List of usage examples for org.joda.time Interval getStart
public DateTime getStart()
From source file:org.filteredpush.qc.date.DateUtils.java
License:Apache License
/** * Test to see whether an eventDate contains a string in an expected ISO format. * /*w w w . jav a 2s . c o m*/ * @param eventDate string to test for expected format. * @return true if eventDate is in an expected format for eventDate, otherwise false. */ public static boolean eventDateValid(String eventDate) { boolean result = false; if (extractDate(eventDate) != null) { result = true; } else { Interval interval = extractInterval(eventDate); if (interval != null) { if (interval.getStart().isBefore(interval.getEnd())) { result = true; } } } return result; }
From source file:org.filteredpush.qc.date.DateUtils.java
License:Apache License
/** * Extract a date from a verbatim date, returning ranges specified to day. * /* w w w. j a v a 2 s . c o m*/ * @param verbatimEventDate a string containing a verbatim event date. * @param yearsBeforeSuspect the value for a year before which parsed years should be considered suspect. * @return an EventResult object containing the extracted date. */ public static EventResult extractDateToDayFromVerbatimER(String verbatimEventDate, int yearsBeforeSuspect) { EventResult result = extractDateFromVerbatimER(verbatimEventDate, yearsBeforeSuspect, null); logger.debug(result.getResultState()); logger.debug(result.getResult()); if (result != null && result.getResultState().equals(EventResult.EventQCResultState.RANGE)) { String dateRange = result.getResult(); try { Interval parseDate = extractDateInterval(dateRange); logger.debug(parseDate); String resultDate = parseDate.getStart().toString("yyyy-MM-dd") + "/" + parseDate.getEnd().toString("yyyy-MM-dd"); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); result.setResultState(EventResult.EventQCResultState.INTERNAL_PREREQISITES_NOT_MET); result.addComment(e.getMessage()); } } return result; }
From source file:org.filteredpush.qc.date.DateUtils.java
License:Apache License
/** * Extract a date from a verbatim date, returning ranges specified to day. * // w w w .j a va2s .c o m * @param verbatimEventDate a string containing a verbatim event date. * @param yearsBeforeSuspect the value for a year before which parsed years should be considered suspect. * @return a map with result and resultState as keys * * @deprecated * @see DateUtils#extractDateToDayFromVerbatimER(String, int) replacement method */ public static Map<String, String> extractDateToDayFromVerbatim(String verbatimEventDate, int yearsBeforeSuspect) { Map<String, String> result = extractDateFromVerbatim(verbatimEventDate, yearsBeforeSuspect); if (result.size() > 0 && result.get("resultState").equals("range")) { String dateRange = result.get("result"); try { Interval parseDate = extractDateInterval(dateRange); logger.debug(parseDate); String resultDate = parseDate.getStart().toString("yyyy-MM-dd") + "/" + parseDate.getEnd().toString("yyyy-MM-dd"); result.put("result", resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } return result; }
From source file:org.filteredpush.qc.date.DateUtils.java
License:Apache License
/** * Given a string that may represent a date or range of dates, or date time or range of date times, * attempt to extract a standard date from that string. * /*from ww w .ja v a 2s . co m*/ * @param verbatimEventDate a string containing a verbatim event date. * @param yearsBeforeSuspect Dates that parse to a year prior to this year are marked as suspect. * @param assumemmddyyyy if true, assume that dates in the form nn-nn-nnnn are mm-dd-yyyy, if false, assume * that these are dd-mm-yyyy, if null, such dates are tested for ambiguity. * * @return an EventResult with a resultState for the nature of the match and result for the resulting date. */ public static EventResult extractDateFromVerbatimER(String verbatimEventDate, int yearsBeforeSuspect, Boolean assumemmddyyyy) { EventResult result = new EventResult(); String resultDate = null; // Remove some common no data comments if (verbatimEventDate != null && verbatimEventDate.contains("[no date]")) { verbatimEventDate = verbatimEventDate.replace("[no date]", ""); } if (verbatimEventDate != null && verbatimEventDate.contains("[no year]")) { verbatimEventDate = verbatimEventDate.replace("[no year]", ""); } // Strip off leading and trailing [] if (verbatimEventDate != null && verbatimEventDate.startsWith("[") && verbatimEventDate.endsWith("]")) { verbatimEventDate = verbatimEventDate.substring(1); verbatimEventDate = verbatimEventDate.substring(0, verbatimEventDate.length() - 1); } if (verbatimEventDate != null && verbatimEventDate.matches(".*\\[[0-9]+\\].*")) { verbatimEventDate = verbatimEventDate.replace("[", "").replace("]", ""); } // Strip off leading and trailing quotation marks if (verbatimEventDate != null && verbatimEventDate.startsWith("\"") && verbatimEventDate != null && verbatimEventDate.endsWith("\"")) { verbatimEventDate = verbatimEventDate.substring(1, verbatimEventDate.length() - 1); } // strip off leading and trailing whitespace if (verbatimEventDate != null && (verbatimEventDate.startsWith(" ") || verbatimEventDate.endsWith(" "))) { verbatimEventDate = verbatimEventDate.trim(); } // strip off trailing period after number if (verbatimEventDate != null && verbatimEventDate.endsWith(".") && verbatimEventDate.matches(".*[0-9]\\.$")) { verbatimEventDate = verbatimEventDate.substring(0, verbatimEventDate.length() - 1); logger.debug(verbatimEventDate); } // Stop before doing work if provided verbatim string is null. if (isEmpty(verbatimEventDate)) { return result; } if (verbatimEventDate.matches("^[0-9]{4}[-][0-9]{2}[-][0-9]{2}/[0-9]{4}[-][0-9]{2}[-][0-9]{2}$")) { // if verbatim date is a ISO formatted range with identical first and last dates (/), use just one. // Example: 1982-12-11/1982-12-11 changed to 1982-12-11 String[] bits = verbatimEventDate.split("/"); if (bits.length == 2 && bits[0].equals(bits[1])) { verbatimEventDate = bits[0]; } } if (verbatimEventDate.matches("^[0-9]{4}[/][0-9]{2}[/][0-9]{2}-[0-9]{4}[/][0-9]{2}[/][0-9]{2}$")) { // if verbatim date is a range with identical first and last dates (-), use just one. // Example: 1982/12/11-1982/12/11 changed to 1982/12/11 String[] bits = verbatimEventDate.split("-"); if (bits.length == 2 && bits[0].equals(bits[1])) { verbatimEventDate = bits[0]; } } if (verbatimEventDate .matches("^[0-9]{1,2}[-. ][0-9]{1,2}[-. ][0-9]{4}/[0-9]{1,2}[-. ][0-9]{1,2}[-. ][0-9]{4}$")) { // if verbatim date is a range with identical first and last dates (/), use just one. // Example: 12-11-1982/12-11-1982 changed to 12-11-1982 String[] bits = verbatimEventDate.split("/"); if (bits.length == 2 && bits[0].equals(bits[1])) { verbatimEventDate = bits[0]; } } if (verbatimEventDate .matches("^[0-9]{1,2}[./ ][0-9]{1,2}[./ ][0-9]{4}[-][0-9]{1,2}[./ ][0-9]{1,2}[./ ][0-9]{4}$")) { // if verbatim date is a range with identical first and last dates (-), use just one. // Example: 12/11/1982-12/11/1982 changed to 12/11/1982 String[] bits = verbatimEventDate.split("-"); if (bits.length == 2 && bits[0].equals(bits[1])) { verbatimEventDate = bits[0]; } } if (verbatimEventDate.matches("^[0-9]{4}[-]([0-9]{1,2}|[A-Za-z]+)[-][0-9]{1,2}.*")) { // Both separators are the same. // Example 1982-02-05 // Example 1982-Feb-05 // Example 1982-02-05 // Example 1982-02-05T05:03:06 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy/MM/dd").getParser(), DateTimeFormat.forPattern("yyyy/MMM/dd").getParser(), DateTimeFormat.forPattern("yyyy-MMM-dd").getParser(), ISODateTimeFormat.dateOptionalTimeParser().getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); DateMidnight parseDate = LocalDate.parse(verbatimEventDate, formatter).toDateMidnight(); resultDate = parseDate.toString("yyyy-MM-dd"); result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (verbatimEventDate.matches("^[0-9]{4}[/]([0-9]{1,2}|[A-Za-z]+)[/][0-9]{1,2}.