use of org.apache.jackrabbit.oak.run.cli.NodeStoreFixture in project jackrabbit-oak by apache.
the class TextExtractorMain method main.
public static void main(String[] args) throws Exception {
OptionParser parser = new OptionParser();
Options opts = new Options();
opts.setCommandName(TikaCommandOptions.NAME);
opts.setSummary("Provides text extraction related operations");
opts.setConnectionString(CommonOptions.DEFAULT_CONNECTION_STRING);
opts.registerOptionsFactory(TikaCommandOptions.FACTORY);
// NodeStore is only required for generate command. So make it optional
opts.parseAndConfigure(parser, args, false);
TikaCommandOptions tikaOpts = opts.getOptionBean(TikaCommandOptions.class);
// If generate then check that NodeStore is specified
if (tikaOpts.generate()) {
opts.checkNonOptions();
}
try (Closer closer = Closer.create()) {
boolean report = tikaOpts.report();
boolean extract = tikaOpts.extract();
boolean generate = tikaOpts.generate();
BlobStore blobStore;
NodeStore nodeStore = null;
File dataFile = tikaOpts.getDataFile();
File storeDir = tikaOpts.getStoreDir();
File tikaConfigFile = tikaOpts.getTikaConfig();
BinaryResourceProvider binaryResourceProvider = null;
BinaryStats stats = null;
String path = tikaOpts.getPath();
if (tikaConfigFile != null) {
checkArgument(tikaConfigFile.exists(), "Tika config file %s does not exist", tikaConfigFile.getAbsolutePath());
}
if (storeDir != null) {
if (storeDir.exists()) {
checkArgument(storeDir.isDirectory(), "Path [%s] specified for storing extracted " + "text content is not a directory", storeDir.getAbsolutePath());
}
}
checkNotNull(dataFile, "Data file not configured with %s", tikaOpts.getDataFileSpecOpt());
if (!generate) {
// For report and extract case we do not need NodeStore access so create BlobStore directly
BlobStoreFixture blobStoreFixture = BlobStoreFixtureProvider.create(opts);
closer.register(blobStoreFixture);
blobStore = checkNotNull(blobStoreFixture).getBlobStore();
} else {
NodeStoreFixture nodeStoreFixture = NodeStoreFixtureProvider.create(opts);
closer.register(nodeStoreFixture);
blobStore = nodeStoreFixture.getBlobStore();
nodeStore = nodeStoreFixture.getStore();
}
checkNotNull(blobStore, "This command requires an external BlobStore configured");
if (generate) {
checkNotNull(dataFile, "Data file path not provided");
log.info("Generated csv data to be stored in {}", dataFile.getAbsolutePath());
BinaryResourceProvider brp = new NodeStoreBinaryResourceProvider(nodeStore, blobStore);
CSVFileGenerator generator = new CSVFileGenerator(dataFile);
generator.generate(brp.getBinaries(path));
}
if (report || extract) {
checkArgument(dataFile.exists(), "Data file %s does not exist", dataFile.getAbsolutePath());
CSVFileBinaryResourceProvider csvProvider = new CSVFileBinaryResourceProvider(dataFile, blobStore);
closer.register(csvProvider);
binaryResourceProvider = csvProvider;
stats = new BinaryStats(tikaConfigFile, binaryResourceProvider);
String summary = stats.getSummary();
log.info(summary);
}
if (extract) {
checkNotNull(storeDir, "Directory to store extracted text content " + "must be specified via %s", tikaOpts.getStoreDirSpecOpt());
checkNotNull(blobStore, "BlobStore found to be null.");
DataStoreTextWriter writer = new DataStoreTextWriter(storeDir, false);
TextExtractor extractor = new TextExtractor(writer);
if (tikaOpts.isPoolSizeDefined()) {
extractor.setThreadPoolSize(tikaOpts.getPoolSize());
}
if (tikaConfigFile != null) {
extractor.setTikaConfig(tikaConfigFile);
}
closer.register(writer);
closer.register(extractor);
extractor.setStats(stats);
log.info("Using path {}", path);
extractor.extract(binaryResourceProvider.getBinaries(path));
extractor.close();
writer.close();
}
}
}
use of org.apache.jackrabbit.oak.run.cli.NodeStoreFixture in project jackrabbit-oak by apache.
the class JsonIndexTest method readWrite.
