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 @ 43983
History | View | Annotate | Download (7.83 KB)
1 | 43983 | jjdelcerro | 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 | public List<String> nextRowValues(); |
||
20 | } |
||
21 | |||
22 | private static class PossibleDataType { |
||
23 | |||
24 | public boolean possibleInt = true; |
||
25 | public boolean possibleFloat = true; |
||
26 | public boolean possibleDouble = true; |
||
27 | public boolean possibleLong = true; |
||
28 | public boolean possibleURL = true; |
||
29 | public boolean possibleDate = true; |
||
30 | public boolean possibleGeometry = true; |
||
31 | } |
||
32 | |||
33 | private final String filename; |
||
34 | |||
35 | public AutomaticDetectionOfTypes() {
|
||
36 | this("(unknown)"); |
||
37 | } |
||
38 | |||
39 | public AutomaticDetectionOfTypes(String filename) { |
||
40 | this.filename = filename;
|
||
41 | } |
||
42 | |||
43 | private String getFullFileName() { |
||
44 | return this.filename; |
||
45 | } |
||
46 | |||
47 | @SuppressWarnings({"UseSpecificCatch", "ResultOfObjectAllocationIgnored"}) |
||
48 | public int[] detect(int columns, Rows rows, boolean isFirstLineHeader, Locale locale) throws IOException { |
||
49 | List<PossibleDataType> possibleDataTypes;
|
||
50 | int[] types = null; |
||
51 | |||
52 | int lineno = 0; |
||
53 | try {
|
||
54 | if (isFirstLineHeader) {
|
||
55 | rows.nextRowValues(); |
||
56 | lineno++; |
||
57 | } |
||
58 | possibleDataTypes = new ArrayList<>(columns); |
||
59 | for (int i = 0; i < columns; i++) { |
||
60 | possibleDataTypes.add(new PossibleDataType());
|
||
61 | } |
||
62 | if (locale == null) { |
||
63 | locale = Locale.getDefault();
|
||
64 | } |
||
65 | DataTypesManager typeManager = ToolsLocator.getDataTypesManager(); |
||
66 | DataTypesManager.CoercionWithLocale toDouble = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.DOUBLE); |
||
67 | DataTypesManager.CoercionWithLocale toFloat = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.FLOAT); |
||
68 | DataTypesManager.CoercionWithLocale toDate = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.DATE); |
||
69 | DataTypesManager.CoercionWithLocale toInt = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.INT); |
||
70 | DataTypesManager.CoercionWithLocale toLong = (DataTypesManager.CoercionWithLocale) typeManager.getCoercion(DataTypes.LONG); |
||
71 | DataTypesManager.Coercion toGeom = typeManager.getCoercion(DataTypes.GEOMETRY); |
||
72 | |||
73 | List<String> row = rows.nextRowValues(); |
||
74 | lineno++; |
||
75 | |||
76 | while (row != null) { |
||
77 | for (int i = 0; i < row.size(); i++) { |
||
78 | while( possibleDataTypes.size()<row.size() ) {
|
||
79 | possibleDataTypes.add(new PossibleDataType());
|
||
80 | } |
||
81 | String rawvalue = row.get(i);
|
||
82 | PossibleDataType possibleDataType = possibleDataTypes.get(i); |
||
83 | if (possibleDataType.possibleDouble) {
|
||
84 | try {
|
||
85 | toDouble.coerce(rawvalue, locale); |
||
86 | possibleDataType.possibleDouble = true;
|
||
87 | } catch (Exception ex) { |
||
88 | possibleDataType.possibleDouble = false;
|
||
89 | } |
||
90 | } |
||
91 | if (possibleDataType.possibleFloat) {
|
||
92 | try {
|
||
93 | toFloat.coerce(rawvalue, locale); |
||
94 | possibleDataType.possibleFloat = true;
|
||
95 | } catch (Exception ex) { |
||
96 | possibleDataType.possibleFloat = false;
|
||
97 | } |
||
98 | } |
||
99 | if (possibleDataType.possibleLong) {
|
||
100 | possibleDataType.possibleLong = isValidLong(rawvalue); |
