Package org.apache.nutch.parse

Examples of org.apache.nutch.parse.Outlink

400
401
402
403
404
405
406
407
408
409
410
            if (target != null && !noFollow && !post)
              try {
               
                URL url = (base.toString().indexOf(';') > 0) ?
                  fixEmbeddedParams(base, target) :  new URL(base, target);
                outlinks.add(new Outlink(url.toString(),
                                         linkText.toString().trim()));
              } catch (MalformedURLException e) {
                // don't care
              }
          }
View Full Code Here
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
      testDOMs[i]= node;
    }
    try {
     answerOutlinks = new Outlink[][]{
         {
           new Outlink("http://www.nutch.org", "anchor"),
         },
         {
           new Outlink("http://www.nutch.org/", "home"),
           new Outlink("http://www.nutch.org/docs/bot.html", "bots"),
         },
         {
           new Outlink("http://www.nutch.org/", "separate this"),
           new Outlink("http://www.nutch.org/docs/ok", "from this"),
         },
         {
           new Outlink("http://www.nutch.org/", "home"),
           new Outlink("http://www.nutch.org/docs/1", "1"),
           new Outlink("http://www.nutch.org/docs/2", "2"),
         },
         {
           new Outlink("http://www.nutch.org/frames/top.html", ""),
           new Outlink("http://www.nutch.org/frames/left.html", ""),
           new Outlink("http://www.nutch.org/frames/invalid.html", ""),
           new Outlink("http://www.nutch.org/frames/right.html", ""),
         },
         {
           new Outlink("http://www.nutch.org/maps/logo.gif", ""),
           new Outlink("http://www.nutch.org/index.html", ""),
           new Outlink("http://www.nutch.org/maps/#bottom", ""),
           new Outlink("http://www.nutch.org/bot.html", ""),
           new Outlink("http://www.nutch.org/docs/index.html", ""),
         },
         {
             new Outlink("http://www.nutch.org/index.html", "whitespace test"),
         },
         {
         },
         {
           new Outlink("http://www.nutch.org/dummy.jsp", "test2"),
         },
         {
         },
         {
           new Outlink("http://www.nutch.org/;x", "anchor1"),
           new Outlink("http://www.nutch.org/g;x", "anchor2"),
           new Outlink("http://www.nutch.org/g;x?y#s", "anchor3")
         },
         {
           new Outlink("http://www.nutch.org/g;something", "anchor1"),
           new Outlink("http://www.nutch.org/g;something?y#s", "anchor2"),
           new Outlink("http://www.nutch.org/;something?y=1", "anchor3"),
           new Outlink("http://www.nutch.org/;something?y=1#s", "anchor4"),
           new Outlink("http://www.nutch.org/?y=1;somethingelse", "anchor5")
         }
      };
  
    } catch (MalformedURLException e) {
       
View Full Code Here
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
      fromUrlCriginalColectionName=SqlSearcher.getCollectionNameOriginal(collection);     
      fromUrlTimestamp=SqlSearcher.getTimestampOriginal(collection);     
    }
   
    for (int i = 0; i < outlinks.length; i++) {
      Outlink outlink = outlinks[i];
      String toUrl = outlink.getToUrl();
          
      if (this.nwIgnoreInternalLinks)
      {
        String toHost = getHost(toUrl);
       
        if (toHost == null || toHost.equals(fromHost)) { // internal link             
          continue;                               // skip it
        }
      }

      if (this.nwUrlNormalizers != null)
      {
        try {         
          toUrl = this.nwUrlNormalizers. normalize(toUrl, URLNormalizers.SCOPE_LINKDB);
        }
        catch (Exception e) {
          LOG.warn("Skipping " + toUrl + ":" + e);
          toUrl = null;
        }
      }
     
      if (toUrl != null && this.nwUrlFilters != null) {
        try {
          toUrl = this.nwUrlFilters.filter(toUrl); // filter the url
          if (toUrl==null) { 
            LOG.info("LINKDB URL FILTERED"); 
          }
        }
        catch (Exception e) {
          LOG.warn("Skipping " + toUrl + ":" + e);
          toUrl = null;
        }
      }
   
      if (toUrl == null) {
        continue;
      }

      inlinks.clear();
   
