annotate backend/src/main/java/org/dive4elements/river/backend/utils/DateGuesser.java @ 8969:fe81eb39080c

Avoid guessing year 1950 for HQ50 etc.
author mschaefer
date Tue, 03 Apr 2018 09:54:26 +0200
parents 5e38e2924c07
children df5c5614e9a7
rev   line source
5844
4dd33b86dc61 Added header to river backend.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
1 /* Copyright (C) 2011, 2012, 2013 by Bundesanstalt für Gewässerkunde
4dd33b86dc61 Added header to river backend.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
2 * Software engineering by Intevation GmbH
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
3 *
5992
4c3ccf2b0304 Removed trailing whitespace.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5895
diff changeset
4 * This file is Free Software under the GNU AGPL (>=v3)
5844
4dd33b86dc61 Added header to river backend.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
5 * and comes with ABSOLUTELY NO WARRANTY! Check out the
5992
4c3ccf2b0304 Removed trailing whitespace.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5895
diff changeset
6 * documentation coming with Dive4Elements River for details.
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
7 */
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
8
8187
3bb1c62ad732 Moved package org.dive4elements.river.utils to org.dive4elements.river.backend.utils.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5992
diff changeset
9 package org.dive4elements.river.backend.utils;
5844
4dd33b86dc61 Added header to river backend.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
10
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
11 import java.util.Calendar;
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
12 import java.util.Date;
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
13 import java.util.regex.Matcher;
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
14 import java.util.regex.Pattern;
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
15
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
16 public final class DateGuesser {
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
17 public static final String [] MONTH = {
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
18 "jan", "feb", "mrz", "apr", "mai", "jun",
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
19 "jul", "aug", "sep", "okt", "nov", "dez"
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
20 };
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
21
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
22 public static final int guessMonth(final String s) {
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
23 final String t = s.toLowerCase();
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
24 for (int i = 0; i < MONTH.length; ++i)
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
25 if (MONTH[i].equals(t)) {
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
26 return i;
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
27 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
28 return -1;
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
29 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
30
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
31 public static final Pattern YYYY_MM_DD =
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
32 Pattern.compile("^(\\d{4})-(\\d{2})-(\\d{2})$");
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
33
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
34 public static final Pattern DD_MM_YYYY =
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
35 Pattern.compile("^(\\d{1,2})\\.(\\d{1,2})\\.(\\d{2,4})$");
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
36
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
37 public static final Pattern MMM_YYYY =
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
38 Pattern.compile("^((\\d{1,2})\\.)?(\\w{3})[\\. -]?((\\d\\d)|(\\d{4}))$");
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
39
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
40 public static final Pattern GARBAGE_YYYY =
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
41 Pattern.compile("^[^\\d=]*((\\d\\d)|([12]\\d{3}))$");
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
42
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
43 public static final Pattern YYYY_MM_DDThh_mm =
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
44 Pattern.compile("^(\\d{4})-(\\d{2})-(\\d{2})T(\\d{2}):(\\d{2})$");
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
45
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
46 public static final Pattern YYYY_MM_DDThh_mm_ss =
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
47 Pattern.compile(
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
48 "^(\\d{4})-(\\d{2})-(\\d{2})T(\\d{2}):(\\d{2}):(\\d{2})$");
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
49
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
50 public static final Pattern DD_MM_YYYYThh_mm =
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
51 Pattern.compile(
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
52 "^(\\d{1,2})\\.(\\d{1,2})\\.(\\d{2,4})T(\\d{1,2}):(\\d{2})$");
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
53
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
54 public static final Pattern DD_MM_YYYYThh_mm_ss =
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
55 Pattern.compile("^(\\d{1,2})\\.(\\d{1,2})\\.(\\d{2,4})"
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
56 + "T(\\d{1,2}):(\\d{2}):(\\d{2})$");
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
57
4763
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
58 public static final Pattern DDMMYY =
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
59 Pattern.compile("^(\\d{2})(\\d{2})(\\d{2})$");
4763
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
60
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
61 private DateGuesser() {
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
62 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
63
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
64 public static final int calendarMonth(final String month) {
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
65 return calendarMonth(Integer.parseInt(month));
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
66 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
67
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
68 public static final int calendarMonth(final int month) {
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
69 return Math.max(Math.min(month-1, 11), 0);
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
70 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
71
4764
1cfda99c024c DateGuesser: doc.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 4763
diff changeset
72 /**
1cfda99c024c DateGuesser: doc.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 4763
diff changeset
73 * Guess date by trying all different patterns.
