Révision 3187
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/projects/antract/ExporterIdentifiantsSujetsMacro.groovy (revision 3187) | ||
|---|---|---|
| 1 |
// Copyright © 2021 ENS-LYON |
|
| 2 |
// Licensed under the terms of the GNU General Public License version 3 (http://www.gnu.org/licenses/gpl-3.0.html) |
|
| 3 |
// @author mdecorde |
|
| 4 |
|
|
| 5 |
// STANDARD DECLARATIONS |
|
| 6 |
package org.txm.macro.projects.antract |
|
| 7 |
|
|
| 8 |
import org.txm.searchengine.cqp.corpus.CQPCorpus |
|
| 9 |
import org.txm.searchengine.cqp.corpus.query.CQLQuery |
|
| 10 |
import org.txm.searchengine.cqp.CQPSearchEngine |
|
| 11 |
|
|
| 12 |
if (!(corpusViewSelection instanceof CQPCorpus)) {
|
|
| 13 |
println "Erreur: la sélection n'est pas un corpus." |
|
| 14 |
return false |
|
| 15 |
} |
|
| 16 |
|
|
| 17 |
CQPCorpus corpus = corpusViewSelection |
|
| 18 |
corpus.compute() |
|
| 19 |
CQPCorpus parentCorpus = corpus.getMainCorpus() |
|
| 20 |
|
|
| 21 |
String parentCorpusName = parentCorpus.getCqpId() |
|
| 22 |
|
|
| 23 |
def struct_prop = null |
|
| 24 |
def finalValues = [] |
|
| 25 |
def ids = null |
|
| 26 |
|
|
| 27 |
if (parentCorpusName.startsWith("AF-VOIX-OFF-")) {
|
|
| 28 |
|
|
| 29 |
struct_prop = corpus.getStructuralUnit("div").getProperty("id")
|
|
| 30 |
values = struct_prop.getValues(corpus) |
|
| 31 |
for (def id : values) {
|
|
| 32 |
if (id.startsWith("AFE")) {
|
|
| 33 |
finalValues << id |
|
| 34 |
} |
|
| 35 |
} |
|
| 36 |
|
|
| 37 |
} else if (parentCorpusName.startsWith("AFNOTICES") || parentCorpusName.startsWith("AF-NOTICES")) {
|
|
| 38 |
|
|
| 39 |
struct_prop = corpus.getStructuralUnit("notice").getProperty("identifiantdelanotice")
|
|
| 40 |
def values = struct_prop.getValues(corpus) |
|
| 41 |
|
|
| 42 |
def struct_prop2 = corpus.getStructuralUnit("notice").getProperty("typedenotice")
|
|
| 43 |
def qr = corpus.query(new CQLQuery("<notice_typedenotice=\"Notice sujet\">[]"), "TMP", false)
|
|
| 44 |
int[] positions = qr.getStarts() |
|
| 45 |
int[] strucs = CQPSearchEngine.getCqiClient().cpos2Struc(struct_prop2.getQualifiedName(), positions) |
|
| 46 |
String sujetsIDS = CQPSearchEngine.getCqiClient().struc2Str(struct_prop.getQualifiedName(), strucs) |
|
| 47 |
for (String id : values) {
|
|
| 48 |
int idx = sujetsIDS.indexOf(id) |
|
| 49 |
if (idx > 0) {
|
|
| 50 |
finalValues << id // this id is actually in the sujetsIDS |
|
| 51 |
} |
|
| 52 |
} |
|
| 53 |
|
|
| 54 |
} else {
|
|
| 55 |
println "Erreur: le corpus n'est pas un corpus d'AF-VOIX-OFF ni AFNOTICE" |
|
| 56 |
return false |
|
| 57 |
} |
|
| 58 |
|
|
| 59 |
ids = finalValues.sort().join("|")
|
|
| 60 |
|
|
| 61 |
println "Liste des identifiants (${finalValues.size()}) : "+ids
|
|
| 62 |
|
|
| 63 |
monitor.syncExec(new Runnable() {
|
|
| 64 |
public void run() {
|
|
| 65 |
IOClipboard.write(ids) |
|
| 66 |
} |
|
| 67 |
}) |
|
| 68 |
|
|
| 69 |
return ids |
|
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/projects/antract/ExporterLeCorpusVersOkapiMacro.groovy (revision 3187) | ||
|---|---|---|
| 1 |
// Copyright © 2021 ENS-LYON |
|
| 2 |
// Licensed under the terms of the GNU General Public License version 3 (http://www.gnu.org/licenses/gpl-3.0.html) |
|
| 3 |
// @author mdecorde |
|
| 4 |
|
|
| 5 |
// STANDARD DECLARATIONS |
|
| 6 |
package org.txm.macro.projects.antract |
|
| 7 |
|
|
| 8 |
import org.txm.searchengine.cqp.corpus.CQPCorpus |
|
| 9 |
import org.txm.searchengine.cqp.corpus.query.CQLQuery |
|
| 10 |
import org.txm.macro.projects.antract.OkapiSaphirAPI |
|
| 11 |
import org.txm.searchengine.cqp.CQPSearchEngine |
|
| 12 |
|
|
| 13 |