      String anchor = outlink.getAnchor();        // truncate long anchors
      
      if (anchor.length() > this.nwMaxAnchorLength) {
        anchor = anchor.substring(0, this.nwMaxAnchorLength);
      }
View Full Code Here
355
356
357
358
359
360
361
362
363
364
365
            }
            if (target != null && !noFollow && !post)
              try {
               
                URL url = URLUtil.resolveURL(base, target);
                outlinks.add(new Outlink(url.toString(),
                                         linkText.toString().trim()));
              } catch (MalformedURLException e) {
                // don't care
              }
          }
View Full Code Here
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
      testDOMs[i]= node;
    }
    try {
      answerOutlinks = new Outlink[][]{
          {
            new Outlink("http://www.nutch.org", "anchor"),
          },
          {
            new Outlink("http://www.nutch.org/", "home"),
            new Outlink("http://www.nutch.org/docs/bot.html", "bots"),
          },
          {
            new Outlink("http://www.nutch.org/", "separate this"),
            new Outlink("http://www.nutch.org/docs/ok", "from this"),
          },
          {
            new Outlink("http://www.nutch.org/", "home"),
            new Outlink("http://www.nutch.org/docs/1", "1"),
            new Outlink("http://www.nutch.org/docs/2", "2"),
          },
          {
            new Outlink("http://www.nutch.org/frames/top.html", ""),
            new Outlink("http://www.nutch.org/frames/left.html", ""),
            new Outlink("http://www.nutch.org/frames/invalid.html", ""),
            new Outlink("http://www.nutch.org/frames/right.html", ""),
          },
          {
            new Outlink("http://www.nutch.org/maps/logo.gif", ""),
            new Outlink("http://www.nutch.org/index.html", ""),
            new Outlink("http://www.nutch.org/maps/#bottom", ""),
            new Outlink("http://www.nutch.org/bot.html", ""),
            new Outlink("http://www.nutch.org/docs/index.html", ""),
          },
          {
            new Outlink("http://www.nutch.org/index.html", "whitespace test"),
          },
          {
          },
          {
            new Outlink("http://www.nutch.org/dummy.jsp", "test2"),
          },
          {
          },
          {
            new Outlink("http://www.nutch.org/;x", "anchor1"),
            new Outlink("http://www.nutch.org/g;x", "anchor2"),
            new Outlink("http://www.nutch.org/g;x?y#s", "anchor3")
          },
          {
            // this is tricky - see RFC3986 section 5.4.1 example 7
            new Outlink("http://www.nutch.org/g", "anchor1"),
            new Outlink("http://www.nutch.org/g?y#s", "anchor2"),
            new Outlink("http://www.nutch.org/;something?y=1", "anchor3"),
            new Outlink("http://www.nutch.org/;something?y=1#s", "anchor4"),
            new Outlink("http://www.nutch.org/;something?y=1;somethingelse", "anchor5")
          },
          {
            new Outlink("http://www.nutch.org/g", ""),
            new Outlink("http://www.nutch.org/g1", ""),
            new Outlink("http://www.nutch.org/g2", "bla bla"),
            new Outlink("http://www.nutch.org/test.gif", "bla bla"),
          }
      };

    } catch (MalformedURLException e) {
View Full Code Here
92
93
94
95
96
97
98
99
100
101
102
            Parse parse = new ParseUtil(this.conf).parse(content).get(content.getUrl());
            ParseData theParseData = parse.getData();
            Outlink[] theOutlinks = theParseData.getOutlinks();
           
            for(int count = 0; count < theOutlinks.length; count++) {
              outLinksList.add(new Outlink(theOutlinks[count].getToUrl(), theOutlinks[count].getAnchor()));
            }
           
            resultText += entry.getName() + " " + parse.getText() + " ";
          } catch (ParseException e) {
            if (LOG.isInfoEnabled()) {
View Full Code Here
58
59
60
61
62
63
64
65
66
67
68
    Assert.assertNotNull(filter);

    NutchDocument doc = new NutchDocument();

    String title = "The Foo Page";
    Outlink[] outlinks = new Outlink[] { new Outlink("http://foo.com/", "Foo") };
    Metadata metaData = new Metadata();
    metaData.add("Language", "en/us");
    ParseData parseData = new ParseData(ParseStatus.STATUS_SUCCESS, title, outlinks, metaData);
    ParseImpl parse = new ParseImpl("this is a sample foo bar page. hope you enjoy it.", parseData);
View Full Code Here
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
        Map<String, String> outlinkMap = new LinkedHashMap<String, String>();