1cfda99c024c DateGuesser: doc.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 4763
diff changeset
74 * Throws IllegalArgumentException if not able to guess.
1cfda99c024c DateGuesser: doc.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 4763
diff changeset
75 * @param s The date to be guessed (e.g. 11.02.2001).
1cfda99c024c DateGuesser: doc.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 4763
diff changeset
76 * @return the parsed Date.
1cfda99c024c DateGuesser: doc.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 4763
diff changeset
77 */
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
78 public static Date guessDate(final String s) {
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
79 if ((s == null) || s.trim().isEmpty()) {
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
80 throw new IllegalArgumentException();
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
81 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
82
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
83 final String t = s.trim();
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
84 Matcher m;
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
85
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
86 m = YYYY_MM_DD.matcher(t);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
87
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
88 if (m.matches()) {
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
89 final Calendar cal = Calendar.getInstance();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
90 final String year = m.group(1);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
91 final String month = m.group(2);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
92 final String day = m.group(3);
5895
4110565baff8 backend: Fixed DateGuesser for DDMMYY.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5882
diff changeset
93 cal.clear();
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
94 cal.set(
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
95 Integer.parseInt(year),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
96 calendarMonth(month),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
97 Integer.parseInt(day),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
98 12, 0, 0);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
99 return cal.getTime();
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
100 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
101
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
102 m = DD_MM_YYYY.matcher(t);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
103
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
104 if (m.matches()) {
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
105 final Calendar cal = Calendar.getInstance();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
106 final String year = m.group(3);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
107 final String month = m.group(2);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
108 final String day = m.group(1);
5895
4110565baff8 backend: Fixed DateGuesser for DDMMYY.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5882
diff changeset
109 cal.clear();
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
110 cal.set(
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
111 Integer.parseInt(year) + (year.length() == 2 ? 1900 : 0),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
112 calendarMonth(month),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
113 Integer.parseInt(day),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
114 12, 0, 0);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
115 return cal.getTime();
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
116 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
117
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
118 m = MMM_YYYY.matcher(t);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
119
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
120 if (m.matches()) {
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
121 final int month = guessMonth(m.group(3));
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
122 if (month >= 0) {
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
123 final Calendar cal = Calendar.getInstance();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
124 final String year = m.group(4);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
125 final String day = m.group(1);
5895
4110565baff8 backend: Fixed DateGuesser for DDMMYY.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5882
diff changeset
126 cal.clear();
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
127 cal.set(
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
128 Integer.parseInt(year) + (year.length() == 2 ? 1900 : 0),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
129 month,
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
130 ((day == null) || day.isEmpty()) ? 15 : Integer.parseInt(day),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
131 12, 0, 0);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
132 return cal.getTime();
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
133 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
134 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
135
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
136 m = YYYY_MM_DDThh_mm.matcher(t);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
137
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
138 if (m.matches()) {
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
139 final Calendar cal = Calendar.getInstance();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
140 final String year = m.group(1);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
141 final String month = m.group(2);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
142 final String day = m.group(3);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
143 final String hour = m.group(4);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
144 final String minute = m.group(5);
5895
4110565baff8 backend: Fixed DateGuesser for DDMMYY.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5882
diff changeset
145 cal.clear();
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
146 cal.set(
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
147 Integer.parseInt(year),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
148 calendarMonth(month),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
149 Integer.parseInt(day),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
150 Integer.parseInt(hour),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
151 Integer.parseInt(minute),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
152 0
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
153 );
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
154 return cal.getTime();
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
155 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
156
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
157 m = YYYY_MM_DDThh_mm_ss.matcher(t);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
158
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
159 if (m.matches()) {
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
160 final Calendar cal = Calendar.getInstance();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
161 final String year = m.group(1);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
162 final String month = m.group(2);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
163 final String day = m.group(3);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
164 final String hour = m.group(4);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
165 final String minute = m.group(5);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
166 final String second = m.group(6);
5895
4110565baff8 backend: Fixed DateGuesser for DDMMYY.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5882
diff changeset
167 cal.clear();
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
168 cal.set(
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
169 Integer.parseInt(year),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
170 calendarMonth(month),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
171 Integer.parseInt(day),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
172 Integer.parseInt(hour),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
