List of usage examples for org.apache.commons.validator.routines UrlValidator UrlValidator
public UrlValidator()
From source file:com.svi.uzabase.logic.ValidationProcess.java
private List<XMLHolder> validateData(List<XMLHolder> xmlBatchHolder) { try {//from w ww . j a v a 2 s . c o m int totalCounter = 0; //Initialize dictionary String dictFileName = "file://./dic/english.jar"; String configFile = "file://./classes/spellCheck.config"; BasicDictionary dictionary = new BasicDictionary(dictFileName); SpellCheckConfiguration configuration = new SpellCheckConfiguration(configFile); BasicSuggester suggester = new BasicSuggester(configuration); suggester.attach(dictionary); // create SpellCheck object based on configuration and specify Suggester SpellCheck spellCheck = new SpellCheck(configuration); spellCheck.setSuggester(suggester); //set for jprogress bar for (XMLHolder h : xmlBatchHolder) { totalCounter += h.size(); } progress = new AtomicInteger(0); total = new AtomicInteger(totalCounter); mf.setJprogressValues(total, progress); //validation process begins here String[] invalidWords = { "corporation", "inc.", "city", "corp.", "st.", "co.", "ltd." }; String[] invalidBoardWords = { "other", "oth" }; String[] validWords = { "loc", "to", "ext", "local" }; String[] invalidCharacters = { ",", "/", "\\", "[", "]", "\"", ":", "^", "{", "}", "%", "+", "#", "(", ")" }; String[] splitter; String tempURL; SimpleDateFormat sdf = new SimpleDateFormat("YYYY/MM/DD"); SimpleDateFormat fiscalYear = new SimpleDateFormat("MM/DD"); sdf.setLenient(false); Set<String> officerList = new HashSet<>(); List<Double> percentOwnership = new ArrayList<>(); UrlValidator urlValidator = new UrlValidator(); Date date = null; for (XMLHolder h : xmlBatchHolder) { for (Field f : h) { mf.loader("Validating fields: ", false); if (!f.getType().equals("none") && !f.getValue().equals("*N/A")) { switch (f.getType()) { case "city": if (f.getValue().isEmpty() || f.getValue().equals("")) { f.add("Address is empty"); } else { if (hasWhiteSpaceTrailing(f.getValue())) { f.add("has trailing white space "); } if (cityList.indexOf(f.getValue()) < 0) { f.add("City not found on list!"); } } break; case "province": if (f.getValue().isEmpty() || f.getValue().equals("")) { f.add("Address is empty"); } else { if (hasWhiteSpaceTrailing(f.getValue())) { f.add("has trailing white space "); } if (provinceList.indexOf(f.getValue()) < 0) { f.add("Province not found on list!"); } } break; case "tel": if (!f.getValue().isEmpty() || !f.getValue().equals("")) { // if (f.getValue().matches("[a-z A-Z]+")) { if (f.getValue().matches(".*[a-zA-Z]+.*")) { for (String s : validWords) { if (!f.getValue().contains(s)) { f.add("Invalid telephone number"); } } } if (f.getValue().replace(" ", "").replace("-", "").length() < 7 || f.getValue().replace(" ", "").replace("-", "").length() > 8) { f.add("Invalid telephone number length"); } if (hasWhiteSpaceTrailing(f.getValue())) { f.add("has trailing white space "); } if (StringUtils.countMatches(f.getValue(), "-") > 2) { f.add("Invalid telephone number"); } for (String c : invalidCharacters) { if (f.getValue().contains(c)) { f.add("Contains invalid character [ " + c + " ]"); break; } } } break; case "fax": if (!f.getValue().isEmpty() || !f.getValue().equals("")) { // if (f.getValue().matches("[a-z A-Z]+")) { if (f.getValue().matches(".*[a-zA-Z]+.*")) { for (String s : validWords) { if (!f.getValue().contains(s)) { f.add("Invalid fax number"); } } } if (f.getValue().replace(" ", "").length() < 6) { f.add("Invalid fax number"); } if (StringUtils.countMatches(f.getValue(), "-") > 1) { f.add("Invalid fax number"); } if (hasWhiteSpaceTrailing(f.getValue())) { f.add("has trailing white space "); } for (String c : invalidCharacters) { if (f.getValue().contains(c)) { f.add("Contains invalid character [ " + c + " ]"); break; } } } break; case "person": if (!f.getValue().isEmpty() || !f.getValue().equals("")) { if (!f.getValue().matches("[a-zA-Z\\.,\\- ()]+")) { f.add("Invalid name"); } if (f.getValue().matches("[a-z ]+")) { f.add("All small caps"); } if (f.getValue().matches("\\w+")) { f.add("Only one word"); } if (f.getValue().replace(" ", "").length() > 30) { f.add("More than 30 characters."); } if (f.getValue().replace(" ", "").length() < 2) { f.add("Invalid name."); } if (hasWhiteSpaceTrailing(f.getValue())) { f.add("has trailing white space "); } } break; case "email": if (!f.getValue().isEmpty() || !f.getValue().equals("")) { if (!EmailValidator.getInstance(true).isValid(f.getValue())) { f.add("Invalid email"); } } break; case "website": if (!f.getValue().isEmpty() || !f.getValue().equals("")) { if (!f.getValue().contains("http")) { tempURL = "http://" + f.getValue(); } else { tempURL = f.getValue(); } if (!urlValidator.isValid(tempURL)) { f.add("Invalid website"); } if (hasWhiteSpaceTrailing(f.getValue())) { f.add("has trailing white space "); } } break; case "name": officerList.add(f.getValue()); if (!f.getValue().isEmpty() || !f.getValue().equals("")) { if (!f.getValue().matches("[a-zA-Z\\.,\\-() ]+")) { f.add("Invalid name"); } if (f.getValue().replace(" ", "").length() > 30) { f.add("More than 50 characters."); } if (f.getValue().matches("[a-z ]+")) { f.add("All small caps"); } if (f.getValue().matches("\\w+")) { f.add("Only one word"); } if (f.getValue().replace(" ", "").length() < 2) { f.add("Invalid name."); } if (hasWhiteSpaceTrailing(f.getValue())) { f.add("has trailing white space "); } for (String s : invalidWords) { if (f.getValue().contains(s)) { f.add("Contains invalid word: " + s); break; } } } break; case "stockholder": officerList.add(f.getValue()); if (!f.getValue().isEmpty() || !f.getValue().equals("")) { if (!f.getValue().matches("[a-zA-Z\\.,\\-() ]+")) { f.add("Invalid name"); } if (f.getValue().replace(" ", "").length() > 30) { f.add("More than 50 characters."); } if (f.getValue().matches("[a-z ]+")) { f.add("All small caps"); } if (f.getValue().matches("\\w+")) { f.add("Only one word"); } if (f.getValue().replace(" ", "").length() < 2) { f.add("Invalid name."); } if (hasWhiteSpaceTrailing(f.getValue())) { f.add("has trailing white space "); } for (String s : invalidWords) { if (f.getValue().contains(s)) { f.add("Contains invalid word: " + s); break; } } } break; case "board": if (!f.getValue().isEmpty() || !f.getValue().equals("")) { if (!f.getValue().matches("[a-zA-Z\\.,\\-() ]+")) { f.add("Invalid position"); } for (String c : invalidCharacters) { if (f.getValue().contains(c)) { f.add("Contains invalid character [ " + c + " ]"); break; } } for (String c : invalidBoardWords) { if (f.getValue().contains(c)) { f.add("Contains invalid word [ " + c + " ]"); break; } } if (f.getValue().equalsIgnoreCase("N") || f.getValue().equalsIgnoreCase("Y")) { f.add("is letter " + f.getValue() + " only"); } if (Character.isLowerCase(f.getValue().charAt(0))) { f.add("starts with a lower case letter"); } spellCheck.setText(f.getValue(), Constants.DOC_TYPE_TEXT, "en"); spellCheck.check(); if (spellCheck.hasMisspelt()) { f.add("word is misspelled."); } } break; case "corporation": if (companyList.indexOf(f.getValue().toUpperCase()) < 0) { f.add("Company name not found on table."); } break; case "sec": if (StringUtils.countMatches(f.getValue(), "-") > 1) { f.add("Invalid SEC number"); } if (f.getValue().replace(" ", "").length() > 9) { f.add("SEC number more than 9 digits."); } if (hasWhiteSpaceTrailing(f.getValue())) { f.add("SEC has trailing white space."); } for (String c : invalidCharacters) { if (f.getValue().contains(c)) { f.add("Contains invalid character [ " + c + " ]"); break; } } break; case "tin": if (f.getValue().isEmpty() || f.getValue().equals("")) { f.add("TIN is empty"); } if (hasWhiteSpaceTrailing(f.getValue())) { f.add("TIN has trailing white space."); } if (!f.getValue().matches("[0-9]+")) { f.add("invalid TIN number"); } if (f.getValue().replace(" ", "").replace("-", "").length() > 12 || f.getValue().replace(" ", "").replace("-", "").length() < 9) { f.add("TIN number invalid length."); } if (StringUtils.countMatches(f.getValue(), "-") > 1) { f.add("Invalid TIN number"); } for (String c : invalidCharacters) { if (f.getValue().contains(c)) { f.add("Contains invalid character [ " + c + " ]"); break; } } break; case "nationality": if (!f.getValue().isEmpty() || !f.getValue().equals("")) { if (nationalityList.indexOf(f.getValue()) < 0) { f.add("nationality is misspelled."); } } break; case "purpose": splitter = f.getValue().split(" "); for (int i = 0; i < splitter.length; i++) { spellCheck.setText(splitter[i], Constants.DOC_TYPE_TEXT, "en"); spellCheck.check(); if (spellCheck.hasMisspelt()) { f.add("word is misspelled. ( " + spellCheck.getMisspelt() + " )"); } } break; case "periodCovered": try { date = sdf.parse(f.getValue()); if (!f.getValue().equals(sdf.format(date))) { f.add("Invalid date format"); } } catch (ParseException ex) { f.add("Invalid date format"); } break; case "fiscalYear": try { date = fiscalYear.parse(f.getValue()); if (!f.getValue().equals(sdf.format(date))) { f.add("Invalid date format"); } } catch (ParseException ex) { f.add("Invalid date format"); } break; case "position": if (f.getValue().contains("\\d+")) { f.add("Invalid position/designation"); } if (f.getValue().replace(" ", "").length() > 10 || f.getValue().replace(" ", "").length() < 3) { f.add("More than 30 characters."); } break; case "shareType": if (f.getValue().toLowerCase().contains("total")) { f.add("Share type contains total."); } if (f.getValue().replace(" ", "").length() > 20) { f.add("Share type More than 20 characters."); } break; case "ownership": percentOwnership.add(Double.parseDouble(f.getValue())); if (Double.parseDouble(f.getValue()) > 100) { f.add("Percent ownership more than 100%"); } break; default: break; } } else if (f.getType().equals("tin") && f.getValue().equals("*N/A")) { f.add("TIN is N/A"); } } } } catch (EncryptedDocumentException | SuggesterException ex) { Logger.getLogger(ValidationProcess.class.getName()).log(Level.SEVERE, null, ex); } return xmlBatchHolder; }
