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 @ 44455
History | View | Annotate | Download (7.81 KB)
| 1 |
package org.gvsig.fmap.dal.store.csv; |
|---|---|
| 2 |
|
| 3 |
import java.io.IOException; |
| 4 |
import java.net.URL; |
| 5 |
import java.util.ArrayList; |
| 6 |
import java.util.List; |
| 7 |
import java.util.Locale; |
| 8 |
import org.gvsig.fmap.dal.DataTypes; |
| 9 |
import org.gvsig.tools.ToolsLocator; |
| 10 |
import org.gvsig.tools.dataTypes.DataTypesManager; |
| 11 |
|
| 12 |
/**
|
| 13 |
*
|
| 14 |
* @author jjdelcerro
|
| 15 |
*/
|
| 16 |
public class AutomaticDetectionOfTypes { |
| 17 |
|
| 18 |
public interface Rows { |
| 19 |
|
| 20 |
public List<String> nextRowValues(); |
| 21 |
} |
| 22 |
|
| 23 |
private static class PossibleDataType { |
| 24 |
|
| 25 |
public boolean possibleInt = true; |
| 26 |
public boolean possibleFloat = true; |
| 27 |
public boolean possibleDouble = true; |
| 28 |
public boolean possibleLong = true; |
| 29 |
public boolean possibleURL = true; |
| 30 |
public boolean possibleDate = true; |
| 31 |
public boolean possibleGeometry = true; |
| 32 |
} |
| 33 |
|
| 34 |
private final String filename; |
| 35 |
|
| 36 |
public AutomaticDetectionOfTypes() {
|
| 37 |
this("(unknown)"); |
| 38 |
} |
| 39 |
|
| 40 |
public AutomaticDetectionOfTypes(String filename) { |
| 41 |
this.filename = filename;
|
| 42 |
} |
| 43 |
|
| 44 |
private String getFullFileName() { |
| 45 |
return this.filename; |
| 46 |
} |
| 47 |
|
| 48 |
@SuppressWarnings({"UseSpecificCatch", "ResultOfObjectAllocationIgnored"}) |
| 49 |
public int[] detect(int columns, Rows rows, boolean isFirstLineHeader, Locale locale) throws IOException { |
| 50 |
List<PossibleDataType> possibleDataTypes;
|
| 51 |
int[] types = null; |
| 52 |
|
| 53 |
int lineno = 0; |
| 54 |
try {
|
| 55 |
if (isFirstLineHeader) {
|
| 56 |
rows.nextRowValues(); |
| 57 |
lineno++; |
| 58 |
} |
| 59 |
possibleDataTypes = new ArrayList<>(columns); |
| 60 |
for (int i = 0; i < columns; i++) { |
| 61 |
possibleDataTypes.add(new PossibleDataType());
|
| 62 |
} |
| 63 |
if (locale == null) { |
| 64 |
locale = Locale.getDefault();
|
| 65 |
} |
| 66 |
DataTypesManager typeManager = ToolsLocator.getDataTypesManager(); |
| 67 |
DataTypesManager.CoercionWithLocale toDouble = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.DOUBLE); |
| 68 |
DataTypesManager.CoercionWithLocale toFloat = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.FLOAT); |
| 69 |
DataTypesManager.CoercionWithLocale toDate = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.DATE); |
| 70 |
DataTypesManager.CoercionWithLocale toInt = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.INT); |
| 71 |
DataTypesManager.CoercionWithLocale toLong = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.LONG); |
| 72 |
DataTypesManager.Coercion toGeom = typeManager.getCoercion(DataTypes.GEOMETRY); |
| 73 |
|
| 74 |
List<String> row = rows.nextRowValues(); |
| 75 |
lineno++; |
| 76 |
|
| 77 |
while (row != null) { |
| 78 |
for (int i = 0; i < row.size(); i++) { |
| 79 |
while( possibleDataTypes.size()<row.size() ) {
|
| 80 |
possibleDataTypes.add(new PossibleDataType());
|
| 81 |
} |
| 82 |
String rawvalue = row.get(i);
|
| 83 |
PossibleDataType possibleDataType = possibleDataTypes.get(i); |
| 84 |
if (possibleDataType.possibleDouble) {
|
| 85 |
try {
|
| 86 |
toDouble.coerce(rawvalue, locale); |
| 87 |
possibleDataType.possibleDouble = true;
|
| 88 |
} catch (Exception ex) { |
| 89 |
possibleDataType.possibleDouble = false;
|
| 90 |
} |
| 91 |
} |
| 92 |
if (possibleDataType.possibleFloat) {
|
| 93 |
try {
|
| 94 |
toFloat.coerce(rawvalue, locale); |
