Révision 3528
| TXM/trunk/org.txm.core/src/java/org/txm/core/results/TXMResult.java (revision 3528) | ||
|---|---|---|
| 1 | 1 |
package org.txm.core.results; |
| 2 | 2 |
|
| 3 |
import java.io.BufferedReader; |
|
| 3 | 4 |
import java.io.File; |
| 4 | 5 |
import java.io.IOException; |
| 5 | 6 |
import java.lang.reflect.Constructor; |
| ... | ... | |
| 1185 | 1186 |
} |
| 1186 | 1187 |
|
| 1187 | 1188 |
/** |
| 1188 |
* |
|
| 1189 |
* load or reload parameters from the preference store node
|
|
| 1189 | 1190 |
* @return |
| 1190 | 1191 |
* @throws Exception |
| 1191 | 1192 |
*/ |
| 1192 |
protected boolean autoLoadParametersFromAnnotations() throws Exception {
|
|
| 1193 |
public boolean autoLoadParametersFromAnnotations() throws Exception {
|
|
| 1193 | 1194 |
return this.autoLoadParametersFromAnnotations(Parameter.COMPUTING) && this.autoLoadParametersFromAnnotations(Parameter.INTERNAL); |
| 1194 | 1195 |
} |
| 1195 | 1196 |
|
| ... | ... | |
| 2076 | 2077 |
} |
| 2077 | 2078 |
clone = this.getClass().getDeclaredConstructor(String.class).newInstance(newNodePath); |
| 2078 | 2079 |
|
| 2079 |
System.out.println("CLONE HAS BEEN COMPUTED? " + clone.hasBeenComputedOnce());
|
|
| 2080 |
System.out.println("CLONE IS DIRTY? " + clone.isDirty());
|
|
| 2080 |
//System.out.println("CLONE HAS BEEN COMPUTED? " + clone.hasBeenComputedOnce());
|
|
| 2081 |
//System.out.println("CLONE IS DIRTY? " + clone.isDirty());
|
|
| 2081 | 2082 |
|
| 2082 | 2083 |
if (all) {
|
| 2083 | 2084 |
for (TXMResult child : this.children) {
|
| ... | ... | |
| 2095 | 2096 |
public boolean importParameters(File parameters) {
|
| 2096 | 2097 |
try {
|
| 2097 | 2098 |
Properties props = new Properties(); |
| 2098 |
props.load(IOUtils.getReader(parameters)); |
|
| 2099 |
BufferedReader r = IOUtils.getReader(parameters); |
|
| 2100 |
props.load(r); |
|
| 2101 |
r.close(); |
|
| 2099 | 2102 |
|
| 2100 | 2103 |
// some tests before... |
| 2101 | 2104 |
String className = props.getProperty(TXMPreferences.CLASS); |
| TXM/trunk/org.txm.libs.deptreeviz/src/org/txm/libs/deptreeviz/UDPrintTree.java (revision 3528) | ||
|---|---|---|
| 8 | 8 |
import java.util.Arrays; |
| 9 | 9 |
import java.util.List; |
| 10 | 10 |
|
| 11 |
|
|
| 12 | 11 |
import io.gitlab.nats.deptreeviz.DepTree; |
| 13 | 12 |
import io.gitlab.nats.deptreeviz.SimpleParse; |
| 14 | 13 |
import io.gitlab.nats.deptreeviz.SimpleWord; |
| ... | ... | |
| 16 | 15 |
public class UDPrintTree {
|
| 17 | 16 |
|
| 18 | 17 |
public static File print(File file, List<String> conll, String[] Tvalues, String[] NTvalues) {
|
| 18 |
|
|
| 19 | 19 |
try {
|
| 20 | 20 |
PrintWriter pw = new PrintWriter(file); |
| 21 | 21 |
|
| ... | ... | |
| 24 | 24 |
List<String> labels = p.getVerticesLabels().get("SYN");
|
| 25 | 25 |
if (Tvalues != null && Tvalues.length == words.size()) {
|
| 26 | 26 |
|
| 27 |
for (int i = 0 ; i < words.size() ; i++) {
|
|
| 27 |
for (int i = 0; i < words.size(); i++) {
|
