Examples of 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

Examples of org.archive.io.warc.TestWARCReader


  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.TestWARCReader

     * @throws Exception
     */
    public void testPlainHttpRecord() throws Exception {
        String payload = "hogehogehogehogehoge";
        WARCRecordInfo recinfo = TestWARCRecordInfo.createHttpResponse(payload);
        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", "text/plain", res.getHeader("Content-Type"));

View Full Code Here

Examples of org.archive.io.warc.TestWARCReader

    public void testPlainChunkedHttpRecord() throws Exception {
        String payload = "hogehogehogehogehoge";
        WARCRecordInfo recinfo = new TestWARCRecordInfo(
                TestWARCRecordInfo.buildHttpResponseBlock("200 OK",
                        "text/plain", payload.getBytes("UTF-8"), 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", "text/plain", res.getHeader("Content-Type"));

View Full Code Here

Examples of org.archive.io.warc.TestWARCReader

    
    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.TestWARCReader

        return createTestHtmlResource(null, timestamp, payloadBytes);
    }
    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.TestWARCReader

    public static final Resource createTest502Resource() throws IOException {
        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

Examples of org.archive.io.warc.TestWARCReader

    }
    
    public void testRenderResource_BasicCapture() throws Exception {
        final String ct = "image/gif";
        WARCRecordInfo recinfo = TestWARCRecordInfo.createHttpResponse(ct, TestWARCRecordInfo.PAYLOAD_GIF);
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        Resource payloadResource = new WarcResource(rec, ar);
        payloadResource.parseHeaders();
        Resource headersResource = payloadResource;
        
        TestServletOutputStream servletOutput = new TestServletOutputStream();

View Full Code Here

Examples of org.archive.io.warc.TestWARCReader

    public void testRenderResource_CompressedCapture() throws Exception {
        final String ct = "image/gif";
        WARCRecordInfo recinfo = new TestWARCRecordInfo(
                TestWARCRecordInfo.buildCompressedHttpResponseBlock(ct,
                        TestWARCRecordInfo.PAYLOAD_GIF));
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        Resource payloadResource = new WarcResource(rec, ar);
        payloadResource.parseHeaders();
        Resource headersResource = payloadResource;
        
        TestServletOutputStream servletOutput = new TestServletOutputStream();

View Full Code Here

Examples of org.archive.io.warc.TestWARCReader

    }    
        
    public void testRenderResource_Redirect() throws Exception {
        String location = "http://www.example.com/index.html";
        WARCRecordInfo recinfo = new TestWARCRecordInfo(TestWARCRecordInfo.buildHttpRedirectResponseBlock(location));
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        Resource payloadResource = new WarcResource(rec, ar);
        payloadResource.parseHeaders();
        
        final String originalUrl = "http://www.example.com/";
        final String captureTimestamp = "20130101123456";

View Full Code Here

Examples of org.archive.io.warc.TestWARCReader

            "</payload>\n";
        final byte[] recordBytes = TestWARCRecordInfo.buildHttpResponseBlock(
            "200 OK", ct, payload.getBytes("UTF-8"), true);
        //System.out.println(new String(recordBytes, "UTF-8"));
        WARCRecordInfo recinfo = new TestWARCRecordInfo(recordBytes);
        TestWARCReader ar = new TestWARCReader(recinfo);
        WARCRecord rec = ar.get(0);
        Resource payloadResource = new WarcResource(rec, ar);
        payloadResource.parseHeaders();
        Resource headersResource = payloadResource;


        TestServletOutputStream servletOutput = new TestServletOutputStream();

View Full Code Here

0 1

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.