def gen_rule_queries(query_dir, experiment_conf, rule_number, rule):
'''
Generate queries for all rules
'''
if rule[TYPE] == FIRST:
gen_first_rule_query(query_dir, rule_number)
elif rule[TYPE] == PREV:
gen_prev_rule_query(query_dir, rule_number, rule, 1)
gen_someprev_rule_query(query_dir, rule_number, rule, 2)
gen_allprev_rule_query(query_dir, rule_number, rule, 3)
query = COND_QUERY.format(rn=rule_number)
filename = query_dir + os.sep + 'r' + str(rule_number) + '.cql'
write_to_txt(filename, query)
# All input attributes (except identifier)
# Get attribute list
att_list = get_attribute_list(experiment_conf[ATT])
# Exclude sequence identifier (A1)
att_list = att_list[1:]
att_list = ', '.join(att_list)
# Attributes non in TUP (transitive tuples)
attnt_list = get_attribute_list(experiment_conf[ATT])
# Exclude sequence identifier (A1) and TUP attributes (A2, A3)
attnt_list = attnt_list[3:]
attnt_list = ', '.join(attnt_list)
# Generate D_i Pref Queries
query = DI_PREF_QUERY.format(att=att_list,
attnt=attnt_list,
cond=rule[COND_SIMPLE],
pref=rule[PREF],
rn=rule_number)
filename = query_dir + os.sep + 'd' + str(rule_number) + '_pref.cql'
write_to_txt(filename, query)
# Generate D_i NonPref Queries
query = \
DI_NONPREF_QUERY.format(att=att_list, attnt=attnt_list,
cond=rule[COND_SIMPLE],
nonpref=rule[NONPREF], rn=rule_number)
filename = query_dir + os.sep + 'd' + str(rule_number) + '_nonpref.cql'
write_to_txt(filename, query)
# Generate D_i Queries
# Get attribute list
att_list = get_attribute_list(experiment_conf[ATT])
# Exclude sequence identifier (A1)
att_list = att_list[1:]
p_att_list = ['p.' + att for att in att_list]
p_att_list = ', '.join(p_att_list)
np_att_list = ['p.' + att + ' AS _' + att for att in att_list]
np_att_list = ', '.join(np_att_list)
ceteris_cond = get_ceteris_attributes(experiment_conf)
ceteris_cond = ['p.' + att + ' = np.' + att for att in ceteris_cond]
ceteris_cond = ' AND '.join(ceteris_cond)
filename = query_dir + os.sep + 'd' + str(rule_number) + '.cql'
query = DI_QUERY.format(p_att=p_att_list,
np_att=np_att_list,
rn=rule_number,
cet_cond=ceteris_cond)
write_to_txt(filename, query)
def gen_conseq_stream(configuration, experiment_conf):
'''
Generate data stream
'''
# Build attribute list
att_list = get_attribute_list(experiment_conf[ATT], include_timestamp=True)
# Get list of sequence identifiers
id_list = gen_sequence_id_list(experiment_conf[NSQ])
# Randomize start timestamp for every identifier
id_start_list = []
for rec in id_list:
start = random.randint(0, experiment_conf[RAN] - 1)
id_start_list.append((rec, start))
# Get maximum timestamp (maximum range + maximum slide)
max_ts = get_max_data_timestamp(configuration[PARAMETER])
# File
filename = get_data_file(configuration, experiment_conf)
# First instant
rec_list = gen_conseq_records(configuration, experiment_conf,
id_start_list, 0)
write_to_csv(filename, att_list, rec_list)
