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 / XMLFileAsList.java @ 47617
History | View | Annotate | Download (14.2 KB)
1 |
/*
|
---|---|
2 |
* To change this license header, choose License Headers in Project Properties.
|
3 |
* To change this template file, choose Tools | Templates
|
4 |
* and open the template in the editor.
|
5 |
*/
|
6 |
package org.gvsig.fmap.dal.store.csv; |
7 |
|
8 |
import java.io.BufferedReader; |
9 |
import java.io.Closeable; |
10 |
import java.io.File; |
11 |
import java.io.IOException; |
12 |
import java.nio.ByteBuffer; |
13 |
import java.nio.CharBuffer; |
14 |
import java.nio.charset.Charset; |
15 |
import java.nio.charset.StandardCharsets; |
16 |
import java.util.AbstractList; |
17 |
import java.util.ArrayList; |
18 |
import java.util.HashMap; |
19 |
import java.util.List; |
20 |
import java.util.Map; |
21 |
import java.util.Objects; |
22 |
import javax.xml.parsers.SAXParser; |
23 |
import javax.xml.parsers.SAXParserFactory; |
24 |
import org.apache.commons.io.FilenameUtils; |
25 |
import org.apache.commons.io.IOUtils; |
26 |
import org.apache.commons.lang3.StringUtils; |
27 |
import org.apache.commons.lang3.tuple.Pair; |
28 |
import org.gvsig.fmap.dal.store.csv.RecordsFile.Record; |
29 |
import org.gvsig.fmap.dal.store.csv.RecordsFile.RecordType; |
30 |
import static org.gvsig.fmap.dal.store.csv.RecordsFile.RecordTypeBuilder.recordTypeBuilder; |
31 |
import org.gvsig.fmap.dal.store.csv.virtualrows.RandomAccessFileReader; |
32 |
import org.gvsig.tools.util.GetItemWithSize64; |
33 |
import org.gvsig.tools.util.ListBuilder; |
34 |
import org.xml.sax.Attributes; |
35 |
import org.xml.sax.InputSource; |
36 |
import org.xml.sax.Locator; |
37 |
import org.xml.sax.SAXException; |
38 |
import org.xml.sax.helpers.DefaultHandler; |
39 |
|
40 |
/**
|
41 |
*
|
42 |
* @author jjdelcerro
|
43 |
*/
|
44 |
@SuppressWarnings("UseSpecificCatch") |
45 |
public class XMLFileAsList |
46 |
extends AbstractList<List<String>> |
47 |
implements Closeable, GetItemWithSize64<List<String>> { |
48 |
|
49 |
private static final int IDXFIELD_LINE = 0; |
50 |
private static final int IDXFIELD_COLUMN = 1; |
51 |
private static final int IDXFIELD_LINEPOS = 2; |
52 |
private static final int IDXFIELD_RECORDPOS = 3; |
53 |
|
54 |
private final RandomAccessFileReader reader; |
55 |
private final String recordPath; |
56 |
private RecordsFile index;
|
57 |
private List<String> fieldPaths; |
58 |
|
59 |
|
60 |
// ruta recordnum ruta recordnum
|
61 |
// public XMLFileAsList(File text, Charset charset, List<Pair<String,Integer>> recordPath, List<Pair<String,Integer>> fieldPaths) throws IOException {
|
62 |
public XMLFileAsList(File text, Charset charset, String recordPath, List<String> fieldPaths) throws IOException { |
63 |
this.reader = new RandomAccessFileReader(text, charset); |
64 |
this.index = null; |
65 |
this.recordPath = recordPath;
|
66 |
} |
67 |
|
68 |
public XMLFileAsList(File text, File index, Charset charset, String recordPath, List<String> fieldPaths) throws IOException { |
69 |
this.reader = new RandomAccessFileReader(text, charset); |
70 |
if (index.exists()) {
|
71 |
// TODO: Force to create index if text newer than index
|
72 |
this.index = new RecordsFileImpl(index); |
73 |
} else {
|
74 |
this.createIndex(index);
|
75 |
} |
76 |
this.recordPath = recordPath;
|
77 |
} |
78 |
|
79 |
public XMLFileAsList(RandomAccessFileReader reader, RecordsFile index, String recordPath, List<String> fieldPaths) throws IOException { |
80 |
this.reader = reader;
|
81 |
this.index = index;
|
82 |
this.recordPath = recordPath;
|
83 |
} |
84 |
|
85 |
@Override
|
86 |
public void close() { |
87 |
IOUtils.closeQuietly(this.reader);
|
88 |
IOUtils.closeQuietly(this.index);
|
89 |
} |
90 |
|
91 |