*")) { // Both separators are the same. // Example 1982/02/05 // Example 1982/Feb/05 // Example 1982-02-05 // Example 1982/02/05T05:03:06 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy/MM/dd").getParser(), DateTimeFormat.forPattern("yyyy/MMM/dd").getParser(), DateTimeFormat.forPattern("yyyy-MMM-dd").getParser(), ISODateTimeFormat.dateOptionalTimeParser().getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); DateMidnight parseDate = LocalDate.parse(verbatimEventDate, formatter).toDateMidnight(); resultDate = parseDate.toString("yyyy-MM-dd"); result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (verbatimEventDate.matches("^[0-9]{4}[.,][0-9]{1,2}[.,][0-9]{1,2}$")) { // Example 1982.02.05 // Example 1982,02,05 // Cases where the 1-2 digit numbers are both smaller than 12 are treated as ambiguous. String resultDateMD = null; String resultDateDM = null; DateMidnight parseDate1 = null; DateMidnight parseDate2 = null; try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy.MM.dd").getParser(), DateTimeFormat.forPattern("yyyy,MM,dd").getParser(), DateTimeFormat.forPattern("yyyy,MM.dd").getParser(), DateTimeFormat.forPattern("yyyy.MM,dd").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); parseDate1 = LocalDate.parse(verbatimEventDate, formatter).toDateMidnight(); resultDateMD = parseDate1.toString("yyyy-MM-dd"); } catch (Exception e) { logger.debug(e.getMessage()); } try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy.dd.MM").getParser(), DateTimeFormat.forPattern("yyyy,dd,MM").getParser(), DateTimeFormat.forPattern("yyyy,dd.MM").getParser(), DateTimeFormat.forPattern("yyyy.dd,MM").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); parseDate2 = LocalDate.parse(verbatimEventDate, formatter).toDateMidnight(); resultDateDM = parseDate2.toString("yyyy-MM-dd"); } catch (Exception e) { logger.debug(e.getMessage()); } if (resultDateMD != null && resultDateDM == null) { result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDateMD); } else if (resultDateMD == null && resultDateDM != null) { result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDateDM); } else if (resultDateMD != null && resultDateDM != null) { if (resultDateMD.equals(resultDateDM)) { result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDateDM); } else { result.setResultState(EventResult.EventQCResultState.AMBIGUOUS); Interval range = null; if (parseDate1.isBefore(parseDate2)) { result.setResult(resultDateMD + "/" + resultDateDM); } else { result.setResult(resultDateDM + "/" + resultDateMD); } } } } if (verbatimEventDate.matches("^[0-9]{1,2}[-/ ][0-9]{4}")) { // Example 02/1982 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("MM-yyyy").getParser(), DateTimeFormat.forPattern("MM/yyyy").getParser(), DateTimeFormat.forPattern("MM yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); DateMidnight parseDate = LocalDate.parse(verbatimEventDate, formatter).toDateMidnight(); resultDate = parseDate.toString("yyyy-MM"); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (verbatimEventDate.matches("^[0-9]{4}[0-9]{1,2}[0-9]{1,2}[?]$")) { // Example: 19720325 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyyMMdd").getParser(), DateTimeFormat.forPattern("yyyyMMdd?").getParser(), ISODateTimeFormat.dateOptionalTimeParser().getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter() .withLocale(Locale.CHINESE); DateMidnight parseDate = LocalDate.parse(verbatimEventDate, formatter).toDateMidnight(); resultDate = parseDate.toString("yyyy-MM-dd"); result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (verbatimEventDate.matches("^[0-9]{4}[-][0-9]{3}/[0-9]{4}[-][0-9]{3}$")) { // Example: 1982-145 try { String[] bits = verbatimEventDate.split("/"); DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy-D").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); LocalDate parseStartDate = LocalDate.parse(bits[0], formatter); LocalDate parseEndDate = LocalDate.parse(bits[1], formatter); resultDate = parseStartDate.toString("yyyy-MM-dd") + "/" + parseEndDate.toString("yyyy-MM-dd"); logger.debug(resultDate); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[0-9]{4}0000$")) { // case 19800000 verbatimEventDate = verbatimEventDate.substring(0, 4); } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[0-9]{4}$")) { // Example: 1962 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy").getParser(), }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); DateMidnight parseDate = LocalDate.parse(verbatimEventDate, formatter).toDateMidnight(); resultDate = parseDate.toString("yyyy"); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[12][0-9]{1}00[']{0,1}s$")) { // Example: 1900s try { String verbatimEventDateDelta = verbatimEventDate.replace("'s", "s"); DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy's").getParser(), }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); DateMidnight parseDate = LocalDate.parse(verbatimEventDateDelta, formatter).toDateMidnight(); DateMidnight endDate = parseDate.plusYears(100).minusDays(1); resultDate = parseDate.toString("yyyy") + "-01-01/" + endDate.toString("yyyy") + "-12-31"; result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[12][0-9]{2}0[']{0,1}s$")) { // Example: 1970s try { String verbatimEventDateDelta = verbatimEventDate.replace("'s", "s"); DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy's").getParser(), }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); DateMidnight parseDate = LocalDate.parse(verbatimEventDateDelta, formatter).toDateMidnight(); DateMidnight endDate = parseDate.plusYears(10).minusDays(1); resultDate = parseDate.toString("yyyy") + "-01-01/" + endDate.toString("yyyy") + "-12-31"; result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[A-Za-z]{3,9}[.]{0,1}[ ]{0,1}[-/ ][0-9]{4}$")) { // Example: Jan-1980 // Example: Jan./1980 // Example: January 1980 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("MMM-yyyy").getParser(), DateTimeFormat.forPattern("MMM/yyyy").getParser(), DateTimeFormat.forPattern("MMM /yyyy").getParser(), DateTimeFormat.forPattern("MMM yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); String cleaned = verbatimEventDate.replace(".", ""); DateMidnight parseDate = LocalDate.parse(cleaned, formatter).toDateMidnight(); resultDate = parseDate.toString("yyyy-MM"); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN)) { // Example: 04/03/1994 (ambiguous) // Example: 04/20/1994 // Example: 20/04/1994 String resultDateMD = null; String resultDateDM = null; DateMidnight parseDate1 = null; DateMidnight parseDate2 = null; if (assumemmddyyyy == null || assumemmddyyyy) { try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("MM/dd/yyyy").getParser(), DateTimeFormat.forPattern("MM/dd yyyy").getParser(), DateTimeFormat.forPattern("MM/dd-yyyy").getParser(), DateTimeFormat.forPattern("MM/dd, yyyy").getParser(), DateTimeFormat.forPattern("MM/dd,yyyy").getParser(), DateTimeFormat.forPattern("MM dd yyyy").getParser(), DateTimeFormat.forPattern("MM-dd-yyyy").getParser(), DateTimeFormat.forPattern("MM.dd.yyyy").getParser(), DateTimeFormat.forPattern("MM. dd. yyyy").getParser(), DateTimeFormat.forPattern("MM. dd. yyyy.").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers) .toFormatter(); parseDate1 = LocalDate.parse(verbatimEventDate, formatter).toDateMidnight(); resultDateMD = parseDate1.toString("yyyy-MM-dd"); } catch (Exception e) { logger.debug(e.getMessage()); } } if (assumemmddyyyy == null || !assumemmddyyyy) { try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("dd/MM/yyyy").getParser(), DateTimeFormat.forPattern("dd/MM yyyy").getParser(), DateTimeFormat.forPattern("dd/MM-yyyy").getParser(), DateTimeFormat.forPattern("dd/MM, yyyy").getParser(), DateTimeFormat.forPattern("dd/MM,yyyy").getParser(), DateTimeFormat.forPattern("dd MM yyyy").getParser(), DateTimeFormat.forPattern("dd-MM-yyyy").getParser(), DateTimeFormat.forPattern("dd.MM.yyyy").getParser(), DateTimeFormat.forPattern("dd. MM. yyyy").getParser(), DateTimeFormat.forPattern("dd. MM. yyyy.").