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 / csvlib / TextFileAsList.java @ 45721
History | View | Annotate | Download (6.41 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.csvlib; |
7 |
|
8 |
import java.io.Closeable; |
9 |
import java.io.DataOutputStream; |
10 |
import java.io.File; |
11 |
import java.io.FileOutputStream; |
12 |
import java.io.IOException; |
13 |
import java.io.RandomAccessFile; |
14 |
import java.nio.LongBuffer; |
15 |
import java.nio.MappedByteBuffer; |
16 |
import java.nio.channels.FileChannel.MapMode; |
17 |
import java.nio.charset.Charset; |
18 |
import java.util.AbstractList; |
19 |
import java.util.Arrays; |
20 |
import java.util.Date; |
21 |
import java.util.List; |
22 |
import org.apache.commons.io.FilenameUtils; |
23 |
import org.apache.commons.io.IOUtils; |
24 |
|
25 |
/**
|
26 |
*
|
27 |
* @author jjdelcerro
|
28 |
*/
|
29 |
public class TextFileAsList extends AbstractList<String> implements Closeable { |
30 |
|
31 |
private final File text; |
32 |
private final File idx; |
33 |
private final Charset charset; |
34 |
private final Bytes buffer; |
35 |
|
36 |
private RandomAccessFile text_raf; |
37 |
|
38 |
private RandomAccessFile idx_raf; |
39 |
private MappedByteBuffer idx_map; |
40 |
private LongBuffer idx_buffer; |
41 |
|
42 |
public TextFileAsList(File text, File idx, Charset charset) { |
43 |
this.text = text;
|
44 |
this.idx = idx;
|
45 |
this.charset = charset;
|
46 |
this.buffer = new Bytes(this.charset); |
47 |
} |
48 |
|
49 |
public static void create_index(File csv, File idx, Charset charset) throws IOException { |
50 |
FileOutputStream idx_fos = null; |
51 |
DataOutputStream idx_dos = null; |
52 |
RandomAccessFile csv_raf = null; |
53 |
try {
|
54 |
idx_fos = new FileOutputStream(idx); |
55 |
idx_dos = new DataOutputStream(idx_fos); |
56 |
csv_raf = new RandomAccessFile(csv, "r"); |
57 |
Bytes buffer = new Bytes(charset);
|
58 |
String line;
|
59 |
long pos = csv_raf.getFilePointer();
|
60 |
while ((line = readLine(csv_raf, buffer)) != null) { |
61 |
idx_dos.writeLong(pos); |
62 |
pos = csv_raf.getFilePointer(); |
63 |
} |
64 |
} finally {
|
65 |
IOUtils.closeQuietly(idx_dos); |
66 |
IOUtils.closeQuietly(idx_fos); |
67 |
IOUtils.closeQuietly(csv_raf); |
68 |
} |
69 |
} |
70 |
|
71 |
public void open() { |
72 |
try {
|
73 |
text_raf = new RandomAccessFile(this.text, "r"); |
74 |
idx_raf = new RandomAccessFile(this.idx, "r"); |
75 |
long idx_size = idx_raf.length();
|
76 |
idx_map = idx_raf.getChannel().map(MapMode.READ_ONLY, 0, idx_size);
|
77 |
idx_buffer = idx_map.asLongBuffer(); |
78 |
} catch (IOException ex) { |
79 |
throw new RuntimeException("Can't open csv/idx files.", ex); |
80 |
} |
81 |
} |
82 |
|
83 |
private long getFilePointer(int lineno) { |
84 |
if (idx_buffer == null) { |
85 |
this.open();
|
86 |
} |
87 |
return idx_buffer.get(lineno);
|
88 |
} |
89 |
|
90 |
@Override
|
91 |
public void close() { |
92 |
IOUtils.closeQuietly(this.text_raf);
|
93 |
IOUtils.closeQuietly(this.idx_raf);
|
94 |
this.idx_buffer = null; |
95 |
this.idx_map = null; |
96 |
} |
97 |
|
98 |
@Override
|
99 |
public String get(int index) { |
100 |
if (text_raf == null) { |
101 |
this.open();
|
102 |
} |
103 |
try {
|
104 |
long pos = this.getFilePointer(index); |
105 |
this.text_raf.seek(pos);
|
106 |
