1 package jalview.ext.ensembl;
3 import jalview.io.FileParse;
4 import jalview.util.StringUtils;
6 import java.io.BufferedReader;
7 import java.io.DataOutputStream;
8 import java.io.IOException;
9 import java.io.InputStream;
10 import java.io.InputStreamReader;
11 import java.net.HttpURLConnection;
12 import java.net.MalformedURLException;
14 import java.util.HashMap;
15 import java.util.List;
18 import javax.ws.rs.HttpMethod;
20 import org.json.simple.JSONArray;
21 import org.json.simple.JSONObject;
22 import org.json.simple.parser.JSONParser;
24 import com.stevesoft.pat.Regex;
27 * Base class for Ensembl REST service clients
31 abstract class EnsemblRestClient extends EnsemblSequenceFetcher
34 * update these constants when Jalview has been checked / updated for
35 * changes to Ensembl REST API
36 * @see https://github.com/Ensembl/ensembl-rest/wiki/Change-log
38 private static final String LATEST_ENSEMBLGENOMES_REST_VERSION = "4.6";
40 private static final String LATEST_ENSEMBL_REST_VERSION = "4.6";
42 private static final String REST_CHANGE_LOG = "https://github.com/Ensembl/ensembl-rest/wiki/Change-log";
44 private static Map<String, EnsemblInfo> domainData;
46 // @see https://github.com/Ensembl/ensembl-rest/wiki/Output-formats
47 private static final String PING_URL = "http://rest.ensembl.org/info/ping.json";
49 private final static long AVAILABILITY_RETEST_INTERVAL = 10000L; // 10 seconds
51 private final static long VERSION_RETEST_INTERVAL = 1000L * 3600; // 1 hr
53 private static final Regex TRANSCRIPT_REGEX = new Regex(
54 "(ENS)([A-Z]{3}|)T[0-9]{11}$");
56 private static final Regex GENE_REGEX = new Regex(
57 "(ENS)([A-Z]{3}|)G[0-9]{11}$");
61 domainData = new HashMap<String, EnsemblInfo>();
62 domainData.put(ENSEMBL_REST, new EnsemblInfo(ENSEMBL_REST,
63 LATEST_ENSEMBL_REST_VERSION));
64 domainData.put(ENSEMBL_GENOMES_REST, new EnsemblInfo(
65 ENSEMBL_GENOMES_REST, LATEST_ENSEMBLGENOMES_REST_VERSION));
68 protected volatile boolean inProgress = false;
71 * Default constructor to use rest.ensembl.org
73 public EnsemblRestClient()
79 * Constructor given the target domain to fetch data from
83 public EnsemblRestClient(String d)
89 * Answers true if the query matches the regular expression pattern for an
90 * Ensembl transcript stable identifier
95 public boolean isTranscriptIdentifier(String query)
97 return query == null ? false : TRANSCRIPT_REGEX.search(query);
101 * Answers true if the query matches the regular expression pattern for an
102 * Ensembl gene stable identifier
107 public boolean isGeneIdentifier(String query)
109 return query == null ? false : GENE_REGEX.search(query);
113 public boolean queryInProgress()
119 public StringBuffer getRawRecords()
125 * Returns the URL for the client http request
129 * @throws MalformedURLException
131 protected abstract URL getUrl(List<String> ids)
132 throws MalformedURLException;
135 * Returns true if client uses GET method, false if it uses POST
139 protected abstract boolean useGetRequest();
142 * Return the desired value for the Content-Type request header
147 * @see https://github.com/Ensembl/ensembl-rest/wiki/HTTP-Headers
149 protected abstract String getRequestMimeType(boolean multipleIds);
152 * Return the desired value for the Accept request header
155 * @see https://github.com/Ensembl/ensembl-rest/wiki/HTTP-Headers
157 protected abstract String getResponseMimeType();
160 * Checks Ensembl's REST 'ping' endpoint, and returns true if response
161 * indicates available, else false
163 * @see http://rest.ensembl.org/documentation/info/ping
166 private boolean checkEnsembl()
168 HttpURLConnection conn = null;
171 // note this format works for both ensembl and ensemblgenomes
172 // info/ping.json works for ensembl only (March 2016)
173 URL ping = new URL(getDomain()
174 + "/info/ping?content-type=application/json");
177 * expect {"ping":1} if ok
179 BufferedReader br = getHttpResponse(ping, null);
180 JSONParser jp = new JSONParser();
181 JSONObject val = (JSONObject) jp.parse(br);
182 String pingString = val.get("ping").toString();
183 return pingString != null;
184 } catch (Throwable t)
186 System.err.println("Error connecting to " + PING_URL + ": "
199 * returns a reader to a Fasta response from the Ensembl sequence endpoint
203 * @throws IOException
205 protected FileParse getSequenceReader(List<String> ids)
208 URL url = getUrl(ids);
210 BufferedReader reader = getHttpResponse(url, ids);
216 FileParse fp = new FileParse(reader, url.toString(), "HTTP_POST");
221 * Writes the HTTP request and gets the response as a reader.