@Test
public void readWrite() throws Exception {
JsonIndexCommand index = new JsonIndexCommand();
try (NodeStoreFixture fixture = memoryFixture()) {
NodeStore store = fixture.getStore();
index.session = JsonIndexCommand.openSession(store, "admin", "admin");
assertCommand(index, combineLines(""), "{'addNode':'/foo', 'node':{'jcr:primaryType': 'nt:unstructured', 'x': 1, 'y':{}}}", "{'session': 'save'}");
assertCommand(index, combineLines("/foo", "/jcr:system", "/oak:index", "/rep:security"), "{'xpath':'/jcr:root/* order by @jcr:path'}");
assertCommand(index, combineLines("/oak:index/counter"), "{'xpath':'/jcr:root//element(*, oak:QueryIndexDefinition)[@type=`counter`] " + "order by @jcr:path'}");
assertCommand(index, combineLines("[nt:unstructured] as [a] /* property test = 1 " + "where ([a].[x] = 1) and (isdescendantnode([a], [/])) */"), "{'addNode':'/oak:index/test', 'node':{ " + "'jcr:primaryType':'oak:QueryIndexDefinition', " + "'type':'property', " + "'reindex':true, " + "'entryCount': 1, " + "'{Name}declaringNodeTypes': ['nt:unstructured'], " + "'{Name}propertyNames':['x'] " + "}}", "{'session':'save'}", "{'xpath':'explain /jcr:root//element(*, nt:unstructured)[@x=1]'}", "{'xpath':'/jcr:root//element(*, nt:unstructured)[@x=2]'}");
assertCommand(index, combineLines("50"), "{'addNode':'/foo/test', 'node':{'jcr:primaryType': 'oak:Unstructured', 'child':{}}}", "{'$x':1}", "{'loop':[" + "{'$p': '/foo/test/child/n', '+': '$x'}, " + "{'addNode': '$p', 'node': {'x': '$x', 'jcr:primaryType': 'nt:unstructured'}}, " + "{'session':'save'}, " + "{'$x': '$x', '+':1}, " + "{'$break': true, 'if': '$x', '=': 100}]}", "{'session':'save'}", "{'xpath':'/jcr:root//element(*, nt:unstructured)[@x<50]', 'quiet':true}", "{'$y':0}", "{'for':'$result', 'do': [{'$y': '$y', '+': 1}]}", "{'print': '$y'}");
assertCommand(index, combineLines("[nt:unstructured] as [a] /* nodeType Filter(query=" + "explain select [jcr:path], [jcr:score], * from [nt:unstructured] as a " + "where [x] = 1 and isdescendantnode(a, '/') /* xpath: " + "/jcr:root//element(*, nt:unstructured)[@x=1] */, path=//*, " + "property=[x=[1]]) where ([a].[x] = 1) and (isdescendantnode([a], [/])) */"), "{'setProperty': '/oak:index/test/type', 'value': 'disabled'}", "{'session':'save'}", "{'xpath':'explain /jcr:root//element(*, nt:unstructured)[@x=1]'}");
assertCommand(index, combineLines("[nt:unstructured] as [a] /* traverse '*' " + "where [a].[x] = 1 */"), "{'removeNode': '/oak:index/nodetype'}", "{'session':'save'}", "{'sql':'explain select * from [nt:unstructured] as [a] where [x]=1'}");
assertCommand(index, combineLines("['/foo': {\n" + " 'jcr:primaryType': 'nt:unstructured', '{Long}x': '1', 'y': {}, 'test': {}\n" + "}]"), "{'xpath':'/jcr:root/foo', 'depth':2}");
index.session.logout();
}
}
use of org.apache.jackrabbit.oak.run.cli.NodeStoreFixture in project jackrabbit-oak by apache.
