Révision 2768
| tmp/org.txm.index.core/src/org/txm/index/core/functions/PartitionIndex.java (revision 2768) | ||
|---|---|---|
| 32 | 32 |
import java.io.FileOutputStream; |
| 33 | 33 |
import java.io.IOException; |
| 34 | 34 |
import java.io.OutputStreamWriter; |
| 35 |
import java.util.ArrayList; |
|
| 35 | 36 |
import java.util.HashMap; |
| 36 | 37 |
import java.util.List; |
| 37 | 38 |
|
| ... | ... | |
| 417 | 418 |
public String getResultType() {
|
| 418 | 419 |
return PartitionIndex.class.getSimpleName(); |
| 419 | 420 |
} |
| 421 |
|
|
| 422 |
/** The novoc. */ |
|
| 423 |
protected static int novoc = 1; |
|
| 424 |
|
|
| 425 |
/** The prefix r. */ |
|
| 426 |
protected static String prefixR = "PartitionIndex"; //$NON-NLS-1$ |
|
| 427 |
|
|
| 428 |
/** |
|
| 429 |
* As r matrix. |
|
| 430 |
* |
|
| 431 |
* TODO MD: restore the asRMatrix method used by the execR&co macros as demonstration |
|
| 432 |
* TODO MD: finalize a export to R mechanism (maybe linked to the export engines to be implemented |
|
| 433 |
* |
|
| 434 |
* @return the string |
|
| 435 |
* @throws RWorkspaceException the r workspace exception |
|
| 436 |
*/ |
|
| 437 |
public String asRMatrix() throws RWorkspaceException {
|
|
| 438 |
String symbol = prefixR+novoc; |
|
| 439 |
|
|
| 440 |
ArrayList<String> colnames = new ArrayList<String>(); |
|
| 441 |
List<String> partnames = getPartNames(); |
|
| 442 |
partnames.add(this.getCorpus().getName()); |
|
| 443 |
|
|
| 444 |
colnames.add("F"); //$NON-NLS-1$
|
|
| 445 |
if (partnames.size() > 1) |
|
| 446 |
for (int j = 0; j < partnames.size(); j++) |
|
| 447 |
colnames.add(partnames.get(j)); |
|
| 448 |
|
|
| 449 |
//System.out.println("cols: "+colnames);
|
|
| 450 |
|
|
| 451 |
String[] keywords = new String[this.lines.size()]; |
|
| 452 |
|
|
| 453 |
int[] freq = new int[this.lines.size()]; |
|
| 454 |
ArrayList<int[]> partfreqs = new ArrayList<int[]>(partnames.size()); |
|
| 455 |
if (partnames.size() > 1) |
|
| 456 |
for (int j = 0; j < partnames.size(); j++) |
|
| 457 |
partfreqs.add(new int[this.lines.size()]); |
|
| 458 |
|
|
| 459 |
for (int i = 0 ; i < lines.size() ; i++) |
|
| 460 |
{
|
|
| 461 |
Line ligne = lines.get(i); |
|
| 462 |
freq[i] = ligne.getFrequency(); |
|
| 463 |
keywords[i] = ligne.toString(); |
|
| 464 |
if (partnames.size() > 1) |
|
| 465 |
for (int j = 0; j < partnames.size(); j++) |
|
| 466 |
{
|
|
| 467 |
partfreqs.get(j)[i] = ligne.getFrequency(j); |
|
| 468 |
} |
|
| 469 |
|
|
| 470 |
} |
|
| 471 |
|
|
| 472 |
RWorkspace rw = RWorkspace.getRWorkspaceInstance(); |
|
| 473 |
if (partnames.size() > 1) |
|
| 474 |
for (int j = 0; j < partnames.size(); j++) |
|
| 475 |
rw.addVectorToWorkspace("vocpartfreqs"+j, partfreqs.get(j)); //$NON-NLS-1$
|
|
| 476 |
rw.addVectorToWorkspace("vocfreq", freq); //$NON-NLS-1$
|
|
| 477 |
rw.addVectorToWorkspace("vockeywords", keywords); //$NON-NLS-1$
|
|
| 478 |
rw.addVectorToWorkspace("voccolnames", colnames.toArray(new String[colnames.size()])); //$NON-NLS-1$
|
|
| 479 |
|
|
| 480 |
int ncol = 1; |
|
| 481 |
if (partnames.size() > 1) |
|
| 482 |
ncol += partnames.size(); |
|
| 483 |
|
|
| 484 |
int nrow = lines.size(); |
|
| 485 |
String partscmd = ""; //$NON-NLS-1$ |
|
| 486 |
if (partnames.size() > 1) |
|
| 487 |
for (int j = 0; j < partnames.size(); j++) |
|
| 488 |
partscmd +=", vocpartfreqs"+j; //$NON-NLS-1$ |
|
| 489 |
|
|
| 490 |
rw.eval(symbol+ "<- matrix(data = c(vocfreq"+partscmd+"), nrow = "+nrow+", ncol = "+ncol+")"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ |
|
| 491 |
rw.eval("colnames("+symbol+" ) <- voccolnames"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 492 |
rw.eval("rownames("+symbol+" ) <- vockeywords"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 493 |
rw.eval(symbol+ "<- list(data="+symbol+")"); //$NON-NLS-1$ //$NON-NLS-2$ |
