Package bixo.parser

Examples of bixo.parser.SimpleParser

191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
        Pipe statusPipe = new Pipe("status pipe", fetchPipe.getStatusTailPipe());
        Pipe contentPipe = new Pipe("content pipe", fetchPipe.getContentTailPipe());
        contentPipe = TupleLogger.makePipe(contentPipe, true);
       
        // Take content and split it into content output plus parse to extract URLs.
        SimpleParser parser;
        if (options.isUseBoilerpipe()) {
            parser = new SimpleParser(new BoilerpipeContentExtractor(), new SimpleLinkExtractor(), new ParserPolicy());
        } else if (options.isGenerateHTML()) {
            parser = new SimpleParser(new HtmlContentExtractor(), new SimpleLinkExtractor(), new ParserPolicy(), true);
        } else {
            parser = new SimpleParser();
        }
       
        parser.setExtractLanguage(false);
        ParsePipe parsePipe = new ParsePipe(contentPipe, parser);

       
        // Create the output map that connects each tail pipe to the appropriate sink, and the
        // list of tail pipes.
View Full Code Here
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
        SimpleHttpFetcher fetcher = new SimpleHttpFetcher(1, policy, new FirefoxUserAgent());
        fetcher.setMaxRetryCount(options.getMaxRetries());
       
        // Give a long timeout for parsing
        ParserPolicy parserPolicy = new ParserPolicy(MAX_PARSE_DURATION);
        SimpleParser parser = new SimpleParser(parserPolicy);

        SimpleParser rawParser = new SimpleParser(parserPolicy, true);
       
        // Create Boilperpipe content extractor
        SimpleParser bpParser = new SimpleParser(new BoilerpipeContentExtractor(), new NullLinkExtractor(), parserPolicy);
       
        if (options.isTraceLogging()) {
            Logger.getRootLogger().setLevel(Level.TRACE);
            System.setProperty("bixo.root.level", "TRACE");
        }
       
        String urls[] = options.getUrls() == null ? null : options.getUrls().split(",");
        boolean interactive = (urls == null);
        int index = 0;
       
        while (interactive || (index < urls.length)) {
          String url;
         
          try {
              if (interactive) {
                System.out.print("URL to fetch: ");
                url = readInputLine();
                if (url.length() == 0) {
                  System.exit(0);
                }
              } else {
                url = args[index++];
              }

              System.out.println("Fetching " + url);
            FetchedDatum result = fetcher.get(new ScoredUrlDatum(url));
            System.out.println(String.format("Fetched %s: headers = %s", result.getUrl(), result.getHeaders()));
            System.out.flush();
           
            // System.out.println("Result = " + result.toString());
            ParsedDatum parsed = parser.parse(result);
            System.out.println(String.format("Parsed %s: lang = %s, size = %d", parsed.getUrl(),
                            parsed.getLanguage(), parsed.getParsedText().length()));
           
            ParsedDatum bpParsed = bpParser.parse(result);
            ParsedDatum rawParsed = rawParser.parse(result);
           
            if (interactive) {
                while (true) {
                    System.out.print("Next action - (d)ump regular, dump (b)oilerpipe, dump (r)aw, (e)xit: ");
View Full Code Here
53
54
55
56
57
58
59
60
61
62
63
        BixoPlatform platform = new BixoPlatform(ParsePipeTest.class, Platform.Local);
       

        Pipe pipe = new Pipe("parse_source");
        ParsePipe parserPipe = new ParsePipe(pipe, new SimpleParser());
        BasePath inputPath = platform.makePath("build/test/ParserPipeTest/in");
        Tap in = platform.makeTap(platform.makeBinaryScheme(FetchedDatum.FIELDS), inputPath);
        BasePath outputPath = platform.makePath("build/test/ParserPipeTest/out");
        Tap out = platform.makeTap(platform.makeBinaryScheme(ParsedDatum.FIELDS), outputPath, SinkMode.REPLACE);
View Full Code Here
240
241
242
243
244
245
246
247
248
249
250
        Pipe statusPipe = new Pipe("status pipe", fetchPipe.getStatusTailPipe());
        Pipe contentPipe = new Pipe("content pipe", fetchPipe.getContentTailPipe());
        contentPipe = TupleLogger.makePipe(contentPipe, true);

        // Create a parser that returns back the raw HTML (cleaned up by Tika) as the parsed content.
        SimpleParser parser = new SimpleParser(new ParserPolicy(), true);
        ParsePipe parsePipe = new ParsePipe(fetchPipe.getContentTailPipe(), parser);
       
        Pipe analyzerPipe = new Pipe("analyzer pipe");
        analyzerPipe = new Each(parsePipe.getTailPipe(), new AnalyzeHtml());
       
View Full Code Here
93
94
95
96
97
98
99
            super.cleanup(flowProcess, operationCall);
        }
    }

    public ParsePipe(Pipe fetcherPipe) {
        this(fetcherPipe, new SimpleParser());
    }
View Full Code Here
TOP

Related Classes of bixo.parser.SimpleParser

  • bixo.config.ParserPolicy
  • bixo.datum.ParsedDatum
  • bixo.examples.crawl.DemoCrawlWorkflow
  • bixo.examples.webmining.DemoWebMiningWorkflow
  • bixo.pipes.ParsePipe
  • bixo.pipes.ParsePipeTest
  • bixo.tools.FetchAndParseTool
  • org.apache.tika.metadata.Metadata
  • org.apache.tika.parser.AutoDetectParser
  • org.apache.tika.parser.ParseContext
    • Copyright © 2018 www.massapicom. All rights reserved.
    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.