import org.txm.backtomedia.preferences.BackToMediaPreferences |
|
| 14 |
|
|
| 15 |
if (!(corpusViewSelection instanceof CQPCorpus)) {
|
|
| 16 |
println "Erreur: la sélection n'est pas un corpus." |
|
| 17 |
return false |
|
| 18 |
} |
|
| 19 |
|
|
| 20 |
@Field @Option(name="titre_du_nouveau_corpus_okapi", usage="Nom du corpus à créer", widget="String", required=false, def="") |
|
| 21 |
def titre_du_nouveau_corpus_okapi |
|
| 22 |
|
|
| 23 |
if (!ParametersDialog.open(this)) return; |
|
| 24 |
|
|
| 25 |
CQPCorpus corpus = corpusViewSelection |
|
| 26 |
corpus.compute() |
|
| 27 |
CQPCorpus parentCorpus = corpus.getMainCorpus() |
|
| 28 |
|
|
| 29 |
String parentCorpusName = parentCorpus.getCqpId() |
|
| 30 |
|
|
| 31 |
def struct_prop = null |
|
| 32 |
def finalValues = [] |
|
| 33 |
def ids = null |
|
| 34 |
|
|
| 35 |
if (parentCorpusName.startsWith("AF-VOIX-OFF-")) {
|
|
| 36 |
ids = gse.run(ExporterIdentifiantsSujetsMacro, ["args":[:], "corpusViewSelection":corpusViewSelection, "monitor":monitor]) |
|
| 37 |
} else if (parentCorpusName.startsWith("AFNOTICES") || parentCorpusName.startsWith("AF-NOTICES")) {
|
|
| 38 |
ids = gse.run(ExporterIdentifiantsSujetsMacro, ["args":[:], "corpusViewSelection":corpusViewSelection, "monitor":monitor]) |
|
| 39 |
} else {
|
|
| 40 |
println "Erreur: le corpus n'est pas un corpus d'AF-VOIX-OFF ni AFNOTICE" |
|
| 41 |
return false |
|
| 42 |
} |
|
| 43 |
|
|
| 44 |
if (!OkapiSaphirAPI.initializeCredentials(BackToMediaPreferences.MEDIA_AUTH_LOGIN, BackToMediaPreferences.MEDIA_AUTH_PASSWORD, |
|
| 45 |
"OKAPI - Antract", "You must authenticate to access the okapi corpus API", "OKAPI - Antract", monitor)) {
|
|
| 46 |
println "Cancel corpus export" |
|
| 47 |
return null; |
|
| 48 |
} |
|
| 49 |
|
|
| 50 |
String user = System.getProperty(BackToMediaPreferences.MEDIA_AUTH_LOGIN) |
|
| 51 |
String password = System.getProperty(BackToMediaPreferences.MEDIA_AUTH_PASSWORD) |
|
| 52 |
|
|
| 53 |
String sessionID = OkapiSaphirAPI.login(user, password) |
|
| 54 |
if (sessionID.length() > 0) {
|
|
| 55 |
String corpusID = OkapiSaphirAPI.create_corpus(sessionID, titre_du_nouveau_corpus_okapi, ids) |
|
| 56 |
if (sessionID.length() > 0 && sessionID.startsWith("http")) {
|
|
| 57 |
println "New Okapi corpus $titre_du_nouveau_corpus_okapi created: "+corpusID |
|
| 58 |
} else {
|
|
| 59 |
println "Error: Okapi corpus not created: "+corpusID |
|
| 60 |
} |
|
| 61 |
} else {
|
|
| 62 |
println "Error: Okapi session not started. Resetting credentials." |
|
| 63 |
System.setProperty(BackToMediaPreferences.MEDIA_AUTH_LOGIN, "") |
|
| 64 |
System.setProperty(BackToMediaPreferences.MEDIA_AUTH_PASSWORD, "") |
|
| 65 |
} |
|
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/projects/antract/ListerIdentifiantsSujetsMacro.groovy (revision 3187) | ||
|---|---|---|
| 1 |
// Copyright © 2021 ENS-LYON |
|
| 2 |
// Licensed under the terms of the GNU General Public License version 3 (http://www.gnu.org/licenses/gpl-3.0.html) |
|
| 3 |
// @author mdecorde |
|
| 4 |
|
|
| 5 |
// STANDARD DECLARATIONS |
|
| 6 |
package org.txm.macro.projects.antract |
|
| 7 |
|
|
| 8 |
import org.txm.searchengine.cqp.corpus.CQPCorpus |
|
| 9 |
import org.txm.searchengine.cqp.corpus.query.CQLQuery |
|
| 10 |
import org.txm.searchengine.cqp.CQPSearchEngine |
|
| 11 |
|
|
| 12 |
if (!(corpusViewSelection instanceof CQPCorpus)) {
|
|
| 13 |
println "Erreur: la sélection n'est pas un corpus." |
|
| 14 |
return false |
|
| 15 |
} |
|
| 16 |
|
|
| 17 |
CQPCorpus corpus = corpusViewSelection |
|
| 18 |
corpus.compute() |
|
| 19 |
CQPCorpus parentCorpus = corpus.getMainCorpus() |
|
| 20 |
|
|
| 21 |
String parentCorpusName = parentCorpus.getCqpId() |
