def gen_cql_env(configuration, experiment_conf, output):
'''
Generate environment files for StremPref
'''
text = get_register_stream(configuration, experiment_conf)
query_dir = get_query_dir(configuration, experiment_conf)
# Environment files for equivalent CQL queries
filename = query_dir + os.sep + 'table_ots.cql'
text += REG_Q_STR.format(qname='table_ots', qfile=filename)
filename = query_dir + os.sep + 'stream_ots.cql'
text += REG_Q_STR.format(qname='stream_ots', qfile=filename)
filename = query_dir + os.sep + 'z.cql'
text += REG_Q_STR.format(qname='z', qfile=filename)
filename = query_dir + os.sep + 'z_prime.cql'
text += REG_Q_STR.format(qname='z_prime', qfile=filename)
filename = query_dir + os.sep + 'p_start.cql'
text += REG_Q_STR.format(qname='p_start', qfile=filename)
filename = query_dir + os.sep + 'p_end.cql'
text += REG_Q_STR.format(qname='p_end', qfile=filename)
filename = query_dir + os.sep + 'p_start_end.cql'
text += REG_Q_STR.format(qname='p_start_end', qfile=filename)
# Final equivalent query
filename = query_dir + os.sep + 'equiv.cql'
if output:
# Get output filename
out_file = get_out_file(configuration, experiment_conf)
text += REG_Q_OUTPUT_STR.format(qname='equiv', qfile=filename,
ofile=out_file)
else:
text += REG_Q_STR.format(qname='equiv', qfile=filename)
filename = get_env_file(configuration, experiment_conf)
write_to_txt(filename, text)
def gen_util_query(configuration, experiment_conf):
'''
Generate single query
'''
op_list = experiment_conf[OPERATOR_LIST]
query = 'SELECT '
# ENDSEQ
if ENDSEQ in op_list:
query += Q_ENDSEQ
# CONSEQ
if CONSEQ in op_list:
query += Q_CONSEQ
# SEQ
query += Q_SEQ.format(ran=experiment_conf[RAN], sli=experiment_conf[SLI])
where_list = []
# MINSEQ
if MINSEQ in op_list:
where_list.append(Q_MINSEQ.format(min=experiment_conf[MIN]))
# MAXSEQ
if MAXSEQ in op_list:
where_list.append(Q_MAXSEQ.format(max=experiment_conf[MAX]))
if len(where_list):
query += '\nWHERE ' + ' AND '.join(where_list)
# Select correct query
rules_dict = gen_rules_dict(configuration, experiment_conf)
pref_str = get_temporal_preferences(rules_dict)
query += Q_BESTSEQ.format(pref=pref_str)
# Store query code
filename = get_query_util_file(configuration, experiment_conf)
write_to_txt(filename, query)
def gen_cql_rpos_spos_queries(query_dir):
'''
Generate RPOS and SPOS queries
'''
filename = query_dir + os.sep + 'rpos.cql'
write_to_txt(filename, CQL_RPOS)
filename = query_dir + os.sep + 'spos.cql'
write_to_txt(filename, CQL_SPOS)
def gen_prev_rule_query(out_dir, rule_number, rule, condition_number):
'''
Generate query for rule with previous condition
'''
filename = out_dir + os.sep + \
'r' + str(rule_number) + '_f' + str(condition_number) + '.cql'
query = PREV_QUERY.format(cond=rule[COND_PREV])
write_to_txt(filename, query)
def gen_cql_z_query(query_dir, experiment_conf):
'''
Consider RANGE and SLIDE and generate Z relation
'''
query = CQL_Z.format(ran=experiment_conf[RAN],
sli=experiment_conf[SLI])
filename = query_dir + os.sep + 'z.cql'
write_to_txt(filename, query)
def gen_seq_query(configuration, experiment_conf):
'''
Generate queries with SEQ operator
'''
query_dir = get_query_dir(configuration, experiment_conf)
filename = query_dir + os.sep + 'seq.cql'
query = SEQ_QUERY.format(ran=experiment_conf[RAN],
sli=experiment_conf[SLI])
write_to_txt(filename, query)
def gen_cql_queries(configuration, experiment_conf):
'''
Generate all CQL queries equivalent to ENDSEQ operator
'''
query_dir = get_query_dir(configuration, experiment_conf)
query = CQL_Z.format(ran=experiment_conf[RAN],
