List of usage examples for java.util.regex Matcher replaceAll
public String replaceAll(Function<MatchResult, String> replacer)
From source file:org.languagetool.JLanguageTool.java
private Map<Integer, String> replaceSoftHyphens(List<String> tokens) { Pattern ignoredCharacterRegex = language.getIgnoredCharactersRegex(); Map<Integer, String> ignoredCharsTokens = new HashMap<>(); if (ignoredCharacterRegex == null) { return ignoredCharsTokens; }//www . j a va2 s . c om for (int i = 0; i < tokens.size(); i++) { Matcher matcher = ignoredCharacterRegex.matcher(tokens.get(i)); if (matcher.find()) { ignoredCharsTokens.put(i, tokens.get(i)); tokens.set(i, matcher.replaceAll("")); } } return ignoredCharsTokens; }
From source file:net.longfalcon.newsj.FetchBinaries.java
@Transactional(propagation = Propagation.REQUIRED, isolation = Isolation.READ_COMMITTED) public long scan(NewsClient nntpClient, Group group, long firstArticle, long lastArticle, String type, boolean compressedHeaders) throws IOException { // this is a hack - tx is not working ATM TransactionStatus transaction = transactionManager .getTransaction(new DefaultTransactionDefinition(TransactionDefinition.PROPAGATION_REQUIRED)); long startHeadersTime = System.currentTimeMillis(); long maxNum = 0; Map<String, Message> messages = new LinkedHashMap<>(MESSAGE_BUFFER + 1); Iterable<NewsArticle> articlesIterable = null; try {/*w w w . j a v a 2 s .c o m*/ if (compressedHeaders) { _log.warn("Compressed Headers setting not currently functional"); articlesIterable = nntpClient.iterateArticleInfo(firstArticle, lastArticle); } else { articlesIterable = nntpClient.iterateArticleInfo(firstArticle, lastArticle); } } catch (IOException e) { _log.error(e.toString()); if (nntpClient.getReplyCode() == 400) { _log.info("NNTP connection timed out. Reconnecting..."); nntpClient = nntpConnectionFactory.getNntpClient(); nntpClient.selectNewsgroup(group.getName()); articlesIterable = nntpClient.iterateArticleInfo(firstArticle, lastArticle); } } Period headersTime = new Period(startHeadersTime, System.currentTimeMillis()); Set<Long> rangeRequested = ArrayUtil.rangeSet(firstArticle, lastArticle); Set<Long> messagesReceived = new HashSet<>(); Set<Long> messagesBlacklisted = new HashSet<>(); Set<Long> messagesIgnored = new HashSet<>(); Set<Long> messagesInserted = new HashSet<>(); Set<Long> messagesNotInserted = new HashSet<>(); // check error codes? long startUpdateTime = System.currentTimeMillis(); if (articlesIterable != null) { for (NewsArticle article : articlesIterable) { long articleNumber = article.getArticleNumberLong(); if (articleNumber == 0) { continue; } messagesReceived.add(articleNumber); Pattern pattern = Defaults.PARTS_SUBJECT_REGEX; String subject = article.getSubject(); Matcher matcher = pattern.matcher(subject); if (ValidatorUtil.isNull(subject) || !matcher.find()) { // not a binary post most likely.. continue messagesIgnored.add(articleNumber); if (_log.isDebugEnabled()) { _log.debug(String.format("Skipping message no# %s : %s", articleNumber, subject)); } continue; } //Filter binaries based on black/white list if (isBlacklisted(article, group)) { messagesBlacklisted.add(articleNumber); continue; } String group1 = matcher.group(1); String group2 = matcher.group(2); if (ValidatorUtil.isNumeric(group1) && ValidatorUtil.isNumeric(group2)) { int currentPart = Integer.parseInt(group1); int maxParts = Integer.parseInt(group2); subject = (matcher.replaceAll("")).trim(); if (!messages.containsKey(subject)) { messages.put(subject, new Message(article, currentPart, maxParts)); } else if (currentPart > 0) { Message message = messages.get(subject); String articleId = article.getArticleId(); String messageId = articleId.substring(1, articleId.length() - 1); int size = article.getSize(); message.addPart(currentPart, messageId, articleNumber, size); messages.put(subject, message); } } } long count = 0; long updateCount = 0; long partCount = 0; maxNum = lastArticle; // add all the requested then remove the ones we did receive. Set<Long> rangeNotRecieved = new HashSet<>(); rangeNotRecieved.addAll(rangeRequested); rangeNotRecieved.removeAll(messagesReceived); if (!type.equals("partrepair")) { _log.info(String.format("Received %d articles of %d requested, %d blacklisted, %d not binaries", messagesReceived.size(), lastArticle - firstArticle + 1, messagesBlacklisted.size(), messagesIgnored.size())); } if (rangeNotRecieved.size() > 0) { switch (type) { case "backfill": // don't add missing articles break; case "partrepair": case "update": default: addMissingParts(rangeNotRecieved, group); break; } _log.info("Server did not return article numbers " + ArrayUtil.stringify(rangeNotRecieved)); } if (!messages.isEmpty()) { long dbUpdateTime = 0; maxNum = firstArticle; //insert binaries and parts into database. when binary already exists; only insert new parts for (Map.Entry<String, Message> entry : messages.entrySet()) { String subject = entry.getKey(); Message message = entry.getValue(); Map<Integer, MessagePart> partsMap = message.getPartsMap(); if (!ValidatorUtil.isNull(subject) && !partsMap.isEmpty()) { String binaryHash = EncodingUtil .md5Hash(subject + message.getFrom() + String.valueOf(group.getId())); Binary binary = binaryDAO.findByBinaryHash(binaryHash); if (binary == null) { long startDbUpdateTime = System.currentTimeMillis(); binary = new Binary(); binary.setName(subject); binary.setFromName(message.getFrom()); binary.setDate(message.getDate().toDate()); binary.setXref(message.getxRef()); binary.setTotalParts(message.getMaxParts()); binary.setGroupId(group.getId()); binary.setBinaryHash(binaryHash); binary.setDateAdded(new Date()); binaryDAO.updateBinary(binary); dbUpdateTime += (System.currentTimeMillis() - startDbUpdateTime); count++; if (count % 500 == 0) { _log.info(String.format("%s bin adds...", count)); } } else { updateCount++; if (updateCount % 500 == 0) { _log.info(String.format("%s bin updates...", updateCount)); } } long binaryId = binary.getId(); if (binaryId == 0) { throw new RuntimeException("ID for binary wasnt set."); } for (MessagePart messagePart : message.getPartsMap().values()) { long articleNumber = messagePart.getArticleNumber(); maxNum = (articleNumber > maxNum) ? articleNumber : maxNum; partCount++; // create part - its possible some bugs are happening here. Part part = new Part(); part.setBinaryId(binaryId); part.setMessageId(messagePart.getMessageId()); part.setNumber(messagePart.getArticleNumber()); part.setPartNumber(messagePart.getPartNumber()); part.setSize(messagePart.getSize()); part.setDateAdded(new Date()); try { long startDbUpdateTime = System.currentTimeMillis(); partDAO.updatePart(part); dbUpdateTime += (System.currentTimeMillis() - startDbUpdateTime); messagesInserted.add(messagePart.getArticleNumber()); } catch (Exception e) { _log.error(e.toString()); messagesNotInserted.add(messagePart.getArticleNumber()); } } } } //TODO: determine whether to add to missing articles if insert failed if (messagesNotInserted.size() > 0) { _log.warn("WARNING: Parts failed to insert"); addMissingParts(messagesNotInserted, group); } Period dbUpdatePeriod = new Period(dbUpdateTime); _log.info("Spent " + _periodFormatter.print(dbUpdatePeriod) + " updating the db"); } Period updateTime = new Period(startUpdateTime, System.currentTimeMillis()); if (!type.equals("partrepair")) { _log.info(count + " new, " + updateCount + " updated, " + partCount + " parts."); _log.info(" " + _periodFormatter.print(headersTime) + " headers, " + _periodFormatter.print(updateTime) + " update."); } transactionManager.commit(transaction); return maxNum; } else { _log.error("Error: Can't get parts from server (msgs not array)\n Skipping group"); return 0; } }
From source file:org.languagetool.tagging.uk.CompoundTagger.java
@Nullable private List<AnalyzedToken> tagMatch(String word, List<AnalyzedToken> leftAnalyzedTokens, List<AnalyzedToken> rightAnalyzedTokens) { List<AnalyzedToken> newAnalyzedTokens = new ArrayList<>(); List<AnalyzedToken> newAnalyzedTokensAnimInanim = new ArrayList<>(); String animInanimNotTagged = null; for (AnalyzedToken leftAnalyzedToken : leftAnalyzedTokens) { String leftPosTag = leftAnalyzedToken.getPOSTag(); if (leftPosTag == null || IPOSTag.contains(leftPosTag, IPOSTag.abbr.getText())) continue; // we don't want to have v_kly for -? // but we do for - if (leftPosTag.startsWith("noun:inanim") && leftPosTag.contains("v_kly")) continue; String leftPosTagExtra = ""; boolean leftNv = false; if (leftPosTag.contains(PosTagHelper.NO_VIDMINOK_SUBSTR)) { leftNv = true;//from w w w . j av a 2 s.c o m leftPosTag = leftPosTag.replace(PosTagHelper.NO_VIDMINOK_SUBSTR, ""); } Matcher matcher = EXTRA_TAGS_DROP.matcher(leftPosTag); if (matcher.find()) { leftPosTag = matcher.replaceAll(""); } matcher = EXTRA_TAGS.matcher(leftPosTag); if (matcher.find()) { leftPosTagExtra += matcher.group(); leftPosTag = matcher.replaceAll(""); } for (AnalyzedToken rightAnalyzedToken : rightAnalyzedTokens) { String rightPosTag = rightAnalyzedToken.getPOSTag(); if (rightPosTag == null // || rightPosTag.contains("v_kly") || rightPosTag.contains(IPOSTag.abbr.getText())) continue; if (rightPosTag.startsWith("noun:inanim") && rightPosTag.contains("v_kly")) continue; String extraNvTag = ""; boolean rightNv = false; if (rightPosTag.contains(PosTagHelper.NO_VIDMINOK_SUBSTR)) { rightNv = true; if (leftNv) { extraNvTag += PosTagHelper.NO_VIDMINOK_SUBSTR; } } Matcher matcherR = EXTRA_TAGS_DROP.matcher(rightPosTag); if (matcherR.find()) { rightPosTag = matcherR.replaceAll(""); } matcherR = EXTRA_TAGS.matcher(rightPosTag); if (matcherR.find()) { rightPosTag = matcherR.replaceAll(""); } if (stripPerfImperf(leftPosTag).equals(stripPerfImperf(rightPosTag)) && (IPOSTag .startsWith(leftPosTag, IPOSTag.numr, IPOSTag.adv, IPOSTag.adj, IPOSTag.verb) || (IPOSTag.startsWith(leftPosTag, IPOSTag.intj) && leftAnalyzedToken.getLemma().equalsIgnoreCase(rightAnalyzedToken.getLemma())))) { String newPosTag = leftPosTag + extraNvTag + leftPosTagExtra; if ((leftPosTag.contains("adjp") && !rightPosTag.contains("adjp")) || (!leftPosTag.contains("adjp") && rightPosTag.contains("adjp"))) { newPosTag = newPosTag.replaceFirst(":&adjp:(actv|pasv):(im)?perf", ""); } String newLemma = leftAnalyzedToken.getLemma() + "-" + rightAnalyzedToken.getLemma(); newAnalyzedTokens.add(new AnalyzedToken(word, newPosTag, newLemma)); } // noun-noun else if (leftPosTag.startsWith(IPOSTag.noun.getText()) && rightPosTag.startsWith(IPOSTag.noun.getText())) { // discard - as noun:anim if (leftAnalyzedToken.getToken().equalsIgnoreCase(rightAnalyzedToken.getToken()) && leftPosTag.contains(TAG_ANIM) && rightPosTag.contains(TAG_ANIM)) continue; String agreedPosTag = getAgreedPosTag(leftPosTag, rightPosTag, leftNv, word); if (agreedPosTag == null && rightPosTag.startsWith("noun:inanim:m:v_naz") && isMinMax(rightAnalyzedToken.getToken())) { agreedPosTag = leftPosTag; } if (agreedPosTag == null && !isSameAnimStatus(leftPosTag, rightPosTag)) { agreedPosTag = tryAnimInanim(leftPosTag, rightPosTag, leftAnalyzedToken.getLemma(), rightAnalyzedToken.getLemma(), leftNv, rightNv, word); if (agreedPosTag == null) { animInanimNotTagged = leftPosTag.contains(":anim") ? "anim-inanim" : "inanim-anim"; } else { newAnalyzedTokensAnimInanim .add(new AnalyzedToken(word, agreedPosTag + extraNvTag + leftPosTagExtra, leftAnalyzedToken.getLemma() + "-" + rightAnalyzedToken.getLemma())); continue; } } if (agreedPosTag != null) { newAnalyzedTokens.add(new AnalyzedToken(word, agreedPosTag + extraNvTag + leftPosTagExtra, leftAnalyzedToken.getLemma() + "-" + rightAnalyzedToken.getLemma())); } } // numr-numr: - else if (leftPosTag.startsWith(IPOSTag.numr.getText()) && rightPosTag.startsWith(IPOSTag.numr.getText())) { String agreedPosTag = getNumAgreedPosTag(leftPosTag, rightPosTag, leftNv); if (agreedPosTag != null) { newAnalyzedTokens.add(new AnalyzedToken(word, agreedPosTag + extraNvTag + leftPosTagExtra, leftAnalyzedToken.getLemma() + "-" + rightAnalyzedToken.getLemma())); } } // noun-numr match else if (IPOSTag.startsWith(leftPosTag, IPOSTag.noun) && IPOSTag.startsWith(rightPosTag, IPOSTag.numr)) { // gender tags match String leftGenderConj = PosTagHelper.getGenderConj(leftPosTag); if (leftGenderConj != null && leftGenderConj.equals(PosTagHelper.getGenderConj(rightPosTag))) { newAnalyzedTokens.add(new AnalyzedToken(word, leftPosTag + extraNvTag + leftPosTagExtra, leftAnalyzedToken.getLemma() + "-" + rightAnalyzedToken.getLemma())); // - ? : -, - if (!leftPosTag.contains(":p:")) { newAnalyzedTokens.add(new AnalyzedToken(word, leftPosTag.replaceAll(":[mfn]:", ":p:") + extraNvTag + leftPosTagExtra, leftAnalyzedToken.getLemma() + "-" + rightAnalyzedToken.getLemma())); } } else { // (with different gender tags): ? (:p:) - (:f:) String agreedPosTag = getNumAgreedPosTag(leftPosTag, rightPosTag, leftNv); if (agreedPosTag != null) { newAnalyzedTokens .add(new AnalyzedToken(word, agreedPosTag + extraNvTag + leftPosTagExtra, leftAnalyzedToken.getLemma() + "-" + rightAnalyzedToken.getLemma())); // - ? : -, - if (!agreedPosTag.contains(":p:")) { newAnalyzedTokens.add(new AnalyzedToken(word, agreedPosTag.replaceAll(":[mfn]:", ":p:") + extraNvTag + leftPosTagExtra, leftAnalyzedToken.getLemma() + "-" + rightAnalyzedToken.getLemma())); } } } } // noun-adj match: -, - // -? noun-adj ? else if (leftPosTag.startsWith(IPOSTag.noun.getText()) && IPOSTag.startsWith(rightPosTag, IPOSTag.numr) || (IPOSTag.startsWith(rightPosTag, IPOSTag.adj) && isJuniorSenior(leftAnalyzedToken, rightAnalyzedToken))) { // if( ! leftPosTag.contains(":prop") // || isJuniorSenior(leftAnalyzedToken, rightAnalyzedToken) ) { // discard - as noun:anim // if( leftAnalyzedToken.getToken().equalsIgnoreCase(rightAnalyzedToken.getToken()) ) // continue; String leftGenderConj = PosTagHelper.getGenderConj(leftPosTag); if (leftGenderConj != null && leftGenderConj.equals(PosTagHelper.getGenderConj(rightPosTag))) { newAnalyzedTokens.add(new AnalyzedToken(word, leftPosTag + extraNvTag + leftPosTagExtra, leftAnalyzedToken.getLemma() + "-" + rightAnalyzedToken.getLemma())); } // } } } } if (!newAnalyzedTokens.isEmpty() && !PosTagHelper.hasPosTagPart(newAnalyzedTokens, ":p:")) { if ((LemmaHelper.hasLemma(leftAnalyzedTokens, LemmaHelper.DAYS_OF_WEEK) && LemmaHelper.hasLemma(rightAnalyzedTokens, LemmaHelper.DAYS_OF_WEEK)) || (LemmaHelper.hasLemma(leftAnalyzedTokens, LemmaHelper.MONTH_LEMMAS) && LemmaHelper.hasLemma(rightAnalyzedTokens, LemmaHelper.MONTH_LEMMAS))) { newAnalyzedTokens.add( new AnalyzedToken(word, newAnalyzedTokens.get(0).getPOSTag().replaceAll(":[mfn]:", ":p:"), newAnalyzedTokens.get(0).getLemma())); } } // remove duplicates newAnalyzedTokens = new ArrayList<>(new LinkedHashSet<>(newAnalyzedTokens)); if (newAnalyzedTokens.isEmpty()) { newAnalyzedTokens = newAnalyzedTokensAnimInanim; } if (animInanimNotTagged != null && newAnalyzedTokens.isEmpty()) { compoundDebugLogger.logUnknownCompound(word + " " + animInanimNotTagged); } return newAnalyzedTokens.isEmpty() ? null : newAnalyzedTokens; }
From source file:com.ibm.bi.dml.test.utils.TestUtils.java
