#!/home/rocky/.pyenv/versions/3.4.8/bin/python from __future__ import print_function import click import pickle def sort_profile_info(path, max_count=1000): profile_info = pickle.load(open(path, "rb")) # Classify unused rules. Some are "unused" because the have nullable # nonterminals and those show up as a different rule. Sothe rule # *is* used just not in the form where a nullable symbol hasn't been # nulled. # And in some cases this is intentional. Uncompyle6 creates such grammar # rules to ensure that positions of certain nonterminals in semantic # actions appear in the same place as similar grammar rules unused_rules = set() # just the grammar rules used_rules = [] # (count, grammar rule) for rule, count in profile_info.items(): if count == 0: unused_rules.add(rule) else: used_rules.append((count, rule)) for count, rule in used_rules: if rule.find("\\e_") > -1: canonic_rule = rule.replace("\\e_", "", 1000) if canonic_rule in unused_rules: unused_rules.remove(canonic_rule) pass pass pass unused_items = [(0, item) for item in sorted(unused_rules)] used_items = sorted(used_rules, reverse=False) return [item for item in unused_items + used_items if item[0] <= max_count] DEFAULT_COVERAGE_FILE = "/tmp/spark-grammar.cover", DEFAULT_COUNT = 100 @click.command() @click.option('--path', type=str, default=DEFAULT_COVERAGE_FILE, help=("grammar coverage file (default %s)" % DEFAULT_COVERAGE_FILE)) @click.option('--max-count', type=int, default=DEFAULT_COUNT, help=("limit output to rules having no more than this many hits (default %d)" % DEFAULT_COUNT)) def run(path, max_count): """Print grammar reduce statistics for a series of spark-parser parses """ for count, rule in sort_profile_info(path, max_count): print("%d: %s" % (count, rule)) pass return if __name__ == '__main__': run()