2 |
26 Feb 07 |
jari |
1 |
package org.tigr.microarray.mev.file.agilent; |
2 |
26 Feb 07 |
jari |
2 |
|
2 |
26 Feb 07 |
jari |
3 |
/* |
2 |
26 Feb 07 |
jari |
* Created on Jan 21, 2004 |
2 |
26 Feb 07 |
jari |
5 |
*/ |
2 |
26 Feb 07 |
jari |
6 |
import java.io.BufferedReader; |
2 |
26 Feb 07 |
jari |
7 |
import java.io.File; |
2 |
26 Feb 07 |
jari |
8 |
import java.io.FileReader; |
2 |
26 Feb 07 |
jari |
9 |
import java.io.IOException; |
2 |
26 Feb 07 |
jari |
10 |
import java.util.StringTokenizer; |
2 |
26 Feb 07 |
jari |
11 |
import java.util.Vector; |
2 |
26 Feb 07 |
jari |
12 |
|
2 |
26 Feb 07 |
jari |
13 |
/** |
2 |
26 Feb 07 |
jari |
* This is a simple utility class that reads a tab-delimited text file into a Vector of Strings |
2 |
26 Feb 07 |
jari |
* where each String is a line from the file. |
2 |
26 Feb 07 |
jari |
16 |
* |
2 |
26 Feb 07 |
jari |
* It exists to deal with blank cells in a spreadsheet. |
2 |
26 Feb 07 |
jari |
18 |
* |
2 |
26 Feb 07 |
jari |
* Where there are null fields in the file, a String 'replacer' is added as a placeholder. |
2 |
26 Feb 07 |
jari |
20 |
* |
2 |
26 Feb 07 |
jari |
* @author vu |
2 |
26 Feb 07 |
jari |
22 |
*/ |
2 |
26 Feb 07 |
jari |
23 |
public class Reader { |
2 |
26 Feb 07 |
jari |
//private Vector vNullLine; |
2 |
26 Feb 07 |
jari |
25 |
private Vector vLine; |
2 |
26 Feb 07 |
jari |
26 |
|
2 |
26 Feb 07 |
jari |
27 |
|
2 |
26 Feb 07 |
jari |
28 |
public Reader() { |
2 |
26 Feb 07 |
jari |
//do nothing |
2 |
26 Feb 07 |
jari |
30 |
}//end constructor() |
2 |
26 Feb 07 |
jari |
31 |
|
2 |
26 Feb 07 |
jari |
32 |
|
2 |
26 Feb 07 |
jari |
33 |
public void readFile(File f) throws IOException { |
2 |
26 Feb 07 |
jari |
34 |
this.vLine = this.readLines(f); |
2 |
26 Feb 07 |
jari |
35 |
} |
2 |
26 Feb 07 |
jari |
36 |
|
2 |
26 Feb 07 |
jari |
37 |
|
2 |
26 Feb 07 |
jari |
38 |
public Vector getVNullLine(String s) { |
2 |
26 Feb 07 |
jari |
39 |
Vector vReturn = this.nullifyLines(this.vLine, s); |
2 |
26 Feb 07 |
jari |
40 |
return vReturn; |
2 |
26 Feb 07 |
jari |
41 |
} |
2 |
26 Feb 07 |
jari |
42 |
|
2 |
26 Feb 07 |
jari |
43 |
|
2 |
26 Feb 07 |
jari |
44 |
public Vector getVNullLineBasedOnHeader(String s) { |
2 |
26 Feb 07 |
jari |
45 |
Vector vReturn = this.nullifyLinesBasedOnHeader(this.vLine, s); |
2 |
26 Feb 07 |
jari |
46 |
return vReturn; |
2 |
26 Feb 07 |
jari |
47 |
} |
2 |
26 Feb 07 |
jari |
48 |
|
2 |
26 Feb 07 |
jari |
49 |
|
2 |
26 Feb 07 |
jari |
50 |
public Vector getVLine() { |
2 |
26 Feb 07 |
jari |
51 |
return this.vLine; |
2 |
26 Feb 07 |
jari |
52 |
} |
2 |
26 Feb 07 |