sli=experiment_conf[SLI])
filename = query_dir + os.sep + 'z.cql'
write_to_txt(filename, query)
gen_cql_final_query(query_dir, experiment_conf)
def gen_someprev_rule_query(out_dir, rule_number, rule, condition_number):
'''
Generate query for rule with some previous condition
'''
filename = out_dir + os.sep + 'm_sp' + str(rule_number) + '.cql'
query = MIN_SOMEPREV_QUERY.format(cond=rule[COND_SOMPREV])
write_to_txt(filename, query)
filename = out_dir + os.sep + \
'r' + str(rule_number) + '_f' + str(condition_number) + '.cql'
query = SOMEPREV_QUERY.format(rn=rule_number)
write_to_txt(filename, query)
def gen_bestseq_query(configuration, experiment_conf):
'''
Generate StreamPref queries with BESTSEQ operator
'''
filename = get_query_dir(configuration, experiment_conf) + \
os.sep + 'bestseq.cql'
rules_dict = gen_rules_dict(configuration, experiment_conf)
pref_str = get_temporal_preferences(rules_dict)
query = BESTSEQ_QUERY.format(ran=experiment_conf[RAN],
sli=experiment_conf[SLI],
pref=pref_str)
write_to_txt(filename, query)
def gen_util_env(configuration, experiment_conf):
'''
Generate environment
'''
text = get_register_stream(configuration, experiment_conf)
# Get query filename
filename = get_query_util_file(configuration, experiment_conf)
# Register query
text += REG_Q_STR.format(qname='util', qfile=filename)
# Get environment filename
filename = get_env_util_file(configuration, experiment_conf)
write_to_txt(filename, text)
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_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_endseq_env(configuration, experiment_conf, output):
'''
Generate environment files for ENDSEQ operator
'''
text = get_register_stream(configuration, experiment_conf)
# Get query filename
query_dir = get_query_dir(configuration, experiment_conf)
filename = query_dir + os.sep + 'endseq.cql'
# Register query
if output:
# Get output filename
out_file = get_out_file(configuration, experiment_conf)
text += REG_Q_OUTPUT_STR.format(qname='endseq', qfile=filename,
ofile=out_file)
else:
text += REG_Q_STR.format(qname='endseq', qfile=filename)
# Get environment filename
filename = get_env_file(configuration, experiment_conf)
write_to_txt(filename, text)
def gen_cql_env(configuration, experiment_conf, output):
'''
Generate environment files for StremPref
'''
# Register stream
text = get_register_stream(configuration,
experiment_conf,
include_tup=True)
query_dir = get_query_dir(configuration, experiment_conf)
filename = query_dir + os.sep + 'z.cql'
text += REG_Q_STR.format(qname='z', qfile=filename)
filename = query_dir + os.sep + 'p_join.cql'
text += REG_Q_STR.format(qname='p_join', qfile=filename)
filename = query_dir + os.sep + 'p.cql'
text += REG_Q_STR.format(qname='p', qfile=filename)
text += gen_reg_rules_queries(configuration, query_dir, experiment_conf)
query_name = 't1'
filename = query_dir + os.sep + query_name + '.cql'
text += REG_Q_STR.format(qname=query_name, qfile=filename)
level = experiment_conf[LEV]
for number in range(2, level + 1):
query_name = 't' + str(number)
filename = query_dir + os.sep + query_name + '.cql'
text += REG_Q_STR.format(qname=query_name, qfile=filename)
query_name = 'id'
filename = query_dir + os.sep + query_name + '.cql'
text += REG_Q_STR.format(qname=query_name, qfile=filename)
query_name = 'equiv'
filename = query_dir + os.sep + query_name + '.cql'
if output:
# Get output filename
out_file = get_out_file(configuration, experiment_conf)
text += REG_Q_OUTPUT_STR.format(qname=query_name,
qfile=filename,
ofile=out_file)
else:
text += REG_Q_STR.format(qname=query_name, qfile=filename)