From source file:com._17od.upm.gui.MainWindow.java
private boolean urlIsValid(String urL) { UrlValidator urlValidator = new UrlValidator(); if (urlValidator.isValid(urL)) { return true; } else {//from ww w. j av a2 s. co m return false; } }
From source file:com._17od.upm.gui.AccountDialog.java
/** * Use com.apache.commons.validator library in order to check the validity * (proper formating, e.x http://www.url.com) of the given URL. * /*from w ww. j av a 2 s . c om*/ * @param urL * @return */ private boolean urlIsValid(String urL) { UrlValidator urlValidator = new UrlValidator(); if (urlValidator.isValid(urL)) { return true; } else { return false; } }
From source file:com.ut.healthelink.service.impl.transactionInManagerImpl.java
@Override public boolean isValidURL(String url) { UrlValidator urlValidator = new UrlValidator(); if (urlValidator.isValid(url)) { return true; } else {// w ww. j a v a 2 s.c om return false; } }
From source file:noThreads.ParseLevel2.java
/** * @param theUrl/*from ww w . ja va 2 s . c om*/ * @param conAttempts * @return * @throws IOException */ public boolean validUrl(String theUrl, int conAttempts) throws IOException { long total_time = 0, endTime = 0; long startTime = System.currentTimeMillis(); URL link = new URL(theUrl); int CONNECT_TIMEOUT = 5000, READ_TIMEOUT = 2000; HttpURLConnection huc = (HttpURLConnection) link.openConnection(); huc.setRequestProperty("User-Agent", userAgent); huc.setConnectTimeout(CONNECT_TIMEOUT); huc.setReadTimeout(READ_TIMEOUT); try { huc.connect(); } catch (java.net.ConnectException e) { print(e.getMessage() + "\n"); if (e.getMessage().equalsIgnoreCase("Connection timed out")) { if (conAttempts != MAX_CONNECTION_ATTEMPTS) { print("Recurrencing validUrl method..."); return validUrl(theUrl, conAttempts + 1); } else { return false; } } else { return false; } } catch (java.net.SocketTimeoutException e) { print(e.getMessage() + "\n"); if (e.getMessage().equalsIgnoreCase("connect timed out")) { if (conAttempts != MAX_CONNECTION_ATTEMPTS) { print("Recurrencing validUrl method..."); return validUrl(theUrl, conAttempts + 1); } else { return false; } } else { return false; } } catch (IOException e) { print(e.getMessage() + "\n"); return false; } UrlValidator urlValidator = new UrlValidator(); if (urlValidator.isValid(theUrl) == true) { print("valid url form"); if (huc.getContentType() != null) { print("Content: " + huc.getContentType()); if (huc.getContentType().equals("text/html") || huc.getContentType().equals("unknown/unknown")) { if (getResposeCode(theUrl, 0) >= java.net.HttpURLConnection.HTTP_BAD_REQUEST) { print("Server Response Code: " + getResposeCode(theUrl, 0)); return false; } } print(huc.getContentType()); endTime = System.currentTimeMillis(); total_time = total_time + (endTime - startTime); print("Total elapsed time is :" + total_time + "\n"); return true; } else {//edw erxetai an den prolavei na diavasei h an einai null to content endTime = System.currentTimeMillis(); total_time = total_time + (endTime - startTime); print("Total elapsed time is :" + total_time + "\n"); if (conAttempts != MAX_CONNECTION_ATTEMPTS) { print("Recurrencing validUrl method..."); return validUrl(theUrl, conAttempts + 1); } else { return false; } } } else { endTime = System.currentTimeMillis(); total_time = total_time + (endTime - startTime); print("Total elapsed time is :" + total_time + "\n"); return false; } }
From source file:org.ambraproject.action.article.MediaCoverageAction.java
/** * Validate the input from the form/* ww w. ja v a2 s. c o m*/ * @return true if everything is ok */ private boolean validateInput() { // TODO handle data better boolean isValid = true; if (StringUtils.isBlank(uri)) { isValid = false; } UrlValidator urlValidator = new UrlValidator(); if (StringUtils.isBlank(link)) { addFieldError("link", "This field is required."); isValid = false; } else if (!urlValidator.isValid(link)) { addFieldError("link", "Invalid Media link URL"); isValid = false; } if (StringUtils.isBlank(name)) { addFieldError("name", "This field is required."); isValid = false; } if (StringUtils.isBlank(email)) { addFieldError("email", "This field is required."); isValid = false; } else if (!EmailValidator.getInstance().isValid(email)) { addFieldError("email", "Invalid e-mail address"); isValid = false; } HttpServletRequest request = ServletActionContext.getRequest(); if (!captchaService.validateCaptcha(request.getRemoteAddr(), captchaChallenge, captchaResponse)) { addFieldError("captcha", "Verification is incorrect. Please try again."); isValid = false; } if (isValid) { this.link = this.link.substring(0, Math.min(this.link.length(), MAX_LENGTH)); this.name = this.name.substring(0, Math.min(this.name.length(), MAX_LENGTH)); this.email = this.email.substring(0, Math.min(this.email.length(), MAX_LENGTH)); if (!StringUtils.isBlank(comment)) { this.comment = this.comment.substring(0, Math.min(this.comment.length(), MAX_LENGTH)); } } return isValid; }
