use of cascading.tap.hadoop.Hfs in project parquet-mr by apache.
the class TestParquetTBaseScheme method testWrite.
@Test
public void testWrite() throws Exception {
Path path = new Path(parquetOutputPath);
JobConf jobConf = new JobConf();
final FileSystem fs = path.getFileSystem(jobConf);
if (fs.exists(path))
fs.delete(path, true);
Scheme sourceScheme = new TextLine(new Fields("first", "last"));
Tap source = new Hfs(sourceScheme, txtInputPath);
Scheme sinkScheme = new ParquetTBaseScheme(Name.class);
Tap sink = new Hfs(sinkScheme, parquetOutputPath);
Pipe assembly = new Pipe("namecp");
assembly = new Each(assembly, new PackThriftFunction());
HadoopFlowConnector hadoopFlowConnector = new HadoopFlowConnector();
Flow flow = hadoopFlowConnector.connect("namecp", source, sink, assembly);
flow.complete();
assertTrue(fs.exists(new Path(parquetOutputPath)));
assertTrue(fs.exists(new Path(parquetOutputPath + "/_metadata")));
assertTrue(fs.exists(new Path(parquetOutputPath + "/_common_metadata")));
}
use of cascading.tap.hadoop.Hfs in project parquet-mr by apache.
the class TestParquetTBaseScheme method doRead.
private void doRead(Scheme sourceScheme) throws Exception {
createFileForRead();
Path path = new Path(txtOutputPath);
final FileSystem fs = path.getFileSystem(new Configuration());
if (fs.exists(path))
fs.delete(path, true);
Tap source = new Hfs(sourceScheme, parquetInputPath);
Scheme sinkScheme = new TextLine(new Fields("first", "last"));
Tap sink = new Hfs(sinkScheme, txtOutputPath);
Pipe assembly = new Pipe("namecp");
assembly = new Each(assembly, new UnpackThriftFunction());
Flow flow = new HadoopFlowConnector().connect("namecp", source, sink, assembly);
flow.complete();
String result = FileUtils.readFileToString(new File(txtOutputPath + "/part-00000"));
assertEquals("Alice\tPractice\nBob\tHope\nCharlie\tHorse\n", result);
}
use of cascading.tap.hadoop.Hfs in project parquet-mr by apache.
the class ParquetScroogeSchemeTest method verifyScroogeRead.
public <T> void verifyScroogeRead(List<TBase> recordsToWrite, Class<T> readClass, String expectedStr, String projectionFilter) throws Exception {
Configuration conf = new Configuration();
deleteIfExist(PARQUET_PATH);
deleteIfExist(TXT_OUTPUT_PATH);
final Path parquetFile = new Path(PARQUET_PATH);
writeParquetFile(recordsToWrite, conf, parquetFile);
Scheme sourceScheme = new ParquetScroogeScheme(new Config().withRecordClass(readClass).withProjectionString(projectionFilter));
Tap source = new Hfs(sourceScheme, PARQUET_PATH);
Scheme sinkScheme = new TextLine(new Fields("first", "last"));
Tap sink = new Hfs(sinkScheme, TXT_OUTPUT_PATH);
Pipe assembly = new Pipe("namecp");
assembly = new Each(assembly, new ObjectToStringFunction());
Flow flow = new HadoopFlowConnector().connect("namecp", source, sink, assembly);
flow.complete();
String result = FileUtils.readFileToString(new File(TXT_OUTPUT_PATH + "/part-00000"));
assertEquals(expectedStr, result);
}
use of cascading.tap.hadoop.Hfs in project Impatient by Cascading.
the class Main method main.
public static void main(String[] args) {
String docPath = args[0];
String wcPath = args[1];
String stopPath = args[2];
String tfidfPath = args[3];
String trapPath = args[4];
String checkPath = args[5];
Properties properties = new Properties();
AppProps.setApplicationJarClass(properties, Main.class);
AppProps.setApplicationName(properties, "Impatient Part 6");
AppProps.addApplicationTag(properties, "tutorial:impatient");
AppProps.addApplicationTag(properties, "technology:Cascading");
FlowConnector flowConnector = new Hadoop2MR1FlowConnector(properties);
// create source and sink taps
Tap docTap = new Hfs(new TextDelimited(true, "\t"), docPath);
Tap wcTap = new Hfs(new TextDelimited(true, "\t"), wcPath);
Fields stop = new Fields("stop");
Tap stopTap = new Hfs(new TextDelimited(stop, true, "\t"), stopPath);
Tap tfidfTap = new Hfs(new TextDelimited(true, "\t"), tfidfPath);
Tap trapTap = new Hfs(new TextDelimited(true, "\t"), trapPath);
Tap checkTap = new Hfs(new TextDelimited(true, "\t"), checkPath);
// use a stream assertion to validate the input data
Pipe docPipe = new Pipe("token");
AssertMatches assertMatches = new AssertMatches("doc\\d+\\s.*");
docPipe = new Each(docPipe, AssertionLevel.STRICT, assertMatches);
// specify a regex operation to split the "document" text lines into a token stream
Fields token = new Fields("token");
Fields text = new Fields("text");
RegexSplitGenerator splitter = new RegexSplitGenerator(token, "[ \\[\\]\\(\\),.]");
Fields fieldSelector = new Fields("doc_id", "token");
docPipe = new Each(docPipe, text, splitter, fieldSelector);