|
| 22 |
|
|
| 23 |
def struct_prop = null |
|
| 24 |
def finalValues = [] |
|
| 25 |
def ids = null |
|
| 26 |
|
|
| 27 |
if (parentCorpusName.startsWith("AF-VOIX-OFF-")) {
|
|
| 28 |
|
|
| 29 |
struct_prop = corpus.getStructuralUnit("div").getProperty("id")
|
|
| 30 |
values = struct_prop.getValues(corpus) |
|
| 31 |
for (def id : values) {
|
|
| 32 |
if (id.startsWith("AFE")) {
|
|
| 33 |
finalValues << id |
|
| 34 |
} |
|
| 35 |
} |
|
| 36 |
|
|
| 37 |
} else if (parentCorpusName.startsWith("AFNOTICES") || parentCorpusName.startsWith("AF-NOTICES")) {
|
|
| 38 |
|
|
| 39 |
struct_prop = corpus.getStructuralUnit("notice").getProperty("identifiantdelanotice")
|
|
| 40 |
def values = struct_prop.getValues(corpus) |
|
| 41 |
|
|
| 42 |
def struct_prop2 = corpus.getStructuralUnit("notice").getProperty("typedenotice")
|
|
| 43 |
def qr = corpus.query(new CQLQuery("<notice_typedenotice=\"Notice sujet\">[]"), "TMP", false)
|
|
| 44 |
int[] positions = qr.getStarts() |
|
| 45 |
int[] strucs = CQPSearchEngine.getCqiClient().cpos2Struc(struct_prop2.getQualifiedName(), positions) |
|
| 46 |
String sujetsIDS = CQPSearchEngine.getCqiClient().struc2Str(struct_prop.getQualifiedName(), strucs) |
|
| 47 |
for (String id : values) {
|
|
| 48 |
int idx = sujetsIDS.indexOf(id) |
|
| 49 |
if (idx > 0) {
|
|
| 50 |
finalValues << id // this id is actually in the sujetsIDS |
|
| 51 |
} |
|
| 52 |
} |
|
| 53 |
|
|
| 54 |
} else {
|
|
| 55 |
println "Erreur: le corpus n'est pas un corpus d'AF-VOIX-OFF ni AFNOTICE" |
|
| 56 |
return false |
|
| 57 |
} |
|
| 58 |
|
|
| 59 |
ids = finalValues.sort().join("|")
|
|
| 60 |
|
|
| 61 |
println "Liste des identifiants (${finalValues.size()}) : "+ids
|
|
| 62 |
|
|
| 63 |
monitor.syncExec(new Runnable() {
|
|
| 64 |
public void run() {
|
|
| 65 |
IOClipboard.write(ids) |
|
| 66 |
} |
|
| 67 |
}) |
|
| 68 |
|
|
| 69 |
return ids |
|
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/projects/antract/OkapiSaphirAPI.groovy (revision 3187) | ||
|---|---|---|
| 11 | 11 |
import java.net.URLEncoder; |
| 12 | 12 |
import java.util.List; |
| 13 | 13 |
import java.util.Map; |
| 14 |
import javax.ws.rs.core.UriBuilder |
|
| 14 | 15 |
|
| 15 | 16 |
import org.eclipse.swt.widgets.Display; |
| 16 | 17 |
import org.eclipse.swt.widgets.Shell; |
| ... | ... | |
| 25 | 26 |
|
| 26 | 27 |
public static final String baseUrl = "https://okapi.ina.fr/antract/api/saphir/"; |
| 27 | 28 |
public static final String okapiUrl = "https://okapi.ina.fr/antract/html/resource_587884135/resource_996999332.html"; |
| 29 |
|
|
| 28 | 30 |
public static boolean debug = true; |
| 29 | 31 |
|
| 32 |
public static String getDirectAccessURL(String id) {
|
|
| 33 |
return OkapiSaphirAPI.okapiUrl+"?linkedObjectUri="+URLEncoder.encode(id, "UTF-8") |
|
| 34 |
} |
|
| 35 |
|
|
| 30 | 36 |
public static String login(String user, String password) {
|
| 31 | 37 |
URL url1; |
| 32 | 38 |
try {
|
| ... | ... | |
| 46 | 52 |
} |
| 47 | 53 |
|
| 48 | 54 |
String sessionID = getOkapiSessionID(conn) |
| 55 |
String content1 = getContent(conn); |
|
| 56 |
if (debug) {
|
|
| 57 |
println "content="+content1 |
|
| 58 |
println "sessionID="+sessionID |
|
| 59 |
} |
|
| 60 |
|
|
| 49 | 61 |
if (sessionID.length() == 0) {
|
| 50 | 62 |
if (debug) System.out.println("Error: not connected no session ID found");
|
| 51 | 63 |
return ""; |
| 52 | 64 |
} |
| 53 | 65 |
|
| 54 |
String content1 = getContent(conn); |
|
| 55 | 66 |
if (!content1.contains("Welcome "+user)) {
|
| 56 | 67 |
if (debug) System.out.println("Error: not connected, welcome message not found: "+content1);
|
| 57 | 68 |
return ""; |
