Package bixo.exceptions

Examples of bixo.exceptions.AbortedFetchException


       
        @Override
        public FetchedDatum get(ScoredUrlDatum scoredUrl) throws BaseFetchException {
            String url = scoredUrl.getUrl();
            if (url.contains("page-6")) {
                throw new AbortedFetchException(url, AbortedFetchReason.SLOW_RESPONSE_RATE);
            } else if (url.contains("page-7")) {
                throw new HttpFetchException(url, "msg", HttpStatus.SC_GONE, new HttpHeaders());
            else if (url.contains("page-8")) {
                throw new IOFetchException(url, new IOException());
            } else if (url.contains("page-9")) {
View Full Code Here


            // mime-types from the server to be processed, set "" as one of the valid mime-types in FetcherPolicy.
            mimeType = HttpUtils.getMimeTypeFromContentType(contentType);
            Set<String> mimeTypes = _fetcherPolicy.getValidMimeTypes();
            if ((mimeTypes != null) && (mimeTypes.size() > 0)) {
                if (!mimeTypes.contains(mimeType)) {
                    throw new AbortedFetchException(url, "Invalid mime-type: " + mimeType, AbortedFetchReason.INVALID_MIMETYPE);
                }
            }
           
            needAbort = false;
        } catch (ClientProtocolException e) {
            // Oleg guarantees that no abort is needed in the case of an IOException (which is is a subclass of)
            needAbort = false;

            // If the root case was a "too many redirects" error, we want to map this to a specific
            // exception that contains the final redirect.
            if (e.getCause() instanceof MyRedirectException) {
                MyRedirectException mre = (MyRedirectException)e.getCause();
                String redirectUrl = url;

                try {
                    redirectUrl = mre.getUri().toURL().toExternalForm();
                } catch (MalformedURLException e2) {
                    LOGGER.warn("Invalid URI saved during redirect handling: " + mre.getUri());
                }

                throw new RedirectFetchException(url, redirectUrl, mre.getReason());
            } else if (e.getCause() instanceof RedirectException) {
                throw new RedirectFetchException(url, extractRedirectedUrl(url, localContext), RedirectExceptionReason.TOO_MANY_REDIRECTS);
            } else {
                throw new IOFetchException(url, e);
            }
        } catch (IOException e) {
            // Oleg guarantees that no abort is needed in the case of an IOException
            needAbort = false;
           
            if (e instanceof ConnectionPoolTimeoutException) {
                // Should never happen, so let's dump some info about the connection pool.
                ThreadSafeClientConnManager cm = (ThreadSafeClientConnManager)_httpClient.getConnectionManager();
                int numConnections = cm.getConnectionsInPool();
                cm.closeIdleConnections(0, TimeUnit.MILLISECONDS);
                LOGGER.error(String.format("Got ConnectionPoolTimeoutException: %d connections before, %d after idle close", numConnections, cm.getConnectionsInPool()));
            }
           
            throw new IOFetchException(url, e);
        } catch (URISyntaxException e) {
            throw new UrlFetchException(url, e.getMessage());
        } catch (IllegalStateException e) {
            throw new UrlFetchException(url, e.getMessage());
        } catch (BaseFetchException e) {
            throw e;
        } catch (Exception e) {
            // Map anything else to a generic IOFetchException
            // TODO KKr - create generic fetch exception
            throw new IOFetchException(url, new IOException(e));
        } finally {
            safeAbort(needAbort, request);
        }
       
        // Figure out how much data we want to try to fetch.
        int maxContentSize = getMaxContentSize(mimeType);
        int targetLength = maxContentSize;
        boolean truncated = false;
        String contentLengthStr = headerMap.getFirst(HttpHeaderNames.CONTENT_LENGTH);
        if (contentLengthStr != null) {
            try {
                int contentLength = Integer.parseInt(contentLengthStr);
                if (contentLength > targetLength) {
                    truncated = true;
                } else {
                    targetLength = contentLength;
                }
            } catch (NumberFormatException e) {
                // Ignore (and log) invalid content length values.
                LOGGER.warn("Invalid content length in header: " + contentLengthStr);
            }
        }