// define "ScrubFunction" to clean up the token stream
Fields scrubArguments = new Fields("doc_id", "token");
docPipe = new Each(docPipe, scrubArguments, new ScrubFunction(scrubArguments), Fields.RESULTS);
// perform a left join to remove stop words, discarding the rows
// which joined with stop words, i.e., were non-null after left join
Pipe stopPipe = new Pipe("stop");
Pipe tokenPipe = new HashJoin(docPipe, token, stopPipe, stop, new LeftJoin());
tokenPipe = new Each(tokenPipe, stop, new RegexFilter("^$"));
tokenPipe = new Retain(tokenPipe, fieldSelector);
// one branch of the flow tallies the token counts for term frequency (TF)
Pipe tfPipe = new Pipe("TF", tokenPipe);
Fields tf_count = new Fields("tf_count");
tfPipe = new CountBy(tfPipe, new Fields("doc_id", "token"), tf_count);
Fields tf_token = new Fields("tf_token");
tfPipe = new Rename(tfPipe, token, tf_token);
// one branch counts the number of documents (D)
Fields doc_id = new Fields("doc_id");
Fields tally = new Fields("tally");
Fields rhs_join = new Fields("rhs_join");
Fields n_docs = new Fields("n_docs");
Pipe dPipe = new Unique("D", tokenPipe, doc_id);
dPipe = new Each(dPipe, new Insert(tally, 1), Fields.ALL);
dPipe = new Each(dPipe, new Insert(rhs_join, 1), Fields.ALL);
dPipe = new SumBy(dPipe, rhs_join, tally, n_docs, long.class);
// one branch tallies the token counts for document frequency (DF)
Pipe dfPipe = new Unique("DF", tokenPipe, Fields.ALL);
Fields df_count = new Fields("df_count");
dfPipe = new CountBy(dfPipe, token, df_count);
Fields df_token = new Fields("df_token");
Fields lhs_join = new Fields("lhs_join");
dfPipe = new Rename(dfPipe, token, df_token);
dfPipe = new Each(dfPipe, new Insert(lhs_join, 1), Fields.ALL);
// example use of a debug, to observe tuple stream; turn off below
dfPipe = new Each(dfPipe, DebugLevel.VERBOSE, new Debug(true));
// join to bring together all the components for calculating TF-IDF
// the D side of the join is smaller, so it goes on the RHS
Pipe idfPipe = new HashJoin(dfPipe, lhs_join, dPipe, rhs_join);
// create a checkpoint, to observe the intermediate data in DF stream
Checkpoint idfCheck = new Checkpoint("checkpoint", idfPipe);
// the IDF side of the join is smaller, so it goes on the RHS
Pipe tfidfPipe = new CoGroup(tfPipe, tf_token, idfCheck, df_token);
// calculate the TF-IDF weights, per token, per document
Fields tfidf = new Fields("tfidf");
String expression = "(double) tf_count * Math.log( (double) n_docs / ( 1.0 + df_count ) )";
ExpressionFunction tfidfExpression = new ExpressionFunction(tfidf, expression, Double.class);
Fields tfidfArguments = new Fields("tf_count", "df_count", "n_docs");
tfidfPipe = new Each(tfidfPipe, tfidfArguments, tfidfExpression, Fields.ALL);
fieldSelector = new Fields("tf_token", "doc_id", "tfidf");
tfidfPipe = new Retain(tfidfPipe, fieldSelector);
tfidfPipe = new Rename(tfidfPipe, tf_token, token);
// keep track of the word counts, which are useful for QA
Pipe wcPipe = new Pipe("wc", tfPipe);
Fields count = new Fields("count");
wcPipe = new SumBy(wcPipe, tf_token, tf_count, count, long.class);
wcPipe = new Rename(wcPipe, tf_token, token);
// additionally, sort by count
wcPipe = new GroupBy(wcPipe, count, count);
// connect the taps, pipes, traps, checkpoints, etc., into a flow
FlowDef flowDef = FlowDef.flowDef().setName("tfidf").addSource(docPipe, docTap).addSource(stopPipe, stopTap).addTailSink(tfidfPipe, tfidfTap).addTailSink(wcPipe, wcTap).addTrap(docPipe, trapTap).addCheckpoint(idfCheck, checkTap);
// set to DebugLevel.VERBOSE for trace, or DebugLevel.NONE in production
flowDef.setDebugLevel(DebugLevel.VERBOSE);
// set to AssertionLevel.STRICT for all assertions, or AssertionLevel.NONE in production
flowDef.setAssertionLevel(AssertionLevel.STRICT);
// write a DOT file and run the flow
Flow tfidfFlow = flowConnector.connect(flowDef);
tfidfFlow.writeDOT("dot/tfidf.dot");
tfidfFlow.complete();
}
use of cascading.tap.hadoop.Hfs in project parquet-mr by apache.
the class ParquetScroogeSchemeTest method doWrite.
private void doWrite() throws Exception {
Path path = new Path(parquetOutputPath);
final FileSystem fs = path.getFileSystem(new Configuration());
if (fs.exists(path))
fs.delete(path, true);
Scheme sourceScheme = new TextLine(new Fields("first", "last"));
Tap source = new Hfs(sourceScheme, txtInputPath);
Scheme sinkScheme = new ParquetScroogeScheme<Name>(Name.class);
Tap sink = new Hfs(sinkScheme, parquetOutputPath);
Pipe assembly = new Pipe("namecp");
assembly = new Each(assembly, new PackThriftFunction());
Flow flow = new HadoopFlowConnector().connect("namecp", source, sink, assembly);
flow.complete();
}
Aggregations