Examples of WARCRecord

edu.cmu.lemurproject.WarcRecord
@author mhoy
it.unimi.dsi.law.warc.io.WarcRecord
net.bpiwowar.mg4j.extensions.warc.WarcRecord
org.archive.io.warc.WARCRecord
org.jwat.warc.WarcRecord
This class represents a parsed WARC record header block including possible validation and format warnings/errors encountered in the process. The payload of the WARC record is accessible through a wrapped payload object. @author nicl

Examples of org.archive.io.warc.WARCRecord

        String ctype = "text/plain";
        WARCRecordInfo recinfo = new TestWARCRecordInfo(
                TestWARCRecordInfo.buildCompressedHttpResponseBlock(ctype,
                        payload.getBytes(), true));
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        WarcResource res = new WarcResource(rec, ar);
        res.parseHeaders();
        
        assertEquals("statusCode", 200, res.getStatusCode());
        assertEquals("content-type", ctype, res.getHeader("Content-Type"));

View Full Code Here

Examples of org.archive.io.warc.WARCRecord

        final String ct = "image/gif";
        WARCRecordInfo recinfo = new TestWARCRecordInfo(block);
        recinfo.setType(WARCRecordType.metadata);
        recinfo.setMimetype(ct);
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        WarcResource res = new WarcResource(rec, ar);
        // must not fail
        res.parseHeaders();
        
        // should return assumed 200

View Full Code Here

Examples of org.archive.io.warc.WARCRecord

     */
    public void testRevisitRecord() throws Exception {
        final String ct = "text/html";
        WARCRecordInfo recinfo = TestWARCRecordInfo.createRevisitHttpResponse(ct, 1345);
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        WarcResource res = new WarcResource(rec, ar);
        res.parseHeaders();
        
        // these are from this record.
        assertEquals("statusCode", 200, res.getStatusCode());

View Full Code Here

Examples of org.archive.io.warc.WARCRecord

     */
    public void testOldRevisitRecord() throws Exception {
        final String ct = "text/html";
        WARCRecordInfo recinfo = TestWARCRecordInfo.createRevisitHttpResponse(ct, 1345, false);
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        WarcResource res = new WarcResource(rec, ar);
        res.parseHeaders();
        
        // should either return special value or throw appropriate exception (TBD)
        int scode = res.getStatusCode();

View Full Code Here

Examples of org.archive.io.warc.WARCRecord

        WARCRecordInfo recinfo = new TestWARCRecordInfo(block);
        recinfo.setType(WARCRecordType.resource);
        recinfo.setUrl("ftp://ftp.example.com/afile.txt");
        recinfo.setMimetype(ct);
        ArchiveReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = (WARCRecord)ar.get(0);
        WarcResource res = new WarcResource(rec, ar);
        res.parseHeaders();
                
        int scode = res.getStatusCode();
        assertEquals("statusCode", 200, scode);

View Full Code Here

Examples of org.archive.io.warc.WARCRecord

  public static Resource createTestTextResource(String ctype, byte[] payloadBytes)
      throws IOException {
    WARCRecordInfo recinfo = TestWARCRecordInfo.createHttpResponse(ctype,
      payloadBytes);
    TestWARCReader ar = new TestWARCReader(recinfo);
    WARCRecord rec = ar.get(0);
    WarcResource resource = new WarcResource(rec, ar);
    resource.parseHeaders();
    return resource;
  }

View Full Code Here

Examples of org.archive.io.warc.WARCRecord

    /* (non-Javadoc)
     * @see org.archive.wayback.util.Adapter#adapt(java.lang.Object)
     */
                @Override
    public WARCRecord adapt(ArchiveRecord o) {
      WARCRecord rec = null;
      if (o instanceof WARCRecord) {
        rec = (WARCRecord) o;
      }
      return rec;
    }

View Full Code Here

Examples of org.archive.io.warc.WARCRecord

    public static Resource createTestHtmlResource(String uri, String timestamp, byte[] payloadBytes) throws IOException {
        TestWARCRecordInfo recinfo = TestWARCRecordInfo.createCompressedHttpResponse("text/html", payloadBytes);
        recinfo.setCreate14DigitDateFromDT14(timestamp);
        if (uri != null) recinfo.setUrl(uri);
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        WarcResource resource = new WarcResource(rec, ar);
        resource.parseHeaders();
        return resource;
    }

View Full Code Here

Examples of org.archive.io.warc.WARCRecord

    }
    public static Resource createTestRevisitResource(String timestamp, int len, boolean withHeader) throws IOException {
        TestWARCRecordInfo recinfo = TestWARCRecordInfo.createRevisitHttpResponse("text/html", len, withHeader);
        recinfo.setCreate14DigitDateFromDT14(timestamp);
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        WarcResource resource = new WarcResource(rec, ar);
        resource.parseHeaders();
        return resource;
    }

View Full Code Here

Examples of org.archive.io.warc.WARCRecord

        byte[] failPayload = "failed\n".getBytes("UTF-8");
        byte[] content = TestWARCRecordInfo.buildHttpResponseBlock("502 Bad Gateway", "text/plain", failPayload);
  
        TestWARCRecordInfo recinfo = new TestWARCRecordInfo(content);
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        WarcResource resource = new WarcResource(rec, ar);
        resource.parseHeaders();
        return resource;
    }

View Full Code Here

0 1 2 3

TOP

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.