| ... | ... | |
| 67 | 78 |
} |
| 68 | 79 |
} |
| 69 | 80 |
|
| 81 |
|
|
| 82 |
|
|
| 70 | 83 |
public static String update_corpus(String sessionID, String corpusID, String ids) {
|
| 71 | 84 |
return create_or_update_corpus(sessionID, corpusID, null, ids); |
| 72 | 85 |
} |
| ... | ... | |
| 75 | 88 |
return create_or_update_corpus(sessionID, null, title, ids); |
| 76 | 89 |
} |
| 77 | 90 |
|
| 78 |
public static String create_or_update_corpus(String sessionID, String corpusID, String title, String ids) {
|
|
| 91 |
protected static String create_or_update_corpus(String sessionID, String corpusID, String title, String ids) {
|
|
| 79 | 92 |
URL url2; |
| 80 | 93 |
String action = ""; |
| 81 | 94 |
try {
|
| ... | ... | |
| 93 | 106 |
return ""; |
| 94 | 107 |
} |
| 95 | 108 |
|
| 96 |
url2 = new URL(baseUrl+"create_corpus"); |
|
| 109 |
if (action == "create") {
|
|
| 110 |
url2 = new URL(baseUrl+"create_corpus?title="+URLEncoder.encode(title, "UTF-8")); |
|
| 111 |
} else if (action == "update") {
|
|
| 112 |
url2 = new URL(baseUrl+"create_corpus?uri="+URLEncoder.encode(corpusID, "UTF-8")); |
|
| 113 |
} |
|
| 97 | 114 |
|
| 98 | 115 |
if (debug) System.out.println("Requesting: "+action+" : "+url2);
|
| 99 | 116 |
HttpURLConnection conn2 = (HttpURLConnection) url2.openConnection(); |
| 100 |
conn2.setRequestMethod("POST");
|
|
| 117 |
conn2.setRequestProperty("content-type", "text/plain; charset=UTF-8");
|
|
| 101 | 118 |
conn2.setRequestProperty("session", sessionID);
|
| 102 |
|
|
| 119 |
conn2.setDoOutput(true); |
|
| 120 |
|
|
| 103 | 121 |
if (debug) System.out.println("Writing POST...");
|
| 104 | 122 |
|
| 105 |
conn2.setDoOutput(true); |
|
| 106 | 123 |
DataOutputStream wr = new DataOutputStream(conn2.getOutputStream()); |
| 107 |
String str = "" |
|
| 108 |
if (action == "create") {
|
|
| 109 |
str += "title="+title |
|
| 110 |
} else if (action == "update") {
|
|
| 111 |
str += "uri="+corpusID |
|
| 112 |
} |
|
| 113 |
str += "&content="+ids |
|
| 114 |
if (debug) println "POST="+str |
|
| 115 |
wr.writeBytes(str); |
|
| 124 |
|
|
| 125 |
if (debug) println "POST="+URLEncoder.encode(ids, "UTF-8") |
|
| 126 |
wr.writeBytes(ids); |
|
| 116 | 127 |
wr.flush(); |
| 117 | 128 |
wr.close(); |
| 118 |
|
|
| 119 |
conn2.connect() |
|
| 120 | 129 |
|
| 121 | 130 |
if (debug) System.out.println("getResponseCode...");
|
| 122 | 131 |
int code2 = conn2.getResponseCode(); |
| 123 | 132 |
if (code2 != HttpURLConnection.HTTP_OK) {
|
| 124 |
if (debug) System.out.println("Error: during corpus "+action+" error='"+code2+"' and command="+url2);
|
|
| 133 |
if (debug) System.out.println("Error: during corpus "+action+" error='"+code2+"', command="+url2+", post="+URLEncoder.encode(ids, "UTF-8"));
|
|
| 125 | 134 |
return ""; |
| 126 | 135 |
} |
| 127 | 136 |
if (debug) System.out.println("getContent...");
|
| 128 |
String content2 = getContent(conn2); |
|
| 137 |
String content2 = getContent(conn2).trim(); |
|
| 138 |
if (content2.startsWith("\"") && content2.endsWith("\"")) {
|
|
| 139 |
content2 = content2.substring(1, content2.length() -1) |
|
| 140 |
} |
|
| 129 | 141 |
|
| 130 | 142 |
conn2.disconnect(); |
| 131 | 143 |
|
| ... | ... | |
| 254 | 266 |
} |
| 255 | 267 |
|
| 256 | 268 |
//TODO remove this and call UsernamePasswordDialog.initializeCredentials |
| 257 |
public static boolean initializeCredentials(String userKey, String passwordKey, String target, String title, String details) {
|
|
| 269 |
public static boolean initializeCredentials(Shell shell, String userKey, String passwordKey, String target, String title, String details) {
|
|
| 258 | 270 |
|
| 259 | 271 |