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers) .toFormatter(); parseDate2 = LocalDate.parse(verbatimEventDate, formatter).toDateMidnight(); resultDateDM = parseDate2.toString("yyyy-MM-dd"); } catch (Exception e) { logger.debug(e.getMessage()); } } if (resultDateMD != null && resultDateDM == null) { result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDateMD); } else if (resultDateMD == null && resultDateDM != null) { result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDateDM); } else if (resultDateMD != null && resultDateDM != null) { if (resultDateMD.equals(resultDateDM)) { result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDateDM); } else { result.setResultState(EventResult.EventQCResultState.AMBIGUOUS); Interval range = null; if (parseDate1.isBefore(parseDate2)) { result.setResult(resultDateMD + "/" + resultDateDM); } else { result.setResult(resultDateDM + "/" + resultDateMD); } } } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^([0-9]{1,2}|[A-Za-z]+)[-/.]([0-9]{1,2}|[A-Za-z]+)[-/. ][0-9]{4}$")) { // Example: 03/Jan/1982 // Example: Jan-03-1982 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("MMM/dd/yyyy").getParser(), DateTimeFormat.forPattern("dd/MMM/yyyy").getParser(), DateTimeFormat.forPattern("MMM/dd yyyy").getParser(), DateTimeFormat.forPattern("dd/MMM yyyy").getParser(), DateTimeFormat.forPattern("MMM-dd-yyyy").getParser(), DateTimeFormat.forPattern("dd-MMM-yyyy").getParser(), DateTimeFormat.forPattern("MMM-dd yyyy").getParser(), DateTimeFormat.forPattern("dd-MMM yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd.yyyy").getParser(), DateTimeFormat.forPattern("dd.MMM.yyyy").getParser(), DateTimeFormat.forPattern("MM.dd.yyyy").getParser(), DateTimeFormat.forPattern("dd.MM.yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); DateMidnight parseDate = LocalDate.parse(verbatimEventDate, formatter.withLocale(Locale.ENGLISH)) .toDateMidnight(); resultDate = parseDate.toString("yyyy-MM-dd"); result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[X*]{2}[-/. ]([0-9]{1,2}|[A-Za-z]+)[-/. ][0-9]{4}$")) { // Example: XX-04-1982 (XX for day) // Example: XX-Jan-1995 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("MMM/yyyy").getParser(), DateTimeFormat.forPattern("MMM yyyy").getParser(), DateTimeFormat.forPattern("MMM-yyyy").getParser(), DateTimeFormat.forPattern("MMM yyyy").getParser(), DateTimeFormat.forPattern("MMM.yyyy").getParser(), DateTimeFormat.forPattern("MM.yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); DateMidnight parseDate = LocalDate .parse(verbatimEventDate.substring(3), formatter.withLocale(Locale.ENGLISH)) .toDateMidnight(); resultDate = parseDate.toString("yyyy-MM"); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[X*]{2}[-/. ][X*]{2,3}[-/. ][0-9]{4}$")) { // Example: XX-XXX-1995 // Example: **-**-1995 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy").getParser(), }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); String yearBit = verbatimEventDate.substring(verbatimEventDate.length() - 4); DateMidnight parseDate = LocalDate.parse(yearBit, formatter.withLocale(Locale.ENGLISH)) .toDateMidnight(); resultDate = parseDate.toString("yyyy"); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (verbatimEventDate.matches("^[0-9]{4}[-][0-9]{3}$")) { // Example: 1994-128 (three digits after year = day of year). if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN)) { try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy-D").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers) .toFormatter(); LocalDate parseDate = LocalDate.parse(verbatimEventDate, formatter); resultDate = parseDate.toString("yyyy-MM-dd"); logger.debug(resultDate); result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN)) { try { // Example: 1983-15 (two digits after year may fall into subsequent blocks). // Example: 1933-Mar DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy/M").getParser(), DateTimeFormat.forPattern("yyyy-M").getParser(), DateTimeFormat.forPattern("yyyy-MMM").getParser(), DateTimeFormat.forPattern("yyyy.MMM").getParser(), DateTimeFormat.forPattern("yyyy.MMM.").getParser(), DateTimeFormat.forPattern("yyyy MMM.").getParser(), DateTimeFormat.forPattern("yyyy MMM").getParser(), DateTimeFormat.forPattern("yyyy. MMM.").getParser(), DateTimeFormat.forPattern("yyyy. MMM").getParser(), DateTimeFormat.forPattern("yyyy/MMM").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); String cleaned = cleanMonth(verbatimEventDate); LocalDate parseDate = LocalDate.parse(cleaned, formatter.withLocale(Locale.ENGLISH)); resultDate = parseDate.toString("yyyy-MM"); // resultDate = parseDate.dayOfMonth().withMinimumValue() + "/" + parseDate.dayOfMonth().withMaximumValue(); logger.debug(resultDate); if (verbatimEventDate.matches("^[0-9]{4}[-][0-9]{2}$")) { String century = verbatimEventDate.substring(0, 2); String startBit = verbatimEventDate.substring(0, 4); String endBit = verbatimEventDate.substring(5, 7); // 1815-16 won't parse here, passes to next block // 1805-06 could be month or abbreviated year // 1805-03 should to be month if (Integer.parseInt(startBit) >= Integer.parseInt(century + endBit)) { result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } else { result.setResultState(EventResult.EventQCResultState.SUSPECT); result.setResult(resultDate); } } else { result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[0-9]{4}[-][0-9]{2}$")) { // Example: 1884-85 (two digits look like year later in century). try { String century = verbatimEventDate.substring(0, 2); String startBit = verbatimEventDate.substring(0, 4); String endBit = verbatimEventDate.substring(5, 7); String assembly = startBit + "/" + century + endBit; logger.debug(assembly); Interval parseDate = Interval.parse(assembly); logger.debug(parseDate); resultDate = parseDate.getStart().toString("yyyy") + "/" + parseDate.getEnd().toString("yyyy"); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[0-9]{4}[0-9]{2}[0-9]{2}$") && !verbatimEventDate.endsWith("0000")) { // Example: 19950315 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyyMMdd").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); DateMidnight parseDate = LocalDate.parse(verbatimEventDate, formatter.withLocale(Locale.ENGLISH)) .toDateMidnight(); resultDate = parseDate.toString("yyyy-MM-dd"); result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN)) { // Example: 1845 try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); LocalDate parseDate = LocalDate.parse(verbatimEventDate, formatter); resultDate = parseDate.dayOfYear().withMinimumValue() + "/" + parseDate.dayOfYear().withMaximumValue(); logger.debug(resultDate); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN)) { // Multiple yyyy-mmm-ddd, mmm-dd-yyyy, dd-mmm-yyyy patterns. try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("yyyy MMM dd").getParser(), DateTimeFormat.forPattern("yyyy MMM. dd").getParser(), DateTimeFormat.forPattern("yyyy, MMM dd").getParser(), DateTimeFormat.forPattern("yyyy, MMM. dd").getParser(), DateTimeFormat.forPattern("yyyy.MMM.dd").getParser(), DateTimeFormat.forPattern("yyyy.MMM.dd.").getParser(), DateTimeFormat.forPattern("yyyy. MMM. dd").getParser(), DateTimeFormat.forPattern("yyyy. MMM. dd.").getParser(), DateTimeFormat.forPattern("yyyy. MMM dd.").getParser(), DateTimeFormat.forPattern("yyyy. MMM dd").getParser(), DateTimeFormat.forPattern("yyyy MMM. dd.").getParser(), DateTimeFormat.forPattern("yyyy: MMM. dd.").getParser(), DateTimeFormat.forPattern("yyyy: MMM. dd").getParser(), DateTimeFormat.forPattern("yyyy: MMM dd").getParser(), DateTimeFormat.forPattern("yyyy:MMM dd").getParser(), DateTimeFormat.forPattern("yyyy:MMM. dd").getParser(), DateTimeFormat.forPattern("yyyy:MMM.dd").getParser(), DateTimeFormat.forPattern("yyyy MMM dd'st'").getParser(), DateTimeFormat.forPattern("yyyy MMM. dd'st'").getParser(), DateTimeFormat.forPattern("yyyy MMM dd'nd'").getParser(), DateTimeFormat.forPattern("yyyy MMM. dd'nd'").getParser(), DateTimeFormat.forPattern("yyyy MMM dd'rd'").getParser(), DateTimeFormat.forPattern("yyyy MMM. dd'rd'").getParser(), DateTimeFormat.forPattern("yyyy MMM dd'th'").getParser(), DateTimeFormat.forPattern("yyyy MMM. dd'th'").getParser(), DateTimeFormat.forPattern("MMM dd, yyyy").getParser(), DateTimeFormat.forPattern("MMM