173 Integer.parseInt(minute),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
174 Integer.parseInt(second)
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
175 );
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
176 return cal.getTime();
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
177 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
178
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
179 m = DD_MM_YYYYThh_mm.matcher(t);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
180
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
181 if (m.matches()) {
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
182 final Calendar cal = Calendar.getInstance();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
183 final String year = m.group(3);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
184 final String month = m.group(2);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
185 final String day = m.group(1);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
186 final String hour = m.group(4);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
187 final String minute = m.group(5);
5895
4110565baff8 backend: Fixed DateGuesser for DDMMYY.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5882
diff changeset
188 cal.clear();
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
189 cal.set(
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
190 Integer.parseInt(year) + (year.length() == 2 ? 1900 : 0),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
191 calendarMonth(month),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
192 Integer.parseInt(day),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
193 Integer.parseInt(hour),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
194 Integer.parseInt(minute),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
195 0
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
196 );
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
197 return cal.getTime();
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
198 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
199
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
200 m = DD_MM_YYYYThh_mm_ss.matcher(t);
4763
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
201
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
202 if (m.matches()) {
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
203 final Calendar cal = Calendar.getInstance();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
204 final String year = m.group(3);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
205 final String month = m.group(2);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
206 final String day = m.group(1);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
207 final String hour = m.group(4);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
208 final String minute = m.group(5);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
209 final String second = m.group(6);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
210 cal.clear();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
211 cal.set(
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
212 Integer.parseInt(year) + (year.length() == 2 ? 1900 : 0),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
213 calendarMonth(month),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
214 Integer.parseInt(day),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
215 Integer.parseInt(hour),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
216 Integer.parseInt(minute),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
217 Integer.parseInt(second)
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
218 );
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
219 return cal.getTime();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
220 }
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
221
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
222 m = DDMMYY.matcher(t);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
223
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
224 if (m.matches()) {
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
225 final Calendar cal = Calendar.getInstance();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
226 final String day = m.group(1);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
227 final String month = m.group(2);
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
228 final String yearS = m.group(3);
4770
28fa48986ae9 DateGuesser: Take current year modulo 100 if compare it with a two digit year value. Removed auto boxing.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 4764
diff changeset
229 int year = Integer.parseInt(yearS);
28fa48986ae9 DateGuesser: Take current year modulo 100 if compare it with a two digit year value. Removed auto boxing.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 4764
diff changeset
230
28fa48986ae9 DateGuesser: Take current year modulo 100 if compare it with a two digit year value. Removed auto boxing.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 4764
diff changeset
231 if (year <= cal.get(Calendar.YEAR) % 100) {
4763
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
232 year += 2000;
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
233 }
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
234 else {
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
235 year += 1900;
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
236 }
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
237 // TODO Warning: two digit year
5895
4110565baff8 backend: Fixed DateGuesser for DDMMYY.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5882
diff changeset
238 cal.clear();
4763
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
239 cal.set(
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
240 year,
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
241 Integer.parseInt(month), // month
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
242 Integer.parseInt(day), // day
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
243 12, 0, 0);
4763
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
244 return cal.getTime();
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
245 }
8195396b23ce DateGuesser: Added and handle DDMMYY pattern.
Felix Wolfsteller <felix.wolfsteller@intevation.de>
parents: 2357
diff changeset
246
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
247 m = GARBAGE_YYYY.matcher(t);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
248
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
249 if (m.matches() && !t.contains("HQ") && !t.contains("HW")) {
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
250 final Calendar cal = Calendar.getInstance();
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
251 final String year = m.group(1);
5895
4110565baff8 backend: Fixed DateGuesser for DDMMYY.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5882
diff changeset
252 cal.clear();
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
253 cal.set(
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
254 Integer.parseInt(year) + (year.length() == 2 ? 1900 : 0),
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
255 5, // month
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
256 15, // day
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
257 12, 0, 0);
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
258 return cal.getTime();
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
259 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
260
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
261 throw new IllegalArgumentException();
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
262 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
263
8969
fe81eb39080c Avoid guessing year 1950 for HQ50 etc.
mschaefer
parents: 8856
diff changeset
264 public static void main(final String [] args) {
2357
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
265 for (int i = 0; i < args.length; ++i) {
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
266 System.out.println(args[i] + ": " + guessDate(args[i]));
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
267 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
268 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
269 }
2e18e12fac11 Added date guesser for WST columns.
Sascha L. Teichmann <sascha.teichmann@intevation.de>
parents:
diff changeset
270 // end of file

http://dive4elements.wald.intevation.org