# For each timestamp
for timestamp in range(1, max_ts + 1):
rec_list = gen_conseq_records(configuration, experiment_conf,
id_start_list, timestamp)
append_to_csv(filename, att_list, rec_list)
def gen_cql_queries(configuration, experiment_conf):
'''
Generate queries with CQL original operators equivalent to BESTSEQ operator
'''
filename = get_tup_file(configuration)
gen_transitive_tup(configuration, filename)
query_dir = get_query_dir(configuration, experiment_conf)
# Generate z query (sequences)
query = Z_QUERY.format(ran=experiment_conf[RAN], sli=experiment_conf[SLI])
filename = query_dir + os.sep + 'z.cql'
write_to_txt(filename, query)
# Generate p_join query (join z positions)
# Get attribute list
att_list = get_attribute_list(experiment_conf[ATT])
# Exclude sequence identifier (A1)
att_list = att_list[1:]
z1_att_list = ['z1.' + att for att in att_list]
z1_att_list = ', '.join(z1_att_list)
z2_att_list = ['z2.' + att + ' AS _' + att for att in att_list]
z2_att_list = ', '.join(z2_att_list)
query = P_JOIN_QUERY.format(z1_att=z1_att_list, z2_att=z2_att_list)
filename = query_dir + os.sep + 'p_join.cql'
write_to_txt(filename, query)
# Generate query p (positions to be compared)
diff_filter = ['NOT ' + att + ' = _' + att for att in att_list]
diff_filter = ' OR '.join(diff_filter)
query = P_QUERY.format(p_filter=diff_filter)
filename = query_dir + os.sep + 'p.cql'
write_to_txt(filename, query)
# Get rule list
rule_list = get_rule_list(configuration, experiment_conf)
# Generate query t1 (identifier of dominant sequences) and
# individual rule queries
query_list = []
for index, rule in enumerate(rule_list):
# Generates queries R_i and D_i for each rule
gen_rule_queries(query_dir, experiment_conf, index + 1, rule)
query = 'SELECT * FROM d' + str(index + 1)
query_list.append(query)
query = '\nUNION\n'.join(query_list) + ';'
filename = query_dir + os.sep + 't1.cql'
write_to_txt(filename, query)
# Generate T_i Queries
gen_cql_transitive_queries(experiment_conf, query_dir)
# Generate ID query
query = ID_QUERY.format(rn=experiment_conf[LEV])
filename = query_dir + os.sep + 'id.cql'
write_to_txt(filename, query)
# Generate query for final result
query = 'SELECT z.* FROM z, id WHERE z.a1 = id.a1;'
filename = query_dir + os.sep + 'equiv.cql'
write_to_txt(filename, query)
def gen_cql_w_query(query_dir, experiment_conf):
'''
Consider RANGE and SLIDE and generate W relation
'''
# Build attribute names list
att_list = get_attribute_list(experiment_conf[ATT])
att_str = ', '.join(att_list)
# W
query = CQL_W.format(att=att_str,
ran=experiment_conf[RAN],
sli=experiment_conf[SLI])
filename = query_dir + os.sep + 'w.cql'
write_to_txt(filename, query)
def gen_cql_final_query(query_dir, experiment_conf):
'''
Generate final query equivalent to SEQ operator for a range parameter
'''
# Get attribute list
att_list = get_attribute_list(experiment_conf[ATT], prefix='w.')