return readLine(text_raf, this.buffer); |
107 |
} catch (IOException ex) { |
108 |
throw new RuntimeException("Can't access to " + index + " element.", ex); |
109 |
} |
110 |
} |
111 |
|
112 |
@Override
|
113 |
public int size() { |
114 |
if (idx_buffer == null) { |
115 |
this.open();
|
116 |
} |
117 |
return this.idx_buffer.capacity(); |
118 |
} |
119 |
|
120 |
@Override
|
121 |
public void clear() { |
122 |
throw new UnsupportedOperationException(); |
123 |
} |
124 |
|
125 |
private static class Bytes { |
126 |
|
127 |
private static final int DEFAULT_CAPACITY = 1024; |
128 |
private final Charset charset; |
129 |
|
130 |
private byte[] buffer = new byte[DEFAULT_CAPACITY]; |
131 |
private int pos = 0; |
132 |
|
133 |
public Bytes(Charset charset) { |
134 |
this.charset = charset;
|
135 |
} |
136 |
|
137 |
public void reset() { |
138 |
this.pos = 0; |
139 |
} |
140 |
|
141 |
public void append(int value) { |
142 |
if (pos >= buffer.length) {
|
143 |
grow(buffer.length + DEFAULT_CAPACITY); |
144 |
} |
145 |
this.buffer[pos++] = (byte) value; |
146 |
} |
147 |
|
148 |
private void grow(int minCapacity) { |
149 |
buffer = Arrays.copyOf(buffer, minCapacity);
|
150 |
} |
151 |
|
152 |
public int length() { |
153 |
return this.pos; |
154 |
} |
155 |
|
156 |
@Override
|
157 |
public String toString() { |
158 |
String s = new String(buffer, 0, pos, charset); |
159 |
return s;
|
160 |
} |
161 |
} |
162 |
|
163 |
private static String readLine(RandomAccessFile raf, Bytes buffer) throws IOException { |
164 |
buffer.reset(); |
165 |
int c = -1; |
166 |
boolean eol = false; |
167 |
|
168 |
while (!eol) {
|
169 |
switch (c = raf.read()) {
|
170 |
case -1: |
171 |
case '\n': |
172 |
eol = true;
|
173 |
break;
|
174 |
case '\r': |
175 |
eol = true;
|
176 |
long cur = raf.getFilePointer();
|
177 |
if ((raf.read()) != '\n') { |
178 |
raf.seek(cur); |
179 |
} |
180 |
break;
|
181 |
default:
|
182 |
buffer.append(c); |
183 |
break;
|
184 |
} |
185 |
} |
186 |
|
187 |
if ((c == -1) && (buffer.length() == 0)) { |
188 |
return null; |
189 |
} |
190 |
return buffer.toString();
|
191 |
} |
192 |
|
193 |
public static void main(String[] args) throws Exception { |
194 |
String fname;
|
195 |
// fname = "/home/jjdelcerro/Descargas/test/origen_coordenadas.csv";
|
196 |
// fname = "/home/jjdelcerro/Descargas/test/esp_poblaciones.csv";
|
197 |
fname = "/home/jjdelcerro/Descargas/test/esp_provincias.csv";
|
198 |
|
199 |
File csv_file = new File(fname); |
200 |
File idx_file = new File(FilenameUtils.removeExtension(csv_file.getAbsolutePath()) + ".idx"); |
201 |
|
202 |
TextFileAsList.create_index(csv_file, idx_file, Charset.forName("UTF-8")); |
203 |
|
204 |
List<String> text = new TextFileAsList(csv_file, idx_file, Charset.forName("UTF-8")); |
205 |
System.out.println("Lines " + text.size()); |
206 |
for (String line : text) { |
207 |
// System.out.println(line);
|
208 |
} |
209 |
// System.out.println("-----------------");
|
210 |
long t1 = new Date().getTime(); |
211 |
for (int i = text.size()-1; i >= 0 ; i--) { |
212 |
String line = text.get(i);
|
213 |
// System.out.println(String.format("%6d: %s", i, line));
|
214 |
} |
215 |
long t2 = new Date().getTime(); |
216 |
System.out.println("Time: "+ (t2-t1) + " ms"); |
217 |
IOUtils.closeQuietly((Closeable) text);
|
218 |
} |
219 |
|
220 |
} |