        // Now finally read in response body, up to targetLength bytes.
        // Note that entity might be null, for zero length responses.
        byte[] content = new byte[0];
        long readRate = 0;
        HttpEntity entity = response.getEntity();
        needAbort = true;

        if (entity != null) {
            InputStream in = null;

            try {
                in = entity.getContent();
                byte[] buffer = new byte[BUFFER_SIZE];
                int bytesRead = 0;
                int totalRead = 0;
                ByteArrayOutputStream out = new ByteArrayOutputStream(DEFAULT_BYTEARRAY_SIZE);

                int readRequests = 0;
                int minResponseRate = _fetcherPolicy.getMinResponseRate();
                // TODO KKr - we need to monitor the rate while reading a
                // single block. Look at HttpClient
                // metrics support for how to do this. Once we fix this, fix
                // the test to read a smaller (< 20K)
                // chuck of data.
                while ((totalRead < targetLength)
                    && ((bytesRead = in.read(buffer, 0, Math.min(buffer.length, targetLength - totalRead))) != -1)) {
                    readRequests += 1;
                    totalRead += bytesRead;
                    out.write(buffer, 0, bytesRead);

                    // Assume read time is at least one millisecond, to avoid DBZ exception.
                    long totalReadTime = Math.max(1, System.currentTimeMillis() - readStartTime);
                    readRate = (totalRead * 1000L) / totalReadTime;

                    // Don't bail on the first read cycle, as we can get a hiccup starting out.
                    // Also don't bail if we've read everything we need.
                    if ((readRequests > 1) && (totalRead < targetLength) && (readRate < minResponseRate)) {
                        throw new AbortedFetchException(url, "Slow response rate of " + readRate + " bytes/sec", AbortedFetchReason.SLOW_RESPONSE_RATE);
                    }

                    // Check to see if we got interrupted.
                    if (Thread.interrupted()) {
                        throw new AbortedFetchException(url, AbortedFetchReason.INTERRUPTED);
                    }
                }

                content = out.toByteArray();
                needAbort = truncated || (in.available() > 0);
            } catch (IOException e) {
                // We don't need to abort if there's an IOException
                throw new IOFetchException(url, e);
            } finally {
                safeAbort(needAbort, request);
                safeClose(in);
            }
        }
       
        // Toss truncated image content.
        if  (   (truncated)
            &&  (!isTextMimeType(mimeType))) {
            throw new AbortedFetchException(url, "Truncated image", AbortedFetchReason.CONTENT_SIZE);
        }

        // Now see if we need to uncompress the content.
        String contentEncoding = headerMap.getFirst(HttpHeaderNames.CONTENT_ENCODING);
        if (contentEncoding != null) {
            if (LOGGER.isTraceEnabled()) {
                fetchTrace.append("; Content-Encoding: " + contentEncoding);
            }

            // TODO KKr We might want to just decompress a truncated gzip
            // containing text (since we have a max content size to save us
            // from any gzip corruption). We might want to break the following
            // out into a separate method, by the way (if not refactor this
            // entire monolithic method).
            //
            try {
                if ("gzip".equals(contentEncoding) || "x-gzip".equals(contentEncoding)) {
                    if (truncated) {
                        throw new AbortedFetchException(url, "Truncated compressed data", AbortedFetchReason.CONTENT_SIZE);
                    } else {
                        ExpandedResult expandedResult = EncodingUtils.processGzipEncoded(content, maxContentSize);
                        truncated = expandedResult.isTruncated();
                        if  (   (truncated)
                            &&  (!isTextMimeType(mimeType))) {
                            throw new AbortedFetchException(url, "Truncated decompressed image", AbortedFetchReason.CONTENT_SIZE);
                        } else {
                            content = expandedResult.getExpanded();
                            if (LOGGER.isTraceEnabled()) {
                                fetchTrace.append("; unzipped to " + content.length + " bytes");
                            }
View Full Code Here

            // While we still have entries, we need to write them out to avoid losing them.
            while (iter.hasNext()) {
                ScoredUrlDatum item = iter.next();
                FetchedDatum result = new FetchedDatum(item);
                process.increment(FetchCounters.URLS_SKIPPED, 1);
                AbortedFetchException status = new AbortedFetchException(item.getUrl(), AbortedFetchReason.INTERRUPTED);
               
                Tuple tuple = result.getTuple();
                tuple.add(status);
               _fetchMgr.collect(tuple);
            }
View Full Code Here

TOP

Related Classes of bixo.exceptions.AbortedFetchException

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.