if (System.getProperty(userKey) == null |
| 260 |
|| System.getProperty(passwordKey) == null) {
|
|
| 272 |
|| System.getProperty(passwordKey) == null |
|
| 273 |
|| System.getProperty(userKey).length() == 0 |
|
| 274 |
|| System.getProperty(passwordKey).length() == 0) {
|
|
| 261 | 275 |
|
| 262 |
Shell shell = Display.getCurrent().getActiveShell(); |
|
| 263 |
if (shell == null) {
|
|
| 264 |
System.out.println("Error: could not open crendetial dialog on current Thread.");
|
|
| 265 |
return false; |
|
| 266 |
} |
|
| 267 |
|
|
| 268 | 276 |
UsernamePasswordDialog dialog = new UsernamePasswordDialog(shell, [userKey != null, passwordKey != null ] as boolean[], target); |
| 269 | 277 |
if (userKey != null) dialog.setUsername(System.getProperty(userKey)); |
| 270 | 278 |
dialog.setTitle(title); |
| ... | ... | |
| 283 | 291 |
} |
| 284 | 292 |
|
| 285 | 293 |
//TODO remove this and call UsernamePasswordDialog.initializeCredentials |
| 286 |
public static boolean initializeCredentials(String userKey, String passwordKey, String target, String title, String details, JobHandler monitor) {
|
|
| 294 |
public static boolean initializeCredentials(Shell shell, String userKey, String passwordKey, String target, String title, String details, JobHandler monitor) {
|
|
| 287 | 295 |
|
| 288 | 296 |
final def ret = [false] |
| 289 | 297 |
|
| ... | ... | |
| 291 | 299 |
public void run() {
|
| 292 | 300 |
System.out.println("Opening credential dialog...");
|
| 293 | 301 |
try {
|
| 294 |
ret[0] = initializeCredentials(userKey, passwordKey, target, title, details); |
|
| 302 |
ret[0] = initializeCredentials(shell, userKey, passwordKey, target, title, details);
|
|
| 295 | 303 |
} catch(Throwable t) {
|
| 296 | 304 |
t.printStackTrace(); |
| 297 | 305 |
} |
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/projects/antract/ImporterUnCorpusOkapiMacro.groovy (revision 3187) | ||
|---|---|---|
| 1 |
// Copyright © 2021 MY_INSTITUTION |
|
| 2 |
// Licensed under the terms of the GNU General Public License version 3 (http://www.gnu.org/licenses/gpl-3.0.html) |
|
| 3 |
// @author mdecorde |
|
| 4 |
|
|
| 5 |
// STANDARD DECLARATIONS |
|
| 6 |
package org.txm.macro |
|
| 7 |
|
|
| 8 |
import org.kohsuke.args4j.* |
|
| 9 |
import groovy.transform.Field |
|
| 10 |
import org.txm.rcp.swt.widget.parameters.* |
|
| 11 |
import org.txm.searchengine.cqp.corpus.CQPCorpus |
|
| 12 |
import org.txm.searchengine.cqp.corpus.Subcorpus |
|
| 13 |
import org.txm.searchengine.cqp.corpus.query.CQLQuery |
|
| 14 |
import org.txm.rcp.views.corpora.CorporaView |
|
| 15 |
|
|
| 16 |
// BEGINNING OF PARAMETERS |
|
| 17 |
|
|
| 18 |
@Field @Option(name="name", usage="Nom du sous-corpus à créer", widget="String", required=true, def="") |
|
| 19 |
def name |
|
| 20 |
@Field @Option(name="identifiants", usage="Liste des identifiants de notice séparés par des '|'", widget="Text", required=true, def="") |
|
| 21 |
def identifiants |
|
| 22 |
|
|
| 23 |
if (!(corpusViewSelection instanceof CQPCorpus)) {
|
|
| 24 |
println "Erreur: la sélection n'est pas un corpus." |
|
| 25 |
return false |
|
| 26 |
} |
|
| 27 |
|
|
| 28 |
CQPCorpus corpus = corpusViewSelection |
|
| 29 |
corpus.compute() |
|
| 30 |
CQPCorpus parentCorpus = corpus.getMainCorpus() |
|
| 31 |
|
|
| 32 |
String parentCorpusName = parentCorpus.getCqpId() |
|
| 33 |
|
|
| 34 |
if (parentCorpusName.startsWith("AF-VOIX-OFF-")) {
|
|
| 35 |
|
|
| 36 |
} else if (parentCorpusName.startsWith("AFNOTICES") || parentCorpusName.startsWith("AF-NOTICES")) {
|
|
| 37 |
|
|
| 38 |
} else {
|
|
| 39 |
println "Erreur: le corpus n'est pas un corpus d'AF-VOIX-OFF ni AFNOTICE" |
|
| 40 |
return false |