jari |
53 |
|
2 |
26 Feb 07 |
jari |
54 |
|
2 |
26 Feb 07 |
jari |
55 |
public Vector nullifyLinesBasedOnHeader( Vector vLines, String replacer ) { |
2 |
26 Feb 07 |
jari |
56 |
Vector vReturn = new Vector(); |
2 |
26 Feb 07 |
jari |
57 |
|
2 |
26 Feb 07 |
jari |
58 |
String header = ( String ) vLines.elementAt( 0 ); |
2 |
26 Feb 07 |
jari |
59 |
StringTokenizer stHeader = new StringTokenizer( header, "\t", false ); |
2 |
26 Feb 07 |
jari |
60 |
int headerKount = stHeader.countTokens(); |
2 |
26 Feb 07 |
jari |
61 |
|
2 |
26 Feb 07 |
jari |
//loop through the lines |
2 |
26 Feb 07 |
jari |
63 |
for(int i=0; i<vLines.size(); i++) { |
2 |
26 Feb 07 |
jari |
//if( i == 141 ) { |
2 |
26 Feb 07 |
jari |
//System.out.println("Stop"); |
2 |
26 Feb 07 |
jari |
66 |
//} |
2 |
26 Feb 07 |
jari |
67 |
|
2 |
26 Feb 07 |
jari |
68 |
String line = (String) vLines.get(i); |
2 |
26 Feb 07 |
jari |
69 |
if( line.trim().equals( "" ) ) { |
2 |
26 Feb 07 |
jari |
//ignore this blank line |
2 |
26 Feb 07 |
jari |
71 |
} else { |
2 |
26 Feb 07 |
jari |
//tokenize this line |
2 |
26 Feb 07 |
jari |
73 |
StringTokenizer st = new StringTokenizer(line, "\t", true); |
2 |
26 Feb 07 |
jari |
74 |
int kount = st.countTokens(); |
2 |
26 Feb 07 |
jari |
75 |
|
2 |
26 Feb 07 |
jari |
76 |
StringBuffer sb = new StringBuffer(); |
2 |
26 Feb 07 |
jari |
77 |
|
2 |
26 Feb 07 |
jari |
//keep track of the last token to compare to the current token |
2 |
26 Feb 07 |
jari |
79 |
String last = ""; |
2 |
26 Feb 07 |
jari |
80 |
String current = null; |
2 |
26 Feb 07 |
jari |
81 |
|
2 |
26 Feb 07 |
jari |
//loop through the tokens |
2 |
26 Feb 07 |
jari |
83 |
for(int j=0; j<kount; j++) { |
2 |
26 Feb 07 |
jari |
84 |
current = st.nextToken(); |
2 |
26 Feb 07 |
jari |
85 |
if(j == 0) { |
2 |
26 Feb 07 |
jari |
//first token in line |
2 |
26 Feb 07 |
jari |
87 |
if(current.equals("\t")) { |
2 |
26 Feb 07 |
jari |
//first cell was a blank cell, so replace with replacer |
2 |
26 Feb 07 |
jari |
89 |
sb.append(replacer); |
2 |
26 Feb 07 |
jari |
90 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
91 |
} |
2 |
26 Feb 07 |
jari |
92 |
else { |
2 |
26 Feb 07 |
jari |
93 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
94 |
} |
2 |
26 Feb 07 |
jari |
95 |
} |
2 |
26 Feb 07 |
jari |
96 |
else { |
2 |
26 Feb 07 |
jari |
//not first token |
2 |
26 Feb 07 |
jari |
98 |
if(j == kount-1) { |
2 |
26 Feb 07 |
jari |
//last token |
2 |
26 Feb 07 |
jari |
100 |
if(current.equals("\t")) { |
2 |
26 Feb 07 |
jari |
//last token was blank cell, so replace with replacer |
2 |