| 95 |
possibleDataType.possibleFloat = true;
|
| 96 |
} catch (Exception ex) { |
| 97 |
possibleDataType.possibleFloat = false;
|
| 98 |
} |
| 99 |
} |
| 100 |
if (possibleDataType.possibleLong) {
|
| 101 |
possibleDataType.possibleLong = isValidLong(rawvalue); |
| 102 |
} |
| 103 |
if (possibleDataType.possibleInt) {
|
| 104 |
possibleDataType.possibleInt = isValidInteger(rawvalue); |
| 105 |
} |
| 106 |
if (possibleDataType.possibleDate) {
|
| 107 |
try {
|
| 108 |
toDate.coerce(rawvalue, locale); |
| 109 |
possibleDataType.possibleDate = true;
|
| 110 |
} catch (Exception ex) { |
| 111 |
possibleDataType.possibleDate = false;
|
| 112 |
} |
| 113 |
} |
| 114 |
if (possibleDataType.possibleURL) {
|
| 115 |
try {
|
| 116 |
new URL((String) rawvalue); |
| 117 |
possibleDataType.possibleURL = true;
|
| 118 |
} catch (Exception ex) { |
| 119 |
possibleDataType.possibleURL = false;
|
| 120 |
} |
| 121 |
} |
| 122 |
if (possibleDataType.possibleGeometry) {
|
| 123 |
try {
|
| 124 |
toGeom.coerce((String) rawvalue);
|
| 125 |
possibleDataType.possibleGeometry = true;
|
| 126 |
} catch (Exception ex) { |
| 127 |
possibleDataType.possibleGeometry = false;
|
| 128 |
} |
| 129 |
} |
| 130 |
} |
| 131 |
row = rows.nextRowValues(); |
| 132 |
} |
| 133 |
int n = 0; |
| 134 |
types = new int[possibleDataTypes.size()]; |
| 135 |
for (PossibleDataType possibleDataType : possibleDataTypes) {
|
| 136 |
if (possibleDataType.possibleInt) {
|
| 137 |
types[n++] = DataTypes.INT; |
| 138 |
continue;
|
| 139 |
} |
| 140 |
if (possibleDataType.possibleLong) {
|
| 141 |
types[n++] = DataTypes.LONG; |
| 142 |
continue;
|
| 143 |
} |
| 144 |
if (possibleDataType.possibleFloat) {
|
| 145 |
// Forzamos los float a double para evitar perder precision
|
| 146 |
types[n++] = DataTypes.DOUBLE; |
| 147 |
continue;
|
| 148 |
} |
| 149 |
if (possibleDataType.possibleDouble) {
|
| 150 |
types[n++] = DataTypes.DOUBLE; |
| 151 |
continue;
|
| 152 |
} |
| 153 |
if (possibleDataType.possibleURL) {
|
| 154 |
types[n++] = DataTypes.URL; |
| 155 |
continue;
|
| 156 |
} |
| 157 |
if (possibleDataType.possibleDate) {
|
| 158 |
types[n++] = DataTypes.DATE; |
| 159 |
continue;
|
| 160 |
} |
| 161 |
if (possibleDataType.possibleGeometry) {
|
| 162 |
types[n++] = DataTypes.GEOMETRY; |
| 163 |
continue;
|
| 164 |
} |
| 165 |
types[n++] = DataTypes.STRING; |
| 166 |
} |
| 167 |
} catch (Exception ex) { |
| 168 |
throw new RuntimeException("Problems reading file '" + this.getFullFileName() + "' near line " + lineno + ".", ex); |
| 169 |
} |
| 170 |
return types;
|
| 171 |
} |
| 172 |
|
| 173 |
@SuppressWarnings("UseSpecificCatch") |
| 174 |
private boolean isValidLong(String s) { |
| 175 |
if (s == null) { |
| 176 |
return true; |
| 177 |
} |
| 178 |
s = s.trim().toLowerCase(); |
| 179 |
if (s.isEmpty()) {
|
| 180 |
return true; |
| 181 |
} |
| 182 |
try {
|
| 183 |
if (s.startsWith("0x")) { |
| 184 |
Long.valueOf(s.substring(2), 16); |
| 185 |
} else {
|
| 186 |
Long.valueOf(s);
|
| 187 |
} |
| 188 |
return true; |
| 189 |
} catch (Exception ex) { |
| 190 |
return false; |
| 191 |
} |
| 192 |
} |
| 193 |
|
| 194 |
@SuppressWarnings("UseSpecificCatch") |
| 195 |
private boolean isValidInteger(String s) { |
| 196 |
if (s == null) { |
| 197 |
return true; |
| 198 |
} |
| 199 |
s = s.trim().toLowerCase(); |
| 200 |
if (s.isEmpty()) {
|
| 201 |
return true; |
| 202 |
} |
| 203 |
try {
|
| 204 |
if (s.startsWith("0x")) { |
| 205 |
Integer.valueOf(s.substring(2), 16); |
| 206 |
} else {
|
| 207 |
Integer.valueOf(s);
|
| 208 |
} |
| 209 |
return true; |
| 210 |
} catch (Exception ex) { |
| 211 |
return false; |
| 212 |
} |
| 213 |
} |
| 214 |
|
| 215 |
} |