Package bixo.datum

Examples of bixo.datum.ParsedDatum

28
29
30
31
32
33
34
35
36
        super(new ParserPolicy());
    }

    @Override
    public ParsedDatum parse(FetchedDatum fetchedDatum) {
        ParsedDatum parsedDatum = new ParsedDatum("url", "127.0.0.1", "someParsedText", "en", "title", new Outlink[0], null);
        parsedDatum.setPayload(fetchedDatum.getPayload());
        return parsedDatum;
    }
View Full Code Here
50
51
52
53
54
55
56
57
58
59
60
        LOGGER.info("Ending creation writable sequence file tuples");
    }

    @Override
    public void operate(FlowProcess flowProcess, FunctionCall<NullContext> funcCall) {
        ParsedDatum datum = new ParsedDatum(funcCall.getArguments());
        Text key = new Text(datum.getUrl());
        Text value = new Text(datum.getTitle() + '\n' + datum.getParsedText());
        Tuple keyVal = new Tuple(key, value);
        funcCall.getOutputCollector().add(keyVal);
    }
View Full Code Here
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
        LOGGER.info("Ending creation of outlink URLs");
    }

    @Override
    public void operate(FlowProcess process, FunctionCall<NullContext> funcCall) {
        ParsedDatum datum = new ParsedDatum(funcCall.getArguments());
        Outlink outlinks[] = datum.getOutlinks();

        // Bump the crawl depth value only on a successful parse
        int crawlDepth = (Integer) datum.getPayloadValue(CrawlDbDatum.CRAWL_DEPTH);
        datum.setPayloadValue(CrawlDbDatum.CRAWL_DEPTH, crawlDepth + 1);

        TupleEntryCollector collector = funcCall.getOutputCollector();

        for (Outlink outlink : outlinks) {
            String url = outlink.getToUrl();
            url = url.replaceAll("[\n\r]", "");
            url = _normalizer.normalize(url);
            if (_validator.isValid(url)) {
                UrlDatum urlDatum = new UrlDatum(url);
                urlDatum.setPayload(datum.getPayload());
                collector.add(urlDatum.getTuple());
            }
        }
    }
View Full Code Here
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
            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: ");
                    String action = readInputLine();
                    if (action.startsWith("e") || (action.length() == 0)) {
                        break;
                        } else if (action.startsWith("d")) {
                            System.out.println("=====================================================================");
                            System.out.println(parsed.getParsedText());
                            System.out.println("=====================================================================");
                        } else if (action.startsWith("b")) {
                            System.out.println("=====================================================================");
                            System.out.println(bpParsed.getParsedText());
                            System.out.println("=====================================================================");
                        } else if (action.startsWith("r")) {
                            System.out.println("=====================================================================");
                            System.out.println(rawParsed.getParsedText());
                            System.out.println("=====================================================================");
                    } else {
                        System.out.println("Unknown command - " + action);
                    }
                }
View Full Code Here
64
65
66
67
68
69
70
71
72
73
74
75
76
77
    ContentBytes content = new ContentBytes(html.getBytes("utf-8"));
    FetchedDatum fetchedDatum = new FetchedDatum(url, url, System.currentTimeMillis(), headers, content, contentType, 0);
   
    // Call parser.parse
    SimpleParser parser = new SimpleParser();
    ParsedDatum parsedDatum = parser.parse(fetchedDatum);
   
    // Verify outlink is correct.
    Outlink[] outlinks = parsedDatum.getOutlinks();
    Assert.assertEquals(2, outlinks.length);
   
    // TODO KKr - reenable this test when Tika parser calls my handler with
    // the <base> element, which is needed to correctly resolve relative links.
    // Assert.assertEquals("http://newdomain.com/link", outlinks[0].getToUrl());
View Full Code Here
96
97
98
99
100
101
102
103
104
105
106
107
108
109
    ContentBytes content = new ContentBytes(html.getBytes("utf-8"));
    FetchedDatum fetchedDatum = new FetchedDatum(url, url, System.currentTimeMillis(), headers, content, contentType, 0);
   
    // Call parser.parse
    SimpleParser parser = new SimpleParser();
    ParsedDatum parsedDatum = parser.parse(fetchedDatum);
   
    // Verify outlink is correct.
    Outlink[] outlinks = parsedDatum.getOutlinks();
    Assert.assertEquals(2, outlinks.length);
   