26 Feb 07 |
jari |
102 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
103 |
sb.append(replacer); |
2 |
26 Feb 07 |
jari |
104 |
} |
2 |
26 Feb 07 |
jari |
105 |
else { |
2 |
26 Feb 07 |
jari |
106 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
107 |
} |
2 |
26 Feb 07 |
jari |
108 |
} |
2 |
26 Feb 07 |
jari |
109 |
else { |
2 |
26 Feb 07 |
jari |
//not first or last token |
2 |
26 Feb 07 |
jari |
111 |
if(current.equals("\t")) { |
2 |
26 Feb 07 |
jari |
112 |
if(last.equals("\t")) { |
2 |
26 Feb 07 |
jari |
//2 tabs in a row, insert null |
2 |
26 Feb 07 |
jari |
114 |
sb.append(replacer); |
2 |
26 Feb 07 |
jari |
115 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
116 |
} |
2 |
26 Feb 07 |
jari |
117 |
else { |
2 |
26 Feb 07 |
jari |
//tab here, skip |
2 |
26 Feb 07 |
jari |
119 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
120 |
} |
2 |
26 Feb 07 |
jari |
121 |
} |
2 |
26 Feb 07 |
jari |
122 |
else { |
2 |
26 Feb 07 |
jari |
//data here |
2 |
26 Feb 07 |
jari |
124 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
125 |
} |
2 |
26 Feb 07 |
jari |
126 |
} |
2 |
26 Feb 07 |
jari |
127 |
} |
2 |
26 Feb 07 |
jari |
128 |
last = current; |
2 |
26 Feb 07 |
jari |
129 |
}//end j |
2 |
26 Feb 07 |
jari |
130 |
|
2 |
26 Feb 07 |
jari |
//before adding, now check number of cells. if less than header line, |
2 |
26 Feb 07 |
jari |
//add nullReplacer to make it up |
2 |
26 Feb 07 |
jari |
133 |
StringTokenizer st2 = new StringTokenizer( sb.toString(), "\t", false ); |
2 |
26 Feb 07 |
jari |
134 |
int tokenKount2 = st2.countTokens(); |
2 |
26 Feb 07 |
jari |
135 |
int diff = headerKount - tokenKount2; |
2 |
26 Feb 07 |
jari |
136 |
for( int d = 0; d < diff; d ++ ) { |
2 |
26 Feb 07 |
jari |
137 |
if( d == 0 ) { |
2 |
26 Feb 07 |
jari |
138 |
sb.append( "\t" ); |
2 |
26 Feb 07 |
jari |
139 |
} |
2 |
26 Feb 07 |
jari |
140 |
|
2 |
26 Feb 07 |
jari |
141 |
sb.append( replacer ); |
2 |
26 Feb 07 |
jari |
142 |
|
2 |
26 Feb 07 |
jari |
143 |
if( d + 1 == diff ) { |
2 |
26 Feb 07 |
jari |
//don't add tab |
2 |
26 Feb 07 |
jari |
145 |
} else { |
2 |
26 Feb 07 |
jari |
146 |
sb.append( "\t" ); |
2 |
26 Feb 07 |
jari |
147 |
} |
2 |
26 Feb 07 |
jari |
148 |
} |
2 |
26 Feb 07 |
jari |
149 |
|
2 |
26 Feb 07 |
jari |
150 |
vReturn.add(sb.toString()); |
2 |
26 Feb 07 |
jari |
151 |
}//end else |
2 |
26 Feb 07 |
jari |
152 |
}//end i |
2 |
26 Feb 07 |
jari |
153 |
|
2 |
26 Feb 07 |
jari |
154 |
return vReturn; |
2 |
26 Feb 07 |
jari |
155 |
} |
2 |
26 Feb 07 |
jari |
156 |
|
2 |
26 Feb 07 |
jari |
157 |
|
2 |
26 Feb 07 |
jari |
