Package org.archive.modules

Examples of org.archive.modules.CrawlURI

68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
    @Override
    protected Collection<TestData> makeData(String content, String uri)
    throws Exception {
        UURI src = UURIFactory.getInstance("http://www.archive.org/start/");
        CrawlURI euri = new CrawlURI(src, null, null, NAVLINK_MISC);
        Recorder recorder = createRecorder(content);
        euri.setContentType("text/css");
        euri.setRecorder(recorder);
        euri.setContentSize(content.length());
       
        // TODO: This test was naively modified to account for the abscense of LINK, but no effort was made to confirm that it is actually testing anything useful
        UURI dest = UURIFactory.getInstance(uri);
        CrawlURI link = euri.createCrawlURI(dest, EMBED_MISC, Hop.EMBED);
        TestData td = new TestData(euri, link);
        return Collections.singleton(td);
    }
View Full Code Here
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
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
    }
   

    @Override
    protected ProcessResult innerProcessResult(CrawlURI puri) {
        CrawlURI curi = (CrawlURI)puri;
       
        // Check if uris should be blocked
        if (getBlockAll()) {
            curi.setFetchStatus(S_BLOCKED_BY_USER);
            return ProcessResult.FINISH;
        }

        // Check if allowed by regular expression
        String regex = getAllowByRegex();
        if (regex != null && !regex.equals("")) {
            if (!TextUtils.matches(regex, curi.toString())) {
                curi.setFetchStatus(S_BLOCKED_BY_USER);
                return ProcessResult.FINISH;
            }
        }

        // Check if blocked by regular expression
        regex = getBlockByRegex();
        if (regex != null && !regex.equals("")) {
            if (TextUtils.matches(regex, curi.toString())) {
                curi.setFetchStatus(S_BLOCKED_BY_USER);
                return ProcessResult.FINISH;
            }
        }

        // Possibly recheck scope
        if (getRecheckScope()) {
            if (!isInScope(curi)) {
                // Scope rejected
                curi.setFetchStatus(S_OUT_OF_SCOPE);
                return ProcessResult.FINISH;
            }
        }
       
        return ProcessResult.PROCEED;
View Full Code Here
332
333
334
335
336
337
338
339
340
341
342
343
344
                if (getExtractorJS() != null) {
                    linkCount += getExtractorJS().considerStrings(ext, curi, url);
                }
            } else {
                int max = ext.getExtractorParameters().getMaxOutlinks();
                CrawlURI relToVia = addRelativeToVia(curi, max, url,
                        LinkContext.EMBED_MISC, Hop.EMBED);
                CrawlURI relToBase = addRelativeToBase(curi, max, url,
                        LinkContext.EMBED_MISC, Hop.EMBED);
                addAnnotations(relToVia, relToBase);
                linkCount++;
            }
        }
View Full Code Here
359
360
361
362
363
364
365
366
367
368
369
370
371
        }

