@Test
public void testJSONFlatten() {
Scheme sourceScheme = new TextLine(new Fields("line"));
Tap input = new Hfs(sourceScheme, "data/input1.json");
Pipe assembly = new Pipe("json_flatten");
JSONSplitter jsonSplitter =
new JSONSplitter(new Fields("name", "age", "phones"), "name", "age", "phones");
assembly =
new Each(assembly, new Fields("line"), jsonSplitter, new Fields("name", "age", "phones"));
JSONFlatten jsonFlatten =
new JSONFlatten(new Fields("phone_number", "phone_type"), "number", "type");
assembly =
new Each(
assembly,
new Fields("phones"),
jsonFlatten,
new Fields("name", "age", "phone_number", "phone_type"));
assembly = new Each(assembly, AssertionLevel.STRICT, new AssertNotNull());
Tap output = new Hfs(new TextLine(), "output/flatten", true);
FlowConnector connector = new HadoopFlowConnector();
Flow flow = connector.connect(input, output, assembly);
flow.complete();
}
@Test
public void testCascadeConnector() {
Pipe copy = new Pipe("copy");
Properties cfg = HdpBootstrap.asProperties(CascadingHadoopSuite.configuration);
FlowDef flow =
new FlowDef()
.addSource(copy, sourceTap())
.addTailSink(copy, new EsTap("cascading-hadoop/cascade-connector"));
FlowConnector connector = new HadoopFlowConnector(cfg);
Flow[] flows = new Flow[] {connector.connect(flow)};
CascadeConnector cascadeConnector = new CascadeConnector(cfg);
cascadeConnector.connect(flows).complete();
}
/**
* Create a Cascading Flow that will parse a set of mbox files and emit a tab-separated text file
* with fields for the msgId, author, email address, etc.
*
* <p>Note this Flow will only run locally, since we're using the cascading.utils LocalPlatform.
*
* @param options Settings for the flow
* @return Flow suitable for execution
* @throws Exception
*/
public static Flow createFlow(ParseEmailArchivesOptions options) throws Exception {
BasePlatform platform = new LocalPlatform(ParseEmailArchivesWorkflow.class);
// We'll read individual file paths from the input file.
BasePath inputPath = platform.makePath(options.getFileList());
Tap sourceTap = platform.makeTap(platform.makeTextScheme(), inputPath);
Pipe emailPipe = new Pipe("emails");
emailPipe = new Each(emailPipe, new Fields("line"), new MboxSplitterFunction());
emailPipe = new Each(emailPipe, new ParseEmail());
BasePath outputPath = platform.makePath(options.getOutputDir());
TextLineScheme scheme = new TextLineScheme(false);
Tap sinkTap = platform.makeTap(scheme, outputPath, SinkMode.REPLACE);
FlowConnector flowConnector = platform.makeFlowConnector();
Flow flow = flowConnector.connect(sourceTap, sinkTap, emailPipe);
return flow;
}
@SuppressWarnings("rawtypes")
public void run() {
TextLine scheme = new TextLine(new Fields("line"));
// Tap input = inputPath.matches("^[^:]+://.*") ? new Hfs(scheme, inputPath) : new Lfs(scheme,
// inputPath);
Tap input = new Hfs(scheme, inputPath);
// extract the tags through regex and save content in group 1 -> as fields tags
String tagJsonRegex = "\"tags\":\\[([^\\]]*)";
Function parse = new RegexParser(new Fields("tags"), tagJsonRegex, new int[] {1});
// for each line get the tags using a regex
Pipe assembly = new Each("import", new Fields("line"), parse, Fields.RESULTS);
// split "tags" into "tag"
Function split = new RegexSplitGenerator(new Fields("tag"), ",");
assembly = new Each(assembly, new Fields("tags"), split);
assembly = new Each(assembly, new Fields("tag"), new RegexFilter(".+"));
// group each tag by name
assembly = new GroupBy(assembly, new Fields("tag"));
// count each tag under "count" field
Aggregator count = new Count(new Fields("count"));
assembly = new Every(assembly, count);
// create a SINK tap to write to the default filesystem
// by default, TextLine writes all fields out
new TextLine(new Fields("tag", "count"));
// Tap output = outputPath.matches("^[^:]+://.*") ? new Hfs(sinkScheme, outputPath,
// SinkMode.REPLACE) : new Lfs(
// sinkScheme, outputPath, SinkMode.REPLACE);