|
| 41 |
} |
|
| 42 |
|
|
| 43 |
// Open the parameters input dialog box |
|
| 44 |
if (!ParametersDialog.open(this)) return; |
|
| 45 |
|
|
| 46 |
println "Création du sous-corpus avec la liste: $identifiants" |
|
| 47 |
String prop = "div_id" |
|
| 48 |
if (parentCorpusName.startsWith("AFNOTICES") || parentCorpusName.startsWith("AF-NOTICES")) {
|
|
| 49 |
prop = "notice_identifiantdelanotice" |
|
| 50 |
} |
|
| 51 |
|
|
| 52 |
String query = "[_.$prop=\"$identifiants\"] expand to $prop" |
|
| 53 |
println "Query=$query" |
|
| 54 |
Subcorpus sub = parentCorpus.createSubcorpus(new CQLQuery(query), name) |
|
| 55 |
|
|
| 56 |
int s = sub.getSize() |
|
| 57 |
if (s > 0) {
|
|
| 58 |
println "$name subcorpus is created." |
|
| 59 |
monitor.syncExec(new Runnable() {
|
|
| 60 |
public void run() {
|
|
| 61 |
CorporaView.refresh(); |
|
| 62 |
CorporaView.expand(parentCorpus); |
|
| 63 |
} |
|
| 64 |
}); |
|
| 65 |
} else {
|
|
| 66 |
println "Error: the created corpus is empty!" |
|
| 67 |
} |
|
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/projects/antract/ExporterSousCorpusSujetsDansOkapiMacro.groovy (revision 3187) | ||
|---|---|---|
| 1 |
// Copyright © 2021 ENS-LYON |
|
| 2 |
// Licensed under the terms of the GNU General Public License version 3 (http://www.gnu.org/licenses/gpl-3.0.html) |
|
| 3 |
// @author mdecorde |
|
| 4 |
|
|
| 5 |
// STANDARD DECLARATIONS |
|
| 6 |
package org.txm.macro.projects.antract |
|
| 7 |
|
|
| 8 |
import org.txm.searchengine.cqp.corpus.CQPCorpus |
|
| 9 |
import org.txm.searchengine.cqp.corpus.query.CQLQuery |
|
| 10 |
import org.txm.macro.projects.antract.OkapiSaphirAPI |
|
| 11 |
import org.txm.searchengine.cqp.CQPSearchEngine |
|
| 12 |
|
|
| 13 |
import org.txm.backtomedia.preferences.BackToMediaPreferences |
|
| 14 |
|
|
| 15 |
if (!(corpusViewSelection instanceof CQPCorpus)) {
|
|
| 16 |
println "Erreur: la sélection n'est pas un corpus." |
|
| 17 |
return false |
|
| 18 |
} |
|
| 19 |
|
|
| 20 |
@Field @Option(name="titre_du_nouveau_corpus_okapi", usage="Nom du corpus à créer", widget="String", required=false, def="") |
|
| 21 |
def titre_du_nouveau_corpus_okapi |
|
| 22 |
|
|
| 23 |
if (!ParametersDialog.open(this)) return; |
|
| 24 |
|
|
| 25 |
CQPCorpus corpus = corpusViewSelection |
|
| 26 |
corpus.compute() |
|
| 27 |
CQPCorpus parentCorpus = corpus.getMainCorpus() |
|
| 28 |
|
|
| 29 |
String parentCorpusName = parentCorpus.getCqpId() |
|
| 30 |
|
|
| 31 |
def struct_prop = null |
|
| 32 |
def finalValues = [] |
|
| 33 |
def ids = null |
|
| 34 |
|
|
| 35 |
if (parentCorpusName.startsWith("AF-VOIX-OFF-")) {
|
|
| 36 |
ids = gse.run(ListerIdentifiantsSujetsMacro, ["args":[:], "corpusViewSelection":corpusViewSelection, "monitor":monitor]) |
|
| 37 |
} else if (parentCorpusName.startsWith("AFNOTICES") || parentCorpusName.startsWith("AF-NOTICES")) {
|
|
| 38 |
ids = gse.run(ListerIdentifiantsSujetsMacro, ["args":[:], "corpusViewSelection":corpusViewSelection, "monitor":monitor]) |
|
| 39 |
} else {
|
|
| 40 |
println "Erreur: le corpus n'est pas un corpus d'AF-VOIX-OFF ni AFNOTICE" |
|
| 41 |
return false |
|
| 42 |
} |
|
| 43 |
|
|
| 44 |
monitor.syncExec(new Runnable() {
|
|
| 45 |
public void run() {
|
|
| 46 |
if (!OkapiSaphirAPI.initializeCredentials(org.eclipse.swt.widgets.Display.getCurrent().getActiveShell(), BackToMediaPreferences.MEDIA_AUTH_LOGIN, BackToMediaPreferences.MEDIA_AUTH_PASSWORD, "OKAPI - Antract", "You must authenticate to access the okapi corpus API", "OKAPI - Antract", monitor)) {
|
|
| 47 |
println "Cancel corpus export" |
|
| 48 |
return null; |
|
| 49 |
} |
|
| 50 |
} |
|
| 51 |
}); |
|
| 52 |
|
|
| 53 |