att_str = ', '.join(att_list)
# List of final position queries
pos_query_list = []
for position in range(1, experiment_conf[RAN] + 1):
pos_query = CQL_PI_FINAL.format(pos=position, att=att_str)
pos_query_list.append(pos_query)
# Equivalent is the union of final positions
query = '\nUNION\n'.join(pos_query_list) + ';'
filename = query_dir + os.sep + 'equiv.cql'
write_to_txt(filename, query)
def gen_cql_final_query(query_dir, experiment_conf):
'''
Generate final query equivalent to ENDSEQ operator
'''
filename = query_dir + os.sep + 'equiv.cql'
if os.path.isfile(filename):
return
range_value = experiment_conf[RAN]
att_list = get_attribute_list(experiment_conf[ATT])
att_str = ', '.join(att_list)
pos_query_list = []
for position in range(1, range_value + 1):
pos_query = CQL_EQUIV.format(att=att_str, ran=position)
pos_query_list.append(pos_query)
query = '\nUNION\n'.join(pos_query_list) + ';'
out_file = open(filename, 'w')
out_file.write(query)
out_file.close()
def gen_stream(configuration, experiment_conf):
'''
Generate a data stream
'''
# Build attribute list
att_list = get_attribute_list(experiment_conf[ATT], include_timestamp=True)
# Get list of sequence identifiers
id_list = gen_sequence_id_list(experiment_conf[NSQ])
# Get maximum timestamp (maximum range + maximum slide)
max_ts = get_max_data_timestamp(configuration[PARAMETER])
filename = get_data_file(configuration, experiment_conf)
# First timestamp
rec_list = gen_records(configuration, experiment_conf, id_list, 0)
write_to_csv(filename, att_list, rec_list)
# For each timestamp
for timestamp in range(1, max_ts + 1):
rec_list = gen_records(configuration, experiment_conf, id_list,
timestamp)
append_to_csv(filename, att_list, rec_list)
def get_register_stream(configuration, experiment_conf, include_tup=False):
'''
Get register steam string
'''
# Get attribute list
att_list = get_attribute_list(experiment_conf[ATT])
att_list = [att + ' ' + INTEGER for att in att_list]
att_str = ', '.join(att_list)
# Get data filename
filename = get_data_file(configuration, experiment_conf)
# Register stream
text = REG_STREAM_STR.format(atts=att_str, dfile=filename)
if include_tup:
text += '\n\n'
# Register tup table
filename = get_tup_file(configuration)
text += REG_TUP_STR.format(dfile=filename)
text += '\n\n' + '#' * 80 + '\n\n'
return text
def gen_cql_transitive_queries(experiment_conf, query_dir):
'''
Generate CQL queries for transitive comparisons
'''
# Generate T_i Queries
# Get attribute list
att_list = get_attribute_list(experiment_conf[ATT])
# Exclude sequence identifier (A1)
att_list = att_list[1:]
p_att_list = ['p.' + att for att in att_list]
p_att_list = ', '.join(p_att_list)
np_att_list = ['np._' + att for att in att_list]
np_att_list = ', '.join(np_att_list)
join_att = ['p._' + att + ' = np.' + att for att in att_list]
join_att = ' AND '.join(join_att)
for level_number in range(2, experiment_conf[LEV] + 1):
filename = query_dir + os.sep + 't' + str(level_number) + '.cql'
prev_level = level_number - 1
query = TI_QUERY.format(prev_n=prev_level,
p_att=p_att_list,
np_att=np_att_list,
p_np_join=join_att)
write_to_txt(filename, query)
def gen_cql_queries(configuration, experiment_conf):
'''
Generate all CQL queries equivalent to CONSEQ operator
'''
query_dir = get_query_dir(configuration, experiment_conf)
filename = query_dir + os.sep + 'table_ots.cql'
write_to_txt(filename, CQL_TABLE_OTS)
filename = query_dir + os.sep + 'stream_ots.cql'
write_to_txt(filename, CQL_STREAM_OTS)
gen_cql_z_query(query_dir, experiment_conf)
filename = query_dir + os.sep + 'z_prime.cql'
write_to_txt(filename, CQL_Z_PRIME)
filename = query_dir + os.sep + 'p_start.cql'
write_to_txt(filename, CQL_P_START)
filename = query_dir + os.sep + 'p_end.cql'
write_to_txt(filename, CQL_P_END)
filename = query_dir + os.sep + 'p_start_end.cql'
write_to_txt(filename, CQL_P_START_END)
filename = query_dir + os.sep + 'equiv.cql'
att_list = get_attribute_list(experiment_conf[ATT], 'z.')
att_list = ', '.join(att_list)
query = CQL_EQUIV.format(zatt=att_list)
write_to_txt(filename, query)