List of makeBinaryScheme() Examples

Examples of makeBinaryScheme()

bixo.config.BixoPlatform.makeBinaryScheme()

Examples of bixo.config.BixoPlatform.makeBinaryScheme()

            BasePath outputPath = platform.makePath(outputDirName);
            Tap pageStatusSinkTap = platform.makeTap(platform.makeTextScheme(), 
                            platform.makePath(outputPath, "page-status"), SinkMode.REPLACE);
            Tap mboxStatusSinkTap = platform.makeTap(platform.makeTextScheme(), 
                            platform.makePath(outputPath, "mbox-status"), SinkMode.REPLACE);
            Tap contentSinkTap = platform.makeTap(platform.makeBinaryScheme(FetchedDatum.FIELDS), 
                            platform.makePath(outputPath, "content"), SinkMode.REPLACE);
            Tap analyzerSinkTap = platform.makeTap(platform.makeTextScheme(), 
                            platform.makePath(outputPath, "analysis"), SinkMode.REPLACE);


            HashMap<String, Tap> sinkTapMap = new HashMap<String, Tap>(2);

View Full Code Here

Examples of bixo.config.BixoPlatform.makeBinaryScheme()


            // Now we should have an output/1-<timestamp>/ directory, where the
            // /urls dir has 11 entries with
            // one being previously crawled, and the other 10 being pending.


            Tap crawldbTap = platform.makeTap(platform.makeBinaryScheme(CrawlDbDatum.FIELDS), crawlDbPath);
            TupleEntryIterator iter = crawldbTap.openForRead(platform.makeFlowProcess());


            int numFetched = 0;
            int numPending = 0;
            while (iter.hasNext()) {

View Full Code Here

Examples of bixo.config.BixoPlatform.makeBinaryScheme()

            flow = DemoCrawlWorkflow.createFlow(curLoopDirPath, crawlDbPath, defaultPolicy, userAgent, urlFilter, options);
            flow.complete();
            // Update crawldb path
            crawlDbPath = platform.makePath(curLoopDirPath, CrawlConfig.CRAWLDB_SUBDIR_NAME);


            crawldbTap = platform.makeTap(platform.makeBinaryScheme(CrawlDbDatum.FIELDS), crawlDbPath);
            iter = crawldbTap.openForRead(platform.makeFlowProcess());


            numFetched = 0;
            numPending = 0;
            int numDepth0 = 0;

View Full Code Here

Examples of bixo.config.BixoPlatform.makeBinaryScheme()

        PartitioningKey groupingKey = new PartitioningKey("key", 1);
        FetchSetDatum pfd = new FetchSetDatum(urls, fetchTime, 1000, groupingKey.getValue(), groupingKey.getRef());
        
        BixoPlatform platform = new BixoPlatform(ScoredUrlDatumTest.class, platformMode);
        BasePath path = platform.makePath("build/test/ScoredUrlDatumTest/testCascadingSerialization/in");
        Tap in = platform.makeTap(platform.makeBinaryScheme(FetchSetDatum.FIELDS), path, SinkMode.REPLACE);
        TupleEntryCollector write = in.openForWrite(platform.makeFlowProcess());
        write.add(pfd.getTuple());
        write.close();
    }
}

View Full Code Here

Examples of bixo.config.BixoPlatform.makeBinaryScheme()

        
        createDataFile(platform, unfetchedDatumsPath, unfetchedDatums);


        
        // create a workflow
        Tap inputSource1 = platform.makeTap(platform.makeBinaryScheme(UrlDatum.FIELDS), fetchedDatumsPath);
        Pipe fetchedPipe = new Pipe("fetched");
        Tap inputSource2 = platform.makeTap(platform.makeBinaryScheme(UrlDatum.FIELDS), unfetchedDatumsPath);
        Pipe unfetchedPipe = new Pipe("unfetched");


        Map<String, Tap> sources = new HashMap<String, Tap>();

View Full Code Here

Examples of bixo.config.BixoPlatform.makeBinaryScheme()


        
        // create a workflow
        Tap inputSource1 = platform.makeTap(platform.makeBinaryScheme(UrlDatum.FIELDS), fetchedDatumsPath);
        Pipe fetchedPipe = new Pipe("fetched");
        Tap inputSource2 = platform.makeTap(platform.makeBinaryScheme(UrlDatum.FIELDS), unfetchedDatumsPath);
        Pipe unfetchedPipe = new Pipe("unfetched");


        Map<String, Tap> sources = new HashMap<String, Tap>();
        sources.put(fetchedPipe.getName(), inputSource1);
        sources.put(unfetchedPipe.getName(), inputSource2);

View Full Code Here

Examples of bixo.config.BixoPlatform.makeBinaryScheme()

        Map<String, Tap> sources = new HashMap<String, Tap>();
        sources.put(fetchedPipe.getName(), inputSource1);
        sources.put(unfetchedPipe.getName(), inputSource2);


        BasePath resultsPath = platform.makePath(workingDirPath, "results");
        Tap resultSink = platform.makeTap(platform.makeBinaryScheme(UrlDatum.FIELDS), resultsPath, SinkMode.REPLACE);


        Pipe resultsPipe = new GroupBy("results pipe", Pipe.pipes(fetchedPipe, unfetchedPipe), 
                        new Fields(UrlDatum.URL_FN));
        resultsPipe = new Every(resultsPipe, new LatestUrlDatumBuffer(), Fields.RESULTS);

View Full Code Here

Examples of bixo.config.BixoPlatform.makeBinaryScheme()


        // Input : the crawldb
        platform.assertPathExists(crawlDbPath, "CrawlDb doesn't exist");


        // Our crawl db is defined by the CrawlDbDatum
        Tap inputSource = platform.makeTap(platform.makeBinaryScheme(CrawlDbDatum.FIELDS), crawlDbPath);
        Pipe importPipe = new Pipe("import pipe");


        // Split into tuples that are to be fetched and that have already been fetched
        SplitterAssembly splitter = new SplitterAssembly(importPipe, new SplitFetchedUnfetchedCrawlDatums());

View Full Code Here

Examples of bixo.config.BixoPlatform.makeBinaryScheme()

        Flow flow = flowConnector.connect(sources, resultSink, resultsPipe);
        flow.complete();
        
        // verify that the resulting pipe has the latest tuple
        
        Tap testSink = platform.makeTap(platform.makeBinaryScheme(UrlDatum.FIELDS), resultsPath);
        TupleEntryIterator reader = testSink.openForRead(platform.makeFlowProcess());
        int count = 0;
        long latest = 0;
        while (reader.hasNext()) {
            TupleEntry next = reader.next();

View Full Code Here

Examples of bixo.config.BixoPlatform.makeBinaryScheme()

        //      crawldb
        //      content
        //      parse
        //      status
        BasePath outCrawlDbPath = platform.makePath(curWorkingDirPath, CrawlConfig.CRAWLDB_SUBDIR_NAME);
        Tap loopCrawldbSink = platform.makeTap(platform.makeBinaryScheme(CrawlDbDatum.FIELDS), outCrawlDbPath, SinkMode.REPLACE);


        BasePath contentDirPath = platform.makePath(curWorkingDirPath, CrawlConfig.CONTENT_SUBDIR_NAME);
        Tap contentSink = platform.makeTap(platform.makeBinaryScheme(FetchedDatum.FIELDS), contentDirPath, SinkMode.REPLACE);


        BasePath parseDirPath = platform.makePath(curWorkingDirPath, CrawlConfig.PARSE_SUBDIR_NAME);

View Full Code Here

0 1 2

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.