private List<String> getRecord(long position, List<String> fieldPaths) { |
92 |
class StopParserSAXException extends SAXException { |
93 |
} |
94 |
|
95 |
class ParseRecordsHandler extends DefaultHandler { |
96 |
|
97 |
Locator locator; |
98 |
List<String> path = new ArrayList<>(); |
99 |
Map<String, String> record = new HashMap<>(); |
100 |
StringBuilder value = new StringBuilder(); |
101 |
List<String> values = new ArrayList<>(); |
102 |
|
103 |
@Override
|
104 |
public void setDocumentLocator(Locator locator) { |
105 |
this.locator = locator;
|
106 |
} |
107 |
|
108 |
@Override
|
109 |
public void startElement(String uri, String localName, String qName, Attributes attributes) throws SAXException { |
110 |
path.add(localName); |
111 |
String path_s = StringUtils.join(path, "/"); |
112 |
if (StringUtils.equalsIgnoreCase(path_s, recordPath)) {
|
113 |
return;
|
114 |
} |
115 |
this.value.setLength(0); |
116 |
// this.value.trimToSize();
|
117 |
} |
118 |
|
119 |
@Override
|
120 |
public void characters(char[] ch, int start, int length) throws SAXException { |
121 |
value.append(new String(ch, start, length)); |
122 |
} |
123 |
|
124 |
@Override
|
125 |
public void endElement(String uri, String localName, String qName) throws SAXException { |
126 |
String path_s = StringUtils.join(path, "/"); |
127 |
|
128 |
if (StringUtils.equalsIgnoreCase(path_s, recordPath)) {
|
129 |
for (String fieldPath : fieldPaths) { |
130 |
values.add(record.get(fieldPath)); |
131 |
} |
132 |
throw new StopParserSAXException(); |
133 |
} else {
|
134 |
for (String fieldPath : fieldPaths) { |
135 |
if (StringUtils.equalsIgnoreCase(path_s, fieldPath)) {
|
136 |
record.put(fieldPath, this.value.toString());
|
137 |
// No break to handle repeated columns
|
138 |
} |
139 |
} |
140 |
} |
141 |
|
142 |
path.remove(path.size() - 1);
|
143 |
} |
144 |
} |
145 |
|
146 |
ParseRecordsHandler handler = null;
|
147 |
try {
|
148 |
Record record = this.index.get64(position);
|
149 |
long recordPosition = record.getLong(IDXFIELD_RECORDPOS);
|
150 |
this.reader.seek(recordPosition);
|
151 |
BufferedReader breader = new BufferedReader(this.reader, 1024 * 8); |
152 |
|
153 |
InputSource is = new InputSource(breader);
|
154 |
SAXParserFactory spf = SAXParserFactory.newInstance(); |
155 |
spf.setNamespaceAware(true);
|
156 |
SAXParser saxParser = spf.newSAXParser();
|
157 |
handler = new ParseRecordsHandler();
|
158 |
|
159 |
saxParser.parse(is, handler); |
160 |
} catch (StopParserSAXException ex) {
|
161 |
if (handler != null) { |
162 |
return handler.values;
|
163 |
} |
164 |
} catch (Exception ex) { |
165 |
throw new RuntimeException("Can't parse record " + position, ex); |
166 |
} |
167 |
return null; // ????? |
168 |
} |
169 |
|
170 |
@Override
|
171 |
public List<String> get(int index) { |
172 |
return this.getRecord(index, fieldPaths); |
173 |
} |
174 |
|
175 |
@Override
|
176 |
public List<String> get64(long position) { |
177 |
return this.getRecord(position, fieldPaths); |
178 |
} |
179 |
|
180 |
@Override
|
181 |
public int size() { |
182 |
return this.index.size(); |
183 |
} |
184 |
|
185 |
@Override
|
186 |
public long size64() { |
187 |
return this.index.size64(); |
188 |
} |
189 |
|
190 |
final public void createIndex(File indexFile) throws IOException { |
191 |
try {
|
192 |
// 1. Creamos el indice vacio
|
193 |
RecordType recordType = recordTypeBuilder() |
194 |
.addLong() // IDXFIELD_LINE
|
195 |
.addLong() // IDXFIELD_COLUMN
|
196 |
.addLong() // IDXFIELD_LINEPOS
|
197 |
.addLong() // IDXFIELD_RECORDPOS
|
198 |
.build(); |
199 |
Record record = recordType.createRecord(); |
200 |
final RecordsFile theIndex = new RecordsFileImpl(); |
201 |
theIndex.create(indexFile, recordType, 0);
|
202 |
|
203 |
// 2. Rellenamos los campos numero de linea y columna de cada registro
|
204 |
// en el indice.