|
| 494 |
|
|
| 495 |
novoc++; |
|
| 496 |
return symbol; |
|
| 497 |
} |
|
| 420 | 498 |
} |
| tmp/org.txm.index.core/src/org/txm/index/core/functions/Index.java (revision 2768) | ||
|---|---|---|
| 1150 | 1150 |
public String getResultType() {
|
| 1151 | 1151 |
return IndexCoreMessages.RESULT_TYPE; |
| 1152 | 1152 |
} |
| 1153 |
|
|
| 1154 |
/** The novoc. */ |
|
| 1155 |
protected static int novoc = 1; |
|
| 1156 |
|
|
| 1157 |
/** The prefix r. */ |
|
| 1158 |
protected static String prefixR = "Index"; //$NON-NLS-1$ |
|
| 1159 |
|
|
| 1160 |
/** |
|
| 1161 |
* As r matrix. |
|
| 1162 |
* |
|
| 1163 |
* TODO MD: restore the asRMatrix method used by the execR&co macros as demonstration |
|
| 1164 |
* TODO MD: finalize a export to R mechanism (maybe linked to the export engines to be implemented |
|
| 1165 |
* |
|
| 1166 |
* @return the string |
|
| 1167 |
* @throws RWorkspaceException the r workspace exception |
|
| 1168 |
*/ |
|
| 1169 |
public String asRMatrix() throws RWorkspaceException {
|
|
| 1170 |
String symbol = prefixR+novoc; |
|
| 1171 |
|
|
| 1172 |
ArrayList<String> colnames = new ArrayList<String>(); |
|
| 1173 |
ArrayList<String> partnames = new ArrayList<String>(); |
|
| 1174 |
partnames.add(this.getCorpus().getName()); |
|
| 1175 |
|
|
| 1176 |
colnames.add("F"); //$NON-NLS-1$
|
|
| 1177 |
if (partnames.size() > 1) |
|
| 1178 |
for (int j = 0; j < partnames.size(); j++) |
|
| 1179 |
colnames.add(partnames.get(j)); |
|
| 1180 |
|
|
| 1181 |
//System.out.println("cols: "+colnames);
|
|
| 1182 |
|
|
| 1183 |
String[] keywords = new String[this.lines.size()]; |
|
| 1184 |
|
|
| 1185 |
int[] freq = new int[this.lines.size()]; |
|
| 1186 |
ArrayList<int[]> partfreqs = new ArrayList<int[]>(partnames.size()); |
|
| 1187 |
if (partnames.size() > 1) |
|
| 1188 |
for (int j = 0; j < partnames.size(); j++) |
|
| 1189 |
partfreqs.add(new int[this.lines.size()]); |
|
| 1190 |
|
|
| 1191 |
for (int i = 0 ; i < lines.size() ; i++) |
|
| 1192 |
{
|
|
| 1193 |
Line ligne = lines.get(i); |
|
| 1194 |
freq[i] = ligne.getFrequency(); |
|
| 1195 |
keywords[i] = ligne.toString(); |
|
| 1196 |
if (partnames.size() > 1) |
|
| 1197 |
for (int j = 0; j < partnames.size(); j++) |
|
| 1198 |
{
|
|
| 1199 |
partfreqs.get(j)[i] = ligne.getFrequency(j); |
|
| 1200 |
} |
|
| 1201 |
|
|
| 1202 |
} |
|
| 1203 |
|
|
| 1204 |
RWorkspace rw = RWorkspace.getRWorkspaceInstance(); |
|
| 1205 |
if (partnames.size() > 1) |
|
| 1206 |
for (int j = 0; j < partnames.size(); j++) |
|
| 1207 |
rw.addVectorToWorkspace("vocpartfreqs"+j, partfreqs.get(j)); //$NON-NLS-1$
|
|
| 1208 |
rw.addVectorToWorkspace("vocfreq", freq); //$NON-NLS-1$
|
|
| 1209 |
rw.addVectorToWorkspace("vockeywords", keywords); //$NON-NLS-1$
|
|
| 1210 |
rw.addVectorToWorkspace("voccolnames", colnames.toArray(new String[colnames.size()])); //$NON-NLS-1$
|
|
| 1211 |
|
|
| 1212 |
int ncol = 1; |
|
| 1213 |
if (partnames.size() > 1) |
|
| 1214 |
ncol += partnames.size(); |
|
| 1215 |
|
|
| 1216 |
int nrow = lines.size(); |
|
| 1217 |
String partscmd = ""; //$NON-NLS-1$ |
|
| 1218 |
if (partnames.size() > 1) |
|
| 1219 |
for (int j = 0; j < partnames.size(); j++) |
|
| 1220 |
partscmd +=", vocpartfreqs"+j; //$NON-NLS-1$ |
|
| 1221 |
|
|
| 1222 |
rw.eval(symbol+ "<- matrix(data = c(vocfreq"+partscmd+"), nrow = "+nrow+", ncol = "+ncol+")"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ |
|
| 1223 |
rw.eval("colnames("+symbol+" ) <- voccolnames"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 1224 |
rw.eval("rownames("+symbol+" ) <- vockeywords"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 1225 |
rw.eval(symbol+ "<- list(data="+symbol+")"); //$NON-NLS-1$ //$NON-NLS-2$ |
|
| 1226 |
|
|
| 1227 |
novoc++; |
|
| 1228 |
return symbol; |
|
| 1229 |
} |
|
| 1153 | 1230 |
} |
Formats disponibles : Unified diff