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 / simplereaders / CSVReaderSuperCSV.java @ 46161
History | View | Annotate | Download (8.73 KB)
1 |
package org.gvsig.fmap.dal.store.csv.simplereaders; |
---|---|
2 |
|
3 |
import java.io.File; |
4 |
import java.io.IOException; |
5 |
import java.io.Reader; |
6 |
import java.util.List; |
7 |
import org.apache.commons.io.FilenameUtils; |
8 |
import org.apache.commons.text.StringEscapeUtils; |
9 |
import org.apache.commons.lang3.StringUtils; |
10 |
import org.gvsig.fmap.dal.store.csv.CSVStoreParameters; |
11 |
import org.gvsig.fmap.dal.store.csv.virtualrows.RandomAccessFileIndex; |
12 |
import org.gvsig.fmap.dal.store.csv.virtualrows.RandomAccessFileReader; |
13 |
import static org.gvsig.fmap.dal.store.csv.virtualrows.RandomAccessFileReader.FILTER_NONE; |
14 |
import org.gvsig.fmap.dal.store.csv.virtualrows.SuperCSVList; |
15 |
import org.gvsig.tools.dynobject.DynObject; |
16 |
import org.gvsig.tools.task.SimpleTaskStatus; |
17 |
import org.gvsig.tools.util.GetItemWithSize64; |
18 |
import org.slf4j.Logger; |
19 |
import org.slf4j.LoggerFactory; |
20 |
import org.supercsv.comment.CommentStartsWith; |
21 |
import org.supercsv.io.CsvListReader; |
22 |
import org.supercsv.prefs.CsvPreference; |
23 |
import org.supercsv.quote.QuoteMode; |
24 |
|
25 |
public class CSVReaderSuperCSV extends AbstractSimpleReader { |
26 |
|
27 |
//
|
28 |
// http://supercsv.sourceforge.net/examples_reading.html
|
29 |
// http://supercsv.sourceforge.net/apidocs/index.html
|
30 |
//
|
31 |
private static final Logger LOGGER = LoggerFactory.getLogger(CSVReaderSuperCSV.class); |
32 |
|
33 |
private CsvListReader reader;
|
34 |
private final CSVStoreParameters parameters; |
35 |
private List<String> nextLine; |
36 |
private int columns; |
37 |
|
38 |
public CSVReaderSuperCSV(Reader in, CSVStoreParameters parameters) { |
39 |
this(parameters);
|
40 |
this.reader = new CsvListReader(in, getCSVPreferences()); |
41 |
} |
42 |
|
43 |
public CSVReaderSuperCSV(CSVStoreParameters parameters) {
|
44 |
this.reader = null; |
45 |
this.parameters = parameters;
|
46 |
this.reader = null; |
47 |
this.nextLine = null; |
48 |
this.columns = 0; |
49 |
} |
50 |
|
51 |
public CSVStoreParameters getParameters() {
|
52 |
return this.parameters; |
53 |
} |
54 |
|
55 |
@Override
|
56 |
public String[] getHeader() throws IOException { |
57 |
return this.reader.getHeader(true); |
58 |
} |
59 |
|
60 |
@Override
|
61 |
public int getColumnsCount() throws IOException { |
62 |
if( this.columns <= 0 ) { |
63 |
this.columns = reader.length();
|
64 |
if( this.columns <= 0 ) { |
65 |
this.nextLine = this.reader.read(); |
66 |
this.columns = reader.length();
|
67 |
} |
68 |
} |
69 |
return this.columns; |
70 |
} |
71 |
|
72 |
@Override
|
73 |
public GetItemWithSize64<List<String>> getVirtualRows(SimpleTaskStatus status) { |
74 |
RandomAccessFileReader theReader = null;
|
75 |
RandomAccessFileIndex theIndex = null;
|
76 |
try {
|
77 |
CSVStoreParameters params = getParameters(); |
78 |
File data_file = CSVStoreParameters.getFile(params);
|
79 |
if( data_file.length()< 10*1024*1024 ) { |
80 |
return null; |
81 |
} |
82 |
|
83 |
String charset = CSVStoreParameters.getCharset(params);
|
84 |
File index_file = new File(FilenameUtils.removeExtension(data_file.getAbsolutePath()) + ".idx"); |
85 |
|
86 |
theReader = new RandomAccessFileReader(data_file, charset);
|
87 |
theIndex = theReader.createOrOpenIndexOfLines(index_file, false, FILTER_NONE, status);
|
88 |
|
89 |
SuperCSVList list = new SuperCSVList(
