 'select %s(%s)' % (self.conf.part_func, pfargs)
have_func = skytools.exists_function(curs, self.conf.part_func, len(PART_FUNC_ARGS))
# backwards compat
if not have_func and self.conf.part_func == PART_FUNC_NEW:
pfcall = 'select %s(%s)' % (PART_FUNC_OLD, pfargs)
have_func = skytools.exists_function(curs, PART_FUNC_OLD, len(PART_FUNC_ARGS))
if have_func:
self.log.debug('check_part.exec: func:%s, args: %s' % (pfcall, vals))
curs.execute(pfcall, vals)
else:
#
# Otherwise create simple clone.
#
# FixMe: differences from create_partitions():
# - check constraints
# - inheritance
#
self.log.debug('part func %s not found, cloning table' % self.conf.part_func)
struct = TableStruct(curs, self.dest_table)
struct.create(curs, T_ALL, dst)
exec_with_vals(self.conf.post_part)
self.log.info("Created table: %s" % dst)
def real_copy(self, tablename, src_curs, dst_curs, column_list):
"""do actual table copy and return tuple with number of bytes and rows
copied
"""
_src_cols = _dst_cols = column_list
condition = ''
if self.conf.skip_fields:
_src_cols = [col for col in column_list
if col not in self.conf.skip_fields]
_dst_cols = _src_cols
if self.conf.field_map:
_src_cols = [col for col in _src_cols if col in self.conf.field_map]
_dst_cols = [self.conf.field_map[col] for col in _src_cols]
if self.encoding_validator:
def _write_hook(obj, data):
return self.encoding_validator.validate_copy(data, _src_cols, tablename)
else:
_write_hook = None
return skytools.full_copy(tablename, src_curs, dst_curs, _src_cols, condition,
dst_tablename = self.dest_table,
dst_column_list = _dst_cols,
write_hook = _write_hook)
