List of usage examples for java.util.regex Pattern CASE_INSENSITIVE
int CASE_INSENSITIVE
To view the source code for java.util.regex Pattern CASE_INSENSITIVE.
Click Source Link
From source file:uk.ac.kcl.it.DeIdentificationPKPartitionWithoutScheduling.java
@Test public void deidentificationPerformanceTest() { dbmsTestUtils.createBasicInputTable(); dbmsTestUtils.createBasicOutputTable(); dbmsTestUtils.createDeIdInputTable(); List<Mutant> mutants = testUtils.insertTestDataForDeidentification(env.getProperty("tblIdentifiers"), env.getProperty("tblInputDocs"), mutatortype); int totalTruePositives = 0; int totalFalsePositives = 0; int totalFalseNegatives = 0; for (Mutant mutant : mutants) { Set<Pattern> mutatedPatterns = new HashSet<>(); mutant.setDeidentifiedString(elasticGazetteerService.deIdentifyString(mutant.getFinalText(), String.valueOf(mutant.getDocumentid()))); Set<String> set = new HashSet<>(mutant.getOutputTokens()); mutatedPatterns.addAll(//w w w . j a v a2 s. co m set.stream().map(string -> Pattern.compile(Pattern.quote(string), Pattern.CASE_INSENSITIVE)) .collect(Collectors.toSet())); List<MatchResult> results = new ArrayList<>(); for (Pattern pattern : mutatedPatterns) { Matcher matcher = pattern.matcher(mutant.getFinalText()); while (matcher.find()) { results.add(matcher.toMatchResult()); } } int truePositives = getTruePositiveTokenCount(mutant); int falsePositives = getFalsePositiveTokenCount(mutant); int falseNegatives = getFalseNegativeTokenCount(mutant); System.out.println("Doc ID " + mutant.getDocumentid() + " has " + falseNegatives + " unmasked identifiers from a total of " + (falseNegatives + truePositives)); System.out.println("Doc ID " + mutant.getDocumentid() + " has " + falsePositives + " inaccurately masked tokens from a total of " + (falsePositives + truePositives)); System.out.println("TP: " + truePositives + " FP: " + falsePositives + " FN: " + falseNegatives); System.out.println("Doc ID precision " + calcPrecision(falsePositives, truePositives)); System.out.println("Doc ID recall " + calcRecall(falseNegatives, truePositives)); System.out.println(mutant.getDeidentifiedString()); System.out.println(mutant.getFinalText()); System.out.println(mutant.getInputTokens()); System.out.println(mutant.getOutputTokens()); System.out.println(); if (env.getProperty("elasticgazetteerTestOutput") != null) { try { try (BufferedWriter bw = new BufferedWriter( new FileWriter(new File(env.getProperty("elasticgazetteerTestOutput") + File.separator + mutant.getDocumentid())))) { bw.write("Doc ID " + mutant.getDocumentid() + " has " + falseNegatives + " unmasked identifiers from a total of " + (falseNegatives + truePositives)); bw.newLine(); bw.write("Doc ID " + mutant.getDocumentid() + " has " + falsePositives + " inaccurately masked tokens from a total of " + (falsePositives + truePositives)); bw.newLine(); bw.write("TP: " + truePositives + " FP: " + falsePositives + " FN: " + falseNegatives); bw.newLine(); bw.write("Doc ID precision " + calcPrecision(falsePositives, truePositives)); bw.newLine(); bw.write("Doc ID recall " + calcRecall(falseNegatives, truePositives)); bw.newLine(); bw.write(mutant.getDeidentifiedString()); bw.newLine(); bw.write(mutant.getFinalText()); bw.newLine(); bw.write(mutant.getInputTokens().toString()); bw.newLine(); bw.write(mutant.getOutputTokens().toString()); } } catch (IOException e) { e.printStackTrace(); } } totalTruePositives += truePositives; totalFalsePositives += falsePositives; totalFalseNegatives += falseNegatives; } System.out.println(); System.out.println(); System.out.println("THIS RUN TP: " + totalTruePositives + " FP: " + totalFalsePositives + " FN: " + totalFalseNegatives); System.out.println("Doc ID precision " + calcPrecision(totalFalsePositives, totalTruePositives)); System.out.println("Doc ID recall " + calcRecall(totalFalseNegatives, totalTruePositives)); if (env.getProperty("elasticgazetteerTestOutput") != null) { try { try (BufferedWriter bw = new BufferedWriter(new FileWriter( new File(env.getProperty("elasticgazetteerTestOutput") + File.separator + "summary")))) { bw.write("THIS RUN TP: " + totalTruePositives + " FP: " + totalFalsePositives + " FN: " + totalFalseNegatives); bw.newLine(); bw.write("Doc ID precision " + calcPrecision(totalFalsePositives, totalTruePositives)); bw.newLine(); bw.write("Doc ID recall " + calcRecall(totalFalseNegatives, totalTruePositives)); } } catch (IOException e) { e.printStackTrace(); } } }