/** * <p>//from w w w . j a v a 2 s . c o m * Replaces variables in a DML or R script with the specified values. A * variable of format ##name## will be replaced where the name is used to * identify the variable in the hashmap containing the belonging value. * </p> * * @param strScriptDirectory * directory which contains the DML script * @param strScriptFile * filename of the DML script * @param variables * hashmap containing all the variables and their replacements * @deprecated Use ParameterBuilder.setVariablesInScript instead */ public static void setVariablesInScript(String strScriptDirectory, String strScriptFile, HashMap<String, String> variables) { try { String strScript = strScriptDirectory + strScriptFile; String strTmpScript = strScript + "t"; BufferedReader in = new BufferedReader(new InputStreamReader(new FileInputStream(strScript))); FileOutputStream out = new FileOutputStream(strTmpScript); PrintWriter pw = new PrintWriter(out); String content; Pattern unresolvedVars = Pattern.compile(_RS + ".*" + _RS); /** * sothat variables, which were not assigned, are replaced by an * empty string */ while ((content = in.readLine()) != null) { for (String variable : variables.keySet()) { Pattern pattern = Pattern.compile(_RS + variable + _RS); Matcher matcher = pattern.matcher(content); while (matcher.find()) { content = content.replaceFirst(matcher.group().replace("$", "\\$"), variables.get(variable)); } } Matcher matcher = unresolvedVars.matcher(content); content = matcher.replaceAll(""); pw.println(content); } pw.close(); out.close(); in.close(); /* * // remove checksum files if created Path crcFile = new * Path(dmlScriptDirectory + "." + dmlScriptFile + ".crc"); if * (fs.exists(crcFile)) fs.delete(crcFile, false); crcFile = new * Path(dmlScriptDirectory + "." + dmlScriptFile + "t.crc"); if * (fs.exists(crcFile)) fs.delete(crcFile, false); */ } catch (IOException e) { e.printStackTrace(); fail("unable to set variables in dml script: " + e.getMessage()); } }
From source file:org.kuali.kfs.module.ld.businessobject.LaborOriginEntry.java
/** * Get lines from string//from w w w.j a v a 2 s . com */ @Override public String getLine() { StringBuilder sb = new StringBuilder(); Map<String, Integer> lMap = getLaborOriginEntryFieldUtil().getFieldLengthMap(); Map<String, Integer> pMap = getLaborOriginEntryFieldUtil().getFieldBeginningPositionMap(); int entryLength = pMap.get(LaborPropertyConstants.SET_ID) + lMap.get(LaborPropertyConstants.SET_ID); if (universityFiscalYear == null) { sb.append(GeneralLedgerConstants.getSpaceUniversityFiscalYear()); } else { sb.append(universityFiscalYear); } sb.append(getField(lMap.get(KFSPropertyConstants.CHART_OF_ACCOUNTS_CODE), chartOfAccountsCode)); sb.append(getField(lMap.get(KFSPropertyConstants.ACCOUNT_NUMBER), accountNumber)); sb.append(getField(lMap.get(KFSPropertyConstants.SUB_ACCOUNT_NUMBER), subAccountNumber)); sb.append(getField(lMap.get(KFSPropertyConstants.FINANCIAL_OBJECT_CODE), financialObjectCode)); sb.append(getField(lMap.get(KFSPropertyConstants.FINANCIAL_SUB_OBJECT_CODE), financialSubObjectCode)); sb.append(getField(lMap.get(KFSPropertyConstants.FINANCIAL_BALANCE_TYPE_CODE), financialBalanceTypeCode)); sb.append(getField(lMap.get(KFSPropertyConstants.FINANCIAL_OBJECT_TYPE_CODE), financialObjectTypeCode)); sb.append( getField(lMap.get(KFSPropertyConstants.UNIVERSITY_FISCAL_PERIOD_CODE), universityFiscalPeriodCode)); sb.append(getField(lMap.get(KFSPropertyConstants.FINANCIAL_DOCUMENT_TYPE_CODE), financialDocumentTypeCode)); sb.append(getField(lMap.get(KFSPropertyConstants.FINANCIAL_SYSTEM_ORIGINATION_CODE), financialSystemOriginationCode)); sb.append(getField(lMap.get(KFSPropertyConstants.DOCUMENT_NUMBER), documentNumber)); String seqNum = ""; if (transactionLedgerEntrySequenceNumber != null) { seqNum = transactionLedgerEntrySequenceNumber.toString(); } // Format to a length of 5 sb.append(StringUtils.leftPad(seqNum.trim(), lMap.get(KFSPropertyConstants.TRANSACTION_ENTRY_SEQUENCE_NUMBER), "0")); // Labor Specified fields sb.append(getField(lMap.get(KFSPropertyConstants.POSITION_NUMBER), positionNumber)); sb.append(getField(lMap.get(KFSPropertyConstants.PROJECT_CODE), projectCode)); sb.append(getField(lMap.get(KFSPropertyConstants.TRANSACTION_LEDGER_ENTRY_DESC), transactionLedgerEntryDescription)); if (transactionLedgerEntryAmount == null) { sb.append(GeneralLedgerConstants.getZeroTransactionLedgerEntryAmount()); } else { String a = transactionLedgerEntryAmount.abs().toString(); if (transactionLedgerEntryAmount.isNegative()) { sb.append("-"); } else { sb.append("+"); } sb.append(GeneralLedgerConstants.getZeroTransactionLedgerEntryAmount().substring(1, lMap.get(KFSPropertyConstants.TRANSACTION_LEDGER_ENTRY_AMOUNT) - a.length())); sb.append(a); } sb.append( getField(lMap.get(KFSPropertyConstants.TRANSACTION_DEBIT_CREDIT_CODE), transactionDebitCreditCode)); sb.append(formatDate(transactionDate)); sb.append( getField(lMap.get(KFSPropertyConstants.ORGANIZATION_DOCUMENT_NUMBER), organizationDocumentNumber)); sb.append(getField(lMap.get(KFSPropertyConstants.ORGANIZATION_REFERENCE_ID), organizationReferenceId)); sb.append(getField(lMap.get(KFSPropertyConstants.REFERENCE_FIN_DOCUMENT_TYPE_CODE), referenceFinancialDocumentTypeCode)); sb.append(getField(lMap.get(KFSPropertyConstants.FIN_SYSTEM_REF_ORIGINATION_CODE), referenceFinancialSystemOriginationCode)); sb.append(getField(lMap.get(KFSPropertyConstants.FINANCIAL_DOCUMENT_REFERENCE_NBR), referenceFinancialDocumentNumber)); sb.append(formatDate(financialDocumentReversalDate)); sb.append(getField(lMap.get(KFSPropertyConstants.TRANSACTION_ENCUMBRANCE_UPDT_CD), transactionEncumbranceUpdateCode)); sb.append(formatDate(transactionPostingDate)); sb.append(formatDate(payPeriodEndDate)); if (transactionTotalHours == null) { sb.append(StringUtils.rightPad("", lMap.get(KFSPropertyConstants.TRANSACTION_TOTAL_HOURS), " ")); } else { String totalhour = getField(lMap.get(KFSPropertyConstants.TRANSACTION_TOTAL_HOURS), transactionTotalHours.toString()); sb.append(StringUtils.leftPad(totalhour.trim(), lMap.get(KFSPropertyConstants.TRANSACTION_TOTAL_HOURS), " ")); } if (payrollEndDateFiscalYear == null) { sb.append(StringUtils.rightPad("", lMap.get(KFSPropertyConstants.PAYROLL_END_DATE_FISCAL_YEAR), " ")); } else { sb.append(getField(lMap.get(KFSPropertyConstants.PAYROLL_END_DATE_FISCAL_YEAR), payrollEndDateFiscalYear.toString())); } sb.append(getField(lMap.get(LaborPropertyConstants.PAYROLL_END_DATE_FISCAL_PERIOD_CODE), payrollEndDateFiscalPeriodCode)); sb.append(getField(lMap.get(KFSPropertyConstants.EMPLID), emplid)); if (employeeRecord == null) { sb.append(StringUtils.rightPad("", lMap.get(KFSPropertyConstants.EMPLOYEE_RECORD), " ")); } else { String empRecord = getField(lMap.get(KFSPropertyConstants.EMPLOYEE_RECORD), employeeRecord.toString()); sb.append(StringUtils.leftPad(empRecord.trim(), lMap.get(KFSPropertyConstants.EMPLOYEE_RECORD), " ")); } sb.append(getField(lMap.get(KFSPropertyConstants.EARN_CODE), earnCode)); sb.append(getField(lMap.get(KFSPropertyConstants.PAY_GROUP), payGroup)); sb.append(getField(lMap.get(LaborPropertyConstants.SALARY_ADMINISTRATION_PLAN), salaryAdministrationPlan)); sb.append(getField(lMap.get(LaborPropertyConstants.GRADE), grade)); sb.append(getField(lMap.get(LaborPropertyConstants.RUN_IDENTIFIER), runIdentifier)); sb.append(getField(lMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_CHART_OF_ACCOUNTS_CODE), laborLedgerOriginalChartOfAccountsCode)); sb.append(getField(lMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_ACCOUNT_NUMBER), laborLedgerOriginalAccountNumber)); sb.append(getField(lMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_SUB_ACCOUNT_NUMBER), laborLedgerOriginalSubAccountNumber)); sb.append(getField(lMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_FINANCIAL_OBJECT_CODE), laborLedgerOriginalFinancialObjectCode)); sb.append(getField(lMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_FINANCIAL_SUB_OBJECT_CODE), laborLedgerOriginalFinancialSubObjectCode)); sb.append(getField(lMap.get(LaborPropertyConstants.HRMS_COMPANY), hrmsCompany)); sb.append(getField(lMap.get(LaborPropertyConstants.SET_ID), setid)); // pad to full length of 295 chars. while (entryLength > sb.toString().length()) { sb.append(' '); } // KFSMI-5958: Don't want any control characters in output files. They potentially disrupt further processing Matcher controlCharacterMatcher = MATCH_CONTROL_CHARACTERS.matcher(sb); String returnString = controlCharacterMatcher.replaceAll(REPLACE_MATCHED_CONTROL_CHARACTERS); return returnString; }
From source file:org.kuali.kfs.module.ld.businessobject.LaborOriginEntry.java
/** * Sets the entries from text file.//from w w w .j a v a 2s. c o m */ @Override public List<Message> setFromTextFileForBatch(String line, int lineNumber) { List<Message> returnList = new ArrayList(); Map<String, Integer> pMap = getLaborOriginEntryFieldUtil().getFieldBeginningPositionMap(); Map<String, Integer> lMap = getLaborOriginEntryFieldUtil().getFieldLengthMap(); int entryLength = pMap.get(LaborPropertyConstants.SET_ID) + lMap.get(LaborPropertyConstants.SET_ID); // KFSMI-5958: Don't want any control characters in output files. They potentially disrupt further processing Matcher controlCharacterMatcher = MATCH_CONTROL_CHARACTERS.matcher(line); line = controlCharacterMatcher.replaceAll(REPLACE_MATCHED_CONTROL_CHARACTERS); // Just in case line = org.apache.commons.lang.StringUtils.rightPad(line, entryLength, ' '); String fiscalYearString = line.substring(pMap.get(KFSPropertyConstants.UNIVERSITY_FISCAL_YEAR), pMap.get(KFSPropertyConstants.CHART_OF_ACCOUNTS_CODE)); if (!GeneralLedgerConstants.getSpaceUniversityFiscalYear().equals(fiscalYearString)) { try { setUniversityFiscalYear(new Integer(fiscalYearString)); } catch (NumberFormatException e) { returnList.add(new Message("Fiscal year '" + fiscalYearString + "' contains an invalid value.", Message.TYPE_FATAL)); setUniversityFiscalYear(null); } } else { setUniversityFiscalYear(null); } setChartOfAccountsCode(getValue(line, pMap.get(KFSPropertyConstants.CHART_OF_ACCOUNTS_CODE), pMap.get(KFSPropertyConstants.ACCOUNT_NUMBER))); setAccountNumber(getValue(line, pMap.get(KFSPropertyConstants.ACCOUNT_NUMBER), pMap.get(KFSPropertyConstants.SUB_ACCOUNT_NUMBER))); // if