dd., yyyy").getParser(), DateTimeFormat.forPattern("MMMdd, yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'st', yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'nd', yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'rd', yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'd', yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'th', yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd, yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd, yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'st', yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'nd', yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'rd', yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'th', yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd,yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'st',yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'nd',yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'rd',yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'd',yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'th',yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd.yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'st'.yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'nd'.yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'rd'.yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'd'.yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd'th'.yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'st'. yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'nd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'rd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'th'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'st'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'nd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'rd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'th'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'st'.yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'nd'.yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'rd'.yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'd'.yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'th'.yyyy").getParser(), DateTimeFormat.forPattern("MMM-dd-yyyy").getParser(), DateTimeFormat.forPattern("MMM-dd yyyy").getParser(), DateTimeFormat.forPattern("MMM-dd, yyyy").getParser(), DateTimeFormat.forPattern("dd-MMM-yyyy").getParser(), DateTimeFormat.forPattern("dd.MMM.yyyy").getParser(), DateTimeFormat.forPattern("dd,MMM,yyyy").getParser(), DateTimeFormat.forPattern("dd.MMM.,yyyy").getParser(), DateTimeFormat.forPattern("dd. MMM.,yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd. yyyy").getParser(), DateTimeFormat.forPattern("MMM, dd yyyy").getParser(), DateTimeFormat.forPattern("MMM, dd. yyyy").getParser(), DateTimeFormat.forPattern("MMM, dd, yyyy").getParser(), DateTimeFormat.forPattern("MMM, dd., yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd, yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd/yyyy").getParser(), DateTimeFormat.forPattern("MMM dd,yyyy").getParser(), DateTimeFormat.forPattern("MMM dd, yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd,yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd-yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd-yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd, yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd., yyyy").getParser(), DateTimeFormat.forPattern("MMM., dd, yyyy").getParser(), DateTimeFormat.forPattern("MMM.,dd, yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'' yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'' yyyy").getParser(), DateTimeFormat.forPattern("dd. MMM. yyyy").getParser(), DateTimeFormat.forPattern("dd. MMM.yyyy").getParser(), DateTimeFormat.forPattern("dd MMM., yyyy").getParser(), DateTimeFormat.forPattern("dd MMM.,yyyy").getParser(), DateTimeFormat.forPattern("dd MMM,.yyyy").getParser(), DateTimeFormat.forPattern("dd MMM,. yyyy").getParser(), DateTimeFormat.forPattern("dd MMM..yyyy").getParser(), DateTimeFormat.forPattern("dd MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd MMM yyyy").getParser(), DateTimeFormat.forPattern("dd MMM,yyyy").getParser(), DateTimeFormat.forPattern("dd MMM.yyyy").getParser(), DateTimeFormat.forPattern("ddMMM.yyyy").getParser(), DateTimeFormat.forPattern("ddMMM. yyyy").getParser(), DateTimeFormat.forPattern("dd.MMM-yyyy").getParser(), DateTimeFormat.forPattern("dd-MMM-yyyy").getParser(), DateTimeFormat.forPattern("dd.MMM yyyy").getParser(), DateTimeFormat.forPattern("dd. MMM yyyy").getParser(), DateTimeFormat.forPattern("dd, MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd, MMM; yyyy").getParser(), DateTimeFormat.forPattern("dd. MMM; yyyy").getParser(), DateTimeFormat.forPattern("dd MMM-yyyy").getParser(), DateTimeFormat.forPattern("dd-MMM yyyy").getParser(), DateTimeFormat.forPattern("ddMMMyyyy").getParser(), DateTimeFormat.forPattern("MMM dd yyyy").getParser(), DateTimeFormat.forPattern("MMM dd/yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'st' yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'nd' yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'rd' yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'd' yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'th' yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'st' yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'nd' yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'rd' yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'd' yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'th' yyyy").getParser(), DateTimeFormat.forPattern("MMMdd yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd yyyy").getParser(), DateTimeFormat.forPattern("dd MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd'st' MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd'nd' MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd'rd' MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd'd' MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd'th MMM', yyyy").getParser(), DateTimeFormat.forPattern("dd MMM., yyyy").getParser(), DateTimeFormat.forPattern("dd'st' MMM., yyyy").getParser(), DateTimeFormat.forPattern("dd'nd' MMM., yyyy").getParser(), DateTimeFormat.forPattern("dd'rd' MMM., yyyy").getParser(), DateTimeFormat.forPattern("dd'th' MMM., yyyy").getParser(), DateTimeFormat.forPattern("dd MMM yyyy").getParser(), DateTimeFormat.forPattern("dd'st' MMM yyyy").getParser(), DateTimeFormat.forPattern("dd'nd' MMM yyyy").getParser(), DateTimeFormat.forPattern("dd'rd' MMM yyyy").getParser(), DateTimeFormat.forPattern("dd'd' MMM yyyy").getParser(), DateTimeFormat.forPattern("dd'th' MMM yyyy").getParser(), DateTimeFormat.forPattern("dd MMM. yyyy").getParser(), DateTimeFormat.forPattern("dd'st' MMM. yyyy").getParser(), DateTimeFormat.forPattern("dd'nd' MMM. yyyy").getParser(), DateTimeFormat.forPattern("dd'rd' MMM. yyyy").getParser(), DateTimeFormat.forPattern("dd'd' MMM. yyyy").getParser(), DateTimeFormat.forPattern("dd'th' MMM. yyyy").getParser(), DateTimeFormat.forPattern("dd'st' MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd'nd' MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd'rd' MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd'd' MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd'th' MMM, yyyy").getParser(), DateTimeFormat.forPattern("dd/MMM/yyyy").getParser(), DateTimeFormat.forPattern("dd/MMM yyyy").getParser(), DateTimeFormat.forPattern("MMM/dd yyyy").getParser(), DateTimeFormat.forPattern("MMM/dd/yyyy").getParser(), DateTimeFormat.forPattern("MMM dd. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'st'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'nd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'rd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'th'. yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd. yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'st'. yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'nd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'rd'. yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'th'. yyyy").getParser(), DateTimeFormat.forPattern("MMM dd.yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd.yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd-yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'st'-yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'nd'-yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'rd'-yyyy").getParser(), DateTimeFormat.forPattern("MMM. dd'th'-yyyy").getParser(), DateTimeFormat.forPattern("MMM dd-yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'st'-yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'nd'-yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'rd'-yyyy").getParser(), DateTimeFormat.forPattern("MMM dd'th'-yyyy").getParser(), DateTimeFormat.forPattern("yyyy-MMM-dd").