Examples of org.jwat.common.Uri

org.jwat.common.Uri
Custom URI parser/validation based on rfc3986. An URI is split into sub components which are in turn validated. The overall reason for implementing this class is the lack of support for %uxxxx encoding which is not part of the specification but nevertheless needs to be validated since they are in use. Other non standard characters are also used in the wild. This implementation is not a complete substitute for the JDK version. It does not yet implement the following methods. - normalize() - parseServerAuthority() - relativize(URI uri) - resolve(String str) - resolve(URI uri) - toASCIIString() @author nicl

            e.printStackTrace();
            Assert.fail("Unexpected exception!");
        }


        String uriStr = "urn:uuid:a1bc4f20-d057-44bd-a299-bdf31f4dfa53";
        Uri uriObj = Uri.create(uriStr);


        /*
         * Non warc headers.
         */


        header = getTestHeader();
        headerLine = header.addHeader("X-Header1", dateObj, dateStr);
        Assert.assertNotNull(headerLine);
        Assert.assertEquals("X-Header1", headerLine.name);
        Assert.assertEquals(dateStr, headerLine.value);
        headerLine = header.addHeader("X-Header2", dateObj, dateStr);
        Assert.assertNotNull(headerLine);
        Assert.assertEquals("X-Header2", headerLine.name);
        Assert.assertEquals(dateStr, headerLine.value);


        /*
         * Duplicate date headers.
         */


        header = getTestHeader();
        headerLine = header.addHeader("WARC-Date", dateObj, dateStr);
        Assert.assertNotNull(headerLine);
        Assert.assertEquals("WARC-Date", headerLine.name);
        Assert.assertEquals(dateStr, headerLine.value);
        errors = header.diagnostics.getErrors();
        warnings = header.diagnostics.getWarnings();
        Assert.assertEquals(0, errors.size());
        Assert.assertEquals(0, warnings.size());
        headerLine = header.addHeader("WARC-Date", dateObj, dateStr);
        Assert.assertNotNull(headerLine);
        Assert.assertEquals("WARC-Date", headerLine.name);
        Assert.assertEquals(dateStr, headerLine.value);
        errors = header.diagnostics.getErrors();
        warnings = header.diagnostics.getWarnings();
        Assert.assertEquals(1, errors.size());
        Assert.assertEquals(0, warnings.size());
        diagnosis = errors.get(0);
        Assert.assertEquals(DiagnosisType.DUPLICATE, diagnosis.type);
        Assert.assertEquals("'WARC-Date' header", diagnosis.entity);
        Assert.assertEquals(1, diagnosis.information.length);


        /*
         * Integer add header.
         */


        cases = generate_invalid_datatype_cases(integerObj, integerStr, WarcConstants.FN_WARC_SEGMENT_NUMBER, WarcConstants.FN_WARC_RECORD_ID);
        test_headeradd_object_cases(cases, WarcConstants.FDT_INTEGER);


        /*
         * Long add header.
         */


        cases = generate_invalid_datatype_cases(longObj, longStr, WarcConstants.FN_CONTENT_LENGTH, WarcConstants.FN_CONTENT_TYPE);
        test_headeradd_object_cases(cases, WarcConstants.FDT_LONG);


        /*
         * Digest add header.
         */


        cases = generate_invalid_datatype_cases(digestObj, digestStr, WarcConstants.FN_WARC_BLOCK_DIGEST, WarcConstants.FN_WARC_SEGMENT_NUMBER);
        test_headeradd_object_cases(cases, WarcConstants.FDT_DIGEST);


        /*
         * ContentType add header.
         */


        cases = generate_invalid_datatype_cases(contentTypeObj, contentTypeStr, WarcConstants.FN_CONTENT_TYPE, WarcConstants.FN_WARC_BLOCK_DIGEST);
        test_headeradd_object_cases(cases, WarcConstants.FDT_CONTENTTYPE);


        /*
         * Date add header.
         */


        cases = generate_invalid_datatype_cases(dateObj, dateStr, WarcConstants.FN_WARC_DATE, WarcConstants.FN_CONTENT_TYPE);
        test_headeradd_object_cases(cases, WarcConstants.FDT_DATE);