chart code is empty while accounts cannot cross charts, then derive chart code from account number AccountService acctserv = SpringContext.getBean(AccountService.class); if (StringUtils.isEmpty(getChartOfAccountsCode()) && StringUtils.isNotEmpty(getAccountNumber()) && !acctserv.accountsCanCrossCharts()) { Account account = acctserv.getUniqueAccountForAccountNumber(getAccountNumber()); if (account != null) { setChartOfAccountsCode(account.getChartOfAccountsCode()); } } setSubAccountNumber(getValue(line, pMap.get(KFSPropertyConstants.SUB_ACCOUNT_NUMBER), pMap.get(KFSPropertyConstants.FINANCIAL_OBJECT_CODE))); setFinancialObjectCode(getValue(line, pMap.get(KFSPropertyConstants.FINANCIAL_OBJECT_CODE), pMap.get(KFSPropertyConstants.FINANCIAL_SUB_OBJECT_CODE))); setFinancialSubObjectCode(getValue(line, pMap.get(KFSPropertyConstants.FINANCIAL_SUB_OBJECT_CODE), pMap.get(KFSPropertyConstants.FINANCIAL_BALANCE_TYPE_CODE))); setFinancialBalanceTypeCode(getValue(line, pMap.get(KFSPropertyConstants.FINANCIAL_BALANCE_TYPE_CODE), pMap.get(KFSPropertyConstants.FINANCIAL_OBJECT_TYPE_CODE))); setFinancialObjectTypeCode(getValue(line, pMap.get(KFSPropertyConstants.FINANCIAL_OBJECT_TYPE_CODE), pMap.get(KFSPropertyConstants.UNIVERSITY_FISCAL_PERIOD_CODE))); setUniversityFiscalPeriodCode(getValue(line, pMap.get(KFSPropertyConstants.UNIVERSITY_FISCAL_PERIOD_CODE), pMap.get(KFSPropertyConstants.FINANCIAL_DOCUMENT_TYPE_CODE))); setFinancialDocumentTypeCode(getValue(line, pMap.get(KFSPropertyConstants.FINANCIAL_DOCUMENT_TYPE_CODE), pMap.get(KFSPropertyConstants.FINANCIAL_SYSTEM_ORIGINATION_CODE))); setFinancialSystemOriginationCode( getValue(line, pMap.get(KFSPropertyConstants.FINANCIAL_SYSTEM_ORIGINATION_CODE), pMap.get(KFSPropertyConstants.DOCUMENT_NUMBER))); setDocumentNumber(getValue(line, pMap.get(KFSPropertyConstants.DOCUMENT_NUMBER), pMap.get(KFSPropertyConstants.TRANSACTION_ENTRY_SEQUENCE_NUMBER))); String sequenceNumberString = line.substring( pMap.get(KFSPropertyConstants.TRANSACTION_ENTRY_SEQUENCE_NUMBER), pMap.get(KFSPropertyConstants.POSITION_NUMBER)); if (!GeneralLedgerConstants.getSpaceTransactionEntrySequenceNumber().equals(sequenceNumberString) && !GeneralLedgerConstants.getZeroTransactionEntrySequenceNumber().equals(sequenceNumberString)) { try { setTransactionLedgerEntrySequenceNumber(new Integer(sequenceNumberString.trim())); } catch (NumberFormatException e) { returnList.add(new Message( "Transaction Sequence Number '" + sequenceNumberString + "' contains an invalid value.", Message.TYPE_FATAL)); setTransactionLedgerEntrySequenceNumber(null); } } else { setTransactionLedgerEntrySequenceNumber(null); } setPositionNumber(getValue(line, pMap.get(KFSPropertyConstants.POSITION_NUMBER), pMap.get(KFSPropertyConstants.PROJECT_CODE))); setProjectCode(getValue(line, pMap.get(KFSPropertyConstants.PROJECT_CODE), pMap.get(KFSPropertyConstants.TRANSACTION_LEDGER_ENTRY_DESC))); setTransactionLedgerEntryDescription( getValue(line, pMap.get(KFSPropertyConstants.TRANSACTION_LEDGER_ENTRY_DESC), pMap.get(KFSPropertyConstants.TRANSACTION_LEDGER_ENTRY_AMOUNT))); String amountString = line.substring(pMap.get(KFSPropertyConstants.TRANSACTION_LEDGER_ENTRY_AMOUNT), pMap.get(KFSPropertyConstants.TRANSACTION_DEBIT_CREDIT_CODE)); if (!amountString.trim().equals(GeneralLedgerConstants.EMPTY_CODE)) { try { setTransactionLedgerEntryAmount(new KualiDecimal(amountString.trim())); } catch (NumberFormatException e) { returnList.add(new Message("Transaction Amount '" + amountString + "' contains an invalid value.", Message.TYPE_FATAL)); setTransactionLedgerEntryAmount(KualiDecimal.ZERO); } } else { returnList.add(new Message("Transaction Amount cannot be blank.", Message.TYPE_FATAL)); setTransactionLedgerEntryAmount(KualiDecimal.ZERO); } setTransactionDebitCreditCode(line.substring(pMap.get(KFSPropertyConstants.TRANSACTION_DEBIT_CREDIT_CODE), pMap.get(KFSPropertyConstants.TRANSACTION_DATE))); String transactionDateString = line.substring(pMap.get(KFSPropertyConstants.TRANSACTION_DATE), pMap.get(KFSPropertyConstants.ORGANIZATION_DOCUMENT_NUMBER)); if (!transactionDateString.trim().equals(GeneralLedgerConstants.EMPTY_CODE)) { try { setTransactionDate(parseDate( org.springframework.util.StringUtils.trimTrailingWhitespace(transactionDateString), false)); } catch (ParseException e) { setTransactionDate(null); returnList.add( new Message("Transaction Date '" + transactionDateString + "' contains an invalid value.", Message.TYPE_FATAL)); } } else { setTransactionDate(null); } setOrganizationDocumentNumber(getValue(line, pMap.get(KFSPropertyConstants.ORGANIZATION_DOCUMENT_NUMBER), pMap.get(KFSPropertyConstants.ORGANIZATION_REFERENCE_ID))); setOrganizationReferenceId(getValue(line, pMap.get(KFSPropertyConstants.ORGANIZATION_REFERENCE_ID), pMap.get(KFSPropertyConstants.REFERENCE_FIN_DOCUMENT_TYPE_CODE))); setReferenceFinancialDocumentTypeCode( getValue(line, pMap.get(KFSPropertyConstants.REFERENCE_FIN_DOCUMENT_TYPE_CODE), pMap.get(KFSPropertyConstants.FIN_SYSTEM_REF_ORIGINATION_CODE))); setReferenceFinancialSystemOriginationCode( getValue(line, pMap.get(KFSPropertyConstants.FIN_SYSTEM_REF_ORIGINATION_CODE), pMap.get(KFSPropertyConstants.FINANCIAL_DOCUMENT_REFERENCE_NBR))); setReferenceFinancialDocumentNumber( getValue(line, pMap.get(KFSPropertyConstants.FINANCIAL_DOCUMENT_REFERENCE_NBR), pMap.get(KFSPropertyConstants.FINANCIAL_DOCUMENT_REVERSAL_DATE))); String