|
90 |
theReader, |
91 |
theIndex, |
92 |
CSVStoreParameters.isFirstLineHeader(getParameters())?1:0 |
93 |
); |
94 |
|
95 |
list.setPreferences(this.getCSVPreferences());
|
96 |
return list;
|
97 |
} catch (IOException ex) { |
98 |
return null; |
99 |
} finally {
|
100 |
// We do not close the index or the reader because we need it to remain open
|
101 |
// IOUtils.closeQuietly(theReader);
|
102 |
// IOUtils.closeQuietly(theIndex);
|
103 |
} |
104 |
} |
105 |
|
106 |
@Override
|
107 |
public List<String> read() throws IOException { |
108 |
List<String> line; |
109 |
if( this.nextLine != null ) { |
110 |
line = this.nextLine;
|
111 |
this.nextLine = null; |
112 |
} else {
|
113 |
line = this.reader.read();
|
114 |
} |
115 |
if( line!=null ) { |
116 |
for (int i = 0; i < line.size(); i++) { |
117 |
String s = line.get(i);
|
118 |
if( s!=null ) { |
119 |
line.set(i, unescapeCRLF(s)); |
120 |
} |
121 |
} |
122 |
} |
123 |
return line;
|
124 |
} |
125 |
|
126 |
@Override
|
127 |
public void close() throws IOException { |
128 |
this.reader.close();
|
129 |
} |
130 |
|
131 |
@Override
|
132 |
public List<String> skip(int lines) throws IOException { |
133 |
if( lines <= 0 ) { |
134 |
return null; |
135 |
} |
136 |
if( this.nextLine != null ) { |
137 |
this.nextLine = null; |
138 |
lines--; |
139 |
} |
140 |
List<String> row = null; |
141 |
for ( int i = 0; i < lines; i++ ) { |
142 |
row = reader.read(); |
143 |
} |
144 |
return row;
|
145 |
} |
146 |
|
147 |
public final CsvPreference getCSVPreferences() { |
148 |
try {
|
149 |
String s;
|
150 |
char quoteChar;
|
151 |
int delimiterChar;
|
152 |
String endOfLineSymbols;
|
153 |
|
154 |
DynObject params = this.getParameters();
|
155 |
|
156 |
CsvPreference.Builder builder; |
157 |
|
158 |
CsvPreference defaultPreference = CSVStoreParameters |
159 |
.getPredefinedCSVPreferences(params); |
160 |
if ( defaultPreference == null ) { |
161 |
defaultPreference = CsvPreference.STANDARD_PREFERENCE; |
162 |
} |
163 |
|
164 |
endOfLineSymbols = CSVStoreParameters.getRecordSeparator(params); |
165 |
if ( StringUtils.isBlank(endOfLineSymbols) ) {
|
166 |
endOfLineSymbols = defaultPreference.getEndOfLineSymbols(); |
167 |
} |
168 |
s = CSVStoreParameters.getQuoteCharacter(params); |
169 |
if ( StringUtils.isBlank(s) ) {
|
170 |
quoteChar = (char) defaultPreference.getQuoteChar();
|
171 |
} else {
|
172 |
quoteChar = s.charAt(0);
|
173 |
} |
174 |
s = CSVStoreParameters.getDelimiter(params); |
175 |
if ( StringUtils.isBlank(s) ) {
|
176 |
delimiterChar = defaultPreference.getDelimiterChar(); |
177 |
} else {
|
178 |
delimiterChar = s.charAt(0);
|
179 |
} |
180 |
|
181 |
builder = new CsvPreference.Builder(quoteChar, delimiterChar,
|
182 |
endOfLineSymbols); |
183 |
|
184 |
s = CSVStoreParameters.getCommentStartMarker(params); |
185 |
if ( !StringUtils.isBlank(s) ) {
|
186 |
CommentStartsWith cs = new CommentStartsWith(s);
|
187 |
builder.skipComments(cs); |
188 |
} |
189 |
|
190 |
builder.surroundingSpacesNeedQuotes(CSVStoreParameters |
191 |
.getSurroundingSpacesNeedQuotes(params)); |
192 |
QuoteMode quoteMode = CSVStoreParameters.getQuoteMode(params); |
193 |
if ( quoteMode != null ) { |
194 |
builder.useQuoteMode(quoteMode); |
195 |
} |
196 |
return builder.build();
|
197 |
} catch (Exception e) { |
198 |
LOGGER.warn("Can't make preferences for CSV '" + getFullFileName()
|
199 |
+ "'.", e);
|
200 |
return null; |
201 |
} |
202 |
} |
203 |
|
204 |
private String getFullFileName() { |
205 |
// Usar solo para mostrar mensajes en el logger.