        /*
         * InetAddress add header.
         */


        cases = generate_invalid_datatype_cases(inetAddressObj, inetAddressStr, WarcConstants.FN_WARC_IP_ADDRESS, WarcConstants.FN_WARC_DATE);
        test_headeradd_object_cases(cases, WarcConstants.FDT_INETADDRESS);


        /*
         * URI add header.
         */


        cases = generate_invalid_datatype_cases(uriObj, "<" + uriStr + ">", WarcConstants.FN_WARC_RECORD_ID, WarcConstants.FN_WARC_IP_ADDRESS);
        test_headeradd_object_cases(cases, WarcConstants.FDT_URI);


        /*
         * Test datatype for headers.
         */


        /*
        WarcConstants.FN_WARC_TYPE
        WarcConstants.FN_WARC_TRUNCATED
        WarcConstants.FN_WARC_FILENAME
        WarcConstants.FN_WARC_PROFILE
        */


        /*
         * Integer.
         */


        String segmentNrStr = "42";
        Integer segmentNrObj = Integer.parseInt(segmentNrStr);
        cases = generate_header_datatype_cases(segmentNrObj, segmentNrStr, WarcConstants.FN_WARC_SEGMENT_NUMBER, "warcSegmentNumberStr", "warcSegmentNumber");
        test_headeradd_object_cases(cases, WarcConstants.FDT_INTEGER);


        /*
         * Long.
         */


        String contentLengthStr = "1234567890123456";
        Long contentLengthObj = Long.parseLong(contentLengthStr);
        cases = generate_header_datatype_cases(contentLengthObj, contentLengthStr, WarcConstants.FN_CONTENT_LENGTH, "contentLengthStr", "contentLength");
        test_headeradd_object_cases(cases, WarcConstants.FDT_LONG);


        String segmentTotalLengthStr = "9876543210987654";
        Long segmentTotalLengthObj = Long.parseLong(segmentTotalLengthStr);
        cases = generate_header_datatype_cases(segmentTotalLengthObj, segmentTotalLengthStr, WarcConstants.FN_WARC_SEGMENT_TOTAL_LENGTH, "warcSegmentTotalLengthStr", "warcSegmentTotalLength");
        test_headeradd_object_cases(cases, WarcConstants.FDT_LONG);


        /*
         * Digest.
         */


        String blockDigestStr = "sha1:Y4N5SWNQBIBIGQ66IFXDMLGJW6FZFV6U";
        WarcDigest blockDigestObj = WarcDigest.parseWarcDigest(blockDigestStr);
        cases = generate_header_datatype_cases(blockDigestObj, blockDigestStr, WarcConstants.FN_WARC_BLOCK_DIGEST, "warcBlockDigestStr", "warcBlockDigest");
        test_headeradd_object_cases(cases, WarcConstants.FDT_DIGEST);


        String payloadDigestStr = "sha1:BCCYP7NW6QIIOSM523Y5XHQKE5KWLMBD";
        WarcDigest payloadDigestObj = WarcDigest.parseWarcDigest(payloadDigestStr);
        cases = generate_header_datatype_cases(payloadDigestObj, payloadDigestStr, WarcConstants.FN_WARC_PAYLOAD_DIGEST, "warcPayloadDigestStr", "warcPayloadDigest");
        test_headeradd_object_cases(cases, WarcConstants.FDT_DIGEST);


        /*
         * ContentType.
         */


        contentTypeStr = "application/http; msgtype=request";
        contentTypeObj = ContentType.parseContentType(contentTypeStr);
        cases = generate_header_datatype_cases(contentTypeObj, contentTypeStr, WarcConstants.FN_CONTENT_TYPE, "contentTypeStr", "contentType");
        test_headeradd_object_cases(cases, WarcConstants.FDT_CONTENTTYPE);


        String identifiedPayloadTypeStr = "application/http; msgtype=response";
        ContentType identifiedPayloadTypeObj = ContentType.parseContentType(identifiedPayloadTypeStr);
        cases = generate_header_datatype_cases(identifiedPayloadTypeObj, identifiedPayloadTypeStr, WarcConstants.FN_WARC_IDENTIFIED_PAYLOAD_TYPE, "warcIdentifiedPayloadTypeStr", "warcIdentifiedPayloadType");
        test_headeradd_object_cases(cases, WarcConstants.FDT_CONTENTTYPE);