        // normalize urls and put into map
        if (outlinkAr != null && outlinkAr.length > 0) {
          for (int i = 0; i < outlinkAr.length; i++) {
            Outlink outlink = outlinkAr[i];
            String toUrl = normalizeUrl(outlink.getToUrl());

            if (filterUrl(toUrl) == null) {
              continue;
            }

            // only put into map if the url doesn't already exist in the map or
            // if it does and the anchor for that link is null, will replace if
            // url is existing
            boolean existingUrl = outlinkMap.containsKey(toUrl);
            if (toUrl != null
              && (!existingUrl || (existingUrl && outlinkMap.get(toUrl) == null))) {
              outlinkMap.put(toUrl, outlink.getAnchor());
            }
          }
        }

        // collect the outlinks under the fetch time
View Full Code Here
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
        Assert.assertTrue("caught exception: " + e, false);
      }
      testDOMs[i] = node;
    }
    answerOutlinks = new Outlink[][] {
        { new Outlink("http://www.nutch.org", "anchor"), },
        { new Outlink("http://www.nutch.org/", "home"),
          new Outlink("http://www.nutch.org/docs/bot.html", "bots"), },
          { new Outlink("http://www.nutch.org/", "separate this"),
            new Outlink("http://www.nutch.org/docs/ok", "from this"), },
            { new Outlink("http://www.nutch.org/", "home"),
              new Outlink("http://www.nutch.org/docs/1", "1"),
              new Outlink("http://www.nutch.org/docs/2", "2"), },
              { new Outlink("http://www.nutch.org/frames/top.html", ""),
                new Outlink("http://www.nutch.org/frames/left.html", ""),
                new Outlink("http://www.nutch.org/frames/invalid.html", ""),
                new Outlink("http://www.nutch.org/frames/right.html", ""), },
                { new Outlink("http://www.nutch.org/maps/logo.gif", ""),
                  new Outlink("http://www.nutch.org/index.html", ""),
                  new Outlink("http://www.nutch.org/maps/#bottom", ""),
                  new Outlink("http://www.nutch.org/bot.html", ""),
                  new Outlink("http://www.nutch.org/docs/index.html", ""), },
                  { new Outlink("http://www.nutch.org/index.html", "whitespace test"), },
                  {},
                  { new Outlink("http://www.nutch.org/dummy.jsp", "test2"), },
                  {},
                  { new Outlink("http://www.nutch.org/;x", "anchor1"),
                    new Outlink("http://www.nutch.org/g;x", "anchor2"),
                    new Outlink("http://www.nutch.org/g;x?y#s", "anchor3") },
                    {
                      // this is tricky - see RFC3986 section 5.4.1 example 7
                      new Outlink("http://www.nutch.org/g", "anchor1"),
                      new Outlink("http://www.nutch.org/g?y#s", "anchor2"),
                      new Outlink("http://www.nutch.org/;something?y=1", "anchor3"),
                      new Outlink("http://www.nutch.org/;something?y=1#s", "anchor4"),
                      new Outlink("http://www.nutch.org/;something?y=1;somethingelse",
                          "anchor5") } };

  }
View Full Code Here
230
231
232
233
234
235
236
237
238
239
240
        }
        url = url.replaceAll("&amp;", "&");
        if (LOG.isTraceEnabled()) {
          LOG.trace(" - outlink from JS: '" + url + "'");
        }
        outlinks.add(new Outlink(url, anchor));
      }
    } catch (Exception ex) {
      // if it is a malformed URL we just throw it away and continue with
      // extraction.
      if (LOG.isErrorEnabled()) { LOG.error("getJSLinks", ex); }
View Full Code Here
TOP

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

  • org.apache.hadoop.io.MapWritable
  • org.apache.nutch.indexer.basic.TestBasicIndexingFilter
  • org.apache.nutch.net.UrlNormalizerFactory
  • org.apache.nutch.net.URLNormalizers
  • org.apache.nutch.parse.html.DOMContentUtils
  • org.apache.nutch.parse.html.TestDOMContentUtils
  • org.apache.nutch.parse.js.JSParseFilter
  • org.apache.nutch.parse.mp3.MetadataCollector
  • org.apache.nutch.parse.rss.RSSParser
  • org.apache.nutch.parse.tika.DOMContentUtils
    • 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.