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers).toFormatter(); String cleaned = cleanMonth(verbatimEventDate); cleaned = cleaned.replace("''", "'"); try { // Specify English locale, or local default will be used LocalDate parseDate = LocalDate.parse(cleaned, formatter.withLocale(Locale.ENGLISH)); resultDate = parseDate.toString("yyyy-MM-dd"); } catch (Exception e) { try { logger.debug(e.getMessage()); LocalDate parseDate = LocalDate.parse(cleaned, formatter.withLocale(Locale.FRENCH)); resultDate = parseDate.toString("yyyy-MM-dd"); } catch (Exception e1) { try { logger.debug(e1.getMessage()); LocalDate parseDate = LocalDate.parse(cleaned, formatter.withLocale(Locale.ITALIAN)); resultDate = parseDate.toString("yyyy-MM-dd"); } catch (Exception e2) { try { logger.debug(e2.getMessage()); LocalDate parseDate = LocalDate.parse(cleaned, formatter.withLocale(Locale.GERMAN)); resultDate = parseDate.toString("yyyy-MM-dd"); } catch (Exception e3) { try { logger.debug(e2.getMessage()); LocalDate parseDate = LocalDate.parse(cleaned, formatter.withLocale(Locale.forLanguageTag("es"))); resultDate = parseDate.toString("yyyy-MM-dd"); } catch (Exception e4) { logger.debug(e2.getMessage()); LocalDate parseDate = LocalDate.parse(cleaned, formatter.withLocale(Locale.forLanguageTag("pt"))); resultDate = parseDate.toString("yyyy-MM-dd"); } } } } } logger.debug(resultDate); result.setResultState(EventResult.EventQCResultState.DATE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } logger.debug(result.getResultState()); if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN)) { // Example: jan.-1992 // Example: January 1992 if (verbatimEventDate.matches(".*[0-9]{4}.*")) { try { DateTimeParser[] parsers = { DateTimeFormat.forPattern("MMM, yyyy").getParser(), DateTimeFormat.forPattern("MMM., yyyy").getParser(), DateTimeFormat.forPattern("MMM.,yyyy").getParser(), DateTimeFormat.forPattern("MMM.-yyyy").getParser(), DateTimeFormat.forPattern("MMM.yyyy").getParser(), DateTimeFormat.forPattern("MMM. yyyy").getParser(), DateTimeFormat.forPattern("MMM-yyyy").getParser(), DateTimeFormat.forPattern("MMM -yyyy").getParser(), DateTimeFormat.forPattern("MMM yyyy").getParser(), DateTimeFormat.forPattern("MMM/yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers) .toFormatter(); String cleaned = cleanMonth(verbatimEventDate); // Strip off a trailing period after a final year if (cleaned.matches("^.*[0-9]{4}[.]$")) { cleaned = cleaned.replaceAll("[.]$", ""); } LocalDate parseDate = LocalDate.parse(cleaned, formatter.withLocale(Locale.ENGLISH)); resultDate = parseDate.toString("yyyy-MM"); logger.debug(resultDate); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[0-9]{4}([- ]+| to |[/ ]+)[0-9]{4}$")) { // Example: 1882-1995 // Example: 1882 to 1885 // Example: 1882/1885 try { String cleaned = verbatimEventDate.replace(" ", ""); cleaned = cleaned.replace("-", "/"); if (cleaned.matches("^[0-9]{4}to[0-9]{4}$")) { int len = verbatimEventDate.length(); int lastYear = len - 4; cleaned = verbatimEventDate.substring(0, 4) + "/" + verbatimEventDate.substring(lastYear, len); } logger.debug(cleaned); Interval parseDate = Interval.parse(cleaned); logger.debug(parseDate); resultDate = parseDate.getStart().toString("yyyy") + "/" + parseDate.getEnd().toString("yyyy"); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate .matches("^[A-Za-z]+[.]{0,1}( and | to |[-][ ]{0,1}| [-] )[A-Za-z]+[.]{0,1}(, |[/ .])[0-9]{4}$")) { logger.debug(verbatimEventDate); // Example: Jan to Feb 1882 // Example: Jan-Feb/1882 verbatimEventDate = verbatimEventDate.replace(", ", " "); if (verbatimEventDate.matches("^[A-Za-z]+[.]{0,1}[-][A-Za-z]+[.]{0,1}[.][0-9]{4}$")) { // transform case with multiple periods to slash before year. verbatimEventDate = verbatimEventDate.substring(0, verbatimEventDate.length() - 5) + "/" + verbatimEventDate.substring(verbatimEventDate.length() - 4); logger.debug(verbatimEventDate); } if (verbatimEventDate.matches("^[A-Za-z]+[.]{0,1}[ ][-][ ]{1}[A-Za-z]+[.]{0,1}[/ .][0-9]{4}$")) { // remove space around dash. verbatimEventDate = verbatimEventDate.replace(" - ", "-"); logger.debug(verbatimEventDate); } if (verbatimEventDate.matches("^[A-Za-z]+[.]{0,1}[-][ ]{1}[A-Za-z]+[.]{0,1}[/ .][0-9]{4}$")) { // remove space trailing after dash. verbatimEventDate = verbatimEventDate.replace("- ", "-"); logger.debug(verbatimEventDate); } if (verbatimEventDate.matches("^[A-Za-z]+[.]{0,1} and {1}[A-Za-z]+[.]{0,1}[/ .][0-9]{4}$")) { // replace and with dash verbatimEventDate = verbatimEventDate.replace(" and ", "-"); logger.debug(verbatimEventDate); } if (verbatimEventDate.matches("^[A-Za-z]+[.]{0,1} to {1}[A-Za-z]+[.]{0,1}[/ .][0-9]{4}$")) { // replace to with dash verbatimEventDate = verbatimEventDate.replace(" to ", "-"); logger.debug(verbatimEventDate); } try { String[] bits = verbatimEventDate.replace(" ", "/").split("-"); if (bits != null && bits.length == 2) { String year = verbatimEventDate.substring(verbatimEventDate.length() - 4, verbatimEventDate.length()); String startBit = bits[0] + "/" + year; DateTimeParser[] parsers = { DateTimeFormat.forPattern("MMM/yyyy").getParser(), DateTimeFormat.forPattern("MMM./yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers) .toFormatter(); LocalDate parseStartDate = LocalDate.parse(cleanMonth(startBit), formatter.withLocale(Locale.ENGLISH)); LocalDate parseEndDate = LocalDate.parse(cleanMonth(bits[1]), formatter.withLocale(Locale.ENGLISH)); resultDate = parseStartDate.toString("yyyy-MM") + "/" + parseEndDate.toString("yyyy-MM"); logger.debug(resultDate); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches( "^[0-9]{1,2}[ /.]{0,1}[A-Za-z]+[.]{0,1}( - |[-])[0-9]{1,2}[ /.]{0,1}[A-Za-z]+[.]{0,1}[/ -.][0-9]{4}$")) { logger.debug(verbatimEventDate); // Example: 05/Jan/1882-03/Feb/1885 if (verbatimEventDate.matches( "^[0-9]{1,2}[ /.]{0,1}[A-Za-z]+[.]{0,1}[-][0-9]{1,2}[ /.]{0,1}[A-Za-z]+[.]{0,1}[-][0-9]{4}$")) { // transform case with multiple dashes to slash before year. verbatimEventDate = verbatimEventDate.substring(0, verbatimEventDate.length() - 5) + "/" + verbatimEventDate.substring(verbatimEventDate.length() - 4); logger.debug(verbatimEventDate); } if (verbatimEventDate.matches( "^[0-9]{1,2}[ /.]{0,1}[A-Za-z]+[.]{0,1}[-][0-9]{1,2}[ /.]{0,1}[A-Za-z]+[.]{0,1}[.][0-9]{4}$")) { // transform case with multiple periods to slash before year. verbatimEventDate = verbatimEventDate.substring(0, verbatimEventDate.length() - 5) + "/" + verbatimEventDate.substring(verbatimEventDate.length() - 4); logger.debug(verbatimEventDate); } try { String[] bits = verbatimEventDate.replace(" - ", "-").replace(" ", "/").split("-"); if (bits != null && bits.length == 2) { String year = verbatimEventDate.substring(verbatimEventDate.length() - 4, verbatimEventDate.length()); String startBit = bits[0] + "/" + year; logger.debug(cleanMonth(startBit)); logger.debug(cleanMonth(bits[1])); DateTimeParser[] parsers = { DateTimeFormat.forPattern("dd MMM/yyyy").getParser(), DateTimeFormat.forPattern("dd.MMM/yyyy").getParser(), DateTimeFormat.forPattern("dd/MMM/yyyy").getParser(), DateTimeFormat.forPattern("ddMMM/yyyy").getParser(), DateTimeFormat.forPattern("dd MMM./yyyy").getParser(), DateTimeFormat.forPattern("dd.MMM./yyyy").getParser(), DateTimeFormat.forPattern("dd/MMM./yyyy").getParser(), DateTimeFormat.forPattern("ddMMM./yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers) .toFormatter(); LocalDate parseStartDate = LocalDate.parse(cleanMonth(startBit), formatter.withLocale(Locale.ENGLISH)); LocalDate parseEndDate = LocalDate.parse(cleanMonth(bits[1]), formatter.withLocale(Locale.ENGLISH)); resultDate = parseStartDate.toString("yyyy-MM-dd") + "/" + parseEndDate.toString("yyyy-MM-dd"); logger.debug(resultDate); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches( "^[A-Za-z]+[.]{0,1}[ ]{0,1}[0-9]{1,2}( - |[-]| to | and | et )[A-Za-z]+[.]