        /*
         * Date.
         */


        dateStr = "2010-06-23T13:33:21Z";
        dateObj = WarcDateParser.getDate(dateStr);
        cases = generate_header_datatype_cases(dateObj, dateStr, WarcConstants.FN_WARC_DATE, "warcDateStr", "warcDate");
        test_headeradd_object_cases(cases, WarcConstants.FDT_DATE);


        String refersToDateStr = "2010-06-23T13:33:21Z";
        Date refersToDateObj = WarcDateParser.getDate(refersToDateStr);
        cases = generate_header_datatype_cases(refersToDateObj, refersToDateStr, WarcConstants.FN_WARC_REFERS_TO_DATE, "warcRefersToDateStr", "warcRefersToDate");
        test_headeradd_object_cases(cases, WarcConstants.FDT_DATE);


        /*
         * InetAddress.
         */


        inetAddressStr = "174.36.20.141";
        inetAddressObj = null;
        try {
            inetAddressObj = InetAddress.getByName(inetAddressStr);
        } catch (UnknownHostException e) {
            e.printStackTrace();
            Assert.fail("Unexpected exception!");
        }
        cases = generate_header_datatype_cases(inetAddressObj, inetAddressStr, WarcConstants.FN_WARC_IP_ADDRESS, "warcIpAddress", "warcInetAddress");
        test_headeradd_object_cases(cases, WarcConstants.FDT_INETADDRESS);


        /*
         * URI.
         */


        String recordIdStr = "urn:uuid:12eab1ec-8615-4f09-b6d2-976d96552073";
        Uri recordIdObj = Uri.create(recordIdStr);
        cases = generate_header_datatype_cases(recordIdObj, "<" + recordIdStr + ">", WarcConstants.FN_WARC_RECORD_ID, "warcRecordIdStr", "warcRecordIdUri");
        test_headeradd_object_cases(cases, WarcConstants.FDT_URI);


        String refersToStr = "urn:uuid:bfa9d26b-ff19-402a-8508-e7ff852d4ded";
        Uri refersToObj = Uri.create(refersToStr);
        cases = generate_header_datatype_cases(refersToObj, "<" + refersToStr + ">", WarcConstants.FN_WARC_REFERS_TO, "warcRefersToStr", "warcRefersToUri");
        test_headeradd_object_cases(cases, WarcConstants.FDT_URI);


        String targetUriStr = "urn:uuid:de715d47-9f36-4cdf-84db-eb3b47dcd0e3";
        Uri targetUriObj = Uri.create(targetUriStr);
        cases = generate_header_datatype_cases(targetUriObj, targetUriStr, WarcConstants.FN_WARC_TARGET_URI, "warcTargetUriStr", "warcTargetUriUri");
        test_headeradd_object_cases(cases, WarcConstants.FDT_URI);


        String warcinfoIdStr = "urn:uuid:1cb0e3be-d9e2-4058-bf00-9775c75a71a6";
        Uri warcinfoIdObj = Uri.create(warcinfoIdStr);
        cases = generate_header_datatype_cases(warcinfoIdObj, "<" + warcinfoIdStr + ">", WarcConstants.FN_WARC_WARCINFO_ID, "warcWarcinfoIdStr", "warcWarcinfoIdUri");
        test_headeradd_object_cases(cases, WarcConstants.FDT_URI);


        String segmentOriginIdStr = "urn:uuid:c4fc410a-4c7b-4bcc-a251-382b1d669f9a";
        Uri segmentOriginIdObj = Uri.create(segmentOriginIdStr);
        cases = generate_header_datatype_cases(segmentOriginIdObj, "<" + segmentOriginIdStr + ">", WarcConstants.FN_WARC_SEGMENT_ORIGIN_ID, "warcSegmentOriginIdStr", "warcSegmentOriginIdUrl");
        test_headeradd_object_cases(cases, WarcConstants.FDT_URI);