# Get environment filename
filename = get_env_file(configuration, experiment_conf)
write_to_txt(filename, text)
def gen_cql_env(configuration, experiment_conf, output):
'''
Generate environment files for StremPref
'''
text = get_register_stream(configuration, experiment_conf)
query_dir = get_query_dir(configuration, experiment_conf)
# Environment files for equivalent CQL queries
# RPOS
filename = query_dir + os.sep + 'rpos.cql'
text += REG_Q_STR.format(qname='rpos', qfile=filename)
# SPOS
filename = query_dir + os.sep + 'spos.cql'
text += REG_Q_STR.format(qname='spos', qfile=filename)
# W
filename = query_dir + os.sep + 'w.cql'
text += REG_Q_STR.format(qname='w', qfile=filename)
# W1 and P1
filename = query_dir + os.sep + 'w1.cql'
text += REG_Q_STR.format(qname='w1', qfile=filename)
filename = query_dir + os.sep + 'p1.cql'
text += REG_Q_STR.format(qname='p1', qfile=filename)
# W_i and P_i
range_value = experiment_conf[RAN]
for pos in range(2, range_value + 1):
filename = query_dir + os.sep + 'w' + str(pos) + '.cql'
text += REG_Q_STR.format(qname='w' + str(pos), qfile=filename)
filename = query_dir + os.sep + 'p' + str(pos) + '.cql'
text += REG_Q_STR.format(qname='p' + str(pos), qfile=filename)
# Final equivalent query
filename = query_dir + os.sep + 'equiv.cql'
if output:
# Get output filename
out_file = get_out_file(configuration, experiment_conf)
text += REG_Q_OUTPUT_STR.format(qname='equiv',
qfile=filename,
ofile=out_file)
else:
text += REG_Q_STR.format(qname='equiv', qfile=filename)
filename = get_env_file(configuration, experiment_conf)
write_to_txt(filename, 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 MAXSEQ operator
'''
query_dir = get_query_dir(configuration, experiment_conf)
query = CQL_Z.format(ran=experiment_conf[RAN], sli=experiment_conf[SLI])
filename = query_dir + os.sep + 'z.cql'
write_to_txt(filename, query)
query = CQL_ZMAX.format(max=experiment_conf[MAX])
filename = query_dir + os.sep + 'zmax.cql'
write_to_txt(filename, query)
filename = query_dir + os.sep + 'equiv.cql'
write_to_txt(filename, CQL_EQUIV)
def gen_allprev_rule_query(out_dir, rule_number, rule, condition_number):
'''
Generate query for rule with all previous condition
'''
filename = out_dir + os.sep + 'nv_ap' + str(rule_number) + '.cql'
query = NONVALID_ALLPREV_QUERY.format(cond=rule[COND_ALLPREV])
write_to_txt(filename, query)
filename = out_dir + os.sep + 'm_ap' + str(rule_number) + '.cql'
query = MIN_ALLPREV_QUERY.format(rn=rule_number)
write_to_txt(filename, query)
filename = out_dir + os.sep + \
'r' + str(rule_number) + '_f' + str(condition_number) + '.cql'
query = ALLPREV_QUERY.format(rn=rule_number)
write_to_txt(filename, query)
def gen_cql_position_queries(query_dir, experiment_conf):
'''
Generate queries to get each position
'''
# Generate W_1
filename = query_dir + os.sep + 'w1.cql'
write_to_txt(filename, CQL_W1)
# W_i
for range_value in range(2, experiment_conf[RAN] + 1):
query = CQL_WI.format(prev=range_value - 1)
filename = query_dir + os.sep + \
'w' + str(range_value) + '.cql'
write_to_txt(filename, query)
# P_i
for range_value in range(1, experiment_conf[RAN] + 1):
query = CQL_PI.format(pos=range_value)
filename = query_dir + os.sep + \
'p' + str(range_value) + '.cql'
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)
def gen_first_rule_query(out_dir, rule_number):
'''
Generate query for rule with first condition
'''
filename = out_dir + os.sep + 'r' + str(rule_number) + '.cql'
write_to_txt(filename, FIRST_QUERY)
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)