{0,1}[ ]{0,1}[0-9]{1,2}[/ .,][ ]{0,1}[0-9]{4}$")) { logger.debug(verbatimEventDate); // Example: Aug. 5 - Sept. 8, 1943 try { String[] bits = verbatimEventDate.replace(" to ", "-").replace(" - ", "-").replace(" and ", "-") .replace(" et ", "-").replace(", ", " ").replace(" ", "/").split("-"); if (bits != null && bits.length == 2) { String year = verbatimEventDate.substring(verbatimEventDate.length() - 4, verbatimEventDate.length()); String startBit = bits[0] + "/" + year; logger.debug(cleanMonth(startBit)); logger.debug(cleanMonth(bits[1])); DateTimeParser[] parsers = { DateTimeFormat.forPattern("MMM/dd/yyyy").getParser(), DateTimeFormat.forPattern("MMM./dd/yyyy").getParser(), DateTimeFormat.forPattern("MMM.dd/yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers) .toFormatter(); LocalDate parseStartDate = LocalDate.parse(cleanMonth(startBit), formatter.withLocale(Locale.ENGLISH)); LocalDate parseEndDate = LocalDate.parse(cleanMonth(bits[1]), formatter.withLocale(Locale.ENGLISH)); resultDate = parseStartDate.toString("yyyy-MM-dd") + "/" + parseEndDate.toString("yyyy-MM-dd"); logger.debug(resultDate); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches( "^[0-9]{1,2}([ ]{0,1}[-][ ]{0,1}| and | et | to )[0-9]{1,2}[ /.]{0,1}[A-Za-z]+[.]{0,1}[/ -.][0-9]{4}$")) { // Example: 11 et 14 VII 1910 // Example: 05-02 Jan./1992 String toCheck = verbatimEventDate; toCheck = toCheck.replace(" - ", "-").replace(" et ", "-").replace(" and ", "-").replace(" to ", "-"); // Note: "and" has different semantics than "to", may imply that a specimen record // represents two occurrences (e.g. flower on one date, fruit on another) rather than // a range, but dwc:eventDate representation for both forms on one event is a range. if (toCheck.matches("^[0-9]{1,2}[-][0-9]{1,2}[ /.]{0,1}[A-Za-z]+[.]{0,1}[-][0-9]{4}$")) { // transform case with multiple dashes to slash before year. toCheck = toCheck.substring(0, toCheck.length() - 5) + "/" + toCheck.substring(toCheck.length() - 4); logger.debug(toCheck); } if (toCheck.matches("^[0-9]{1,2}[-][0-9]{1,2}[ /.]{0,1}[A-Za-z]+[.]{0,1}[.][0-9]{4}$")) { // transform case with multiple periods to slash before year. toCheck = toCheck.substring(0, toCheck.length() - 5) + "/" + toCheck.substring(toCheck.length() - 4); logger.debug(toCheck); } try { String[] bits = toCheck.replace(" ", "/").split("-"); if (bits != null && bits.length == 2) { String year = toCheck.substring(toCheck.length() - 4, toCheck.length()); logger.debug(cleanMonth(bits[1])); DateTimeParser[] parsers = { DateTimeFormat.forPattern("dd MMM/yyyy").getParser(), DateTimeFormat.forPattern("dd.MMM/yyyy").getParser(), DateTimeFormat.forPattern("dd/MMM/yyyy").getParser(), DateTimeFormat.forPattern("ddMMM/yyyy").getParser(), DateTimeFormat.forPattern("dd MMM./yyyy").getParser(), DateTimeFormat.forPattern("dd.MMM./yyyy").getParser(), DateTimeFormat.forPattern("dd/MMM./yyyy").getParser(), DateTimeFormat.forPattern("ddMMM./yyyy").getParser() }; DateTimeFormatter formatter = new DateTimeFormatterBuilder().append(null, parsers) .toFormatter(); LocalDate parseEndDate = LocalDate.parse(cleanMonth(bits[1]), formatter.withLocale(Locale.ENGLISH)); String startMonthYear = parseEndDate.toString("MMM/yyyy"); String startBit = bits[0] + "/" + startMonthYear; logger.debug(startBit); LocalDate parseStartDate = LocalDate.parse(startBit, formatter.withLocale(Locale.ENGLISH)); resultDate = parseStartDate.toString("yyyy-MM-dd") + "/" + parseEndDate.toString("yyyy-MM-dd"); logger.debug(resultDate); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[0-9]{2}[-. ]XXX[-. ][0-9]{4}$")) { // Example: 05-XXX-1884 try { String start = verbatimEventDate.substring(verbatimEventDate.length() - 4) + "-01-" + verbatimEventDate.substring(0, 2); String end = verbatimEventDate.substring(verbatimEventDate.length() - 4) + "-12-" + verbatimEventDate.substring(0, 2); EventResult compositeResult = DateUtils.extractDateFromVerbatimER(start + "/" + end, yearsBeforeSuspect, assumemmddyyyy); logger.debug(compositeResult.getResultState()); if (compositeResult.getResultState().equals(EventResult.EventQCResultState.RANGE)) { result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(compositeResult.getResult()); logger.debug(result.getResult()); } } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN) && verbatimEventDate.matches("^[0-9]{4}-[0-9]{2}/[0-9]{4}-[0-9]{2}$")) { // Example: 1885-03/1886-04 try { Interval parseDate = Interval.parse(verbatimEventDate); logger.debug(parseDate); resultDate = parseDate.getStart().toString("yyyy-MM") + "/" + parseDate.getEnd().toString("yyyy-MM"); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN)) { // Example: 1995-03-05/1996-05-08 try { Interval parseDate = Interval.parse(verbatimEventDate); logger.debug(parseDate); resultDate = parseDate.getStart().toString("yyyy-MM-dd") + "/" + parseDate.getEnd().toString("yyyy-MM-dd"); result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultDate); } catch (Exception e) { logger.debug(e.getMessage()); } } if (result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN)) { // Example: Jan,15-18 1882 // Example: Jan. 17 and 18 1882 String cleaned = verbatimEventDate.trim(); if (verbatimEventDate.matches("^[A-Za-z.]+[ ,]+[0-9]{1,2} and [0-9]{0,2}[ ,]+[0-9]{4}$")) { // Note: "and" has different semantics than "to", may imply that a specimen record // represents two occurrences (e.g. flower on one date, fruit on another) rather than // a range, but dwc:eventDate representation for both forms on one event is a range. cleaned = cleaned.replace(" and ", " to "); } if (verbatimEventDate.matches("^[A-Za-z.]+[ ,]+[0-9]{1,2}-[0-9]{0,2}[ ,]+[0-9]{4}$")) { cleaned = cleaned.replace("-", " to "); } if (cleaned.contains(" to ")) { String[] bits = cleaned.split(" to "); String yearRegex = ".*([0-9]{4}).*"; Matcher yearMatcher = Pattern.compile(yearRegex).matcher(cleaned); String monthRegex = "([A-Za-z.]+).*"; Matcher monthMatcher = Pattern.compile(monthRegex).matcher(cleaned); if (yearMatcher.matches() && monthMatcher.matches()) { String year = yearMatcher.group(1); String month = monthMatcher.group(1); if (bits.length == 2) { if (!bits[0].contains(year)) { bits[0] = bits[0] + " " + year; } if (!bits[1].contains(year)) { bits[1] = bits[1] + " " + year; } if (!bits[1].contains(month)) { bits[1] = month + " " + bits[1]; } Map<String, String> resultBit0 = DateUtils.extractDateFromVerbatim(bits[0]); if (resultBit0.size() > 0 && resultBit0.get("resultState").equals("date")) { Map<String, String> resultBit1 = DateUtils.extractDateFromVerbatim(bits[1]); if (resultBit1.size() > 0 && resultBit1.get("resultState").equals("date")) { result.setResultState(EventResult.EventQCResultState.RANGE); result.setResult(resultBit0.get("result") + "/" + resultBit1.get("result")); } } logger.debug(bits[0]); logger.debug(bits[1]); } } } } // Now test to see if result is sane. if (result != null && !result.getResultState().equals(EventResult.EventQCResultState.NOT_RUN)) { Interval testExtract = DateUtils.extractDateInterval(result.getResult()); if (testExtract == null || testExtract.getStart().getYear() < yearsBeforeSuspect) { result.setResultState(EventResult.EventQCResultState.SUSPECT); logger.debug(result.getResult()); logger.debug(testExtract); } else { logger.debug(result.getResult()); } if (!verbatimEventDate.matches(".*[0-9]{4}.*") && yearsBeforeSuspect > 999) { result = new EventResult(); logger.debug(result.getResult()); } } return result; }
From source file:org.filteredpush.qc.date.DateUtils.java
License:Apache License
