︙ | | | ︙ | |
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
|
s += "sentence: " + this.sSentence0 + "\n";
s += "now: " + this.sSentence + "\n";
for (let dToken of this.lToken) {
s += `#${dToken["i"]}\t${dToken["nStart"]}:${dToken["nEnd"]}\t${dToken["sValue"]}\t${dToken["sType"]}`;
if (dToken.hasOwnProperty("lMorph")) {
s += "\t" + dToken["lMorph"].toString();
}
if (dToken.hasOwnProperty("tags")) {
s += "\t" + dToken["tags"].toString();
}
s += "\n";
}
return s;
}
parse (sCountry="${country_default}", bDebug=false, dOptions=null, bContext=false) {
|
|
|
|
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
|
s += "sentence: " + this.sSentence0 + "\n";
s += "now: " + this.sSentence + "\n";
for (let dToken of this.lToken) {
s += `#${dToken["i"]}\t${dToken["nStart"]}:${dToken["nEnd"]}\t${dToken["sValue"]}\t${dToken["sType"]}`;
if (dToken.hasOwnProperty("lMorph")) {
s += "\t" + dToken["lMorph"].toString();
}
if (dToken.hasOwnProperty("aTags")) {
s += "\t" + dToken["aTags"].toString();
}
s += "\n";
}
return s;
}
parse (sCountry="${country_default}", bDebug=false, dOptions=null, bContext=false) {
|
︙ | | | ︙ | |
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
|
// update <sSentence> and retokenize
this.sSentence = sSentence;
let lNewToken = Array.from(_oTokenizer.genTokens(sSentence, true));
for (let dToken of lNewToken) {
if (this.dTokenPos.gl_get(dToken["nStart"], {}).hasOwnProperty("lMorph")) {
dToken["lMorph"] = this.dTokenPos.get(dToken["nStart"])["lMorph"];
}
if (this.dTokenPos.gl_get(dToken["nStart"], {}).hasOwnProperty("tags")) {
dToken["tags"] = this.dTokenPos.get(dToken["nStart"])["tags"];
}
}
this.lToken = lNewToken;
this.dTokenPos.clear();
for (let dToken of this.lToken) {
if (dToken["sType"] != "INFO") {
this.dTokenPos.set(dToken["nStart"], dToken);
|
|
|
|
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
|
// update <sSentence> and retokenize
this.sSentence = sSentence;
let lNewToken = Array.from(_oTokenizer.genTokens(sSentence, true));
for (let dToken of lNewToken) {
if (this.dTokenPos.gl_get(dToken["nStart"], {}).hasOwnProperty("lMorph")) {
dToken["lMorph"] = this.dTokenPos.get(dToken["nStart"])["lMorph"];
}
if (this.dTokenPos.gl_get(dToken["nStart"], {}).hasOwnProperty("aTags")) {
dToken["aTags"] = this.dTokenPos.get(dToken["nStart"])["aTags"];
}
}
this.lToken = lNewToken;
this.dTokenPos.clear();
for (let dToken of this.lToken) {
if (dToken["sType"] != "INFO") {
this.dTokenPos.set(dToken["nStart"], dToken);
|
︙ | | | ︙ | |
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
|
}
}
}
}
}
}
// token tags
if (dToken.hasOwnProperty("tags") && dNode.hasOwnProperty("<tags>")) {
for (let sTag in dToken["tags"]) {
if (dNode["<tags>"].hasOwnProperty(sTag)) {
if (bDebug) {
console.log(" MATCH: /" + sTag);
}
yield { "iNode1": iNode1, "dNode": dGraph[dNode["<tags>"][sTag]] };
bTokenFound = true;
}
|
|
|
|
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
|
}
}
}
}
}
}
// token tags
if (dToken.hasOwnProperty("aTags") && dNode.hasOwnProperty("<tags>")) {
for (let sTag in dToken["aTags"]) {
if (dNode["<tags>"].hasOwnProperty(sTag)) {
if (bDebug) {
console.log(" MATCH: /" + sTag);
}
yield { "iNode1": iNode1, "dNode": dGraph[dNode["<tags>"][sTag]] };
bTokenFound = true;
}
|
︙ | | | ︙ | |
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
|
}
}
else if (cActionType == "/") {
// Tag
let nTokenStart = (eAct[0] > 0) ? nTokenOffset + eAct[0] : nLastToken + eAct[0];
let nTokenEnd = (eAct[1] > 0) ? nTokenOffset + eAct[1] : nLastToken + eAct[1];
for (let i = nTokenStart; i <= nTokenEnd; i++) {
if (this.lToken[i].hasOwnProperty("tags")) {
this.lToken[i]["tags"].add(...sWhat.split("|"))
} else {
this.lToken[i]["tags"] = new Set(sWhat.split("|"));
}
}
if (bDebug) {
console.log(` TAG: ${sWhat} > [${this.lToken[nTokenStart]["sValue"]}:${this.lToken[nTokenEnd]["sValue"]}]`);
}
if (!this.dTags.has(sWhat)) {
this.dTags.set(sWhat, [nTokenStart, nTokenStart]);
|
|
|
|
|
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
|
}
}
else if (cActionType == "/") {
// Tag
let nTokenStart = (eAct[0] > 0) ? nTokenOffset + eAct[0] : nLastToken + eAct[0];
let nTokenEnd = (eAct[1] > 0) ? nTokenOffset + eAct[1] : nLastToken + eAct[1];
for (let i = nTokenStart; i <= nTokenEnd; i++) {
if (this.lToken[i].hasOwnProperty("aTags")) {
this.lToken[i]["aTags"].add(...sWhat.split("|"))
} else {
this.lToken[i]["aTags"] = new Set(sWhat.split("|"));
}
}
if (bDebug) {
console.log(` TAG: ${sWhat} > [${this.lToken[nTokenStart]["sValue"]}:${this.lToken[nTokenEnd]["sValue"]}]`);
}
if (!this.dTags.has(sWhat)) {
this.dTags.set(sWhat, [nTokenStart, nTokenStart]);
|
︙ | | | ︙ | |
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
|
if (dToken["i"] < dTags.get(sTag)[1]) {
return true;
}
return false;
}
function g_tag (dToken, sTag) {
return dToken.hasOwnProperty("tags") && dToken["tags"].has(sTag);
}
function g_space_between_tokens (dToken1, dToken2, nMin, nMax=null) {
let nSpace = dToken2["nStart"] - dToken1["nEnd"]
if (nSpace < nMin) {
return false;
}
|
|
|
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
|
if (dToken["i"] < dTags.get(sTag)[1]) {
return true;
}
return false;
}
function g_tag (dToken, sTag) {
return dToken.hasOwnProperty("aTags") && dToken["aTags"].has(sTag);
}
function g_space_between_tokens (dToken1, dToken2, nMin, nMax=null) {
let nSpace = dToken2["nStart"] - dToken1["nEnd"]
if (nSpace < nMin) {
return false;
}
|
︙ | | | ︙ | |