From source file:de.unirostock.sems.cbarchive.web.HttpImporter.java
private File downloadFile() throws ImporterException { try {//from w w w.j av a2 s. c o m tempFile = File.createTempFile(Fields.TEMP_FILE_PREFIX, ".omex"); HttpResponse getResponse = client.execute(new HttpGet(remoteUrl)); // check if file exists if (getResponse.getStatusLine().getStatusCode() != 200) { LOGGER.warn(getResponse.getStatusLine().getStatusCode(), " ", getResponse.getStatusLine().getReasonPhrase(), " while download ", remoteUrl); throw new ImporterException(String.valueOf(getResponse.getStatusLine().getStatusCode()) + " " + getResponse.getStatusLine().getReasonPhrase() + " while download"); } HttpEntity entity = getResponse.getEntity(); if (entity == null) { LOGGER.error("No content returned while donwloading remote file ", remoteUrl); throw new ImporterException("No content returned while donwloading remote file " + remoteUrl); } // for name suggestions Header dispositionHeader = getResponse.getFirstHeader("Content-Disposition"); if (dispositionHeader != null && dispositionHeader.getValue() != null && dispositionHeader.getValue().isEmpty() == false) { // disposition header is present -> extract name // inline; filename=\"{0}.{1}\" Matcher matcher = Pattern .compile("filename=\\\"?(([a-zA-Z0-9-_\\+]+).(\\w+))\\\"?", Pattern.CASE_INSENSITIVE) .matcher(dispositionHeader.getValue()); if (matcher.find()) { suggestedName = matcher.group(1); for (int i = 0; i < matcher.groupCount(); i++) LOGGER.debug(i, ": ", matcher.group(i)); } } else { // when not -> take the last part of the url Matcher matcher = Pattern.compile("\\/(([a-zA-Z0-9-_\\+]+).(\\w+))$", Pattern.CASE_INSENSITIVE) .matcher(remoteUrl); if (matcher.find()) { suggestedName = matcher.group(1); for (int i = 0; i < matcher.groupCount(); i++) LOGGER.debug(i, ": ", matcher.group(i)); } } // download it OutputStream output = new FileOutputStream(tempFile); IOUtils.copy(entity.getContent(), output); // check against quota if (length != tempFile.length()) { LOGGER.warn("Content-Length (", length, ") and downloaded length (", tempFile.length(), ") are different."); length = tempFile.length(); checkQuotas(); } return tempFile; } catch (IOException e) { LOGGER.error(e, "Exception while download file from ", remoteUrl); throw new ImporterException("Exception while download remote file", e); } }
From source file:com.redhat.rhn.domain.channel.NewChannelHelper.java
/** * Verifies a potential name for a channel * @param name the name of the channel/* w ww.j a va 2 s . com*/ * @return true if it is correct, false otherwise */ public static boolean verifyName(String name) { if (name.length() < 6) { return false; } Pattern pattern = Pattern.compile("^(rhn|red\\s*hat).*", Pattern.CASE_INSENSITIVE); Matcher match = pattern.matcher(name); if (match.matches()) { return false; } pattern = Pattern.compile("^[a-z][\\w\\d\\s\\-\\.\\'\\(\\)\\/\\_]*$", Pattern.CASE_INSENSITIVE); match = pattern.matcher(name); if (!match.matches()) { return false; } return true; }
From source file:com.mycompany.myproject.xemailservice.tests.XEmailIntegrationIT.java
@Test public void testGetFormFields() { RequestExecutor exec = null;//from w w w . j av a 2 s .c o m try { String action = "mcm/components/emailserviceactions/actions/addSubscriber"; String path = "/_jcr_content.emailservice.json?operation=getFormFields&cfgpath=" + XConfigPath + "&actionType=" + action; exec = authorAdmin.http.doGet(path, SC_OK); String content = exec.getContent(); assertNotNull(content); final Pattern p = Pattern.compile("\"type\":\"email\"", Pattern.CASE_INSENSITIVE); if (!p.matcher(content).find()) fail("Couldnot retreive Attribute--Email"); } catch (Exception e) { Assert.fail("Could not retreive default attribute email from XMailService" + e.getMessage()); } }
From source file:wuit.common.crawler.WebSit.Crawler.java
public static void matchValues(String content, String filter, List<KeyValue> list) { if (list == null) list = new ArrayList<KeyValue>(); try {//from w w w. jav a2 s . c o m Matcher m = Pattern.compile(filter, Pattern.CASE_INSENSITIVE | Pattern.MULTILINE).matcher(content); while (m.find()) { if (m.group().isEmpty()) continue; KeyValue value = new KeyValue(); value.value = m.group(); value.start = m.start(); value.end = m.end(); list.add(value); } } catch (Exception e) { System.out.println("Crawler Utitles matchValues :" + e.getMessage()); } }