From source file:org.apache.streams.urls.LinkResolver.java
public void unwindLink(String url) { Objects.requireNonNull(linkDetails); Objects.requireNonNull(url);//from www . j a v a 2 s. c o m // Check url validity UrlValidator urlValidator = new UrlValidator(); if (!urlValidator.isValid(url)) { linkDetails.setLinkStatus(LinkDetails.LinkStatus.MALFORMED_URL); return; } // Check to see if they wound up in a redirect loop, // IE: 'A' redirects to 'B', then 'B' redirects to 'A' if ((linkDetails.getRedirectCount() != null && linkDetails.getRedirectCount() > 0 && (linkDetails.getOriginalURL().equals(url) || linkDetails.getRedirects().contains(url))) || (linkDetails.getRedirectCount() != null && linkDetails.getRedirectCount() > MAX_ALLOWED_REDIRECTS)) { linkDetails.setLinkStatus(LinkDetails.LinkStatus.LOOP); return; } if (!linkDetails.getOriginalURL().equals(url)) linkDetails.getRedirects().add(url); HttpURLConnection connection = null; // Store where the redirected link will go (if there is one) String reDirectedLink = null; try { // Turn the string into a URL URL thisURL = new URL(url); // Be sensitive to overloading domains STREAMS-77 try { String host = thisURL.getHost().toLowerCase(); if (!domainsSensitiveTo.contains(host)) { domainsSensitiveTo.add(host); long domainWait = LinkResolverHelperFunctions.waitTimeForDomain(thisURL.getHost()); if (domainWait > 0) { LOGGER.debug("Waiting for domain: {}", domainWait); Thread.sleep(domainWait); } } } catch (Exception e) { // noOp } connection = (HttpURLConnection) new URL(url).openConnection(); // now we are going to pretend that we are a browser... // This is the way my mac works. if (!BOTS_ARE_OK.contains(thisURL.getHost())) { connection.addRequestProperty("Host", thisURL.getHost()); // Bots are not 'ok', so we need to spoof the headers for (String k : SPOOF_HTTP_HEADERS.keySet()) connection.addRequestProperty(k, SPOOF_HTTP_HEADERS.get(k)); // the test to seattlemamadoc.com prompted this change. // they auto detect bots by checking the referrer chain and the 'user-agent' // this broke the t.co test. t.co URLs are EXPLICITLY ok with bots // there is a list for URLS that behave this way at the top in BOTS_ARE_OK // smashew 2013-13-2013 if (linkDetails.getRedirectCount() > 0 && BOTS_ARE_OK.contains(thisURL.getHost())) connection.addRequestProperty("Referrer", linkDetails.getOriginalURL()); } connection.setReadTimeout(DEFAULT_HTTP_TIMEOUT); connection.setConnectTimeout(DEFAULT_HTTP_TIMEOUT); // we want to follow this behavior on our own to ensure that we are getting to the // proper place. This is especially true with links that are wounded by special // link winders, // IE: connection.setInstanceFollowRedirects(false); if (linkDetails.getCookies() != null) for (String cookie : linkDetails.getCookies()) connection.addRequestProperty("Cookie", cookie.split(";", 1)[0]); connection.connect(); linkDetails.setFinalResponseCode((long) connection.getResponseCode()); Map<String, List<String>> headers = createCaseInsensitiveMap(connection.getHeaderFields()); /* * If they want us to set cookies, well, then we will set cookies * Example URL: * http://nyti.ms/1bCpesx *****************************************************************/ if (headers.containsKey(SET_COOKIE_IDENTIFIER)) linkDetails.getCookies().add(headers.get(SET_COOKIE_IDENTIFIER).get(0)); switch (linkDetails.getFinalResponseCode().intValue()) { /* * W3C HTTP Response Codes: * http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html */ case 200: // HTTP OK linkDetails.setFinalURL(connection.getURL().toString()); linkDetails.setDomain(new URL(linkDetails.getFinalURL()).getHost()); linkDetails.setLinkStatus(LinkDetails.LinkStatus.SUCCESS); break; case 300: // Multiple choices case 301: // URI has been moved permanently case 302: // Found case 303: // Primarily for a HTTP Post case 304: // Not Modified case 306: // This status code is unused but in the redirect block. case 307: // Temporary re-direct /* * Author: * Smashew * * Date: 2013-11-15 * * Note: * It is possible that we have already found our final URL. In * the event that we have found our final URL, we are going to * save this URL as long as it isn't the original URL. * We are still going to ask the browser to re-direct, but in the * case of yet another redirect, seen with the redbull test * this can be followed by a 304, a browser, by W3C standards would * still render the page with it's content, but for us to assert * a success, we are really hoping for a 304 message. *******************************************************************/ if (!linkDetails.getOriginalURL().toLowerCase() .equals(connection.getURL().toString().toLowerCase())) linkDetails.setFinalURL(connection.getURL().toString()); if (!headers.containsKey(LOCATION_IDENTIFIER)) { LOGGER.info("Headers: {}", headers); linkDetails.setLinkStatus(LinkDetails.LinkStatus.REDIRECT_ERROR); } else { linkDetails.setRedirected(Boolean.TRUE); linkDetails.setRedirectCount(linkDetails.getRedirectCount() + 1); reDirectedLink = connection.getHeaderField(LOCATION_IDENTIFIER); } break; case 305: // User must use the specified proxy (deprecated by W3C) break; case 401: // Unauthorized (nothing we can do here) linkDetails.setLinkStatus(LinkDetails.LinkStatus.UNAUTHORIZED); break; case 403: // HTTP Forbidden (Nothing we can do here) linkDetails.setLinkStatus(LinkDetails.LinkStatus.FORBIDDEN); break; case 404: // Not Found (Page is not found, nothing we can do with a 404) linkDetails.setLinkStatus(LinkDetails.LinkStatus.NOT_FOUND); break; case 500: // Internal Server Error case 501: // Not Implemented case 502: // Bad Gateway case 503: // Service Unavailable case 504: // Gateway Timeout case 505: // Version not supported linkDetails.setLinkStatus(LinkDetails.LinkStatus.HTTP_ERROR_STATUS); break; default: LOGGER.info("Unrecognized HTTP Response Code: {}", linkDetails.getFinalResponseCode()); linkDetails.setLinkStatus(LinkDetails.LinkStatus.NOT_FOUND); break; } } catch (MalformedURLException e) { // the URL is trash, so, it can't load it. linkDetails.setLinkStatus(LinkDetails.LinkStatus.MALFORMED_URL); } catch (IOException ex) { // there was an issue we are going to set to error. linkDetails.setLinkStatus(LinkDetails.LinkStatus.ERROR); } catch (Exception ex) { // there was an unknown issue we are going to set to exception. linkDetails.setLinkStatus(LinkDetails.LinkStatus.EXCEPTION); } finally { // if the connection is not null, then we need to disconnect to close any underlying resources if (connection != null) connection.disconnect(); } // If there was a redirection, then we have to keep going // Placing this code here should help to satisfy ensuring that the connection object // is closed successfully. if (reDirectedLink != null) unwindLink(reDirectedLink); }
From source file:org.jaqpot.core.service.resource.AlgorithmResource.java
@POST
@Produces({ MediaType.APPLICATION_JSON, "text/uri-list" })
@Path("/{id}")
@ApiOperation(value = "Creates Model", notes = "Applies Dataset and Parameters on Algorithm and creates Model.", response = Task.class)
@org.jaqpot.core.service.annotations.Task
public Response trainModel(@ApiParam(name = "title", required = true) @FormParam("title") String title,
@ApiParam(name = "description", required = true) @FormParam("description") String description,
@ApiParam(name = "dataset_uri", defaultValue = DEFAULT_DATASET) @FormParam("dataset_uri") String datasetURI,
@ApiParam(name = "prediction_feature", defaultValue = DEFAULT_PRED_FEATURE) @FormParam("prediction_feature") String predictionFeature,
@FormParam("parameters") String parameters,
@ApiParam(name = "transformations", defaultValue = DEFAULT_TRANSFORMATIONS) @FormParam("transformations") String transformations,
@ApiParam(name = "scaling", defaultValue = STANDARIZATION) @FormParam("scaling") String scaling, //, allowableValues = SCALING + "," + STANDARIZATION
@ApiParam(name = "doa", defaultValue = DEFAULT_DOA) @FormParam("doa") String doa,
@PathParam("id") String algorithmId, @HeaderParam("subjectid") String subjectId)
throws QuotaExceededException, ParameterIsNullException, ParameterInvalidURIException,
ParameterTypeException, ParameterRangeException, ParameterScopeException {
UrlValidator urlValidator = new UrlValidator();
Algorithm algorithm = algorithmHandler.find(algorithmId);
if (algorithm == null) {
throw new NotFoundException("Could not find Algorithm with id:" + algorithmId);
}//from ww w .jav a2s . c om
//Dataset validation should happen only in regression and classification algorithms
if (algorithm.getOntologicalClasses().contains("ot:Regression")
|| algorithm.getOntologicalClasses().contains("ot:Classification")) {
if (datasetURI == null) {
throw new ParameterIsNullException("datasetURI");
}
if (!urlValidator.isValid(datasetURI)) {
throw new ParameterInvalidURIException("Not valid Dataset URI.");
}
String datasetId = datasetURI.split("dataset/")[1];
Dataset datasetMeta = datasetHandler.findMeta(datasetId);
if (datasetMeta.getTotalRows() != null && datasetMeta.getTotalRows() < 2) {
throw new BadRequestException("Cannot train model on dataset with less than 2 rows.");
}
}
//Prediction validation should not happen in enm:NoTarget algorithms
if (algorithm.getOntologicalClasses().contains("enm:NoTarget")) {
if (predictionFeature == null) {
throw new ParameterIsNullException("predictionFeature");
}
if (!urlValidator.isValid(predictionFeature)) {
throw new ParameterInvalidURIException("Not valid Prediction Feature URI.");
}
}
if (title == null) {
throw new ParameterIsNullException("title");
}
if (description == null) {
throw new ParameterIsNullException("description");
}
User user = userHandler.find(securityContext.getUserPrincipal().getName());
long modelCount = modelHandler.countAllOfCreator(user.getId());
int maxAllowedModels = new UserFacade(user).getMaxModels();
if (modelCount > maxAllowedModels) {
LOG.info(String.format("User %s has %d models while maximum is %d", user.getId(), modelCount,
maxAllowedModels));
throw new QuotaExceededException("Dear " + user.getId()
