svn-gvsig-desktop / trunk / org.gvsig.desktop / org.gvsig.desktop.compat.cdc / org.gvsig.fmap.dal / org.gvsig.fmap.dal.file / org.gvsig.fmap.dal.file.csv / src / main / java / org / gvsig / fmap / dal / store / csv / AutomaticDetectionOfTypes.java @ 45775
History | View | Annotate | Download (12.1 KB)
| 1 |
package org.gvsig.fmap.dal.store.csv; |
|---|---|
| 2 |
|
| 3 |
import java.io.IOException; |
| 4 |
import java.math.BigDecimal; |
| 5 |
import java.net.URL; |
| 6 |
import java.util.ArrayList; |
| 7 |
import java.util.List; |
| 8 |
import java.util.Locale; |
| 9 |
import org.apache.commons.lang3.StringUtils; |
| 10 |
import org.gvsig.fmap.dal.DataTypes; |
| 11 |
import org.gvsig.tools.ToolsLocator; |
| 12 |
import org.gvsig.tools.dataTypes.DataTypeUtils; |
| 13 |
import org.gvsig.tools.dataTypes.DataTypesManager; |
| 14 |
import org.gvsig.tools.dataTypes.Coercion; |
| 15 |
import org.gvsig.tools.dataTypes.CoercionContext; |
| 16 |
|
| 17 |
/**
|
| 18 |
*
|
| 19 |
* @author jjdelcerro
|
| 20 |
*/
|
| 21 |
public class AutomaticDetectionOfTypes { |
| 22 |
|
| 23 |
public interface Rows { |
| 24 |
|
| 25 |
public List<String> nextRowValues(); |
| 26 |
} |
| 27 |
|
| 28 |
public interface DetectedValue { |
| 29 |
public int getType(); |
| 30 |
public int getDisplaySize(); |
| 31 |
public int getPrecision(); |
| 32 |
public int getScale(); |
| 33 |
public boolean isBlank(); |
| 34 |
} |
| 35 |
|
| 36 |
private static class DetectedValueImpl implements DetectedValue { |
| 37 |
|
| 38 |
private int type; |
| 39 |
private int displaySize; |
| 40 |
private int integerDigits; |
| 41 |
private int decimalDigits; |
| 42 |
private boolean blank; |
| 43 |
|
| 44 |
@Override
|
| 45 |
public int getType() { |
| 46 |
return this.type; |
| 47 |
} |
| 48 |
|
| 49 |
@Override
|
| 50 |
public int getDisplaySize() { |
| 51 |
return this.displaySize; |
| 52 |
} |
| 53 |
|
| 54 |
@Override
|
| 55 |
public int getPrecision() { |
| 56 |
return this.decimalDigits + this.integerDigits; |
| 57 |
} |
| 58 |
|
| 59 |
@Override
|
| 60 |
public int getScale() { |
| 61 |
return this.decimalDigits; |
| 62 |
} |
| 63 |
|
| 64 |
public boolean isBlank() { |
| 65 |
return this.blank; |
| 66 |
} |
| 67 |
} |
| 68 |
|
| 69 |
private static class PossibleDataType { |
| 70 |
|
| 71 |
public boolean possibleInt = true; |
| 72 |
public boolean possibleFloat = true; |
| 73 |
public boolean possibleDouble = true; |
| 74 |
public boolean possibleDecimal = true; |
| 75 |
public boolean possibleLong = true; |
| 76 |
public boolean possibleURL = true; |
| 77 |
public boolean possibleDate = true; |
| 78 |
public boolean possibleTime = true; |
| 79 |
public boolean possibleTimestamp = true; |
| 80 |
public boolean possibleGeometry = true; |
| 81 |
} |
| 82 |
|
| 83 |
private final String filename; |
| 84 |
|
| 85 |
public AutomaticDetectionOfTypes() {
|
