Examples with H2Util org.apache.tika.eval.db.H2Util used on opensource projects

Example 1 with H2Util

use of org.apache.tika.eval.db.H2Util in project tika by apache.

the class ProfilerBatchTest method setUp.

@BeforeClass
public static void setUp() throws Exception {
    Path inputRoot = Paths.get(new ComparerBatchTest().getClass().getResource("/test-dirs/extractsA").toURI());
    dbDir = Files.createTempDirectory(inputRoot, "tika-test-db-dir-");
    Map<String, String> args = new HashMap<>();
    Path db = dbDir.resolve("profiler_test");
    args.put("-db", db.toString());
    //for debugging, you can use this to select only one file pair to load
    //args.put("-includeFilePat", "file8.*");
    /* BatchProcessTestExecutor ex = new BatchProcessTestExecutor(COMPARER_PROCESS_CLASS, args,
                "/single-file-profiler-crawl-input-config.xml");
        StreamStrings streamStrings = ex.execute();
        System.out.println(streamStrings.getErrString());
        System.out.println(streamStrings.getOutString());*/
    H2Util dbUtil = new H2Util(db);
    conn = dbUtil.getConnection();
}

Example 2 with H2Util

use of org.apache.tika.eval.db.H2Util in project tika by apache.

the class ResultsReporterTest method setUp.

@Before
public void setUp() throws Exception {
    configFile = Paths.get(this.getClass().getResource("/reports.xml").toURI());
    tmpDir = Files.createTempDirectory("tika-eval-report-test-");
    connection = new H2Util(tmpDir.resolve(dbName)).getConnection();
    String sql = "CREATE TABLE test_table (ID LONG PRIMARY KEY, STRING VARCHAR(32))";
    Statement st = connection.createStatement();
    st.execute(sql);
    sql = "INSERT into test_table values ( 100000, 'the quick brown')";
    st.execute(sql);
    sql = "INSERT into test_table values (123456789, 'fox jumped over')";
    st.execute(sql);
    connection.commit();
}

Example 3 with H2Util

use of org.apache.tika.eval.db.H2Util in project tika by apache.

the class EvalConsumersBuilder method build.

@Override
public ConsumersManager build(Node node, Map<String, String> runtimeAttributes, ArrayBlockingQueue<FileResource> queue) {
    List<FileResourceConsumer> consumers = new LinkedList<>();
    int numConsumers = BatchProcessBuilder.getNumConsumers(runtimeAttributes);
    Map<String, String> localAttrs = XMLDOMUtil.mapifyAttrs(node, runtimeAttributes);
    Path db = getPath(localAttrs, "db");
    String jdbcConnectionString = localAttrs.get("jdbc");
    Path langModelDir = getPath(localAttrs, "langModelDir");
    try {
        if (langModelDir == null) {
            LanguageIDWrapper.loadBuiltInModels();
        } else {
            LanguageIDWrapper.loadModels(langModelDir);
        }
    } catch (IOException e) {
        throw new RuntimeException(e);
    }
    Path commonTokens = getPath(localAttrs, "commonTokens");
    String defaultLangCode = localAttrs.get("defaultLangCode");
    if (defaultLangCode == null || "".equals(defaultLangCode)) {
        defaultLangCode = "en";
    }
    //can be null, in which case will load from memory
    try {
        AbstractProfiler.loadCommonTokens(commonTokens, defaultLangCode);
    } catch (IOException e) {
        throw new RuntimeException(e);
    }
    JDBCUtil jdbcUtil = null;
    if (db != null) {
        jdbcUtil = new H2Util(db);
    } else if (jdbcConnectionString != null) {
        jdbcUtil = new JDBCUtil(jdbcConnectionString, localAttrs.get("jdbcDriver"));
    } else {
        throw new RuntimeException("Must specify: -db or -jdbc");
    }
    EvalConsumerBuilder consumerBuilder = ClassLoaderUtil.buildClass(EvalConsumerBuilder.class, PropsUtil.getString(localAttrs.get("consumerBuilderClass"), null));
    if (consumerBuilder == null) {
        throw new RuntimeException("Must specify consumerBuilderClass in config file");
    }
    boolean forceDrop = PropsUtil.getBoolean(localAttrs.get("drop"), false);
    MimeBuffer mimeBuffer = null;
    try {
        mimeBuffer = consumerBuilder.init(queue, localAttrs, jdbcUtil, forceDrop);
    } catch (IOException | SQLException e) {
        throw new RuntimeException(e);
    }
    for (int i = 0; i < numConsumers; i++) {
        try {
            consumers.add(consumerBuilder.build());
        } catch (IOException | SQLException e) {
            throw new RuntimeException(e);
        }
    }
    DBConsumersManager manager;
    try {
        manager = new DBConsumersManager(jdbcUtil, mimeBuffer, consumers);
    } catch (SQLException e) {
        throw new RuntimeException(e);
    }
    consumerBuilder.addErrorLogTablePairs(manager);
    return manager;
}