From source file:gsn.storage.SQLUtils.java
public static StringBuilder newRewrite(CharSequence query, CharSequence tableNameToRename, CharSequence replaceTo) { // Selecting strings between pair of "" : (\"[^\"]*\") // Selecting tableID.tableName or tableID.* : (\\w+(\\.(\w+)|\\*)) // The combined pattern is : (\"[^\"]*\")|(\\w+\\.((\\w+)|\\*)) Matcher matcher = pattern.matcher(query); StringBuffer result = new StringBuffer(); while (matcher.find()) { if (matcher.group(2) == null) continue; String tableName = matcher.group(3); if (tableName.equals(tableNameToRename)) { // $4 means that the 4th group of the match should be appended to the // string (the forth group contains the field name). if (replaceTo != null) matcher.appendReplacement(result, new StringBuilder(replaceTo).append("$4").toString()); }/*from w w w .ja v a2s . co m*/ } String toReturn = matcher.appendTail(result).toString().toLowerCase(); int indexOfFrom = toReturn.indexOf(" from ") >= 0 ? toReturn.indexOf(" from ") + " from ".length() : 0; int indexOfWhere = (toReturn.lastIndexOf(" where ") > 0 ? (toReturn.lastIndexOf(" where ")) : toReturn.length()); String selection = toReturn.substring(indexOfFrom, indexOfWhere); Pattern fromClausePattern = Pattern.compile("\\s*(\\w+)\\s*", Pattern.CASE_INSENSITIVE); Matcher fromClauseMather = fromClausePattern.matcher(selection); result = new StringBuffer(); while (fromClauseMather.find()) { if (fromClauseMather.group(1) == null) continue; String tableName = fromClauseMather.group(1); if (tableName.equals(tableNameToRename) && replaceTo != null) fromClauseMather.appendReplacement(result, replaceTo.toString() + " "); } String cleanFromClause = fromClauseMather.appendTail(result).toString(); String finalResult = StringUtils.replace(toReturn, selection, cleanFromClause); return new StringBuilder(finalResult); }
From source file:csiro.pidsvc.core.Settings.java
public boolean isNewVersionAvailable() { if (_manifest == null) return false; String content = Http.simpleGetRequest(getProperty("buildRepository")); Pattern re = Pattern.compile("href=\"pidsvc-(\\d+\\.\\d+)(?:-SNAPSHOT)?\\.(.+?)\\.war\"", Pattern.CASE_INSENSITIVE); Matcher m = re.matcher(content); try {/*from w ww . j av a 2 s . co m*/ if (m.find()) { String currentVersion = _manifest.getMainAttributes().getValue("Implementation-Build"); String newVersion = m.group(2); if (!currentVersion.isEmpty() && !newVersion.equalsIgnoreCase(currentVersion)) return true; } } catch (Exception e) { _logger.debug(e); } return false; }
From source file:org.shredzone.cilla.service.impl.PageServiceImpl.java
@Override public boolean isAcceptedResponse(Page page, String response) { if (!page.isRestricted()) { // If the page is not restricted, the response is always acceptable return true; }//from w w w . j a v a2 s . c om Pattern pattern = Pattern.compile(page.getResponsePattern(), Pattern.CASE_INSENSITIVE); return (pattern.matcher(response).matches()); }
From source file:com.hangum.tadpole.engine.restful.RESTfulAPIUtils.java
/** * user validate/* w w w . jav a 2s .co m*/ * * @param url * @return */ public static boolean validateURL(String url) { Pattern p = Pattern.compile("[/][-A-Za-z0-9+&@#/%=~_()|]{2}", Pattern.CASE_INSENSITIVE); Matcher m = p.matcher(url); return m.find(); }
From source file:net.duckling.ddl.service.resource.impl.FolderPathServiceImpl.java
private String getResourceName(List<Resource> rs, String fileName, String itemType) { if (rs == null || rs.isEmpty()) { return fileName; }/*from www . jav a 2 s.c o m*/ int max = 0; String reg = getPattenName(fileName, itemType, "\\((\\d+)\\)"); try { Pattern pattern = Pattern.compile(reg, Pattern.CASE_INSENSITIVE); for (Resource file : rs) { Matcher ma = pattern.matcher(file.getTitle()); if (ma.matches()) { String id = ma.group(1); try { int tmp = Integer.parseInt(id); if (tmp > max) { max = tmp; } } catch (RuntimeException e) { } } } } catch (RuntimeException e) { LOG.error("", e); return fileName; } max++; return getQueryName(fileName, itemType, "(" + max + ")"); }