| 86 |
this("(unknown)"); |
| 87 |
} |
| 88 |
|
| 89 |
public AutomaticDetectionOfTypes(String filename) { |
| 90 |
this.filename = filename;
|
| 91 |
} |
| 92 |
|
| 93 |
private String getFullFileName() { |
| 94 |
return this.filename; |
| 95 |
} |
| 96 |
|
| 97 |
@SuppressWarnings({"UseSpecificCatch", "ResultOfObjectAllocationIgnored"}) |
| 98 |
public DetectedValue[] detect(int columns, |
| 99 |
Rows rows, |
| 100 |
boolean isFirstLineHeader,
|
| 101 |
Locale locale
|
| 102 |
) throws IOException { |
| 103 |
List<PossibleDataType> possibleDataTypes;
|
| 104 |
DetectedValueImpl[] detectedValues = new DetectedValueImpl [columns]; |
| 105 |
|
| 106 |
int lineno = 0; |
| 107 |
try {
|
| 108 |
if (isFirstLineHeader) {
|
| 109 |
rows.nextRowValues(); |
| 110 |
lineno++; |
| 111 |
} |
| 112 |
possibleDataTypes = new ArrayList<>(columns); |
| 113 |
for (int i = 0; i < columns; i++) { |
| 114 |
possibleDataTypes.add(new PossibleDataType());
|
| 115 |
detectedValues[i] = new DetectedValueImpl();
|
| 116 |
} |
| 117 |
if (locale == null) { |
| 118 |
locale = Locale.getDefault();
|
| 119 |
} |
| 120 |
DataTypesManager typeManager = ToolsLocator.getDataTypesManager(); |
| 121 |
Coercion toDecimal = typeManager.getCoercion(DataTypes.DECIMAL); |
| 122 |
Coercion toDouble = typeManager.getCoercion(DataTypes.DOUBLE); |
| 123 |
Coercion toFloat = typeManager.getCoercion(DataTypes.FLOAT); |
| 124 |
Coercion toDate = typeManager.getCoercion(DataTypes.DATE); |
| 125 |
Coercion toTime = typeManager.getCoercion(DataTypes.TIME); |
| 126 |
Coercion toTimestamp = typeManager.getCoercion(DataTypes.TIMESTAMP); |
| 127 |
Coercion toInt = typeManager.getCoercion(DataTypes.INT); |
| 128 |
Coercion toLong = typeManager.getCoercion(DataTypes.LONG); |
| 129 |
Coercion toGeom = typeManager.getCoercion(DataTypes.GEOMETRY); |
| 130 |
|
| 131 |
CoercionContext coercionContext = DataTypeUtils.coerceContextLocale(locale); |
| 132 |
List<String> row = rows.nextRowValues(); |
| 133 |
lineno++; |
| 134 |
|
| 135 |
while (row != null) { |
| 136 |
for (int i = 0; i < row.size(); i++) { |
| 137 |
while( possibleDataTypes.size()<row.size() ) {
|
| 138 |
possibleDataTypes.add(new PossibleDataType());
|
| 139 |
} |
| 140 |
String rawvalue = row.get(i);
|
| 141 |
if( rawvalue == null ) { |
| 142 |
continue;
|
| 143 |
} |
| 144 |
PossibleDataType possibleDataType = possibleDataTypes.get(i); |
| 145 |
DetectedValueImpl detectedValue = detectedValues[i]; |
| 146 |
if( detectedValue.blank ) {
|
| 147 |
detectedValue.blank = StringUtils.isBlank(rawvalue); |
| 148 |
} |
| 149 |
int displaySize = rawvalue.length();
|
| 150 |
if( displaySize>detectedValue.displaySize ) {
|
| 151 |
detectedValue.displaySize = displaySize; |
| 152 |
} |
| 153 |
if (possibleDataType.possibleDecimal) {
|
| 154 |
try {
|
| 155 |
BigDecimal decimal = (BigDecimal) toDecimal.coerce(rawvalue, coercionContext); |