        Object[][] concurrentHeaders = new Object[][] {
                {Uri.create("urn:uuid:173a3db1-9ba4-496e-9eaf-6e550a62fd88"), "<urn:uuid:173a3db1-9ba4-496e-9eaf-6e550a62fd88>"},
                {Uri.create("urn:uuid:660b74e7-076e-4698-abba-4eeeb8e09bf1"), "<urn:uuid:660b74e7-076e-4698-abba-4eeeb8e09bf1>"},
                {Uri.create("urn:uuid:e6c0d888-b384-4ef4-a698-6166bc0875b8"), "<urn:uuid:e6c0d888-b384-4ef4-a698-6166bc0875b8>"}
        };
        cases = generate_multivalue_cases(concurrentHeaders, WarcConstants.FN_WARC_CONCURRENT_TO, "warcConcurrentToList");
        test_headeradd_multivalue_object_cases(cases, WarcConstants.FDT_URI);


        String warcProfileStr = WarcConstants.PROFILE_IDENTICAL_PAYLOAD_DIGEST;
        Uri warcProfileObj = Uri.create(warcProfileStr);
        cases = generate_header_datatype_cases(warcProfileObj, warcProfileStr, WarcConstants.FN_WARC_PROFILE, "warcProfileStr", "warcProfileUri");
        test_headeradd_object_cases(cases, WarcConstants.FDT_URI);


        warcProfileStr = "urn:uuid:de715d47-9f36-4cdf-84db-eb3b47dcd0e3";
        warcProfileObj = Uri.create(warcProfileStr);
        cases = generate_header_datatype_cases(warcProfileObj, warcProfileStr, WarcConstants.FN_WARC_PROFILE, "warcProfileStr", "warcProfileUri");
        test_headeradd_object_cases(cases, WarcConstants.FDT_URI);


        String warcRefersToTargetUriStr = "urn:uuid:de715d47-9f36-4cdf-84db-eb3b47dcd0e3";
        Uri warcRefersToTargetUriObj = Uri.create(warcRefersToTargetUriStr);
        cases = generate_header_datatype_cases(warcRefersToTargetUriObj, warcRefersToTargetUriStr, WarcConstants.FN_WARC_REFERS_TO_TARGET_URI, "warcRefersToTargetUriStr", "warcRefersToTargetUriUri");
        test_headeradd_object_cases(cases, WarcConstants.FDT_URI);


        /*
        public List<WarcConcurrentTo> warcConcurrentToList = new LinkedList<WarcConcurrentTo>();

View Full Code Here

        }
        /*
         * URI.
         */
        String recordIdStr = "urn:uuid:12eab1ec-8615-4f09-b6d2-976d96552073";
        Uri recordIdObj = Uri.create(recordIdStr);
        String refersToStr = "urn:uuid:bfa9d26b-ff19-402a-8508-e7ff852d4ded";
        Uri refersToObj = Uri.create(refersToStr);
        String targetUriStr = "urn:uuid:de715d47-9f36-4cdf-84db-eb3b47dcd0e3";
        Uri targetUriObj = Uri.create(targetUriStr);
        String warcinfoIdStr = "urn:uuid:1cb0e3be-d9e2-4058-bf00-9775c75a71a6";
        Uri warcinfoIdObj = Uri.create(warcinfoIdStr);
        String segmentOriginIdStr = "urn:uuid:c4fc410a-4c7b-4bcc-a251-382b1d669f9a";
        Uri segmentOriginIdObj = Uri.create(segmentOriginIdStr);
        String refersToTargetUriStr = "urn:uuid:9c0433e9-7365-4228-a4b6-501d3c1ebc50";
        Uri refersToTargetUriObj = Uri.create(refersToTargetUriStr);
        /*
         * ConcurrentTo.
         */
        Object[][] concurrentHeaders = new Object[][] {
                {Uri.create("urn:uuid:173a3db1-9ba4-496e-9eaf-6e550a62fd88"), "urn:uuid:173a3db1-9ba4-496e-9eaf-6e550a62fd88"},