225 * written as Json POST body if more than one
227 * @throws IOException
228 * if response code was not 200, or other I/O error
230 protected BufferedReader getHttpResponse(URL url, List<String> ids)
233 // long now = System.currentTimeMillis();
234 HttpURLConnection connection = (HttpURLConnection) url.openConnection();
237 * POST method allows multiple queries in one request; it is supported for
238 * sequence queries, but not for overlap
240 boolean multipleIds = ids != null && ids.size() > 1;
241 connection.setRequestMethod(multipleIds ? HttpMethod.POST
243 connection.setRequestProperty("Content-Type",
244 getRequestMimeType(multipleIds));
245 connection.setRequestProperty("Accept", getResponseMimeType());
247 connection.setUseCaches(false);
248 connection.setDoInput(true);
249 connection.setDoOutput(multipleIds);
253 writePostBody(connection, ids);
256 int responseCode = connection.getResponseCode();
258 if (responseCode != 200)
261 * note: a GET request for an invalid id returns an error code e.g. 415
262 * but POST request returns 200 and an empty Fasta response
264 System.err.println("Response code " + responseCode + " for " + url);
268 InputStream response = connection.getInputStream();
270 // System.out.println(getClass().getName() + " took "
271 // + (System.currentTimeMillis() - now) + "ms to fetch");
273 checkRateLimits(connection);
275 BufferedReader reader = null;
276 reader = new BufferedReader(new InputStreamReader(response, "UTF-8"));
281 * Inspect response headers for any sign of server overload and respect any
282 * 'retry-after' directive
284 * @see https://github.com/Ensembl/ensembl-rest/wiki/Rate-Limits
287 void checkRateLimits(HttpURLConnection connection)
289 // number of requests allowed per time interval:
290 String limit = connection.getHeaderField("X-RateLimit-Limit");
291 // length of quota time interval in seconds:
292 // String period = connection.getHeaderField("X-RateLimit-Period");
293 // seconds remaining until usage quota is reset:
294 String reset = connection.getHeaderField("X-RateLimit-Reset");
295 // number of requests remaining from quota for current period:
296 String remaining = connection.getHeaderField("X-RateLimit-Remaining");
297 // number of seconds to wait before retrying (if remaining == 0)
298 String retryDelay = connection.getHeaderField("Retry-After");
303 EnsemblInfo info = domainData.get(getDomain());
304 if (retryDelay != null)
306 System.err.println("Ensembl REST service rate limit exceeded, wait "
307 + retryDelay + " seconds before retrying");
310 info.retryAfter = System.currentTimeMillis()
311 + (1000 * Integer.valueOf(retryDelay));
312 } catch (NumberFormatException e)
314 System.err.println("Unexpected value for Retry-After: "
322 // System.out.println(String.format(
323 // "%s Ensembl requests remaining of %s (reset in %ss)",
324 // remaining, limit, reset));
329 * Rechecks if Ensembl is responding, unless the last check was successful and
330 * the retest interval has not yet elapsed. Returns true if Ensembl is up,
331 * else false. Also retrieves and saves the current version of Ensembl data
332 * and REST services at intervals.