revDateStr = line.substring(pMap.get(KFSPropertyConstants.FINANCIAL_DOCUMENT_REVERSAL_DATE), pMap.get(KFSPropertyConstants.TRANSACTION_ENCUMBRANCE_UPDT_CD)); if (!revDateStr.trim().equals(GeneralLedgerConstants.EMPTY_CODE)) { try { setFinancialDocumentReversalDate( parseDate(org.springframework.util.StringUtils.trimTrailingWhitespace(revDateStr), false)); } catch (ParseException e) { setFinancialDocumentReversalDate(null); returnList.add(new Message("Reversal Date '" + revDateStr + "' contains an invalid value.", Message.TYPE_FATAL)); } } else { setFinancialDocumentReversalDate(null); } setTransactionEncumbranceUpdateCode( line.substring(pMap.get(KFSPropertyConstants.TRANSACTION_ENCUMBRANCE_UPDT_CD), pMap.get(KFSPropertyConstants.TRANSACTION_POSTING_DATE))); String postDateStr = line.substring(pMap.get(KFSPropertyConstants.TRANSACTION_POSTING_DATE), pMap.get(KFSPropertyConstants.PAY_PERIOD_END_DATE)); if (!postDateStr.trim().equals(GeneralLedgerConstants.EMPTY_CODE)) { try { setTransactionPostingDate( parseDate(org.springframework.util.StringUtils.trimTrailingWhitespace(postDateStr), false)); } catch (ParseException e) { setTransactionPostingDate(null); returnList.add( new Message("Transaction Posting Date '" + postDateStr + "' contains an invalid value.", Message.TYPE_FATAL)); } } else { setTransactionPostingDate(null); } String payPeriodDateStr = line.substring(pMap.get(KFSPropertyConstants.PAY_PERIOD_END_DATE), pMap.get(KFSPropertyConstants.TRANSACTION_TOTAL_HOURS)); if (!payPeriodDateStr.trim().equals(GeneralLedgerConstants.EMPTY_CODE)) { try { setPayPeriodEndDate(parseDate( org.springframework.util.StringUtils.trimTrailingWhitespace(payPeriodDateStr), false)); } catch (ParseException e) { setPayPeriodEndDate(null); returnList.add( new Message("Pay Period End Date '" + payPeriodDateStr + "' contains an invalid value.", Message.TYPE_FATAL)); } } else { setPayPeriodEndDate(null); } String transTotHrsStr = line.substring(pMap.get(KFSPropertyConstants.TRANSACTION_TOTAL_HOURS), pMap.get(KFSPropertyConstants.PAYROLL_END_DATE_FISCAL_YEAR)); if (!transTotHrsStr.trim().equals(GeneralLedgerConstants.EMPTY_CODE)) { try { setTransactionTotalHours(new BigDecimal(transTotHrsStr.trim())); } catch (NumberFormatException e) { setTransactionTotalHours(null); returnList.add(new Message( "Transaction Total Hours '" + transTotHrsStr.trim() + "' contains an invalid value.", Message.TYPE_FATAL)); } } else { setTransactionTotalHours(null); } String payEndFisYrStr = line.substring(pMap.get(KFSPropertyConstants.PAYROLL_END_DATE_FISCAL_YEAR), pMap.get(LaborPropertyConstants.PAYROLL_END_DATE_FISCAL_PERIOD_CODE)); if (!GeneralLedgerConstants.getSpaceUniversityFiscalYear().equals(payEndFisYrStr)) { try { setPayrollEndDateFiscalYear( new Integer(org.springframework.util.StringUtils.trimTrailingWhitespace(payEndFisYrStr))); } catch (NumberFormatException e) { returnList.add(new Message( "Payroll End Date Fiscal Year '" + payEndFisYrStr + "' contains an invalid value.", Message.TYPE_FATAL)); setPayrollEndDateFiscalYear(null); } } else { setPayrollEndDateFiscalYear(null); } setPayrollEndDateFiscalPeriodCode( getValue(line, pMap.get(LaborPropertyConstants.PAYROLL_END_DATE_FISCAL_PERIOD_CODE), pMap.get(KFSPropertyConstants.EMPLID))); setEmplid(getValue(line, pMap.get(KFSPropertyConstants.EMPLID), pMap.get(KFSPropertyConstants.EMPLOYEE_RECORD))); String empRecordStr = line.substring(pMap.get(KFSPropertyConstants.EMPLOYEE_RECORD), pMap.get(KFSPropertyConstants.EARN_CODE)); if (!empRecordStr.trim().equals(GeneralLedgerConstants.EMPTY_CODE)) { try { setEmployeeRecord(new Integer(empRecordStr.trim())); } catch (NumberFormatException e) { returnList .add(new Message("Employee Record '" + empRecordStr.trim() + "' contains an invalid value.", Message.TYPE_FATAL)); setEmployeeRecord(null); } } else { setEmployeeRecord(null); } setEarnCode( getValue(line, pMap.get(KFSPropertyConstants.EARN_CODE), pMap.get(KFSPropertyConstants.PAY_GROUP))); setPayGroup(getValue(line, pMap.get(KFSPropertyConstants.PAY_GROUP), pMap.get(LaborPropertyConstants.SALARY_ADMINISTRATION_PLAN))); setSalaryAdministrationPlan(getValue(line, pMap.get(LaborPropertyConstants.SALARY_ADMINISTRATION_PLAN), pMap.get(LaborPropertyConstants.GRADE))); setGrade(getValue(line, pMap.get(LaborPropertyConstants.GRADE), pMap.get(LaborPropertyConstants.RUN_IDENTIFIER))); setRunIdentifier(getValue(line, pMap.get(LaborPropertyConstants.RUN_IDENTIFIER), pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_CHART_OF_ACCOUNTS_CODE))); setLaborLedgerOriginalChartOfAccountsCode( getValue(line, pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_CHART_OF_ACCOUNTS_CODE), pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