Example 4 with H2Util

use of org.apache.tika.eval.db.H2Util in project tika by apache.

the class ResultsReporter method main.

public static void main(String[] args) throws Exception {
    DefaultParser defaultCLIParser = new DefaultParser();
    CommandLine commandLine = null;
    try {
        commandLine = defaultCLIParser.parse(OPTIONS, args);
    } catch (ParseException e) {
        System.out.println(e.getMessage());
        USAGE();
        return;
    }
    JDBCUtil dbUtil = null;
    if (commandLine.hasOption("db")) {
        Path db = Paths.get(commandLine.getOptionValue("db"));
        if (!H2Util.databaseExists(db)) {
            throw new RuntimeException("I'm sorry, but I couldn't find this h2 database: " + db + "\nMake sure not to include the .mv.db at the end.");
        }
        dbUtil = new H2Util(db);
    } else if (commandLine.hasOption("jdbc")) {
        String driverClass = null;
        if (commandLine.hasOption("jdbcdriver")) {
            driverClass = commandLine.getOptionValue("jdbcdriver");
        }
        dbUtil = new JDBCUtil(commandLine.getOptionValue("jdbc"), driverClass);
    } else {
        System.err.println("Must specify either -db for the default in-memory h2 database\n" + "or -jdbc for a full jdbc connection string");
        USAGE();
        return;
    }
    try (Connection c = dbUtil.getConnection()) {
        Path tmpReportsFile = null;
        try {
            ResultsReporter resultsReporter = null;
            String reportsFile = commandLine.getOptionValue("rf");
            if (reportsFile == null) {
                tmpReportsFile = getDefaultReportsConfig(c);
                resultsReporter = ResultsReporter.build(tmpReportsFile);
            } else {
                resultsReporter = ResultsReporter.build(Paths.get(reportsFile));
            }
            Path reportsRootDirectory = Paths.get(commandLine.getOptionValue("rd", "reports"));
            if (Files.isDirectory(reportsRootDirectory)) {
                LOG.warn("'Reports' directory exists.  Will overwrite existing reports.");
            }
            resultsReporter.execute(c, reportsRootDirectory);
        } finally {
            if (tmpReportsFile != null) {
                Files.delete(tmpReportsFile);
            }
        }
    }
}

Example 5 with H2Util

use of org.apache.tika.eval.db.H2Util in project tika by apache.

the class XMLErrorLogUpdater method main.

public static void main(String[] args) throws Exception {
    XMLErrorLogUpdater writer = new XMLErrorLogUpdater();
    Path xmlLogFileA = Paths.get(args[0]);
    Path xmlLogFileB = Paths.get(args[1]);
    Path db = Paths.get(args[2]);
    JDBCUtil dbUtil = new H2Util(db);
    Connection connection = dbUtil.getConnection();
    writer.update(connection, ExtractComparer.EXTRACT_EXCEPTION_TABLE_A, xmlLogFileA);
    writer.update(connection, ExtractComparer.EXTRACT_EXCEPTION_TABLE_B, xmlLogFileB);
    connection.commit();
    connection.close();
}