||
101 | } |
||
102 | if (possibleDataType.possibleInt) {
|
||
103 | possibleDataType.possibleInt = isValidInteger(rawvalue); |
||
104 | } |
||
105 | if (possibleDataType.possibleDate) {
|
||
106 | try {
|
||
107 | toDate.coerce(rawvalue, locale); |
||
108 | possibleDataType.possibleDate = true;
|
||
109 | } catch (Exception ex) { |
||
110 | possibleDataType.possibleDate = false;
|
||
111 | } |
||
112 | } |
||
113 | if (possibleDataType.possibleURL) {
|
||
114 | try {
|
||
115 | new URL((String) rawvalue); |
||
116 | possibleDataType.possibleURL = true;
|
||
117 | } catch (Exception ex) { |
||
118 | possibleDataType.possibleURL = false;
|
||
119 | } |
||
120 | } |
||
121 | if (possibleDataType.possibleGeometry) {
|
||
122 | try {
|
||
123 | toGeom.coerce((String) rawvalue);
|
||
124 | possibleDataType.possibleGeometry = true;
|
||
125 | } catch (Exception ex) { |
||
126 | possibleDataType.possibleGeometry = false;
|
||
127 | } |
||
128 | } |
||
129 | } |
||
130 | row = rows.nextRowValues(); |
||
131 | } |
||
132 | int n = 0; |
||
133 | types = new int[possibleDataTypes.size()]; |
||
134 | for (PossibleDataType possibleDataType : possibleDataTypes) {
|
||
135 | if (possibleDataType.possibleInt) {
|
||
136 | types[n++] = DataTypes.INT; |
||
137 | continue;
|
||
138 | } |
||
139 | if (possibleDataType.possibleLong) {
|
||
140 | types[n++] = DataTypes.LONG; |
||
141 | continue;
|
||
142 | } |
||
143 | if (possibleDataType.possibleFloat) {
|
||
144 | // Forzamos los float a double para evitar perder precision
|
||
145 | types[n++] = DataTypes.DOUBLE; |
||
146 | continue;
|
||
147 | } |
||
148 | if (possibleDataType.possibleDouble) {
|
||
149 | types[n++] = DataTypes.DOUBLE; |
||
150 | continue;
|
||
151 | } |
||
152 | if (possibleDataType.possibleURL) {
|
||
153 | types[n++] = DataTypes.URL; |
||
154 | continue;
|
||
155 | } |
||
156 | if (possibleDataType.possibleDate) {
|
||
157 | types[n++] = DataTypes.DATE; |
||
158 | continue;
|
||
159 | } |
||
160 | if (possibleDataType.possibleGeometry) {
|
||
161 | types[n++] = DataTypes.GEOMETRY; |
||
162 | continue;
|
||
163 | } |
||
164 | types[n++] = DataTypes.STRING; |
||
165 | } |
||
166 | } catch (Exception ex) { |
||
167 | throw new RuntimeException("Problems reading file '" + this.getFullFileName() + "' near line " + lineno + ".", ex); |
||
168 | } |
||
169 | return types;
|
||
170 | } |
||
171 | |||
172 | @SuppressWarnings("UseSpecificCatch") |
||
173 | private boolean isValidLong(String s) { |
||
174 | if (s == null) { |
||
175 | return true; |
||
176 | } |
||
177 | s = s.trim().toLowerCase(); |
||
178 | if (s.isEmpty()) {
|
||
179 | return true; |
||
180 | } |
||
181 | try {
|
||
182 | if (s.startsWith("0x")) { |
||
183 | Long.valueOf(s.substring(2), 16); |
||
184 | } else {
|
||
185 | Long.valueOf(s);
|
||
186 | } |
||
187 | return true; |
||
188 | } catch (Exception ex) { |
||
189 | return false; |
||
190 | } |
||
191 | } |
||
192 | |||
193 | @SuppressWarnings("UseSpecificCatch") |
||
194 | private boolean isValidInteger(String s) { |
||
195 | if (s == null) { |
||
196 | return true; |
||
197 | } |
||
198 | s = s.trim().toLowerCase(); |
||
199 | if (s.isEmpty()) {
|
||
200 | return true; |
||
201 | } |
||
202 | try {
|
||
203 | if (s.startsWith("0x")) { |
||
204 | Integer.valueOf(s.substring(2), 16); |
||
205 | } else {
|
||
206 | Integer.valueOf(s);
|
||
207 | } |
||
208 | return true; |
||
209 | } catch (Exception ex) { |
||
210 | return false; |
||
211 | } |
||
212 | } |
||
213 | |||
214 | } |