|
205 |
// Para ello nos recorremos el XML y vamos escribiendo en el indice.
|
206 |
this.reader.rewind();
|
207 |
InputSource is = new InputSource(reader);
|
208 |
SAXParserFactory spf = SAXParserFactory.newInstance(); |
209 |
spf.setNamespaceAware(true);
|
210 |
SAXParser saxParser = spf.newSAXParser();
|
211 |
|
212 |
List<String> path = new ArrayList<>(); |
213 |
// TODO: Aqui habria que quedarse con la primera aparicion de
|
214 |
// los campos solicitados que no pertenezcan al registro principal.
|
215 |
// Probablemente habria que guardar esos valores en disco junto al
|
216 |
// fichero de indice.
|
217 |
saxParser.parse(is, new DefaultHandler() {
|
218 |
Locator locator; |
219 |
|
220 |
@Override
|
221 |
public void setDocumentLocator(Locator locator) { |
222 |
this.locator = locator;
|
223 |
} |
224 |
|
225 |
@Override
|
226 |
public void startElement(String uri, String localName, String qName, Attributes attributes) throws SAXException { |
227 |
path.add(localName); |
228 |
String path_s = StringUtils.join(path, "/"); |
229 |
|
230 |
if (StringUtils.equalsIgnoreCase(path_s, recordPath)) {
|
231 |
record.setLong(IDXFIELD_LINE, locator.getLineNumber() - 1);
|
232 |
record.setLong(IDXFIELD_COLUMN, locator.getColumnNumber() - 1);
|
233 |
record.setLong(IDXFIELD_LINEPOS, 0);
|
234 |
record.setLong(IDXFIELD_RECORDPOS, 0);
|
235 |
theIndex.add(record); |
236 |
} |
237 |
} |
238 |
|
239 |
@Override
|
240 |
public void endElement(String uri, String localName, String qName) throws SAXException { |
241 |
path.remove(path.size() - 1);
|
242 |
} |
243 |
}); |
244 |
|
245 |
// 3. Nos recorremos el indice y calculamos la posicion de la linea
|
246 |
// para cada registro.
|
247 |
this.reader.rewind();
|
248 |
PositionCalculator positionCalculator = new PositionCalculator(reader);
|
249 |
|
250 |
for (int i = 0; i < theIndex.size(); i++) { |
251 |
Record r = theIndex.get(i); |
252 |
positionCalculator.next(r.getLong(IDXFIELD_LINE), r.getLong(IDXFIELD_COLUMN)); |
253 |
r.setLong(IDXFIELD_LINEPOS, positionCalculator.getLinePosition()); |
254 |
r.setLong(IDXFIELD_RECORDPOS, positionCalculator.getColumnPosition()); |
255 |
theIndex.set(i, record); |
256 |
} |
257 |
this.index = new RecordsFileImpl(indexFile); |
258 |
} catch (Exception ex) { |
259 |
throw new IOException("Can't create index " + Objects.toString(indexFile), ex); |
260 |
} |
261 |
} |
262 |
|
263 |
private static class CharacterSize { |
264 |
|
265 |
private final Charset charset; |
266 |
private final CharBuffer charBuffer; |
267 |
|
268 |
public CharacterSize(Charset charset) { |
269 |
this.charset = charset;
|
270 |
this.charBuffer = CharBuffer.allocate(1); |
271 |
} |
272 |
|
273 |
public int size(char ch) { |
274 |
this.charBuffer.put(0, ch); |
275 |
this.charBuffer.position(0); |
276 |
ByteBuffer buffer = this.charset.encode(this.charBuffer); |
277 |
return buffer.limit();
|
278 |
} |
279 |
|
280 |
public int size(char[] cbuf, int off, int len) { |
281 |
CharBuffer cb = CharBuffer.wrap(cbuf, off, len); |
282 |
ByteBuffer byteBuffer = this.charset.encode(cb); |
283 |
return byteBuffer.limit();
|
284 |
} |
285 |
|
286 |
public int size(char[] cbuf) { |
287 |
CharBuffer cb = CharBuffer.wrap(cbuf, 0, cbuf.length); |
288 |