158 |
/** |
2 |
26 Feb 07 |
jari |
* This method exists to deal with the StringTokenizer's inability to recognize a |
2 |
26 Feb 07 |
jari |
* blank cell in a tab-delim file. StringTokinzer just ignores it, so I wrote this to |
2 |
26 Feb 07 |
jari |
* replace a blank cell with a 'replacer' String. |
2 |
26 Feb 07 |
jari |
162 |
* |
2 |
26 Feb 07 |
jari |
* Note: does not work when last cells in line are left blank |
2 |
26 Feb 07 |
jari |
164 |
* |
2 |
26 Feb 07 |
jari |
* @param vLines |
2 |
26 Feb 07 |
jari |
* @param replacer |
2 |
26 Feb 07 |
jari |
* @return Vector of Lines from file where blank cells are replaced with 'replacer' |
2 |
26 Feb 07 |
jari |
168 |
*/ |
2 |
26 Feb 07 |
jari |
169 |
public Vector nullifyLines (Vector vLines, String replacer) { |
2 |
26 Feb 07 |
jari |
170 |
Vector vReturn = new Vector(); |
2 |
26 Feb 07 |
jari |
171 |
|
2 |
26 Feb 07 |
jari |
//loop through the lines |
2 |
26 Feb 07 |
jari |
173 |
for(int i=0; i<vLines.size(); i++) { |
2 |
26 Feb 07 |
jari |
174 |
String line = (String) vLines.get(i); |
2 |
26 Feb 07 |
jari |
175 |
|
2 |
26 Feb 07 |
jari |
176 |
if( line.trim().equals( "" ) ) { |
2 |
26 Feb 07 |
jari |
//ignore blank line |
2 |
26 Feb 07 |
jari |
178 |
} else { |
2 |
26 Feb 07 |
jari |
179 |
|
2 |
26 Feb 07 |
jari |
//tokenize this line |
2 |
26 Feb 07 |
jari |
181 |
StringTokenizer st = new StringTokenizer( line, "\t", true ); |
2 |
26 Feb 07 |
jari |
182 |
int kount = st.countTokens(); |
2 |
26 Feb 07 |
jari |
183 |
|
2 |
26 Feb 07 |
jari |
184 |
StringBuffer sb = new StringBuffer(); |
2 |
26 Feb 07 |
jari |
185 |
|
2 |
26 Feb 07 |
jari |
//keep track of the last token to compare to the current token |
2 |
26 Feb 07 |
jari |
187 |
String last = ""; |
2 |
26 Feb 07 |
jari |
188 |
String current = null; |
2 |
26 Feb 07 |
jari |
189 |
|
2 |
26 Feb 07 |
jari |
//loop through the tokens |
2 |
26 Feb 07 |
jari |
191 |
for(int j=0; j<kount; j++) { |
2 |
26 Feb 07 |
jari |
192 |
current = st.nextToken(); |
2 |
26 Feb 07 |
jari |
193 |
if(j == 0) { |
2 |
26 Feb 07 |
jari |
//first token in line |
2 |
26 Feb 07 |
jari |
195 |
if(current.equals("\t")) { |
2 |
26 Feb 07 |
jari |
//first cell was a blank cell, so replace with replacer |
2 |
26 Feb 07 |
jari |
197 |
sb.append(replacer); |
2 |
26 Feb 07 |
jari |
198 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
199 |
} |
2 |
26 Feb 07 |
jari |
200 |
else { |
2 |
26 Feb 07 |
jari |
201 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
202 |
} |
2 |
26 Feb 07 |
jari |
203 |
} |
2 |
26 Feb 07 |
jari |
204 |
else { |
2 |
26 Feb 07 |