        public void considerStringAsUri(String str) throws IOException {
            if (UriUtils.isVeryLikelyUri(str)) {
                int max = ext.getExtractorParameters().getMaxOutlinks();
                CrawlURI relToVia = addRelativeToVia(curi, max, str,
                        LinkContext.SPECULATIVE_MISC, Hop.SPECULATIVE);
                CrawlURI relToBase = addRelativeToBase(curi, max, str,
                        LinkContext.SPECULATIVE_MISC, Hop.SPECULATIVE);
                addAnnotations(relToVia, relToBase);
                linkCount++;
            }
        }
View Full Code Here
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
     *
     * @throws URIException
     */
    public void testFormsLinkGet() throws URIException {
        UURI uuri = UURIFactory.getInstance("http://www.example.org");
        CrawlURI curi = new CrawlURI(uuri);
        CharSequence cs =
          "<form name=\"testform\" method=\"GET\" action=\"redirect_me?form=true\"> " +
          "  <INPUT TYPE=CHECKBOX NAME=\"checked[]\" VALUE=\"1\" CHECKED> "+
          "  <INPUT TYPE=CHECKBOX NAME=\"unchecked[]\" VALUE=\"1\"> " +
          "  <select name=\"selectBox\">" +
          "    <option value=\"selectedOption\" selected>option1</option>" +
          "    <option value=\"nonselectedOption\">option2</option>" +
          "  </select>" +
          "  <input type=\"submit\" name=\"test\" value=\"Go\">" +
          "</form>";  
        getExtractor().extract(curi, cs);
        curi.getOutLinks();
        assertTrue(CollectionUtils.exists(curi.getOutLinks(), new Predicate() {
            public boolean evaluate(Object object) {
                return ((CrawlURI) object).getURI().indexOf(
                        "/redirect_me?form=true&checked[]=1&unchecked[]=&selectBox=selectedOption&test=Go")>=0;
            }
        }));
View Full Code Here
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
     *
     * @throws URIException
     */
    public void testFormsLinkIgnorePost() throws URIException {
        UURI uuri = UURIFactory.getInstance("http://www.example.org");
        CrawlURI curi = new CrawlURI(uuri);
        CharSequence cs =
            "<form name=\"testform\" method=\"POST\" action=\"redirect_me?form=true\"> " +
            "  <INPUT TYPE=CHECKBOX NAME=\"checked[]\" VALUE=\"1\" CHECKED> "+
            "  <INPUT TYPE=CHECKBOX NAME=\"unchecked[]\" VALUE=\"1\"> " +
            "  <select name=\"selectBox\">" +
            "    <option value=\"selectedOption\" selected>option1</option>" +
            "    <option value=\"nonselectedOption\">option2</option>" +
            "  </select>" +
            "  <input type=\"submit\" name=\"test\" value=\"Go\">" +
            "</form>";  
        getExtractor().extract(curi, cs);
        curi.getOutLinks();
        assertTrue(!CollectionUtils.exists(curi.getOutLinks(), new Predicate() {
            public boolean evaluate(Object object) {
                return ((CrawlURI) object).getURI().indexOf(
                        "/redirect_me?form=true&checked[]=1&unchecked[]=&selectBox=selectedOption&test=Go")>=0;
            }
        }));
View Full Code Here
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
     *
     * @throws URIException
     */
    public void testFormsLinkFindPost() throws URIException {
        UURI uuri = UURIFactory.getInstance("http://www.example.org");
        CrawlURI curi = new CrawlURI(uuri);
        CharSequence cs =
            "<form name=\"testform\" method=\"POST\" action=\"redirect_me?form=true\"> " +
            "  <INPUT TYPE=CHECKBOX NAME=\"checked[]\" VALUE=\"1\" CHECKED> "+
            "  <INPUT TYPE=CHECKBOX NAME=\"unchecked[]\" VALUE=\"1\"> " +
            "  <select name=\"selectBox\">" +
            "    <option value=\"selectedOption\" selected>option1</option>" +
            "    <option value=\"nonselectedOption\">option2</option>" +
            "  </select>" +
            "  <input type=\"submit\" name=\"test\" value=\"Go\">" +
            "</form>";
        getExtractor().setExtractOnlyFormGets(false);
        getExtractor().extract(curi, cs);
        curi.getOutLinks();
        assertTrue(CollectionUtils.exists(curi.getOutLinks(), new Predicate() {
            public boolean evaluate(Object object) {
                return ((CrawlURI) object).getURI().indexOf(
                        "/redirect_me?form=true&checked[]=1&unchecked[]=&selectBox=selectedOption&test=Go")>=0;
            }
        }));
View Full Code Here
136
137
138
139
140
141
142
143
144
145
        }));
    }
   
    public void testMultipleAttributesPerElement() throws URIException {
        UURI uuri = UURIFactory.getInstance("http://www.example.org");
        CrawlURI curi = new CrawlURI(uuri);
        CharSequence cs = "<a src=\"http://www.example.com/\" href=\"http://www.archive.org/\"> ";
        getExtractor().extract(curi, cs);
        assertTrue("not all links found", curi.getOutLinks().size() == 2);
    }
View Full Code Here
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
    @Override
    protected Collection<TestData> makeData(String content, String destURI)
    throws Exception {
        List<TestData> result = new ArrayList<TestData>();
        UURI src = UURIFactory.getInstance("http://www.archive.org/start/");
        CrawlURI euri = new CrawlURI(src, null, null,
            LinkContext.SPECULATIVE_MISC);
        Recorder recorder = createRecorder(content, "UTF-8");
        euri.setContentType("text/xml");
        euri.setRecorder(recorder);
        euri.setContentSize(content.length());
               
        UURI dest = UURIFactory.getInstance(destURI);
        CrawlURI link = euri.createCrawlURI(dest, LinkContext.SPECULATIVE_MISC, Hop.SPECULATIVE);
        result.add(new TestData(euri, link));
       
        return result;
    }
View Full Code Here
129
130
131
132
133
134
135
136
137
138
139
            while ( true ) {
                ArchiveUtils.continueCheck();
               
                setStep(Step.ABOUT_TO_GET_URI, null);

                CrawlURI curi = controller.getFrontier().next();
               
               
                synchronized(this) {
                    ArchiveUtils.continueCheck();
                    setCurrentCuri(curi);
View Full Code Here
TOP

Related Classes of org.archive.modules.CrawlURI

  • org.archive.crawler.datamodel.CandidateURITest
  • org.archive.crawler.datamodel.CrawlURITest
  • org.archive.crawler.deciderules.ClassKeyMatchesRegexDecideRule
  • org.archive.crawler.framework.ToeThread
  • org.archive.crawler.frontier.AbstractFrontier
  • org.archive.crawler.frontier.AMQPUrlReceiver$UrlConsumer
  • org.archive.crawler.frontier.BdbMultipleWorkQueues
  • org.archive.crawler.frontier.BdbMultipleWorkQueuesTest
  • org.archive.crawler.frontier.BdbWorkQueue
  • org.archive.crawler.frontier.FrontierJournal
    • 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.