|
| 28 | 28 |
SimpleWord w = words.get(i); |
| 29 | 29 |
|
| 30 | 30 |
Field fword = SimpleWord.class.getDeclaredField("word");
|
| 31 | 31 |
fword.setAccessible(true); |
| 32 |
fword.set(w, w.getWord()+"/"+Tvalues[i]);
|
|
| 32 |
fword.set(w, w.getWord() + "/" + Tvalues[i]);
|
|
| 33 | 33 |
} |
| 34 | 34 |
} |
| 35 | 35 |
|
| 36 | 36 |
if (NTvalues != null && NTvalues.length == words.size()) {
|
| 37 |
ArrayList<String> tmp = (ArrayList)labels; |
|
| 37 |
ArrayList<String> tmp = (ArrayList) labels;
|
|
| 38 | 38 |
tmp.clear(); |
| 39 |
for (int i = 0 ; i < words.size() ; i++) {
|
|
| 39 |
for (int i = 0; i < words.size(); i++) {
|
|
| 40 | 40 |
tmp.add(NTvalues[i]); |
| 41 | 41 |
} |
| 42 | 42 |
} |
| 43 | 43 |
DepTree<SimpleParse, SimpleWord> dt = new DepTree<SimpleParse, SimpleWord>(p); |
| 44 |
//dt.setShowingReferent(true); |
|
| 45 |
//dt.setShowingCat(true); |
|
| 44 |
// dt.setShowingReferent(true);
|
|
| 45 |
// dt.setShowingCat(true);
|
|
| 46 | 46 |
dt.writeTree(pw); |
| 47 |
//w.close(); |
|
| 47 |
// w.close();
|
|
| 48 | 48 |
return file; |
| 49 | 49 |
} |
| 50 | 50 |
catch (Throwable e) {
|
| ... | ... | |
| 55 | 55 |
} |
| 56 | 56 |
|
| 57 | 57 |
public static void main(String[] args) throws IOException {
|
| 58 |
File file = File.createTempFile("txm", ".svg", new File("/home/mdecorde"));
|
|
| 59 |
|
|
| 60 |
List<String> conll = Arrays.asList( |
|
| 61 |
"1 Ce ce PRON _ Number=Sing|Person=3|PronType=Dem 2 nsubj _ _", |
|
| 62 |
"2 doit devoir VERB _ Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin 0 root _ _", |
|
| 63 |
"3 être être AUX _ VerbForm=Inf 5 cop _ _", |
|
| 64 |
"4 un un DET _ Definite=Ind|Gender=Masc|Number=Sing|PronType=Art 5 det _ _", |
|
| 65 |
"5 général général NOUN _ Gender=Masc|Number=Sing 2 xcomp _ _", |
|
| 66 |
"6 étranger étranger ADJ _ Gender=Masc|Number=Sing 5 amod _ _", |
|
| 67 |
"7 . . PUNCT _ _ 2 punct _ _" |
|
| 68 |
); |
|
| 69 | 58 |
|
| 70 |
|
|
| 71 |
System.out.println("FILE: "+print(file, conll, null, null));
|
|
| 59 |
File file = File.createTempFile("txm", ".svg", new File("/home/mdecorde"));
|
|
| 60 |
|
|
| 61 |
List<String> conll = Arrays.asList( |
|
| 62 |
"1 Ce ce PRON _ Number=Sing|Person=3|PronType=Dem 2 nsubj _ _", |
|
| 63 |
"2 doit devoir VERB _ Mood=Ind|Number=Sing|Person=3|Tense=Pres|VerbForm=Fin 0 root _ _", |
|
| 64 |
"3 être être AUX _ VerbForm=Inf 5 cop _ _", |
|
| 65 |
"4 un un DET _ Definite=Ind|Gender=Masc|Number=Sing|PronType=Art 5 det _ _", |
|
| 66 |
"5 général général NOUN _ Gender=Masc|Number=Sing 2 xcomp _ _", |
|
| 67 |
"6 étranger étranger ADJ _ Gender=Masc|Number=Sing 5 amod _ _", |
|
| 68 |
"7 . . PUNCT _ _ 2 punct _ _"); |
|
| 69 |
|
|
| 70 |
|
|
| 71 |
System.out.println("FILE: " + print(file, conll, null, null));
|
|
| 72 | 72 |
} |
| 73 | 73 |
|
| 74 | 74 |
} |
| TXM/trunk/org.txm.groovy.core/src/groovy/org/txm/macro/pdf/PDF2TXTMacro.groovy (revision 3528) | ||
|---|---|---|
| 1 |
package org.txm.macro.pdf |
