Mercurial > dive4elements > river
comparison flys-backend/src/main/java/de/intevation/flys/importer/AtFileParser.java @ 485:6b231041dc18
Importer: Try to extract time ranges from at files.
flys-backend/trunk@1811 c6561f87-3c4e-4783-a992-168aeb5c3f6f
| author | Sascha L. Teichmann <sascha.teichmann@intevation.de> |
|---|---|
| date | Tue, 03 May 2011 17:09:38 +0000 |
| parents | d980e545ccab |
| children | 8ea09ec7f0c8 |
comparison
equal
deleted
inserted
replaced
| 484:a9e9a8a44d19 | 485:6b231041dc18 |
|---|---|
| 12 import org.apache.log4j.Logger; | 12 import org.apache.log4j.Logger; |
| 13 | 13 |
| 14 import de.intevation.flys.importer.ImportDischargeTable; | 14 import de.intevation.flys.importer.ImportDischargeTable; |
| 15 import de.intevation.flys.importer.ImportDischargeTableValue; | 15 import de.intevation.flys.importer.ImportDischargeTableValue; |
| 16 | 16 |
| 17 import java.util.regex.Pattern; | |
| 18 import java.util.regex.Matcher; | |
| 19 | |
| 20 import java.util.Date; | |
| 21 import java.util.Calendar; | |
| 17 | 22 |
| 18 public class AtFileParser { | 23 public class AtFileParser { |
| 19 | 24 |
| 20 public static final String ENCODING = "ISO-8859-1"; | 25 public static final String ENCODING = "ISO-8859-1"; |
| 21 | 26 |
| 22 private static Logger logger = Logger.getLogger(AtFileParser.class); | 27 private static Logger logger = Logger.getLogger(AtFileParser.class); |
| 23 | 28 |
| 24 private static NumberFormat nf = NumberFormat.getInstance(); | 29 private static NumberFormat nf = NumberFormat.getInstance(); |
| 25 | 30 |
| 31 | |
| 32 // regular expression from hell to find out time range | |
| 33 public static final Pattern DATE_LINE = Pattern.compile( | |
| 34 "^\\*\\s*Abflu[^t]+tafel?\\s*([^\\d]+)" + | |
| 35 "(\\d{1,2})?\\.?(\\d{1,2})?\\.?(\\d{2,4})\\s*(?:(?:bis)|-)?\\s*" + | |
| 36 "(?:(\\d{1,2})?\\.?(\\d{1,2})?\\.?(\\d{2,4}))?\\s*.*$"); | |
| 26 | 37 |
| 27 public AtFileParser() { | 38 public AtFileParser() { |
| 28 } | 39 } |
| 29 | 40 |
| 30 | 41 |
| 40 | 51 |
| 41 boolean beginning = true; | 52 boolean beginning = true; |
| 42 | 53 |
| 43 ImportDischargeTable dischargeTable = new ImportDischargeTable(); | 54 ImportDischargeTable dischargeTable = new ImportDischargeTable(); |
| 44 | 55 |
| 56 Date from = null; | |
| 57 Date to = null; | |
| 58 | |
| 45 try { | 59 try { |
| 46 br = new BufferedReader( | 60 br = new BufferedReader( |
| 47 new InputStreamReader( | 61 new InputStreamReader( |
| 48 new FileInputStream(file), ENCODING)); | 62 new FileInputStream(file), ENCODING)); |
| 49 | 63 |
| 50 while ((line = br.readLine()) != null) { | 64 while ((line = br.readLine()) != null) { |
| 65 | |
| 51 String tmp = line.trim(); | 66 String tmp = line.trim(); |
| 52 | 67 |
| 53 if (tmp.length() == 0) { | 68 if (tmp.length() == 0) { |
| 69 continue; | |
| 70 } | |
| 71 | |
| 72 Matcher m = DATE_LINE.matcher(tmp); | |
| 73 if (m.matches()) { | |
| 74 from = guessDate(m.group(1), m.group(2), m.group(3)); | |
| 75 to = guessDate(m.group(4), m.group(5), m.group(6)); | |
| 76 if (from == null) { | |
| 77 Date t = from; from = to; to = t; | |
| 78 } | |
| 54 continue; | 79 continue; |
| 55 } | 80 } |
| 56 | 81 |
| 57 if (tmp.startsWith("#! name=")) { | 82 if (tmp.startsWith("#! name=")) { |
| 58 // XXX Skip the name, because we don't know where to save | 83 // XXX Skip the name, because we don't know where to save |
| 111 | 136 |
| 112 logger.info("Finished parsing AT file: " + file); | 137 logger.info("Finished parsing AT file: " + file); |
| 113 | 138 |
| 114 return dischargeTable; | 139 return dischargeTable; |
| 115 } | 140 } |
| 141 | |
| 142 public static Date guessDate(String day, String month, String year) { | |
| 143 if (day == null && month == null && year == null) { | |
| 144 return null; | |
| 145 } | |
| 146 | |
| 147 int dayI = 15; | |
| 148 if (day != null) { | |
| 149 try { | |
| 150 dayI = Integer.parseInt(day.trim()); | |
| 151 } | |
| 152 catch (NumberFormatException nfe) { | |
| 153 } | |
| 154 } | |
| 155 | |
| 156 int monthI = 6; | |
| 157 if (month != null) { | |
| 158 try { | |
| 159 monthI = Integer.parseInt(month.trim()); | |
| 160 } | |
| 161 catch (NumberFormatException nfe) { | |
| 162 } | |
| 163 } | |
| 164 | |
| 165 int yearI = 1900; | |
| 166 if (year != null) { | |
| 167 try { | |
| 168 yearI = Integer.parseInt(year.trim()); | |
| 169 if (yearI < 100) { | |
| 170 if (yearI < 20) { | |
| 171 yearI += 2000; | |
| 172 } | |
| 173 else { | |
| 174 yearI += 1900; | |
| 175 } | |
| 176 } | |
| 177 } | |
| 178 catch (NumberFormatException nfe) { | |
| 179 } | |
| 180 } | |
| 181 | |
| 182 Calendar cal = Calendar.getInstance(); | |
| 183 cal.set(yearI, monthI-1, dayI); | |
| 184 return cal.getTime(); | |
| 185 } | |
| 116 } | 186 } |
| 117 // vim:set ts=4 sw=4 si et sta sts=4 fenc=utf8 : | 187 // vim:set ts=4 sw=4 si et sta sts=4 fenc=utf8 : |