    Assert.assertEquals("http://newdomain.com/link1", outlinks[0].getToUrl());
    Assert.assertEquals("link1", outlinks[0].getAnchor());
        // TODO KKr - reenable this test when Tika changes are submitted:
View Full Code Here
128
129
130
131
132
133
134
135
136
137
138
139
140
141
    ContentBytes content = new ContentBytes(html.getBytes("utf-8"));
    FetchedDatum fetchedDatum = new FetchedDatum(url, url, System.currentTimeMillis(), headers, content, contentType, 0);
   
    // Call parser.parse
    SimpleParser parser = new SimpleParser();
    ParsedDatum parsedDatum = parser.parse(fetchedDatum);
   
    // Verify outlink is correct.
    Outlink[] outlinks = parsedDatum.getOutlinks();
    Assert.assertEquals(2, outlinks.length);
   
    Assert.assertEquals("http://olddomain.com/redirected/link1", outlinks[0].getToUrl());
    Assert.assertEquals("link1", outlinks[0].getAnchor());
    Assert.assertEquals("http://domain.com/link2", outlinks[1].getToUrl());
View Full Code Here
157
158
159
160
161
162
163
164
165
166
167
168
169
170
    ContentBytes content = new ContentBytes(html.getBytes("utf-8"));
    FetchedDatum fetchedDatum = new FetchedDatum(url, redirectedUrl, System.currentTimeMillis(), headers, content, contentType, 0);
   
    // Call parser.parse
    SimpleParser parser = new SimpleParser();
    ParsedDatum parsedDatum = parser.parse(fetchedDatum);
   
    // Verify outlink is correct.
    Outlink[] outlinks = parsedDatum.getOutlinks();
    Assert.assertEquals(2, outlinks.length);
   
    Assert.assertEquals("http://newdomain.com/link1", outlinks[0].getToUrl());
    Assert.assertEquals("link1", outlinks[0].getAnchor());
    Assert.assertEquals("http://domain.com/link2", outlinks[1].getToUrl());
View Full Code Here
185
186
187
188
189
190
191
192
193
194
195
196
197
198
        ContentBytes content = new ContentBytes(html.getBytes("utf-8"));
        FetchedDatum fetchedDatum = new FetchedDatum(url, url, System.currentTimeMillis(), headers, content, contentType, 0);
       
        // Call parser.parse
        SimpleParser parser = new SimpleParser();
        ParsedDatum parsedDatum = parser.parse(fetchedDatum);
       
        // Verify outlinks are correct (and we only get the a href ones).
        Outlink[] outlinks = parsedDatum.getOutlinks();
        Assert.assertEquals(2, outlinks.length);
       
        Assert.assertEquals("http://newdomain.com/link1", outlinks[0].getToUrl());
        Assert.assertEquals("link1", outlinks[0].getAnchor());
        Assert.assertEquals("http://domain.com/link2", outlinks[1].getToUrl());
View Full Code Here
216
217
218
219
220
221
222
223
224
225
226
227
228
229
        // Call parser.parse
        ParserPolicy policy = new ParserPolicy( ParserPolicy.DEFAULT_MAX_PARSE_DURATION,
                                                BaseLinkExtractor.ALL_LINK_TAGS,
                                                BaseLinkExtractor.ALL_LINK_ATTRIBUTE_TYPES);
        SimpleParser parser = new SimpleParser(policy);
        ParsedDatum parsedDatum = parser.parse(fetchedDatum);
       
        // Verify outlinks are correct (and we only get the a href ones).
        Outlink[] outlinks = parsedDatum.getOutlinks();
        Assert.assertEquals(7, outlinks.length);
       
        Assert.assertEquals("http://newdomain.com/favicon.ico", outlinks[0].getToUrl());
        Assert.assertEquals("http://newdomain.com/link1", outlinks[1].getToUrl());
        Assert.assertEquals("link1", outlinks[1].getAnchor());
View Full Code Here
TOP

Related Classes of bixo.datum.ParsedDatum

  • bixo.examples.crawl.CreateUrlDatumFromOutlinksFunction
  • bixo.examples.crawl.CreateWritableSeqFileData
  • bixo.parser.DOMParser
  • bixo.parser.FakeParser
  • bixo.parser.SimpleParser
  • bixo.parser.SimpleParserTest
  • bixo.parser.TikaCallable
  • bixo.pipes.ParsePipe$ParseFunction
  • bixo.tools.FetchAndParseTool
  • cascading.tuple.Fields
    • 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.