Package org.apache.nutch.parse

Examples of org.apache.nutch.parse.Parse

51
52
53
54
55
56
57
58
59
60
61
62
    try {
      ParseUtil parser = new ParseUtil(NutchConfiguration.create());
      /* loop through the test documents and validate result */
      for (int t = 0; t < docs.length; t++) {
        Content content = getContent(docs[t]);
        Parse parse = parser.parse(content);
        assertEquals(metalanguages[t], (String) parse.getData().getParseMeta().get(Metadata.LANGUAGE));
      }
    } catch (Exception e) {
      e.printStackTrace(System.out);
      fail(e.toString());
    }
View Full Code Here
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
   * @see #SAMPLE_DIR
   * @throws Exception
   */
  public void testContent() throws Exception {

    Parse parse = new ParseUtil(NutchConfiguration.create())
                        .parseByExtensionId("parse-mspowerpoint", this.content);

    ParseData data = parse.getData();
    String text = parse.getText();

    assertTrue("No content extracted length ==0", text.length() > 0);
   
    this.dumpToFile(this.testFile.getName(), data, text);

View Full Code Here
163
164
165
166
167
168
169
170
171
172
173
174
175
176
   * @see #SAMPLE_DIR
   * @throws Exception
   */
  public void testMeta() throws Exception {

    Parse parse = new ParseUtil(NutchConfiguration.create())
                        .parseByExtensionId("parse-mspowerpoint", content);
   
    ParseData data = parse.getData();

    final FileExtensionFilter titleFilter = new FileExtensionFilter(
        this.testFile.getName() + ".meta");
    final File[] titleFiles = this.sampleDir.listFiles(titleFilter);

View Full Code Here
60
61
62
63
64
65
66
67
68
69
70
71
    byte[] bytes = out.toByteArray();
    Configuration conf = NutchConfiguration.create();

    Content content =
      new Content(url, url, bytes, contentType, new Metadata(), conf);
    Parse parse = new ParseUtil(conf).parseByExtensionId("parse-html",content);

    Metadata metadata = parse.getData().getParseMeta();
    assertEquals(license, metadata.get("License-Url"));
    assertEquals(location, metadata.get("License-Location"));
    assertEquals(type, metadata.get("Work-Type"));
  }
View Full Code Here
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
  public void testId3v2() throws ProtocolException, ParseException {

    String urlString;
    Protocol protocol;
    Content content;
    Parse parse;

    Configuration conf = NutchConfiguration.create();
    urlString = "file:" + sampleDir + fileSeparator + id3v2;
    protocol = new ProtocolFactory(conf).getProtocol(urlString);
    content = protocol.getProtocolOutput(new Text(urlString), new CrawlDatum())
                      .getContent();
    parse = new ParseUtil(conf).parseByExtensionId("parse-mp3", content);
    Metadata metadata = parse.getData().getParseMeta();
    assertEquals("postgresql comment id3v2", metadata.get("COMM-Text"));
    assertEquals("postgresql composer id3v2", metadata.get("TCOM-Text"));
    assertEquals("02", metadata.get("TRCK-Text"));
    assertEquals("http://localhost/", metadata.get("WCOP-URL Link"));
    assertEquals("postgresql artist id3v2", metadata.get("TPE1-Text"));
    assertEquals("(28)", metadata.get("TCON-Text"));
    assertEquals("2004", metadata.get("TYER-Text"));
    assertEquals("postgresql title id3v2", metadata.get("TIT2-Text"));
    assertEquals("postgresql album id3v2", metadata.get("TALB-Text"));
    assertEquals("postgresql encoded by id3v2", metadata.get("TENC-Text"));

    assertEquals("postgresql title id3v2 - "
        + "postgresql album id3v2 - "
        + "postgresql artist id3v2", parse.getData().getTitle());
    assertEquals("http://localhost/", parse.getData().getOutlinks()[0].getToUrl());

  }
View Full Code Here
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
  public void testId3v1() throws ProtocolException, ParseException {

    String urlString;
    Protocol protocol;
    Content content;
    Parse parse;

    Configuration conf = NutchConfiguration.create();
    urlString = "file:" + sampleDir + fileSeparator + id3v1;
    protocol = new ProtocolFactory(conf).getProtocol(urlString);
    content = protocol.getProtocolOutput(new Text(urlString), new CrawlDatum())
                      .getContent();
    parse = new ParseUtil(conf).parseByExtensionId("parse-mp3", content);