+ ", your quota has been exceeded; you already have " + modelCount + " models. "
+ "No more than " + maxAllowedModels + " are allowed with your subscription.");
}
Map<String, Object> options = new HashMap<>();
options.put("title", title);
options.put("description", description);
options.put("dataset_uri", datasetURI);
options.put("prediction_feature", predictionFeature);
options.put("subjectid", subjectId);
options.put("algorithmId", algorithmId);
options.put("parameters", parameters);
options.put("base_uri", uriInfo.getBaseUri().toString());
options.put("creator", securityContext.getUserPrincipal().getName());
Map<String, String> transformationAlgorithms = new LinkedHashMap<>();
if (transformations != null && !transformations.isEmpty()) {
transformationAlgorithms.put(uriInfo.getBaseUri().toString() + "algorithm/pmml",
"{\"transformations\" : \"" + transformations + "\"}");
}
if (scaling != null && !scaling.isEmpty()) {
transformationAlgorithms.put(scaling, "");
}
if (doa != null && !doa.isEmpty()) {
transformationAlgorithms.put(doa, "");
}
if (!transformationAlgorithms.isEmpty()) {
String transformationAlgorithmsString = serializer.write(transformationAlgorithms);
LOG.log(Level.INFO, "Transformations:{0}", transformationAlgorithmsString);
options.put("transformations", transformationAlgorithmsString);
}
parameterValidator.validate(parameters, algorithm.getParameters());
//return Response.ok().build();
Task task = trainingService.initiateTraining(options, securityContext.getUserPrincipal().getName());
return Response.ok(task).build();
}
From source file:org.jaqpot.core.service.resource.DatasetResource.java
@POST @Path("/{id}/qprf") @ApiOperation("Creates QPRF Report") @Authorize// ww w .jav a 2s .c om public Response createQPRFReport( @ApiParam(value = "Authorization token") @HeaderParam("subjectid") String subjectId, @PathParam("id") String id, @FormParam("substance_uri") String substanceURI, @FormParam("title") String title, @FormParam("description") String description) throws QuotaExceededException { User user = userHandler.find(securityContext.getUserPrincipal().getName()); long reportCount = reportHandler.countAllOfCreator(user.getId()); int maxAllowedReports = new UserFacade(user).getMaxReports(); if (reportCount > maxAllowedReports) { LOG.info(String.format("User %s has %d algorithms while maximum is %d", user.getId(), reportCount, maxAllowedReports)); throw new QuotaExceededException("Dear " + user.getId() + ", your quota has been exceeded; you already have " + reportCount + " reports. " + "No more than " + maxAllowedReports + " are allowed with your subscription."); } Dataset ds = datasetHandler.find(id); if (ds == null) { throw new NotFoundException("Dataset with id:" + id + " was not found on the server."); } if (ds.getByModel() == null || ds.getByModel().isEmpty()) { throw new BadRequestException("Selected dataset was not produced by a valid model."); } Model model = modelHandler.find(ds.getByModel()); if (model == null) { throw new BadRequestException("Selected dataset was not produced by a valid model."); } String datasetURI = model.getDatasetUri(); if (datasetURI == null || datasetURI.isEmpty()) { throw new BadRequestException( "The model that created this dataset does not point to a valid training dataset."); } Dataset trainingDS = client.target(datasetURI).request().accept(MediaType.APPLICATION_JSON) .header("subjectid", subjectId).get(Dataset.class); if (trainingDS == null) { throw new BadRequestException( "The model that created this dataset does not point to a valid training dataset."); } if (model.getTransformationModels() != null) { for (String transModelURI : model.getTransformationModels()) { Model transModel = modelHandler.find(transModelURI.split("model/")[1]); if (transModel == null) { throw new NotFoundException( "Transformation model with id:" + transModelURI + " was not found."); } try { trainingDS = jpdiClient .predict(trainingDS, transModel, trainingDS.getMeta(), UUID.randomUUID().toString()) .get(); } catch (InterruptedException ex) { LOG.log(Level.SEVERE, "JPDI Training procedure interupted", ex); throw new InternalServerErrorException("JPDI Training procedure interupted", ex); } catch (ExecutionException ex) { LOG.log(Level.SEVERE, "Training procedure execution error", ex.getCause()); throw new InternalServerErrorException("JPDI Training procedure error", ex.getCause()); } catch (CancellationException ex) { throw new InternalServerErrorException("Procedure was cancelled"); } } } List<String> retainableFeatures = new ArrayList<>(model.getIndependentFeatures()); retainableFeatures.addAll(model.getDependentFeatures()); trainingDS.getDataEntry().parallelStream().forEach(dataEntry -> { dataEntry.getValues().keySet().retainAll(retainableFeatures); }); DataEntry dataEntry = ds.getDataEntry().stream() .filter(de -> de.getCompound().getURI().equals(substanceURI)).findFirst() .orElseThrow(() -> new BadRequestException("")); trainingDS.getDataEntry().add(dataEntry); trainingDS.getMeta().setCreators(new