ByteBuffer byteBuffer = this.charset.encode(cb); |
289 |
return byteBuffer.limit();
|
290 |
} |
291 |
|
292 |
public int size(String s) { |
293 |
CharBuffer cb = CharBuffer.wrap(s); |
294 |
ByteBuffer byteBuffer = this.charset.encode(cb); |
295 |
return byteBuffer.limit();
|
296 |
} |
297 |
} |
298 |
|
299 |
private static class PositionCalculator { |
300 |
|
301 |
private final RandomAccessFileReader reader; |
302 |
private final BufferedReader breader; |
303 |
private long currentPosition; |
304 |
private long currentColumn; |
305 |
private long currentLine; |
306 |
private long currentColumnPosition; |
307 |
private long currentLinePosition; |
308 |
|
309 |
private final char[] ch; |
310 |
private final CharacterSize characterSize; |
311 |
|
312 |
public PositionCalculator(RandomAccessFileReader reader) {
|
313 |
this.reader = reader;
|
314 |
this.breader = new BufferedReader(this.reader, 1024 * 8); |
315 |
this.currentPosition = this.reader.getCurrentPosition(); |
316 |
this.currentLine = 0; |
317 |
this.currentColumn = 0; |
318 |
this.ch = new char[1]; |
319 |
this.characterSize = new CharacterSize(this.reader.getCharset()); |
320 |
} |
321 |
|
322 |
public boolean next(long line, long column) throws IOException { |
323 |
while (this.currentLine < line) { |
324 |
if (breader.read(this.ch, 0, 1) < 1) { |
325 |
return false; |
326 |
} |
327 |
char c = ch[0]; |
328 |
this.currentPosition += characterSize.size(c);
|
329 |
if (c == '\n') { |
330 |
this.currentLine++;
|
331 |
this.currentColumn = 0; |
332 |
} |
333 |
} |
334 |
this.currentLinePosition = this.currentPosition; |
335 |
while (this.currentColumn < column) { |
336 |
if (breader.read(this.ch, 0, 1) < 1) { |
337 |
return false; |
338 |
} |
339 |
char c = ch[0]; |
340 |
this.currentPosition += characterSize.size(c);
|
341 |
if (c == '\n') { |
342 |
// Uff, esto seria un error, ya que la linea corriente no
|
343 |
// tiene tantas columnas.
|
344 |
throw new IOException("Illegal column number " + column + " for line " + line); |
345 |
} |
346 |
this.currentColumn++;
|
347 |
} |
348 |
this.currentColumnPosition = this.currentPosition; |
349 |
return true; |
350 |
} |
351 |
|
352 |
public long getLinePosition() { |
353 |
return this.currentLinePosition; |
354 |
} |
355 |
|
356 |
public long getColumnPosition() { |
357 |
return this.currentColumnPosition; |
358 |
} |
359 |
|
360 |
} |
361 |
|
362 |
public static void main(String[] args) throws Exception { |
363 |
final String XMLFILE1 = "/home/jjdelcerro/datos/geodata/vector/gml/navarra.gml"; |
364 |
final String XMLFILE2 = "/home/jjdelcerro/datos/geodata/vector/gml/Municipis/Municipis.gml"; |
365 |
|
366 |
String gmlfile = XMLFILE1;
|
367 |
XMLFileAsList gml = new XMLFileAsList(
|
368 |
new File(gmlfile), |
369 |
new File(FilenameUtils.removeExtension(gmlfile)+".gmlidx"), |
370 |
StandardCharsets.UTF_8, |
371 |
"FeatureCollection/featureMember",
|
372 |
ListBuilder.create( |
373 |
"FEATURE/fid",
|
374 |
"FEATURE/MERINDAD",
|
375 |
"FEATURE/Cnt_MERIND",
|
376 |
"FEATURE/Nombre"
|
377 |
) |
378 |
); |
379 |
System.out.println("File: "+gmlfile); |
380 |
for (int i = 0; i < gml.size(); i++) { |
381 |
List<String> item = gml.get(i); |
382 |
System.out.println(item);
|
383 |
} |
384 |
} |
385 |
} |