_ACCOUNT_NUMBER))); setLaborLedgerOriginalAccountNumber( getValue(line, pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_ACCOUNT_NUMBER), pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_SUB_ACCOUNT_NUMBER))); setLaborLedgerOriginalSubAccountNumber( getValue(line, pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_SUB_ACCOUNT_NUMBER), pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_FINANCIAL_OBJECT_CODE))); setLaborLedgerOriginalFinancialObjectCode( getValue(line, pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_FINANCIAL_OBJECT_CODE), pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_FINANCIAL_SUB_OBJECT_CODE))); setLaborLedgerOriginalFinancialSubObjectCode( getValue(line, pMap.get(LaborPropertyConstants.LABORLEDGER_ORIGINAL_FINANCIAL_SUB_OBJECT_CODE), pMap.get(LaborPropertyConstants.HRMS_COMPANY))); setHrmsCompany(getValue(line, pMap.get(LaborPropertyConstants.HRMS_COMPANY), pMap.get(LaborPropertyConstants.SET_ID))); setSetid(getValue(line, pMap.get(LaborPropertyConstants.SET_ID), entryLength)); return returnList; }
From source file:com.threerings.getdown.data.Application.java
/** * If appbase_domain property is set, replace the domain on the provided string. *//* w w w . j ava 2 s.c o m*/ protected String replaceDomain(String appbase) { String appbaseDomain = SysProps.appbaseDomain(); if (appbaseDomain != null) { Matcher m = Pattern.compile("(http://[^/]+)(.*)").matcher(appbase); appbase = m.replaceAll(appbaseDomain + "$2"); } return appbase; }
From source file:com.juick.android.MessagesFragment.java
private ArrayList<JuickMessage> filterMessages(ArrayList<JuickMessage> messages) { log("filterMessages start"); SharedPreferences sp = PreferenceManager .getDefaultSharedPreferences(JuickAdvancedApplication.instance.getApplicationContext()); boolean filterNonRussian = sp.getBoolean("filterNonRussian", false); Set<String> filteredOutUsers1 = JuickMessagesAdapter.getFilteredOutUsers(parent); log("filterMessages got filtered out users"); for (Iterator<JuickMessage> iterator = messages.iterator(); iterator.hasNext();) { JuickMessage message = iterator.next(); if (filteredOutUsers1.contains(message.User.UName)) { iterator.remove();//from w w w .j a va 2 s . com continue; } if (message.getRID() == 0) { // don't check comments if (XMPPService.getAnyJuboMessageFilter() != null) { if (!XMPPService.getAnyJuboMessageFilter().allowMessage(message)) { iterator.remove(); continue; } } if (XMPPService.getAnyJuickBlacklist() != null) { if (!XMPPService.getAnyJuickBlacklist().allowMessage(message)) { iterator.remove(); continue; } } } if (filterNonRussian) { String text = message.Text; int nRussian = 0; while (true) { boolean replaced = false; Matcher matcher = httpURL.matcher(text); if (matcher.find()) { try { text = matcher.replaceAll(""); replaced = true; } catch (Exception e) { // } } if (!replaced) break; } final int limit = text.length(); for (int i = 0; i < limit; i++) { int charCode = (int) text.charAt(i); if (russians.get(charCode)) { nRussian++; break; } } if (!text.contains("No description")) { if (nRussian == 0 && limit > 30) { iterator.remove(); continue; } } } } log("filterMessages end"); return messages; }
From source file:org.apache.hadoop.hive.service.HSSessionItem.java
public String removen(String in) { if (in == null) { return ""; }/*from w w w . ja v a 2s . com*/ l4j.info("input: " + in); Pattern p = Pattern.compile("\\n"); Matcher m = p.matcher(in); String ttt = m.replaceAll(""); l4j.debug("output: " + ttt); return ttt; }
From source file:org.apache.tika.eval.AbstractProfiler.java
void getExceptionStrings(Metadata metadata, Map<Cols, String> data) { String fullTrace = metadata.get(TikaCoreProperties.TIKA_META_EXCEPTION_PREFIX + "runtime"); if (fullTrace == null) { fullTrace = metadata.get(RecursiveParserWrapper.EMBEDDED_EXCEPTION); }//from w w w .j a v a 2 s. c om if (fullTrace != null) { //check for "expected" exceptions...exceptions //that can't be fixed. //Do not store trace for "expected" exceptions Matcher matcher = ACCESS_PERMISSION_EXCEPTION.matcher(fullTrace); if (matcher.find()) { data.put(Cols.PARSE_EXCEPTION_ID, Integer.toString(EXCEPTION_TYPE.ACCESS_PERMISSION.ordinal())); return; } matcher = ENCRYPTION_EXCEPTION.matcher(fullTrace); if (matcher.find()) { data.put(Cols.PARSE_EXCEPTION_ID, Integer.toString(EXCEPTION_TYPE.ENCRYPTION.ordinal())); return; } data.put(Cols.PARSE_EXCEPTION_ID, Integer.toString(EXCEPTION_TYPE.RUNTIME.ordinal())); data.put(Cols.ORIG_STACK_TRACE, fullTrace); //TikaExceptions can have object ids, as in the "@2b1ea6ee" in: //org.apache.tika.exception.TikaException: TIKA-198: Illegal //IOException from org.apache.tika.parser.microsoft.OfficeParser@2b1ea6ee //For reporting purposes, let's snip off the object id so that we can more //easily count exceptions. String sortTrace = ExceptionUtils.trimMessage(fullTrace); matcher = CAUSED_BY_SNIPPER.matcher(sortTrace); sortTrace = matcher.replaceAll("$1"); sortTrace = sortTrace.replaceAll("org.apache.tika.", "o.a.t."); data.put(Cols.SORT_STACK_TRACE, sortTrace); } }