/** * Identify whether an event date is consistent with its atomic parts. * //www. j a v a 2 s . c o m * @param eventDate dwc:eventDate string to compare with atomic parts. * @param startDayOfYear dwc:startDayOfYear for comparison with eventDate * @param endDayOfYear dwc:endDayOfYear for comparison with eventDate * @param year dwc:year for comparison with eventDate * @param month dwc:month for comparison with eventDate * @param day dwc:day for comparison with eventDate * * @return true if consistent, or if eventDate is empty, or if all * atomic parts are empty, otherwise false. */ public static boolean isConsistent(String eventDate, String startDayOfYear, String endDayOfYear, String year, String month, String day) { if (isEmpty(eventDate) || (isEmpty(startDayOfYear) && isEmpty(endDayOfYear) && isEmpty(year) && isEmpty(month) && isEmpty(day))) { return true; } // TODO: Add support for eventTime boolean result = false; result = isConsistent(eventDate, year, month, day); logger.debug(result); if ((result || (!isEmpty(eventDate) && isEmpty(year) && isEmpty(month) && isEmpty(day))) && (!isEmpty(startDayOfYear) || !isEmpty(endDayOfYear))) { if (endDayOfYear == null || endDayOfYear.trim().length() == 0 || startDayOfYear.trim().equals(endDayOfYear.trim())) { int startDayInt = -1; try { startDayInt = Integer.parseInt(startDayOfYear); } catch (NumberFormatException e) { logger.debug(e.getMessage()); logger.debug(startDayOfYear + " is not an integer."); result = false; } if (DateUtils.extractDate(eventDate) != null && DateUtils.extractDate(eventDate).getDayOfYear() == startDayInt) { result = true; } else { result = false; } } else { int startDayInt = -1; int endDayInt = -1; try { startDayInt = Integer.parseInt(startDayOfYear); endDayInt = Integer.parseInt(endDayOfYear); } catch (NumberFormatException e) { logger.debug(e.getMessage()); result = false; } Interval eventDateInterval = DateUtils.extractDateInterval(eventDate); logger.debug(eventDateInterval); int endDayOfInterval = eventDateInterval.getEnd().getDayOfYear(); // midnight on the next day, so subtract 1 to get the same integer day. if (eventDateInterval.getStart().getDayOfYear() == startDayInt && endDayOfInterval == endDayInt) { result = true; } else { result = false; } } } return result; }
From source file:org.filteredpush.qc.date.DateUtils.java
License:Apache License
/** * Test if an event date specifies a duration of one day or less. * //from w w w . ja v a 2 s.c o m * @param eventDate to test. * @return true if duration is one day or less. */ public static boolean specificToDay(String eventDate) { boolean result = false; if (!isEmpty(eventDate)) { Interval eventDateInterval = extractInterval(eventDate); logger.debug(eventDateInterval); logger.debug(eventDateInterval.toDuration()); if (eventDateInterval.toDuration().getStandardDays() < 1l) { result = true; } else if (eventDateInterval.toDuration().getStandardDays() == 1l && eventDateInterval.getStart().getDayOfYear() == eventDateInterval.getEnd().getDayOfYear()) { result = true; } } return result; }
From source file:org.filteredpush.qc.date.DateUtils.java
License:Apache License
/** * Given an event date (which may represent a date range), check to see if the * eventDate contains a leap day. Returns false if provided a null or invalid * eventDate value./*from w ww . j a v a 2 s.co m*/ * * @param eventDate to check for a leap day * @return true if a leap day is present in the eventDate range, otherwise false. */ public static boolean includesLeapDay(String eventDate) { boolean result = false; if (!DateUtils.isEmpty(eventDate) && DateUtils.eventDateValid(eventDate)) { Interval interval = extractInterval(eventDate); Integer sYear = interval.getStart().getYear(); Integer eYear = interval.getEnd().getYear(); String startYear = Integer.toString(sYear).trim(); String endYear = Integer.toString(eYear).trim(); String leapDay = startYear + "-02-29"; logger.debug(leapDay); if (DateUtils.eventDateValid(leapDay)) { if (interval.contains(DateUtils.extractInterval(leapDay))) { result = true; } } // Range spanning more than one year, check last year if (!endYear.equals(startYear)) { leapDay = endYear + "-02-29"; logger.debug(leapDay); if (DateUtils.eventDateValid(leapDay)) { if (interval.contains(DateUtils.extractInterval(leapDay))) { result = true; } } } // Ranges of more than two years, check intermediate years if (eYear > sYear + 1) { for (int testYear = sYear + 1; testYear < eYear; testYear++) { leapDay = Integer.toString(testYear).trim() + "-02-29"; logger.debug(leapDay); if (DateUtils.eventDateValid(leapDay)) { if (interval.contains(DateUtils.extractInterval(leapDay))) { result = true; } } } } } return result; }
From source file:org.filteredpush.qc.date.DateUtils.java
License:Apache License
/** * Count the number of leap days present in an event date * /* w w w. j av a 2 s . c o m*/ * @param eventDate to check for leap days * @return number of leap days present in eventDate, 0 if no leap days are present or * if eventDate does not contain a date. */ public static int countLeapDays(String eventDate) { int result = 0; if (!DateUtils.isEmpty(eventDate) && DateUtils.eventDateValid(eventDate)) { Interval interval = extractInterval(eventDate); Integer sYear = interval.getStart().getYear(); Integer eYear = interval.getEnd().getYear(); String startYear = Integer.toString(sYear).trim(); String endYear = Integer.toString(eYear).trim(); String leapDay = startYear + "-02-29"; logger.debug(leapDay); if (DateUtils.eventDateValid(leapDay)) { if (interval.contains(DateUtils.extractInterval(leapDay))) { result = 1; } } // Range spanning more than one year, check last year if (!endYear.equals(startYear)) { leapDay = endYear + "-02-29"; logger.debug(leapDay); if (DateUtils.eventDateValid(leapDay)) { if (interval.contains(DateUtils.extractInterval(leapDay))) { result++; } } } // Ranges of more than two years, check intermediate years if (eYear > sYear + 1) { for (int testYear = sYear + 1; testYear < eYear; testYear++) { leapDay = Integer.toString(testYear).trim() + "-02-29"; logger.debug(leapDay); if (DateUtils.eventDateValid(leapDay)) { if (interval.contains(DateUtils.extractInterval(leapDay))) { result++; } } } } } return result; }
From source file:org.filteredpush.qc.date.DwCEventDQ.java
License:Apache License
/** * Given a set of event terms, examine the content of eventDate, and if it is correctly formatted and * can be interpreted, fill in any empty of the following terms (year, month, day, startDayOfYear, endDayOfYear) * with appropriate values./*from w w w.j a va2 s . co m*/ * *TG2-AMENDMENT_EVENT_FROM_EVENTDATE 710fe118-17e1-440f-b428-88ba3f547d6d * * @param eventDate to examine * @param year to check for emptyness * @param month to check for emptyness * @param day to check for emptyness * @param startDayOfYear to check for emptyness * @param endDayOfYear to check for emptyness * @return an EventDQAmmendment which may contain a proposed amendment for key dwc:day. */ @Provides(value = "urn:uuid:710fe118-17e1-440f-b428-88ba3f547d6d") @Amendment(label = "AMENDMENT_EVENT_FROM_EVENTDATE", description = "One or more empty component terms of the dwc:Event class (dwc:year, dwc:month, dwc:day, dwc:startDayOfYear, dwc:endDayOfYear) have been filled in from a valid value in the term dwc:eventDate.") @Specification(value = "One or more empty component terms of the dwc:Event class (dwc:year, dwc:month, dwc:day, dwc:startDayOfYear, dwc:endDayOfYear) have been filled in from a valid value in the term dwc:eventDate. The field dwc:eventDate is not EMPTY and contains a valid ISO 8601:2004(E). Run this amendment after any other amendment which may affect dwc:eventDate.") public static DQResponse<AmendmentValue> fillInEventFromEventDate( @Consulted(value = "dwc:eventDate") String eventDate, @ActedUpon(value = "dwc:year") String year, @ActedUpon(value = "dwc:month") String month, @ActedUpon(value = "dwc:day") String day, @ActedUpon(value = "dwc:startDayOfYear") String startDayOfYear, @ActedUpon(value = "dwc:endDayOfYear") String endDayOfYear) { DQResponse<AmendmentValue> result = new DQResponse<AmendmentValue>(); if (DateUtils.isEmpty(eventDate)) { result.setResultState(ResultState.INTERNAL_PREREQUISITES_NOT_MET); result.addComment("No value for dwc:eventDate was provided, no data to fill in from."); } else { if (!DateUtils.eventDateValid(eventDate)) { result.setResultState(ResultState.INTERNAL_PREREQUISITES_NOT_MET); result.addComment("Provided value for dwc:eventDate [" + eventDate + "] could not be interpreted."); } else { boolean isRange = false; if (DateUtils.isRange(eventDate)) { isRange = true; } Interval interval = DateUtils.extractInterval(eventDate); if (interval == null) { result.setResultState(ResultState.INTERNAL_PREREQUISITES_NOT_MET); result.addComment("Provided value for dwc:eventDate [" + eventDate + "] appears to be correctly formatted, but could not be interpreted as a valid