336 protected boolean isEnsemblAvailable()
338 EnsemblInfo info = domainData.get(getDomain());
340 long now = System.currentTimeMillis();
343 * check if we are waiting for 'Retry-After' to expire
345 if (info.retryAfter > now)
347 System.err.println("Still " + (1 + (info.retryAfter - now) / 1000)
348 + " secs to wait before retrying Ensembl");
357 * recheck if Ensembl is up if it was down, or the recheck period has elapsed
359 boolean retestAvailability = (now - info.lastAvailableCheckTime) > AVAILABILITY_RETEST_INTERVAL;
360 if (!info.restAvailable || retestAvailability)
362 info.restAvailable = checkEnsembl();
363 info.lastAvailableCheckTime = now;
367 * refetch Ensembl versions if the recheck period has elapsed
369 boolean refetchVersion = (now - info.lastVersionCheckTime) > VERSION_RETEST_INTERVAL;
372 checkEnsemblRestVersion();
373 checkEnsemblDataVersion();
374 info.lastVersionCheckTime = now;
377 return info.restAvailable;
381 * Constructs, writes and flushes the POST body of the request, containing the
382 * query ids in JSON format
386 * @throws IOException
388 protected void writePostBody(HttpURLConnection connection,
389 List<String> ids) throws IOException
392 StringBuilder postBody = new StringBuilder(64);
393 postBody.append("{\"ids\":[");
395 for (String id : ids)
399 postBody.append(",");
402 postBody.append("\"");
403 postBody.append(id.trim());
404 postBody.append("\"");
406 postBody.append("]}");
407 byte[] thepostbody = postBody.toString().getBytes();
408 connection.setRequestProperty("Content-Length",
409 Integer.toString(thepostbody.length));
410 DataOutputStream wr = new DataOutputStream(connection.getOutputStream());
411 wr.write(thepostbody);
417 * Fetches and checks Ensembl's REST version number
421 private void checkEnsemblRestVersion()
423 EnsemblInfo info = domainData.get(getDomain());
425 JSONParser jp = new JSONParser();
429 url = new URL(getDomain()
430 + "/info/rest?content-type=application/json");
431 BufferedReader br = getHttpResponse(url, null);
432 JSONObject val = (JSONObject) jp.parse(br);
433 String version = val.get("release").toString();
434 String majorVersion = version.substring(0, version.indexOf("."));
435 String expected = info.expectedRestVersion;
436 String expectedMajorVersion = expected.substring(0,
437 expected.indexOf("."));
438 info.restMajorVersionMismatch = false;
442 * if actual REST major version is ahead of what we expect,
443 * record this in case we want to warn the user
445 if (Float.valueOf(majorVersion) > Float
446 .valueOf(expectedMajorVersion))
448 info.restMajorVersionMismatch = true;
450 } catch (NumberFormatException e)
452 System.err.println("Error in REST version: " + e.toString());
456 * check if REST version is later than what Jalview has tested against,
457 * if so warn; we don't worry if it is earlier (this indicates Jalview has
458 * been tested in advance against the next pending REST version)
460 boolean laterVersion = StringUtils.compareVersions(version, expected) == 1;
463 System.err.println(String.format(
464 "Expected %s REST version %s but found %s, see %s",
465 getDbSource(), expected, version, REST_CHANGE_LOG));
467 info.restVersion = version;
468 } catch (Throwable t)
470 System.err.println("Error checking Ensembl REST version: "
475 public boolean isRestMajorVersionMismatch()
477 return domainData.get(getDomain()).restMajorVersionMismatch;
481 * Fetches and checks Ensembl's data version number
485 private void checkEnsemblDataVersion()
487 JSONParser jp = new JSONParser();
491 url = new URL(getDomain()
492 + "/info/data?content-type=application/json");
493 BufferedReader br = getHttpResponse(url, null);
494 JSONObject val = (JSONObject) jp.parse(br);
495 JSONArray versions = (JSONArray) val.get("releases");
496 domainData.get(getDomain()).dataVersion = versions.get(0).toString();
497 } catch (Throwable t)
499 System.err.println("Error checking Ensembl data version: "
504 public String getEnsemblDataVersion()
506 return domainData.get(getDomain()).dataVersion;
510 public String getDbVersion()
512 return getEnsemblDataVersion();