View Full Code Here

     * @param nullable allow empty or null value
     * @return an URL object holding the value of the specified string or null,
     * if unable to parse the value as an URL object
     */
    protected Uri parseUri(String uriStr, UriProfile uriProfile, String field, boolean nullable) {
        Uri uri = null;
        if ((uriStr != null) && (uriStr.length() != 0)) {
            try {
                uri = new Uri(uriStr, uriProfile);
            } catch (Exception e) {
                // Invalid URI.
                addInvalidExpectedError("'" + field + "' value",
                        uriStr,
                        e.getMessage());
            }
            if (uri != null) {
                String scheme = uri.getScheme();
                if (scheme == null) {
                    uri = null;
                    // Relative URI.
                    addInvalidExpectedError("'" + field + "' value",
                            uriStr,

View Full Code Here

     * @param bLtGt something
     * @return an URI object holding the value of the specified string or null,
     * if unable to parse the value as an URI object
     */
    protected Uri parseUri(String uriStr, boolean bLtGt, UriProfile uriProfile, String field) {
        Uri uri = null;
        String uriStrClean = uriStr;
        int ltGtBf = 0;
        if (uriStrClean != null && uriStrClean.length() != 0) {
            int fIdx = 0;
            int tIdx = uriStrClean.length();
            if (uriStrClean.startsWith("<")) {
                ltGtBf |= 2;
                ++fIdx;
            }
            if (uriStrClean.endsWith(">")) {
                ltGtBf |= 1;
                --tIdx;
            }
            if (ltGtBf != 0) {
                uriStrClean = uriStrClean.substring(fIdx, tIdx);
            }
            if (bLtGt) {
                switch (ltGtBf) {
                case 2:
                    addInvalidExpectedError("'" + field + "' value", uriStr, "Missing trailing '>' character");
                    break;
                case 1:
                    addInvalidExpectedError("'" + field + "' value", uriStr, "Missing leading '<' character");
                    break;
                case 0:
                    addInvalidExpectedError("'" + field + "' value", uriStr, "Missing encapsulating '<' and '>' characters");
                    break;
                case 3:
                default:
                    break;
                }
            } else {
                switch (ltGtBf) {
                case 2:
                    addInvalidExpectedError("'" + field + "' value", uriStr, "Unexpected leading '<' character");
                    break;
                case 1:
                    addInvalidExpectedError("'" + field + "' value", uriStr, "Unexpected trailing '>' character");
                    break;
                case 3:
                    addInvalidExpectedError("'" + field + "' value", uriStr, "Unexpected encapsulating '<' and '>' characters");
                    break;
                case 0:
                default:
                    break;
                }
            }
            try {
                uri = new Uri(uriStrClean, uriProfile);
            } catch (Exception e) {
                // Invalid URI.
                addInvalidExpectedError("'" + field + "' value",
                        uriStrClean,
                        e.getMessage());
            }
            if (uri != null) {
                String scheme = uri.getScheme();
                if (scheme == null) {
                    uri = null;
                    // Relative URI.
                    addInvalidExpectedError("'" + field + "' value",
                            uriStrClean,

View Full Code Here

                    contentTypeStr = fieldValue;
                    contentType = fieldParsers.parseContentType(fieldValue,
                            WarcConstants.FN_CONTENT_TYPE);
                    break;
                case WarcConstants.FN_IDX_WARC_CONCURRENT_TO:
                    Uri tmpUri = fieldParsers.parseUri(fieldValue, URI_LTGT,
                            uriProfile, WarcConstants.FN_WARC_CONCURRENT_TO);
                    if (fieldValue != null && fieldValue.trim().length() > 0) {
                        warcConcurrentTo = new WarcConcurrentTo();
                        warcConcurrentTo.warcConcurrentToStr = fieldValue;
                        warcConcurrentTo.warcConcurrentToUri = tmpUri;

View Full Code Here

TOP

Related Classes of org.jwat.common.Uri

org.jwat.arc.ArcFieldParsers

org.jwat.warc.TestWarcHeaderAddDatatyped

org.jwat.warc.TestWarcWriter_Headers

org.jwat.warc.WarcFieldParsers

org.jwat.warc.WarcHeader

java.net.URISyntaxException

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.