Examples of org.archive.io.warc.TestWARCReader

org.archive.io.warc.TestWARCReader
Fixture WARCReader.
It works as ArchiveReader reading from WARC file with just one WARC record at offset 0 (there's no "warcinfo" record).

Content of the record is customized through {@link WARCRecordInfo}. ( {@link TestWARCRecordInfo} offers commonly-used default values and convenient factorymethods.

Typical test code would be:
```
 String payload = "hogehogehogehogehoge"; WARCRecordInfo recinfo = TestWARCRecordInfo.createHttpResponse(payload); TestWARCReader ar = new TestWARCReader(recinfo); WARCRecord rec = (WARCRecord)ar.get(0); 
```
@contributor kenji

        result.setCaptureTimestamp("20100101123456");
    }
    
    public static Resource createTestHtmlResource(byte[] payloadBytes) throws IOException {
        WARCRecordInfo recinfo = TestWARCRecordInfo.createCompressedHttpResponse("text/html", 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

    super.setUp();
  }
  
  public static Resource createTestJSResource(byte[] payloadBytes) throws IOException {
    WARCRecordInfo recinfo = TestWARCRecordInfo.createHttpResponse("text/javascript", 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

        result.setOriginalUrl("http://www.example.com/");
    }
    
    public static Resource createTestHtmlResource(byte[] payloadBytes) throws IOException {
        WARCRecordInfo recinfo = TestWARCRecordInfo.createCompressedHttpResponse("text/html", 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

        return resource;
    }
    public static Resource createTestRevisitResource(byte[] payloadBytes, boolean withHeader, boolean gzipContent) throws IOException {
        WARCRecordInfo recinfo = TestWARCRecordInfo.createRevisitHttpResponse(
                "text/html", payloadBytes.length, withHeader, gzipContent);
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        WarcResource resource = new WarcResource(rec, ar);
        resource.parseHeaders();
        return resource;
    }

View Full Code Here

        String payload = "hogehogehogehogehoge";
        String ctype = "text/plain";
        WARCRecordInfo recinfo = new TestWARCRecordInfo(
                TestWARCRecordInfo.buildCompressedHttpResponseBlock(ctype,
                        payload.getBytes()));
        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

        String payload = "hogehogehogehogehoge";
        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

        };
        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

     * @throws Exception
     */
    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

     * @see StandardCharsetDetector#getCharset(org.archive.wayback.core.Resource, org.archive.wayback.core.Resource, org.archive.wayback.core.WaybackRequest)
     */
    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

        final byte[] block = "blahblahblah\n".getBytes();
        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

0 1

TOP

Related Classes of org.archive.io.warc.TestWARCReader

com.google.common.io.CountingInputStream

org.archive.util.anvl.ANVLRecord

org.archive.wayback.archivalurl.ArchivalUrlCSSReplayRendererTest

org.archive.wayback.archivalurl.ArchivalURLJSStringTransformerReplayRendererTest

org.archive.wayback.archivalurl.ArchivalUrlSAXRewriteReplayRendererTest

org.archive.wayback.replay.SelectorReplayDispatcherTest

org.archive.wayback.replay.TransparentReplayRendererTest

org.archive.wayback.resourcestore.resourcefile.WarcResourceTest

org.archive.wayback.webapp.AccessPointTest

java.io.OutputStreamWriter

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.