HashSet<>(Arrays.asList(user.getId()))); Map<String, Object> parameters = new HashMap<>(); UrlValidator urlValidator = new UrlValidator(); if (urlValidator.isValid(substanceURI)) { Dataset structures = client.target(substanceURI + "/structures").request() .accept(MediaType.APPLICATION_JSON).header("subjectid", subjectId).get(Dataset.class); List<Map<String, String>> structuresList = structures.getDataEntry().stream().map(de -> { String compound = de.getCompound().getURI(); String casrn = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23CASRNDefault")) .orElse("").toString(); String einecs = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23EINECSDefault")) .orElse("").toString(); String iuclid5 = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23IUCLID5_UUIDDefault")) .orElse("").toString(); String inchi = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23InChI_stdDefault")) .orElse("").toString(); String reach = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23REACHRegistrationDateDefault")) .orElse("").toString(); String iupac = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23IUPACNameDefault")) .orElse("").toString(); Map<String, String> structuresMap = new HashMap<>(); structuresMap.put("Compound", compound); structuresMap.put("CasRN", casrn); structuresMap.put("EC number", einecs); structuresMap.put("REACH registration date", reach); structuresMap.put("IUCLID 5 Reference substance UUID", iuclid5); structuresMap.put("Std. InChI", inchi); structuresMap.put("IUPAC name", iupac); return structuresMap; }).collect(Collectors.toList()); parameters.put("structures", structuresList); } else { List<Map<String, String>> structuresList = new ArrayList<>(); Map<String, String> structuresMap = new HashMap<>(); structuresMap.put("Compound", ""); structuresMap.put("CasRN", ""); structuresMap.put("EC number", ""); structuresMap.put("REACH registration date", ""); structuresMap.put("IUCLID 5 Reference substance UUID", ""); structuresMap.put("Std. InChI", ""); structuresMap.put("IUPAC name", ""); structuresList.add(structuresMap); parameters.put("structures", structuresList); } parameters.put("predictedFeature", model.getPredictedFeatures().stream().findFirst() .orElseThrow(() -> new BadRequestException("Model does not have a valid predicted feature"))); parameters.put("algorithm", algorithmHandler.find(model.getAlgorithm().getId())); parameters.put("substanceURI", substanceURI); if (model.getLinkedModels() != null && !model.getLinkedModels().isEmpty()) { Model doa = modelHandler.find(model.getLinkedModels().get(0).split("model/")[1]); if (doa != null) { parameters.put("doaURI", doa.getPredictedFeatures().get(0)); parameters.put("doaMethod", doa.getAlgorithm().getId()); } } TrainingRequest request = new TrainingRequest(); request.setDataset(trainingDS); request.setParameters(parameters); request.setPredictionFeature(model.getDependentFeatures().stream().findFirst() .orElseThrow(() -> new BadRequestException("Model does not have a valid prediction feature"))); Report report = client.target("http://147.102.82.32:8094/pws/qprf").request() .header("Content-Type", MediaType.APPLICATION_JSON).accept(MediaType.APPLICATION_JSON) .post(Entity.json(request), Report.class); report.setMeta(MetaInfoBuilder.builder().addTitles(title).addDescriptions(description) .addCreators(securityContext.getUserPrincipal().getName()).build()); report.setId(new ROG(true).nextString(15)); report.setVisible(Boolean.TRUE); reportHandler.create(report); return Response.ok(report).build(); }
From source file:org.jaqpot.core.service.resource.DatasetResource.java
@POST @Path("/{id}/qprf-dummy") @ApiOperation("Creates QPRF Report") @Authorize/* w w w . j a va 2s. c o m*/ public Response createQPRFReportDummy( @ApiParam(value = "Authorization token") @HeaderParam("subjectid") String subjectId, @PathParam("id") String id, @FormParam("substance_uri") String substanceURI, @FormParam("title") String title, @FormParam("description") String description) { Dataset ds = datasetHandler.find(id); if (ds == null) { throw new NotFoundException("Dataset with id:" + id + " was not found on the server."); } if (ds.getByModel() == null || ds.getByModel().isEmpty()) { throw new BadRequestException("Selected dataset was not produced by a valid model."); } Model model = modelHandler.find(ds.getByModel()); if (model == null) { throw new BadRequestException("Selected dataset was not produced by a valid model."); } String datasetURI = model.getDatasetUri(); if (datasetURI == null || datasetURI.isEmpty()) { throw new BadRequestException( "The model that created this dataset does not point to a valid training dataset."); } Dataset trainingDS = client.target(datasetURI).request().accept(MediaType.APPLICATION_JSON) .header("subjectid", subjectId).get(Dataset.class); if (trainingDS == null) { throw new BadRequestException( "The model that created this dataset does not point to a valid training dataset."); } if (model.getTransformationModels() != null) { for (String