| 156 |
possibleDataType.possibleDecimal = true;
|
| 157 |
if( decimal.scale() > detectedValue.decimalDigits ) {
|
| 158 |
detectedValue.decimalDigits = decimal.scale(); |
| 159 |
} |
| 160 |
int integerDigits = decimal.precision() - decimal.scale();
|
| 161 |
if( integerDigits>detectedValue.integerDigits ) {
|
| 162 |
detectedValue.integerDigits = integerDigits; |
| 163 |
} |
| 164 |
} catch (Exception ex) { |
| 165 |
possibleDataType.possibleDecimal = false;
|
| 166 |
} |
| 167 |
} |
| 168 |
if (possibleDataType.possibleDouble) {
|
| 169 |
try {
|
| 170 |
toDouble.coerce(rawvalue, coercionContext); |
| 171 |
possibleDataType.possibleDouble = true;
|
| 172 |
} catch (Exception ex) { |
| 173 |
possibleDataType.possibleDouble = false;
|
| 174 |
} |
| 175 |
} |
| 176 |
if (possibleDataType.possibleFloat) {
|
| 177 |
try {
|
| 178 |
toFloat.coerce(rawvalue, coercionContext); |
| 179 |
possibleDataType.possibleFloat = true;
|
| 180 |
} catch (Exception ex) { |
| 181 |
possibleDataType.possibleFloat = false;
|
| 182 |
} |
| 183 |
} |
| 184 |
if (possibleDataType.possibleLong) {
|
| 185 |
possibleDataType.possibleLong = isValidLong(rawvalue); |
| 186 |
} |
| 187 |
if (possibleDataType.possibleInt) {
|
| 188 |
possibleDataType.possibleInt = isValidInteger(rawvalue); |
| 189 |
} |
| 190 |
if (possibleDataType.possibleDate) {
|
| 191 |
try {
|
| 192 |
toDate.coerce(rawvalue, coercionContext); |
| 193 |
possibleDataType.possibleDate = true;
|
| 194 |
} catch (Exception ex) { |
| 195 |
possibleDataType.possibleDate = false;
|
| 196 |
} |
| 197 |
} |
| 198 |
if (possibleDataType.possibleTime) {
|
| 199 |
try {
|
| 200 |
toTime.coerce(rawvalue, coercionContext); |
| 201 |
possibleDataType.possibleTime = true;
|
| 202 |
} catch (Exception ex) { |
| 203 |
possibleDataType.possibleTime = false;
|
| 204 |
} |
| 205 |
} |
| 206 |
if (possibleDataType.possibleTimestamp) {
|
| 207 |
try {
|
| 208 |
toTimestamp.coerce(rawvalue, coercionContext); |
| 209 |
possibleDataType.possibleTimestamp = true;
|
| 210 |
} catch (Exception ex) { |
| 211 |
possibleDataType.possibleTimestamp = false;
|
| 212 |
} |
| 213 |
} |
| 214 |
if (possibleDataType.possibleURL) {
|
| 215 |
try {
|
| 216 |
new URL((String) rawvalue); |
| 217 |
possibleDataType.possibleURL = true;
|
| 218 |
} catch (Exception ex) { |
| 219 |
possibleDataType.possibleURL = false;
|
| 220 |
} |
| 221 |
} |
| 222 |
|
| 223 |
if (possibleDataType.possibleGeometry) {
|
| 224 |
try {
|
| 225 |
toGeom.coerce((String) rawvalue);
|
| 226 |
possibleDataType.possibleGeometry = true;
|
| 227 |
} catch (Exception ex) { |
| 228 |
possibleDataType.possibleGeometry = false;
|
| 229 |
} |
| 230 |
} |
| 231 |
} |
| 232 |
row = rows.nextRowValues(); |
| 233 |
lineno++; |
| 234 |
} |
| 235 |
int n = 0; |
| 236 |
for (PossibleDataType possibleDataType : possibleDataTypes) {
