experiment.py 文件源码

def analyze_lucene_index(self):
        results = IndexCharacteristics("Lucene", self.ingestion_thread_count, self.thread_counts)
        results.index_type = "Lucene"

        # Don't know how to determine bits per posting for Lucene.
        results.bits_per_posting = math.nan

        with open(self.lucene_build_index_log, 'r') as myfile:
            build_index_log = myfile.read()
            results.total_ingestion_time = \
                float(re.findall("Ingested \d+ chunk files in (\d+\.?\d+) seconds.", build_index_log)[0])

        for i, threads in enumerate(self.thread_counts):
            run_queries_log = self.lucene_run_queries_log[i]
            with open(run_queries_log, 'r') as myfile:
                data = myfile.read()
                results.append_float_field("qps", "QPS:", data)
                results.append_float_field("mps", "MPS:", data)
                results.append_float_field("mpq", "MPQ:", data)
                results.append_float_field("mean_query_latency", "Mean query latency:", data)
                results.append_float_field("planning_overhead", r"Planning overhead:", data)

        # Lucene false positive rate is always zero.
        results.false_positive_rate = 0;
        results.false_negative_rate = 0;

        return results


    ###########################################################################
    #
    # MG4J
    #
    ###########################################################################