def compute_optimal_matching(fact_stores, partners, guards, match_steps):
if len(partners) == 0:
return match_steps
curr_max_score = -100
curr_guard_steps = []
curr_guards_rest = []
curr_best_partner = None
curr_propagated = None
curr_index = -1
curr_free_terms = None
for index in xrange(0, len(partners)):
propagated, partner = partners[index]
num_of_joins = 0
num_of_free = 0
num_of_const = 0
free_terms = []
for term in partner.get_terms():
if term.is_var():
if term.is_binded():
num_of_joins += 1
else:
free_terms.append(term)
num_of_free += 1
elif term.is_const():
num_of_const += 1
partner.exist_bind_terms()
guard_steps, guards_rest = schedule_guards(guards)
guard_count = len(guard_steps)
new_max_score = num_of_joins * 10 + guard_count + num_of_const - num_of_free
if curr_max_score <= new_max_score:
curr_max_score = new_max_score
curr_guard_steps = guard_steps
curr_guards_rest = guards_rest
curr_best_partner = partner
curr_propagated = propagated
curr_index = index
curr_free_terms = free_terms
for term in free_terms:
term.unbind()
lookup_info = fact_stores[curr_best_partner.sym_id].generate_lookup(
curr_best_partner)
best_match_step = {
'is_lookup': True,
'propagated': curr_propagated,
'lookup_index': lookup_info['lookup_index'],
'fact_pat': make_fact_pat(curr_best_partner),
'filter': lookup_info['filter'],
'free_terms': curr_free_terms
}
match_steps.append(best_match_step)
for curr_guard_step in curr_guard_steps:
match_steps.append(curr_guard_step)
curr_best_partner.exist_bind_terms()
return compute_optimal_matching(
fact_stores, partners[:curr_index] + partners[(curr_index + 1):],
curr_guards_rest, match_steps)
def compute_optimal_matching(fact_stores, partners, guards, match_steps):
if len(partners) == 0:
return match_steps
curr_max_score = -100
curr_guard_steps = []
curr_guards_rest = []
curr_best_partner = None
curr_propagated = None
curr_index = -1
curr_free_terms = None
for index in xrange(0,len(partners)):
propagated,partner = partners[index]
num_of_joins = 0
num_of_free = 0
num_of_const = 0
free_terms = []
for term in partner.get_terms():
if term.is_var():
if term.is_binded():
num_of_joins += 1
else:
free_terms.append(term)
num_of_free += 1
elif term.is_const():
num_of_const += 1
partner.exist_bind_terms()
guard_steps,guards_rest = schedule_guards(guards)
guard_count = len(guard_steps)
new_max_score = num_of_joins*10 + guard_count + num_of_const - num_of_free
if curr_max_score <= new_max_score:
curr_max_score = new_max_score
curr_guard_steps = guard_steps
curr_guards_rest = guards_rest
curr_best_partner = partner
curr_propagated = propagated
curr_index = index
curr_free_terms = free_terms
for term in free_terms:
term.unbind()
lookup_info = fact_stores[curr_best_partner.sym_id].generate_lookup(curr_best_partner)
best_match_step = { 'is_lookup' : True
, 'propagated' : curr_propagated
, 'lookup_index' : lookup_info['lookup_index']
, 'fact_pat' : make_fact_pat(curr_best_partner)
, 'filter' : lookup_info['filter']
, 'free_terms' : curr_free_terms }
match_steps.append( best_match_step )
for curr_guard_step in curr_guard_steps:
match_steps.append( curr_guard_step )
curr_best_partner.exist_bind_terms()
return compute_optimal_matching(fact_stores, partners[:curr_index]+partners[(curr_index+1):], curr_guards_rest, match_steps)
def interpret_rule(rule, fact_stores):
rule_entries = map(lambda s: (False, s), rule.simplify()) + map(
lambda p: (True, p), rule.propagate())
rule_lhs = map(lambda t: t[1], rule_entries)
guards = rule.guards()
variables = rule.get_vars()
if len(rule.simplify()) == 0:
has_no_simplify = True
else:
has_no_simplify = False
rule_name = get_all_rule_classes()[rule.rule_id].__name__
interp_rule = []
for i in xrange(0, len(rule_entries)):
propagated, rule_entry = rule_entries[i]
partners = rule_entries[:i] + rule_entries[(i + 1):]
rule_entry.exist_bind_terms()
early_guard_steps, guards_rest = schedule_guards(guards)
match_steps = compute_optimal_matching(fact_stores, partners,
guards_rest, early_guard_steps)
for var in variables:
var.unbind()
interp_rule.append({
'rule_id': rule.rule_id,
'occ_id': i,
'propagated': propagated,
'entry': make_fact_pat(rule_entry),
'match_steps': match_steps,
'has_no_simplify': has_no_simplify,
'rhs': rule.consequents,
'exist_locs': rule.get_exist_locs,
'has_exist_locs': len(rule.exist_locations) > 0
})
return interp_rule
def interpret_rule(rule, fact_stores):
rule_entries = map(lambda s: (False,s),rule.simplify()) + map(lambda p: (True,p),rule.propagate())
rule_lhs = map(lambda t: t[1],rule_entries)
guards = rule.guards()
variables = rule.get_vars()
if len(rule.simplify()) == 0:
has_no_simplify = True
else:
has_no_simplify = False
rule_name = get_all_rule_classes()[rule.rule_id].__name__
interp_rule = []
for i in xrange(0,len(rule_entries)):
propagated,rule_entry = rule_entries[i]
partners = rule_entries[:i] + rule_entries[(i+1):]
rule_entry.exist_bind_terms()
early_guard_steps,guards_rest = schedule_guards(guards)
match_steps = compute_optimal_matching(fact_stores, partners, guards_rest, early_guard_steps)
for var in variables:
var.unbind()
interp_rule.append( { 'rule_id' : rule.rule_id,
'occ_id' : i,
'propagated' : propagated,
'entry' : make_fact_pat(rule_entry),
'match_steps' : match_steps,
'has_no_simplify' : has_no_simplify,
'rhs' : rule.consequents,
'exist_locs' : rule.get_exist_locs,
'has_exist_locs' : len(rule.exist_locations) > 0 } )
return interp_rule