Package org.apache.nutch.parse

Examples of org.apache.nutch.parse.Outlink

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
125
126
127
128
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
163
164
165
166
167
168
                if (r.getLink() != null) {
                    try {
                        // get the outlink
      if (r.getDescription()!= null ) {
          theOutlinks.add(new Outlink(r.getLink(), r.getDescription(), getConf()));
      } else {
          theOutlinks.add(new Outlink(r.getLink(), "", getConf()));
      }
                    } catch (MalformedURLException e) {
                        if (LOG.isWarnEnabled()) {
                          LOG.warn("MalformedURL: " + r.getLink());
                          LOG.warn("Attempting to continue processing outlinks");
                          e.printStackTrace(LogUtil.getWarnStream(LOG));
                        }
                        continue;
                    }
                }

                // now get the descriptions of all the underlying RSS Items and
                // then index them too
                for (int j = 0; j < r.getItems().size(); j++) {
                    RSSItem theRSSItem = (RSSItem) r.getItems().get(j);
                    indexText.append(theRSSItem.getDescription());
                    indexText.append(" ");

                    String whichLink = null;

                    if (theRSSItem.getPermalink() != null)
                        whichLink = theRSSItem.getPermalink();
                    else
                        whichLink = theRSSItem.getLink();

                    if (whichLink != null) {
                        try {
          if (theRSSItem.getDescription()!=null) {
        theOutlinks.add(new Outlink(whichLink, theRSSItem.getDescription(), getConf()));
          } else {
        theOutlinks.add(new Outlink(whichLink, "", getConf()));
          }
                        } catch (MalformedURLException e) {
                            if (LOG.isWarnEnabled()) {
                              LOG.warn("MalformedURL: " + whichLink);
                              LOG.warn("Attempting to continue processing outlinks");
View Full Code Here
102
103
104
105
106
107
108
109
110
111
112
            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
227
228
229
230
231
232
233
234
235
236
237
        }
        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
235
236
237
238
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
          + "'");

    }
    answerOutlinks = new Outlink[][] {
        // 0
        { new Outlink("http://www.nutch.org", "anchor"), },
        // 1
        {
          new Outlink("http://www.nutch.org/", "home"),
          new Outlink("http://www.nutch.org/docs/bot.html",
                "bots"), },
        // 2
        {
          new Outlink("http://www.nutch.org/", "separate this"),
          new Outlink("http://www.nutch.org/docs/ok", "from this"), },
       
        // 3 
        {   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"), },
        // 4 
        {
          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",""),
        },
        // 5
        {
          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", "")
        },
        // 6
        { new Outlink("http://www.nutch.org/index.html",
            "whitespace test"),
        },
        // 7
        {},
        // 8
        { new Outlink("http://www.nutch.org/dummy.jsp", "test2"), },
        // 9
        {},
        // 10
        {
         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")
        },
        // 11
        {
         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") }
        };

  }
View Full Code Here
397
398
399
400
401
402
403
404
405
406
407
            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
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
100
101
102
103
104
105
106
107
108
109
110
            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
398
399
400
401
402
403
404
405
406
407
408
            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
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.