jari |
//not first token |
2 |
26 Feb 07 |
jari |
206 |
if(j == kount-1) { |
2 |
26 Feb 07 |
jari |
//last token |
2 |
26 Feb 07 |
jari |
208 |
if(current.equals("\t")) { |
2 |
26 Feb 07 |
jari |
//last token was blank cell, so replace with replacer |
2 |
26 Feb 07 |
jari |
210 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
211 |
sb.append(replacer); |
2 |
26 Feb 07 |
jari |
212 |
} |
2 |
26 Feb 07 |
jari |
213 |
else { |
2 |
26 Feb 07 |
jari |
214 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
215 |
} |
2 |
26 Feb 07 |
jari |
216 |
} |
2 |
26 Feb 07 |
jari |
217 |
else { |
2 |
26 Feb 07 |
jari |
//not first or last token |
2 |
26 Feb 07 |
jari |
219 |
if(current.equals("\t")) { |
2 |
26 Feb 07 |
jari |
220 |
if(last.equals("\t")) { |
2 |
26 Feb 07 |
jari |
//2 tabs in a row, insert null |
2 |
26 Feb 07 |
jari |
222 |
sb.append(replacer); |
2 |
26 Feb 07 |
jari |
223 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
224 |
} |
2 |
26 Feb 07 |
jari |
225 |
else { |
2 |
26 Feb 07 |
jari |
//tab here, skip |
2 |
26 Feb 07 |
jari |
227 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
228 |
} |
2 |
26 Feb 07 |
jari |
229 |
} |
2 |
26 Feb 07 |
jari |
230 |
else { |
2 |
26 Feb 07 |
jari |
//data here |
2 |
26 Feb 07 |
jari |
232 |
sb.append(current); |
2 |
26 Feb 07 |
jari |
233 |
} |
2 |
26 Feb 07 |
jari |
234 |
} |
2 |
26 Feb 07 |
jari |
235 |
} |
2 |
26 Feb 07 |
jari |
236 |
last = current; |
2 |
26 Feb 07 |
jari |
237 |
}//end j |
2 |
26 Feb 07 |
jari |
238 |
vReturn.add(sb.toString()); |
2 |
26 Feb 07 |
jari |
239 |
}//end else |
2 |
26 Feb 07 |
jari |
240 |
}//end i |
2 |
26 Feb 07 |
jari |
241 |
|
2 |
26 Feb 07 |
jari |
242 |
return vReturn; |
2 |
26 Feb 07 |
jari |
243 |
}//end parseLines() |
2 |
26 Feb 07 |
jari |
244 |
|
2 |
26 Feb 07 |
jari |
245 |
|
2 |
26 Feb 07 |
jari |
246 |
public Vector readLines(File f) throws IOException { |
2 |
26 Feb 07 |
jari |
247 |
Vector vReturn = new Vector(); |
2 |
26 Feb 07 |
jari |
248 |
String s; |
2 |
26 Feb 07 |
jari |
249 |
|
2 |
26 Feb 07 |
jari |
250 |
FileReader fr = new FileReader(f); |
2 |
26 Feb 07 |
jari |
251 |
BufferedReader br = new BufferedReader(fr); |
2 |
26 Feb 07 |
jari |
252 |
|
2 |
26 Feb 07 |
jari |
253 |
while( (s = br.readLine()) != null ) { |
2 |
26 Feb 07 |
jari |
254 |
vReturn.add(s); |
2 |
26 Feb 07 |
jari |
255 |
} |
2 |
26 Feb 07 |
jari |
256 |
fr.close(); |
2 |
26 Feb 07 |
jari |
257 |
br.close(); |
2 |
26 Feb 07 |
jari |
258 |
|
2 |
26 Feb 07 |
jari |
259 |
return vReturn; |
2 |
26 Feb 07 |
jari |
260 |
} |
2 |
26 Feb 07 |
jari |
261 |
} |