Tap output = new Lfs(scheme, outputPath, SinkMode.REPLACE);
// wire the existing Hadoop config into HadoopFlow
Properties properties = ConfigurationUtils.asProperties(hadoopConfiguration);
FlowConnector flowConnector = new HadoopFlowConnector(properties);
FlowConnectorProps.setDebugLevel(properties, DebugLevel.VERBOSE);
Flow flow = flowConnector.connect("hashtagcount", input, output, assembly);
flow.start();
flow.complete();
}
/** Using a local FileTap try a simple flow dumping output to stdout. */
public static void main(String[] args) throws Exception {
if (args.length != 2) {
throw new Exception("first argument is mainframe data file, second is output dir");
}
Properties properties = new Properties();
AppProps.addApplicationTag(properties, "tutorials");
AppProps.addApplicationTag(properties, "cluster:development");
AppProps.setApplicationName(properties, "Cascading-Copybook Hadoop");
String inPath = args[0];
String outPath = args[1];
Pipe copyPipe = new Pipe("testPipe");
// Source turns the mainframe data into a flat Tuple
Scheme sourceScheme = new CustdatScheme();
Tap inTap = new Hfs(sourceScheme, inPath);
// Tuples are written to a csv
Scheme sinkScheme = new TextDelimited(new Custdat(), false, ",");
Tap outTap = new Hfs(sinkScheme, outPath, SinkMode.REPLACE);
FlowDef flowDef =
FlowDef.flowDef()
.addSource(copyPipe, inTap)
.addTailSink(copyPipe, outTap)
.setDebugLevel(DebugLevel.VERBOSE)
.setName("Cascading Cobol");
AppProps.setApplicationJarClass(properties, Main.class);
FlowConnector flowConnector = new Hadoop2MR1FlowConnector(properties);
flowConnector.connect(flowDef).complete();
}
public static void main(String[] args) {
String dataset = args[0];
String outPath = args[1];
Properties properties = new Properties();
properties.setProperty("mapred.max.map.failures.percent", "5");
AppProps.setApplicationJarClass(properties, Main.class);
FlowConnector flowConnector = new Hadoop2MR1FlowConnector(properties);
Fields fmFields =
new Fields(
"loanid",
"monthly_period",
"servicer_name",
"curr_interest_rate",
"unpaid_principal_balance",
"loan_age",
"remaining_months_legal_maturity",
"adjusted_remaining_months_maturity",
"maturity_date",
"metropolitan_statistical_area",
"current_loan_delinq_status",
"mod_flag",
"zero_bal_code",
"zero_bal_effective_date",
"repurchase_indicator");
// create the source tap */
Tap inTap = new Hfs(new TextDelimited(fmFields, false, "|"), dataset);
// create the sink tap
Tap outTap =
new Hfs(
new TextDelimited(
new Fields("monthly_period", "avg-unpaid_principal_balance"), false, "\t"),
outPath);
/** Specify a new Pipe */
Pipe copyPipe = new Pipe("unique copy");
// Deal with duplicates in data set
Pipe mypipe =
new Unique(
copyPipe,
new Fields("loanid", "monthly_period", "unpaid_principal_balance"),
Unique.Include.NO_NULLS);
// define "ScrubFunction" to clean up the token stream
Fields monthArguments = new Fields("loanid", "monthly_period", "unpaid_principal_balance");
// mypipe = new Each( mypipe, monthArguments, new getMonth( monthArguments ), Fields.RESULTS );
// Remove null, i.e. we discad nullsto calculate average
Filter filter = new RegexFilter("(\\d+)\\.(\\d+)");
mypipe = new Each(mypipe, filter);
// Return the Month in [ January - December] , given a number [ 1 - 12 ]
mypipe = new Each(mypipe, monthArguments, new getMonth(monthArguments));
// Group by month, this need to be by month
mypipe = new GroupBy(mypipe, new Fields("monthly_period"), new Fields("monthly_period"));
Fields groupingFields = new Fields("monthly_period");
Fields valueField = new Fields("unpaid_principal_balance");
Fields avgField = new Fields("avg-unpaid_principal_balance");
mypipe = new AverageBy(mypipe, groupingFields, valueField, avgField);
/** connect the taps, pipes, etc., into a flow */
FlowDef flowDef = FlowDef.flowDef().addSource(copyPipe, inTap).addTailSink(mypipe, outTap);
/** run the flow */
flowConnector.connect(flowDef).complete();
}