|
206 |
String s;
|
207 |
try {
|
208 |
s = getParameters().getFile().getAbsolutePath(); |
209 |
} catch (Exception e2) { |
210 |
s = "(unknow)";
|
211 |
} |
212 |
return s;
|
213 |
} |
214 |
|
215 |
@Override
|
216 |
public int getLine() { |
217 |
if( this.reader==null ) { |
218 |
return 0; |
219 |
} |
220 |
return this.reader.getLineNumber(); |
221 |
} |
222 |
|
223 |
@Override
|
224 |
public List<String> nextRowValues() { |
225 |
try {
|
226 |
return this.read(); |
227 |
} catch (IOException ex) { |
228 |
throw new RuntimeException(ex); |
229 |
} |
230 |
} |
231 |
|
232 |
public static String escapeCRLF(String s) { |
233 |
if( s==null ) { |
234 |
return s;
|
235 |
} |
236 |
String s1 = s;
|
237 |
s1 = StringUtils.replace(s1, "\\", "\\\\"); |
238 |
s1 = StringUtils.replace(s1, "\n", "\\n"); |
239 |
s1 = StringUtils.replace(s1, "\r", "\\r"); |
240 |
return s1;
|
241 |
} |
242 |
|
243 |
public static String unescapeCRLF(String s) { |
244 |
if( s==null || s.indexOf('\\')==-1 ) { |
245 |
return s;
|
246 |
} |
247 |
String s1 = s;
|
248 |
s1 = s.replaceAll("(?:^\\\\n)|(?:([^\\\\])\\\\n)","$1\n"); |
249 |
s1 = s1.replaceAll("(?:^\\\\r)|(?:([^\\\\])\\\\n)","$1\r"); |
250 |
s1 = StringUtils.replace(s1, "\\\\", "\\"); |
251 |
return s1;
|
252 |
} |
253 |
|
254 |
|
255 |
public static void main(String[] args) { |
256 |
String s0 = "\\n{\\n \"ANGULO\":\"0.000\",\\n \"\tEXTO\":\"RAVAL ROIG\\\\r\\\\n - \\\\r\\\\nVIRGEN DEL SOCORRO\",\\n \"LINK_POLIGONO\":\"HVCSGISCODE_ENT_11230100000000001\",\\n \"GEOMETRY\":\"00000000014125fe9b57b4a23441503411cb1c432d\"\\n}"; |
257 |
System.out.println("#"+s0+"#"); |
258 |
String s1 = s0.replaceAll("(?:^\\\\n)|(?:([^\\\\])\\\\n)","$1\n"); |
259 |
System.out.println("#"+s1+"#"); |
260 |
String s2 = s1.replaceAll("([^\\\\])\\\\r","$1{r}"); |
261 |
System.out.println("#"+s2+"#"); |
262 |
String s3 = StringEscapeUtils.unescapeCsv(s0);
|
263 |
System.out.println("#"+s3+"#"); |
264 |
} |
265 |
} |