|
| 237 |
if (possibleDataType.possibleInt) {
|
| 238 |
detectedValues[n++].type = DataTypes.INT; |
| 239 |
continue;
|
| 240 |
} |
| 241 |
if (possibleDataType.possibleLong) {
|
| 242 |
detectedValues[n++].type = DataTypes.LONG; |
| 243 |
continue;
|
| 244 |
} |
| 245 |
if (possibleDataType.possibleDecimal) {
|
| 246 |
// Preferimos un Decimal que un Float/Double
|
| 247 |
detectedValues[n++].type = DataTypes.DECIMAL; |
| 248 |
continue;
|
| 249 |
} |
| 250 |
if (possibleDataType.possibleFloat) {
|
| 251 |
// Forzamos los float a double para evitar perder precision
|
| 252 |
detectedValues[n++].type = DataTypes.DOUBLE; |
| 253 |
continue;
|
| 254 |
} |
| 255 |
if (possibleDataType.possibleDouble) {
|
| 256 |
detectedValues[n++].type = DataTypes.DOUBLE; |
| 257 |
continue;
|
| 258 |
} |
| 259 |
if (possibleDataType.possibleURL) {
|
| 260 |
detectedValues[n++].type = DataTypes.URL; |
| 261 |
continue;
|
| 262 |
} |
| 263 |
if (possibleDataType.possibleDate) {
|
| 264 |
detectedValues[n++].type = DataTypes.DATE; |
| 265 |
continue;
|
| 266 |
} |
| 267 |
if (possibleDataType.possibleTime) {
|
| 268 |
detectedValues[n++].type = DataTypes.TIME; |
| 269 |
continue;
|
| 270 |
} |
| 271 |
if (possibleDataType.possibleTimestamp) {
|
| 272 |
detectedValues[n++].type = DataTypes.TIMESTAMP; |
| 273 |
continue;
|
| 274 |
} |
| 275 |
if (possibleDataType.possibleGeometry) {
|
| 276 |
detectedValues[n++].type = DataTypes.GEOMETRY; |
| 277 |
continue;
|
| 278 |
} |
| 279 |
detectedValues[n++].type = DataTypes.STRING; |
| 280 |
} |
| 281 |
} catch (Throwable ex) { |
| 282 |
throw new RuntimeException("Problems reading file '" + this.getFullFileName() + "' near line " + lineno + ".", ex); |
| 283 |
} |
| 284 |
return detectedValues;
|
| 285 |
} |
| 286 |
|
| 287 |
@SuppressWarnings("UseSpecificCatch") |
| 288 |
private boolean isValidLong(String s) { |
| 289 |
if (s == null) { |
| 290 |
return true; |
| 291 |
} |
| 292 |
s = s.trim().toLowerCase(); |
| 293 |
if (s.isEmpty()) {
|
| 294 |
return true; |
| 295 |
} |
| 296 |
try {
|
| 297 |
if (s.startsWith("0x")) { |
| 298 |
Long.valueOf(s.substring(2), 16); |
| 299 |
} else {
|
| 300 |
Long.valueOf(s);
|
| 301 |
} |
| 302 |
return true; |
| 303 |
} catch (Exception ex) { |
| 304 |
return false; |
| 305 |
} |
| 306 |
} |
| 307 |
|
| 308 |
@SuppressWarnings("UseSpecificCatch") |
| 309 |
private boolean isValidInteger(String s) { |
| 310 |
if (s == null) { |
| 311 |
return true; |
| 312 |
} |
| 313 |
s = s.trim().toLowerCase(); |
| 314 |
if (s.isEmpty()) {
|
| 315 |
return true; |
| 316 |
} |
| 317 |
try {
|
| 318 |
if (s.startsWith("0x")) { |
| 319 |
Integer.valueOf(s.substring(2), 16); |
| 320 |
} else {
|
| 321 |
Integer.valueOf(s);
|
| 322 |
} |
| 323 |
return true; |
| 324 |
} catch (Exception ex) { |
| 325 |
return false; |
| 326 |
} |
| 327 |
} |
| 328 |
|
| 329 |
} |