the class IndexCommand method connectInReadWriteModeAndImportIndex.
private void connectInReadWriteModeAndImportIndex(IndexOptions indexOpts, File indexDir) throws Exception {
try (Closer closer = Closer.create()) {
configureCustomizer(opts, closer, false);
NodeStoreFixture fixture = NodeStoreFixtureProvider.create(opts);
closer.register(fixture);
IndexHelper indexHelper = createIndexHelper(fixture, indexOpts, closer);
importIndex(indexHelper, indexDir);
}
}
use of org.apache.jackrabbit.oak.run.cli.NodeStoreFixture in project jackrabbit-oak by apache.
the class IndexCommand method execute.
@Override
public void execute(String... args) throws Exception {
OptionParser parser = new OptionParser();
opts = new Options();
opts.setCommandName(NAME);
opts.setSummary(summary);
opts.setConnectionString(CommonOptions.DEFAULT_CONNECTION_STRING);
opts.registerOptionsFactory(IndexOptions.FACTORY);
opts.parseAndConfigure(parser, args);
indexOpts = opts.getOptionBean(IndexOptions.class);
// Clean up before setting up NodeStore as the temp
// directory might be used by NodeStore for cache stuff like persistentCache
setupDirectories(indexOpts);
setupLogging(indexOpts);
logCliArgs(args);
boolean success = false;
try {
if (indexOpts.isReindex() && opts.getCommonOpts().isReadWrite()) {
performReindexInReadWriteMode(indexOpts);
} else {
try (Closer closer = Closer.create()) {
configureCustomizer(opts, closer, true);
NodeStoreFixture fixture = NodeStoreFixtureProvider.create(opts);
closer.register(fixture);
execute(fixture, indexOpts, closer);
tellReportPaths();
}
}
success = true;
} catch (Throwable e) {
log.error("Error occurred while performing index tasks", e);
e.printStackTrace(System.err);
if (disableExitOnError) {
throw e;
}
} finally {
shutdownLogging();
}
if (!success) {
System.exit(1);
}
}
use of org.apache.jackrabbit.oak.run.cli.NodeStoreFixture in project jackrabbit-oak by apache.
the class Console method main.
public static void main(String[] args) throws Exception {
OptionParser parser = new OptionParser();
OptionSpec quiet = parser.accepts("quiet", "be less chatty");
OptionSpec shell = parser.accepts("shell", "run the shell after executing files");
Options opts = new Options();
OptionSet options = opts.parseAndConfigure(parser, args);
int code = 0;
try (NodeStoreFixture fixture = NodeStoreFixtureProvider.create(opts)) {
List<String> nonOptions = opts.getCommonOpts().getNonOptions();
List<String> scriptArgs = nonOptions.size() > 1 ? nonOptions.subList(1, nonOptions.size()) : Collections.emptyList();
IO io = new IO();
if (options.has(quiet)) {
io.setVerbosity(IO.Verbosity.QUIET);
}
if (!opts.getCommonOpts().isReadWrite()) {
io.out.println("Repository connected in read-only mode. Use '--read-write' for write operations");
}
GroovyConsole console = new GroovyConsole(ConsoleSession.create(fixture.getStore(), fixture.getWhiteboard()), new IO(), fixture);
if (!scriptArgs.isEmpty()) {
if (!options.has(shell)) {
Preferences.verbosity = IO.Verbosity.QUIET;
}
code = console.execute(scriptArgs);
}
if (scriptArgs.isEmpty() || options.has(shell)) {
code = console.run();
}
}
System.exit(code);
}
Aggregations