transModelURI : model.getTransformationModels()) { Model transModel = modelHandler.find(transModelURI.split("model/")[1]); if (transModel == null) { throw new NotFoundException( "Transformation model with id:" + transModelURI + " was not found."); } try { trainingDS = jpdiClient .predict(trainingDS, transModel, trainingDS.getMeta(), UUID.randomUUID().toString()) .get(); } catch (InterruptedException ex) { LOG.log(Level.SEVERE, "JPDI Training procedure interupted", ex); throw new InternalServerErrorException("JPDI Training procedure interupted", ex); } catch (ExecutionException ex) { LOG.log(Level.SEVERE, "Training procedure execution error", ex.getCause()); throw new InternalServerErrorException("JPDI Training procedure error", ex.getCause()); } catch (CancellationException ex) { throw new InternalServerErrorException("Procedure was cancelled"); } } } List<String> retainableFeatures = new ArrayList<>(model.getIndependentFeatures()); retainableFeatures.addAll(model.getDependentFeatures()); trainingDS.getDataEntry().parallelStream().forEach(dataEntry -> { dataEntry.getValues().keySet().retainAll(retainableFeatures); }); DataEntry dataEntry = ds.getDataEntry().stream() .filter(de -> de.getCompound().getURI().equals(substanceURI)).findFirst() .orElseThrow(() -> new BadRequestException("")); trainingDS.getDataEntry().add(dataEntry); Map<String, Object> parameters = new HashMap<>(); UrlValidator urlValidator = new UrlValidator(); if (urlValidator.isValid(substanceURI)) { Dataset structures = client.target(substanceURI + "/structures").request() .accept(MediaType.APPLICATION_JSON).header("subjectid", subjectId).get(Dataset.class); List<Map<String, String>> structuresList = structures.getDataEntry().stream().map(de -> { String compound = de.getCompound().getURI(); String casrn = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23CASRNDefault")) .orElse("").toString(); String einecs = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23EINECSDefault")) .orElse("").toString(); String iuclid5 = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23IUCLID5_UUIDDefault")) .orElse("").toString(); String inchi = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23InChI_stdDefault")) .orElse("").toString(); String reach = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23REACHRegistrationDateDefault")) .orElse("").toString(); String iupac = Optional.ofNullable(de.getValues().get( "https://apps.ideaconsult.net/enmtest/feature/http%3A%2F%2Fwww.opentox.org%2Fapi%2F1.1%23IUPACNameDefault")) .orElse("").toString(); Map<String, String> structuresMap = new HashMap<>(); structuresMap.put("Compound", compound); structuresMap.put("CasRN", casrn); structuresMap.put("EC number", einecs); structuresMap.put("REACH registration date", reach); structuresMap.put("IUCLID 5 Reference substance UUID", iuclid5); structuresMap.put("Std. InChI", inchi); structuresMap.put("IUPAC name", iupac); return structuresMap; }).collect(Collectors.toList()); parameters.put("structures", structuresList); } else { List<Map<String, String>> structuresList = new ArrayList<>(); Map<String, String> structuresMap = new HashMap<>(); structuresMap.put("Compound", ""); structuresMap.put("CasRN", ""); structuresMap.put("EC number", ""); structuresMap.put("REACH registration date", ""); structuresMap.put("IUCLID 5 Reference substance UUID", ""); structuresMap.put("Std. InChI", ""); structuresMap.put("IUPAC name", ""); structuresList.add(structuresMap); parameters.put("structures", structuresList); } parameters.put("predictedFeature", model.getPredictedFeatures().stream().findFirst() .orElseThrow(() -> new BadRequestException("Model does not have a valid predicted feature"))); parameters.put("algorithm", algorithmHandler.find(model.getAlgorithm().getId())); parameters.put("substanceURI", substanceURI); if (model.getLinkedModels() != null && !model.getLinkedModels().isEmpty()) { Model doa = modelHandler.find(model.getLinkedModels().get(0).split("model/")[1]); if (doa != null) { parameters.put("doaURI", doa.getPredictedFeatures().get(0)); parameters.put("doaMethod", doa.getAlgorithm().getId()); } } TrainingRequest request = new TrainingRequest(); request.setDataset(trainingDS); request.setParameters(parameters); request.setPredictionFeature(model.getDependentFeatures().stream().findFirst() .orElseThrow(() -> new BadRequestException("Model does not have a valid prediction feature"))); return Response.ok(request).build(); // Report report = client.target("http://147.102.82.32:8094/pws/qprf") // .request() // .header("Content-Type", MediaType.APPLICATION_JSON) // .accept(MediaType.APPLICATION_JSON) // .post(Entity.json(request), Report.class); // // report.setMeta(MetaInfoBuilder.builder() // .addTitles(title) // .addDescriptions(description) // .addCreators(securityContext.getUserPrincipal().getName()) // .build() // ); // report.setId(new ROG(true).nextString(15)); // report.setVisible(Boolean.TRUE); // reportHandler.create(report); // // return Response.ok(report).build(); }