|
|
| 54 |
String user = System.getProperty(BackToMediaPreferences.MEDIA_AUTH_LOGIN) |
|
| 55 |
String password = System.getProperty(BackToMediaPreferences.MEDIA_AUTH_PASSWORD) |
|
| 56 |
|
|
| 57 |
String sessionID = OkapiSaphirAPI.login(user, password) |
|
| 58 |
if (sessionID.length() > 0) {
|
|
| 59 |
String corpusID = OkapiSaphirAPI.create_corpus(sessionID, titre_du_nouveau_corpus_okapi, ids) |
|
| 60 |
if (corpusID.length() > 0 && corpusID.startsWith("http")) {
|
|
| 61 |
println "New Okapi corpus $titre_du_nouveau_corpus_okapi created with ID "+corpusID |
|
| 62 |
println "Direct access link: "+OkapiSaphirAPI.getDirectAccessURL(corpusID) |
|
| 63 |
} else {
|
|
| 64 |
println "Error: Okapi corpus not created (return value="+corpusID+")" |
|
| 65 |
} |
|
| 66 |
} else {
|
|
| 67 |
println "Error: Okapi session not started. Resetting credentials." |
|
| 68 |
System.setProperty(BackToMediaPreferences.MEDIA_AUTH_LOGIN, "") |
|
| 69 |
System.setProperty(BackToMediaPreferences.MEDIA_AUTH_PASSWORD, "") |
|
| 70 |
} |
|
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/projects/antract/AjouterDesSujetsAuCorpusOkapiMacro.groovy (revision 3187) | ||
|---|---|---|
| 17 | 17 |
return false |
| 18 | 18 |
} |
| 19 | 19 |
|
| 20 |
@Field @Option(name="liste_des_identifiants", usage="Listes des identifiants sujets à ajouter", widget="String", required=false, def="") |
|
| 21 |
def liste_des_identifiants |
|
| 22 |
|
|
| 20 | 23 |
@Field @Option(name="identifiant_du_corpus_okapi_a_modifier", usage="Identifiant Okapi du corpus à modifier", widget="String", required=false, def="") |
| 21 | 24 |
def identifiant_du_corpus_okapi_a_modifier |
| 22 | 25 |
|
| ... | ... | |
| 30 | 33 |
|
| 31 | 34 |
def struct_prop = null |
| 32 | 35 |
def finalValues = [] |
| 33 |
def ids = null |
|
| 34 | 36 |
|
| 35 |
if (parentCorpusName.startsWith("AF-VOIX-OFF-")) {
|
|
| 36 |
ids = gse.run(ExporterIdentifiantsSujetsMacro, ["args":[:], "corpusViewSelection":corpusViewSelection, "monitor":monitor]) |
|
| 37 |
} else if (parentCorpusName.startsWith("AFNOTICES") || parentCorpusName.startsWith("AF-NOTICES")) {
|
|
| 38 |
ids = gse.run(ExporterIdentifiantsSujetsMacro, ["args":[:], "corpusViewSelection":corpusViewSelection, "monitor":monitor]) |
|
| 39 |
} else {
|
|
| 40 |
println "Erreur: le corpus n'est pas un corpus d'AF-VOIX-OFF ni AFNOTICE" |
|
| 41 |
return false |
|
| 42 |
} |
|
| 43 |
|
|
| 44 | 37 |
// set Okapi user&password |
| 45 | 38 |
if (OkapiSaphirAPI.initializeCredentials(BackToMediaPreferences.MEDIA_AUTH_LOGIN, BackToMediaPreferences.MEDIA_AUTH_PASSWORD, |
| 46 | 39 |
"OKAPI - Antract", "You must authenticate to access the okapi corpus API", "OKAPI - Antract", monitor)) {
|
| ... | ... | |
| 58 | 51 |
// start the Okapi session |
| 59 | 52 |
String sessionID = OkapiSaphirAPI.login(user, password) |
| 60 | 53 |
if (sessionID.length() > 0) {
|
| 61 |
String corpusID = OkapiSaphirAPI.update_corpus(sessionID, identifiant_du_corpus_okapi_a_modifier, ids)
|
|
| 54 |
String corpusID = OkapiSaphirAPI.update_corpus(sessionID, identifiant_du_corpus_okapi_a_modifier, liste_des_identifiants)
|
|
| 62 | 55 |
if (sessionID.length() > 0 && sessionID.startsWith("http")) {
|
| 63 |
println "Okapi corpus $identifiant_du_corpus_okapi_a_modifier updated with "+ids
|
|
| 56 |
println "Okapi corpus $identifiant_du_corpus_okapi_a_modifier updated with "+liste_des_identifiants
|
|
| 64 | 57 |
} else {
|
| 65 | 58 |
println "Error: Okapi corpus not updated: "+sessionID |
| 66 | 59 |
} |
| tmp/org.txm.groovy.core/src/groovy/org/txm/macro/projects/antract/WordProperty2WordMacro.groovy (revision 3187) | ||
|---|---|---|