|
| 2 |
// STANDARD DECLARATIONS |
|
| 3 |
|
|
| 4 |
import org.kohsuke.args4j.* |
|
| 5 |
import groovy.transform.Field |
|
| 6 |
import org.txm.rcp.swt.widget.parameters.* |
|
| 7 |
import org.apache.pdfbox.Loader |
|
| 8 |
import org.apache.pdfbox.text.PDFTextStripper |
|
| 9 |
|
|
| 10 |
// WARNING THIS MACRO NEEDS THE INSTALLATION OF THE PDFBOX (https://pdfbox.apache.org) Java library: https://dlcdn.apache.org/pdfbox/3.0.0-alpha3/pdfbox-app-3.0.0-alpha3.jar (standalone library) |
|
| 11 |
|
|
| 12 |
// BEGINNING OF PARAMETERS |
|
| 13 |
|
|
| 14 |
@Field @Option(name="input_file", usage=".pdf input file", widget="File", required=true, def="") |
|
| 15 |
def input_file |
|
| 16 |
|
|
| 17 |
@Field @Option(name="input_dir", usage="The directory containing the .pdf files to read", widget="Folder", required=false, def="") |
|
| 18 |
def input_dir |
|
| 19 |
|
|
| 20 |
// Open the parameters input dialog box |
|
| 21 |
if (!ParametersDialog.open(this)) return |
|
| 22 |
|
|
| 23 |
// END OF PARAMETERS |
|
| 24 |
|
|
| 25 |
if (input_dir != null && input_dir.exists() && input_dir.isDirectory()) {
|
|
| 26 |
|
|
| 27 |
nFiles = 0 |
|
| 28 |
input_dir.eachFileMatch(~/.*.pdf/) { f ->
|
|
| 29 |
name = f.getName() |
|
| 30 |
println "Processing $name..." |
|
| 31 |
idx = name.lastIndexOf(".")
|
|
| 32 |
if (idx > 0) name = name.substring(0, idx) |
|
| 33 |
outputFile = new File(f.getParentFile(), name + ".txt") |
|
| 34 |
|
|
| 35 |
pdfFile = f.getAbsolutePath() |
|
| 36 |
if (pdfFile.toUpperCase().endsWith(".PDF")) {
|
|
| 37 |
textFile = pdfFile.substring(0, pdfFile.length() - 3) + "txt" |
|
| 38 |
try {
|
|
| 39 |
outputFile.withWriter("UTF-8") { writer ->
|
|
| 40 |
doc = Loader.loadPDF(f) |
|
| 41 |
strip = new PDFTextStripper() |
|
| 42 |
writer.print strip.getText(doc) |
|
| 43 |
doc.close() |
|
| 44 |
writer.close() |
|
| 45 |
nFiles++ |
|
| 46 |
} |
|
| 47 |
} catch (Exception e) {
|
|
| 48 |
e.printStackTrace() |
|
| 49 |
} |
|
| 50 |
} |
|
| 51 |
} |
|
| 52 |
println "Done processed $nFiles files." |
|
| 53 |
} else if (input_file != null && input_file.exists() && input_file.isFile()) {
|
|
| 54 |
|
|
| 55 |
name = input_file.getName() |
|
| 56 |
idx = name.lastIndexOf(".")
|
|
| 57 |
if (idx > 0) name = name.substring(0, idx) |
|
| 58 |
outputFile = new File(input_file.getParentFile(), name + ".txt") |
|
| 59 |
|
|
| 60 |
pdfFile = input_file.getAbsolutePath() |
|
| 61 |
if (pdfFile.toUpperCase().endsWith(".PDF")) {
|
|
| 62 |
textFile = pdfFile.substring(0, pdfFile.length() - 3) + "txt" |
|
| 63 |
try {
|
|
| 64 |
outputFile.withWriter("UTF-8") { writer ->
|
|
| 65 |
doc = Loader.loadPDF(input_file) |
|
| 66 |
strip = new PDFTextStripper() |
|
| 67 |
writer.print strip.getText(doc) |
|
| 68 |
doc.close() |
|
| 69 |
writer.close() |
|
| 70 |
} |
|
| 71 |
} catch (Exception e) {
|
|
| 72 |
e.printStackTrace() |
|
| 73 |
} |
|
| 74 |
} |
|
| 75 |
} |
|
Formats disponibles : Unified diff