date."); } else { Map<String, String> values = new HashMap<>(); if (DateUtils.isEmpty(day)) { String newDay = Integer.toString(interval.getStart().getDayOfMonth()); values.put("dwc:day", newDay); result.setResultState(ResultState.FILLED_IN); if (isRange) { result.addComment( "Added day [" + newDay + "] from start day of range [" + eventDate + "]."); } else { result.addComment("Added day [" + newDay + "] from eventDate [" + eventDate + "]."); } } if (DateUtils.isEmpty(month)) { String newMonth = Integer.toString(interval.getStart().getMonthOfYear()); values.put("dwc:month", newMonth); result.setResultState(ResultState.FILLED_IN); if (isRange) { result.addComment("Added month [" + newMonth + "] from start month of eventDate [" + eventDate + "]."); } else { result.addComment("Added month [" + newMonth + "] from eventDate [" + eventDate + "]."); } } if (DateUtils.isEmpty(month)) { String newMonth = Integer.toString(interval.getStart().getMonthOfYear()); values.put("dwc:month", newMonth); result.setResultState(ResultState.FILLED_IN); if (isRange) { result.addComment("Added month [" + newMonth + "] from start month of eventDate [" + eventDate + "]."); } else { result.addComment("Added month [" + newMonth + "] from eventDate [" + eventDate + "]."); } } if (DateUtils.isEmpty(year)) { String newYear = Integer.toString(interval.getStart().getYear()); values.put("dwc:year", newYear); result.setResultState(ResultState.FILLED_IN); if (isRange) { result.addComment("Added year [" + newYear + "] from start month of eventDate [" + eventDate + "]."); } else { result.addComment("Added year [" + newYear + "] from eventDate [" + eventDate + "]."); } } if (DateUtils.isEmpty(startDayOfYear)) { String newDay = Integer.toString(interval.getStart().getDayOfYear()); values.put("dwc:startDayOfYear", newDay); result.setResultState(ResultState.FILLED_IN); if (isRange) { result.addComment("Added startDayOfYear [" + newDay + "] from start day of eventDate [" + eventDate + "]."); } else { result.addComment( "Added startDayOfYear [" + newDay + "] from eventDate [" + eventDate + "]."); } } if (DateUtils.isEmpty(endDayOfYear)) { String newDay = Integer.toString(interval.getEnd().getDayOfYear()); values.put("dwc:endDayOfYear", newDay); result.setResultState(ResultState.FILLED_IN); if (isRange) { result.addComment("Added endDayOfYear [" + newDay + "] from end day of eventDate [" + eventDate + "]."); } else { result.addComment( "Added endDayOfYear [" + newDay + "] from eventDate [" + eventDate + "]."); } } result.setValue(new AmendmentValue(values)); // Time could also be populated, but we probably don't want to. Here is a minimal implementation, // which illustrates some issues in implementation (using zulu time or not, dealing with time in ranges...) //if (DateUtils.isEmpty(eventTime)) { // if (DateUtils.containsTime(eventDate)) { // String newTime = DateUtils.extractZuluTime(eventDate); // result.addResult("dwc:endDayOfYear", newTime ); // result.setResultState(ResultState.FILLED_IN); // result.addComment("Added eventTime ["+ newTime +"] from eventDate ["+eventDate+"]."); // } //} if (!result.getResultState().equals(ResultState.FILLED_IN)) { result.setResultState(ResultState.NO_CHANGE); result.addComment("No changes proposed, all candidate fields to fill in contain values."); } } // end interval extraction } // end format validity check } return result; }
From source file:org.filteredpush.qc.date.DwCEventDQ.java
License:Apache License
/** * Given an eventDate check to see if that event date falls entirely outside a range from a * specified lower bound (1700-01-01 by default) and the present. * /*from w w w . j a v a 2 s . c o m*/ * TODO: This may or may not be consistent with the standard test, current implementation here fails * only if the eventDate has no overlap with the lowerBound to the present, not if the eventDate has * any overlap outside the lowerBound to the present. * TG2-VALIDATION_EVENTDATE_OUTOFRANGE ? probably not * * * @param eventDate to check * @param lowerBound integer representing the year to use as the lower boundary, if null, then uses 1700 * @param useLowerBound if false, no lower limit, otherwise uses supplied lower bound. * @return an DQValidationResponse object describing whether the provided value is in range. */ @Provides(value = "urn:uuid:bc8d1ffb-d074-4b4b-919c-f0c8fbe1a618") // new guid, probably not the test as specified @Validation(label = "VALIDATION_EVENTDATE_WHOLLYOUTOFRANGE", description = "The range of dwc:eventDate does not fall entirely into the future and optionally does not fall entirely before a date designated when the test is run") @Specification(value = "The range of dwc:eventDate is not entirely the future and optionally does not entirely fall before a date designated when the test is run The field dwc:eventDate is not EMPTY.") public static DQResponse<ComplianceValue> isEventDateInRange( @ActedUpon(value = "dwc:eventDate") String eventDate, @Parameter(name = "whollyOutOfRangeLowerBound") Integer lowerBound, @Parameter(name = "whollyOutOfRangeUseLowerBound") Boolean useLowerBound) { DQResponse<ComplianceValue> result = new DQResponse<ComplianceValue>(); // TODO: Implementation may be too tightly bound to year, may need to extract first/last day for finer granularity test if (lowerBound == null) { lowerBound = 1700; } if (useLowerBound == null) { useLowerBound = true; } Integer upperBound = LocalDateTime.now().getYear(); if (DateUtils.isEmpty(eventDate)) { result.addComment("No value provided for dwc:eventDate."); result.setResultState(ResultState.INTERNAL_PREREQUISITES_NOT_MET); } else { if (!DateUtils.eventDateValid(eventDate)) { result.addComment( "Value provided for dwc:eventDate [" + eventDate + "] not recognized as a valid date."); result.setResultState(ResultState.INTERNAL_PREREQUISITES_NOT_MET); } else { int startYear = 0; Interval interval = DateUtils.extractInterval(eventDate); if (DateUtils.isRange(eventDate)) { int endYear = interval.getEnd().getYear(); startYear = interval.getStart().getYear(); if (useLowerBound) { if (endYear < lowerBound || startYear > upperBound) { result.setValue(ComplianceValue.NOT_COMPLIANT); result.addComment("Provided value for dwc:eventDate '" + eventDate + "' is not a range spanning part of the range " + lowerBound.toString() + " to " + upperBound.toString() + " (current year)."); } else { result.setValue(ComplianceValue.COMPLIANT); result.addComment("Provided value for dwc:eventDate '" + eventDate + "' is a range spanning at least part of " + lowerBound.toString() + " to " + upperBound.toString() + " (current year)."); } } else { if (startYear > upperBound) { result.setValue(ComplianceValue.NOT_COMPLIANT); result.addComment("Provided value for dwc:eventDate '" + eventDate + "' is not a range spanning part of the range " + lowerBound.toString() + " to " + upperBound.toString() + " (current year)."); } else { result.setValue(ComplianceValue.COMPLIANT); result.addComment("Provided value for dwc:eventDate '" + eventDate + "' is a range spanning at least part of " + lowerBound.toString() + " to " + upperBound.toString() + " (current year)."); } } } else { startYear = interval.getStart().getYear(); if (useLowerBound) { if (startYear < lowerBound || startYear > upperBound) { result.setValue(ComplianceValue.NOT_COMPLIANT); result.addComment("Provided value for dwc:eventDate '" + eventDate + "' does not have a year in the range " + lowerBound.toString() + " to " + upperBound.toString() + " (current year)."); } else { result.setValue(ComplianceValue.COMPLIANT); result.addComment("Provided value for dwc:eventDate '" + eventDate + "' does not have a year in the range " + lowerBound.toString() + " to " + upperBound.toString() + " (current year)."); } } else { if (startYear > upperBound) { result.setValue(ComplianceValue.NOT_COMPLIANT); result.addComment("Provided value for dwc:eventDate '" + eventDate + "' is not after " + upperBound.toString() + " (current year)."); } else { result.setValue(ComplianceValue.COMPLIANT); result.addComment("Provided value for dwc:eventDate '" + eventDate + "' is after " + upperBound.toString() + " (current year)."); } } } result.setResultState(ResultState.RUN_HAS_RESULT); } } return result; }