| 1 |
|
|
| 2 |
package org.txm.macro.annotation |
|
| 3 |
|
|
| 4 |
// STANDARD DECLARATIONS |
|
| 5 |
import org.txm.rcpapplication.swt.widget.parameters.* |
|
| 6 |
import org.txm.searchengine.cqp.corpus.* |
|
| 7 |
import org.txm.xml.DOMIdentityHook |
|
| 8 |
import org.txm.xml.LocalNameHookActivator |
|
| 9 |
import org.txm.xml.XMLProcessor |
|
| 10 |
import org.txm.objects.* |
|
| 11 |
import org.txm.rcp.utils.JobHandler |
|
| 12 |
import org.txm.rcp.views.corpora.CorporaView |
|
| 13 |
import org.eclipse.core.runtime.Status |
|
| 14 |
import org.txm.annotation.kr.core.KRAnnotationEngine |
|
| 15 |
import org.txm.annotation.kr.core.repository.* |
|
| 16 |
import org.txm.annotation.kr.rcp.commands.SaveAnnotations |
|
| 17 |
import org.txm.annotation.kr.rcp.concordance.WordAnnotationToolbar |
|
| 18 |
import org.txm.importer.StaxIdentityParser |
|
| 19 |
|
|
| 20 |
def scriptName = this.class.getSimpleName() |
|
| 21 |
|
|
| 22 |
if (!(corpusViewSelection instanceof MainCorpus)) {
|
|
| 23 |
println "** $scriptName: please select a Corpus in the Corpus view." |
|
| 24 |
return 0 |
|
| 25 |
} |
|
| 26 |
|
|
| 27 |
MainCorpus mcorpus = corpusViewSelection |
|
| 28 |
Project project = mcorpus.getProject() |
|
| 29 |
File txmDir = new File(project.getProjectDirectory(), "txm/"+mcorpus.getName()) |
|
| 30 |
|
|
| 31 |
@Field @Option(name="outputDirectory", usage="Directory of resulting XML-TXM files", widget="Folder", required=true, def="result_directory") |
|
| 32 |
File outputDirectory |
|
| 33 |
|
|
| 34 |
@Field @Option(name="word_property", usage="The word property to move", widget="String", required=true, def="type") |
|
| 35 |
String word_property |
|
| 36 |
|
|
| 37 |
//@Field @Option(name="word_property_value_to_ignore", usage="The annotation type", widget="String", required=true, def="__undef__") |
|
| 38 |
//String word_property_value_to_ignore |
|
| 39 |
|
|
| 40 |
// Open the parameters input dialog box |
|
| 41 |
if (!ParametersDialog.open(this)) return |
|
| 42 |
|
|
| 43 |
if (!outputDirectory.exists()) {
|
|
| 44 |
println "Error: output directory not found: "+outputDirectory |
|
| 45 |
return false |
|
| 46 |
} |
|
| 47 |
|
|
| 48 |
for (Text text : project.getChildren(Text.class)) {
|
|
| 49 |
println "Text: "+text |
|
| 50 |
|
|
| 51 |
File orig = text.getXMLTXMFile() |
|
| 52 |
if (!orig.exists()) {
|
|
| 53 |
println "Warning: no XML-TXM file found for "+text |
|
| 54 |
continue // next ! |
|
| 55 |
} |
|
| 56 |
|
|
| 57 |
File result = new File(outputDirectory, orig.getName()) |
|
| 58 |
|
|
| 59 |
if (false && result.lastModified() > orig.lastModified()) {
|
|
| 60 |
println "Skipping $text: result file is more recent: $result" |
|
| 61 |
continue // next ! |
|
| 62 |
} |
|
| 63 |
|
|
| 64 |
XMLProcessor xp = new XMLProcessor(orig); |
|
| 65 |
LocalNameHookActivator activator = new LocalNameHookActivator("w");
|
|
| 66 |
new DOMIdentityHook("word", activator, xp) {
|
|
| 67 |
@Override |
|
| 68 |
public void processDom() {
|
|
| 69 |
//println dom |
|
| 70 |
use(groovy.xml.dom.DOMCategory) {
|
|
| 71 |
def form = dom.form[0] |
|
| 72 |
def anatype = dom.ana.findAll(){ a -> a['@type'] == "#$word_property"}
|
|
| 73 |
// println form.text()+" -> $anatype" |
|
| 74 |
if (anatype.size() > 0) { // keep the word
|
|
| 75 |
anatype = anatype[0] |
|
| 76 |
def s = form.text() |
|
| 77 |
form.setTextContent(anatype.text()) |
|
| 78 |
anatype['@type'] = "#form" |
|
| 79 |
anatype.setTextContent(s) |
|
| 80 |
} else {
|
|
| 81 |
dom = null; |
|
| 82 |
} |
|
| 83 |
} |
|
| 84 |
} |
|
| 85 |
} |
|
| 86 |
xp.process(result) |
|
| 87 |
} |
|
Formats disponibles : Unified diff