    Metadata metadata = parse.getData().getParseMeta();
    assertEquals("postgresql comment id3v1", metadata.get("COMM-Text"));
    assertEquals("postgresql artist id3v1", metadata.get("TPE1-Text"));
    assertEquals("(28)", metadata.get("TCON-Text"));
    assertEquals("2004", metadata.get("TYER-Text"));
    assertEquals("postgresql title id3v1", metadata.get("TIT2-Text"));
    assertEquals("postgresql album id3v1", metadata.get("TALB-Text"));

    assertEquals("postgresql title id3v1 - "
        + "postgresql album id3v1 - "
        + "postgresql artist id3v1", parse.getData().getTitle());

  }
View Full Code Here
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
  public void testNone() throws ProtocolException, ParseException {
    String urlString;
    Protocol protocol;
    Content content;
    Parse parse;

    Configuration conf = NutchConfiguration.create();
    urlString = "file:" + sampleDir + fileSeparator + none;
    protocol = new ProtocolFactory(conf).getProtocol(urlString);
    content = protocol.getProtocolOutput(new Text(urlString), new CrawlDatum())
                      .getContent();
    parse = new ParseUtil(conf).parseByExtensionId("parse-mp3", content);
//    Metadata metadata = parse.getData().getParseMeta();
    if (parse.getData().getStatus().isSuccess()) {
      fail("Expected ParseException");
    }
  }
View Full Code Here
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
  protected void tearDown() {}

  public void testIt() throws ProtocolException, ParseException {
    String urlString;
    Content content;
    Parse parse;
    Configuration conf = NutchConfiguration.create();
    Protocol protocol;
    ProtocolFactory factory = new ProtocolFactory(conf);
    OOParser parser = new OOParser();
    parser.setConf(conf);

    for (int i=0; i<sampleFiles.length; i++) {
      urlString = "file:" + sampleDir + fileSeparator + sampleFiles[i];

      protocol = factory.getProtocol(urlString);
      content = protocol.getProtocolOutput(new Text(urlString), new CrawlDatum()).getContent();

      parse = parser.getParse(content);

      String text = parse.getText().replaceAll("[ \t\r\n]+", " ");
      assertTrue(expectedText.equals(text));
    }
  }
View Full Code Here
160
161
162
163
164
165
166
167
      title = script.substring(0, idx);
    }
    ParseData pd = new ParseData(ParseStatus.STATUS_SUCCESS, title, outlinks,
                                 c.getMetadata());
    pd.setConf(this.conf);
    Parse parse = new ParseImpl(script, pd);
    return parse;
  }
View Full Code Here
344
345
346
347
348
349
350
351
352
353
354
355
356
  private static String getUrlContent(String url, Configuration conf) {
    Protocol protocol;
    try {
      protocol = new ProtocolFactory(conf).getProtocol(url);
      Content content = protocol.getProtocolOutput(new Text(url), new CrawlDatum()).getContent();
      Parse parse = new ParseUtil(conf).parse(content);
      System.out.println("text:" + parse.getText());
      return parse.getText();

    } catch (ProtocolNotFound e) {
      e.printStackTrace();
    } catch (ProtocolException e) {
      e.printStackTrace();
View Full Code Here
TOP

Related Classes of org.apache.nutch.parse.Parse

  • com.atlantbh.nutch.filter.index.omit.OmitIndexingFilterTest
  • com.atlantbh.nutch.filter.xpath.XPathHtmlParserFilterTest
  • com.atlantbh.nutch.filter.xpath.XPathIndexingFilterTest
  • com.atlantbh.nutch.index.alternativedataflow.AlternativeDataFlowIndexingFilterTest
  • com.atlantbh.nutch.index.alternativedataflow.flow.CsvDataFlowTest
  • org.apache.nutch.analysis.lang.HTMLLanguageParser
  • org.apache.nutch.analysis.lang.LanguageIdentifier
  • org.apache.nutch.analysis.lang.TestHTMLLanguageParser
  • org.apache.nutch.indexer.field.BasicFields$Extractor
  • org.apache.nutch.indexer.IndexerMapReduce
    • 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.