testdata.java

一套java版本的搜索引擎源码

    }

    // padd number from left
    static String format(int n, String col) {
      String res = padd + n;
      return res.substring(res.length() - col.length());
    }

    // padd string from right
    static String format(String s, String col) {
      return (s + padd).substring(0,col.length());
    }

    /**
     * Prepare a list of benchmark data, using all possible combinations of
     * benchmark parameters.
     *
     * @param sources   list of directories containing different source document
     *                  collections
     * @param analyzers of analyzers to use.
     */
    public static TestData[] getAll(File[] sources, Analyzer[] analyzers)
    {
        List res = new ArrayList(50);
        TestData ref = new TestData();
        for (int q = 0; q < analyzers.length; q++)
        {
            for (int m = 0; m < sources.length; m++)
            {
                for (int i = 0; i < MAX_BUFFERED_DOCS_COUNTS.length; i++)
                {
                    for (int k = 0; k < MERGEFACTOR_COUNTS.length; k++)
                    {
                        for (int n = 0; n < Constants.BOOLEANS.length; n++)
                        {
                            for (int p = 0; p < Constants.BOOLEANS.length; p++)
                            {
                                ref.id = "td-" + q + m + i + k + n + p;
                                ref.source = sources[m];
                                ref.analyzer = analyzers[q];
                                ref.maxBufferedDocs = MAX_BUFFERED_DOCS_COUNTS[i];
                                ref.mergeFactor = MERGEFACTOR_COUNTS[k];
                                ref.compound = Constants.BOOLEANS[n].booleanValue();
                                ref.optimize = Constants.BOOLEANS[p].booleanValue();
                                try
                                {
                                    res.add(ref.clone());
                                }
                                catch (Exception e)
                                {
                                    e.printStackTrace();
                                }
                            }
                        }
                    }
                }
            }
        }
        return (TestData[]) res.toArray(new TestData[0]);
    }

    /**
     * Similar to {@link #getAll(java.io.File[], org.apache.lucene.analysis.Analyzer[])} but only uses
     * maxBufferedDocs of 10 and 100 and same for mergeFactor, thus reducing the number of permutations significantly.
     * It also only uses compund file and optimize is always true.
     *
     * @param sources
     * @param analyzers
     * @return An Array of {@link TestData}
     */
    public static TestData[] getTestDataMinMaxMergeAndMaxBuffered(File[] sources, Analyzer[] analyzers)
    {
        List res = new ArrayList(50);
        TestData ref = new TestData();
        for (int q = 0; q < analyzers.length; q++)
        {
            for (int m = 0; m < sources.length; m++)
            {
                ref.id = "td-" + q + m + "_" + 10 + "_" + 10;
                ref.source = sources[m];
                ref.analyzer = analyzers[q];
                ref.maxBufferedDocs = 10;
                ref.mergeFactor = 10;//MERGEFACTOR_COUNTS[k];
                ref.compound = true;
                ref.optimize = true;
                try
                {
                    res.add(ref.clone());
                }
                catch (Exception e)
                {
                    e.printStackTrace();
                }
                ref.id = "td-" + q + m  + "_" + 10 + "_" + 100;
                ref.source = sources[m];
                ref.analyzer = analyzers[q];
                ref.maxBufferedDocs = 10;
                ref.mergeFactor = 100;//MERGEFACTOR_COUNTS[k];
                ref.compound = true;
                ref.optimize = true;
                try
                {
                    res.add(ref.clone());
                }
                catch (Exception e)
                {
                    e.printStackTrace();
                }
                ref.id = "td-" + q + m + "_" + 100 + "_" + 10;
                ref.source = sources[m];
                ref.analyzer = analyzers[q];
                ref.maxBufferedDocs = 100;
                ref.mergeFactor = 10;//MERGEFACTOR_COUNTS[k];
                ref.compound = true;
                ref.optimize = true;
                try
                {
                    res.add(ref.clone());
                }
                catch (Exception e)
                {
                    e.printStackTrace();
                }
                ref.id = "td-" + q + m + "_" + 100 + "_" + 100;
                ref.source = sources[m];
                ref.analyzer = analyzers[q];
                ref.maxBufferedDocs = 100;
                ref.mergeFactor = 100;//MERGEFACTOR_COUNTS[k];
                ref.compound = true;
                ref.optimize = true;
                try
                {
                    res.add(ref.clone());
                }
                catch (Exception e)
                {
                    e.printStackTrace();
                }
            }
        }
        return (TestData[]) res.toArray(new TestData[0]);
    }

    protected Object clone()
    {
        TestData cl = new TestData();
        cl.id = id;
        cl.compound = compound;
        cl.heap = heap;
        cl.mergeFactor = mergeFactor;
        cl.maxBufferedDocs = maxBufferedDocs;
        cl.optimize = optimize;
        cl.source = source;
        cl.directory = directory;
        cl.analyzer = analyzer;
        // don't clone runData
        return cl;
    }

    public String toString()
    {
        StringBuffer res = new StringBuffer();
        res.append("#-- ID: ").append(id).append(", ").append(new Date()).append(", heap=").append(heap).append(" --\n");
        res.append("# source=").append(source).append(", directory=").append(directory).append("\n");
        res.append("# maxBufferedDocs=").append(maxBufferedDocs).append(", mergeFactor=").append(mergeFactor);
        res.append(", compound=").append(compound).append(", optimize=").append(optimize).append("\n");
        if (queries != null)
        {
            res.append(QueryData.getLabels()).append("\n");
            for (int i = 0; i < queries.length; i++)
            {
                res.append("# ").append(queries[i].toString()).append("\n");
            }
        }
        return res.toString();
    }

    public Analyzer getAnalyzer()
    {
        return analyzer;
    }

    public void setAnalyzer(Analyzer analyzer)
    {
        this.analyzer = analyzer;
    }

    public boolean isCompound()
    {
        return compound;
    }

    public void setCompound(boolean compound)
    {
        this.compound = compound;
    }

    public Directory getDirectory()
    {
        return directory;
    }

    public void setDirectory(Directory directory)
    {
        this.directory = directory;
    }

    public long getHeap()
    {
        return heap;
    }

    public void setHeap(long heap)
    {
        this.heap = heap;
    }

    public String getId()
    {
        return id;
    }

    public void setId(String id)
    {
        this.id = id;
    }

    public int getMaxBufferedDocs()
    {
        return maxBufferedDocs;
    }

    public void setMaxBufferedDocs(int maxBufferedDocs)
    {
        this.maxBufferedDocs = maxBufferedDocs;
    }

    public int getMergeFactor()
    {
        return mergeFactor;
    }

    public void setMergeFactor(int mergeFactor)
    {
        this.mergeFactor = mergeFactor;
    }

    public boolean isOptimize()
    {
        return optimize;
    }

    public void setOptimize(boolean optimize)
    {
        this.optimize = optimize;
    }

    public QueryData[] getQueries()
    {
        return queries;
    }

    public void setQueries(QueryData[] queries)
    {
        this.queries = queries;
    }

    public Vector getRunData()
    {
        return runData;
    }

    public void setRunData(Vector runData)
    {
        this.runData = runData;
    }

    public File getSource()
    {
        return source;
    }

    public void setSource(File source)
    {
        this.source = source;
    }
}