@aim-packages/subtitle 0.0.11 → 0.0.13
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/index.cjs.js +13 -13
- package/dist/index.d.ts +11 -13
- package/dist/index.es.js +360 -347
- package/package.json +1 -1
package/dist/index.es.js
CHANGED
|
@@ -5,8 +5,8 @@ function _(n, e = 2) {
|
|
|
5
5
|
return n.toString().padStart(e, "0");
|
|
6
6
|
}
|
|
7
7
|
function b(n) {
|
|
8
|
-
const e = Math.floor(n / 3600), a = Math.floor(n % 3600 / 60), t = n % 3600 % 60, i = Math.floor(t % 1 * 1e3), r = _(e),
|
|
9
|
-
return `${r}:${
|
|
8
|
+
const e = Math.floor(n / 3600), a = Math.floor(n % 3600 / 60), t = n % 3600 % 60, i = Math.floor(t % 1 * 1e3), r = _(e), u = _(a), s = _(Math.floor(t)), d = _(i, 3);
|
|
9
|
+
return `${r}:${u}:${s}.${d}`;
|
|
10
10
|
}
|
|
11
11
|
function H(n) {
|
|
12
12
|
if (!n)
|
|
@@ -50,17 +50,17 @@ function a2(n, e) {
|
|
|
50
50
|
N.index = h, i.length + t.length * 10 + N.text.length + 50 > e && (a.push(t), t = [], i = ""), i += N.text, t.push(N);
|
|
51
51
|
}
|
|
52
52
|
t.length > 0 && a.push(t);
|
|
53
|
-
const r = [],
|
|
53
|
+
const r = [], u = [], s = [], d = [], O = [0];
|
|
54
54
|
for (let h = 0; h < n.length; h++) {
|
|
55
55
|
const N = n[h];
|
|
56
|
-
N.index = h,
|
|
57
|
-
`).length +
|
|
58
|
-
`)),
|
|
59
|
-
`)),
|
|
56
|
+
N.index = h, s.join(`
|
|
57
|
+
`).length + s.length * 10 + N.text.length + 10 > e && (r.push(s.join(`
|
|
58
|
+
`)), u.push(d.join(`
|
|
59
|
+
`)), s.length = 0, d.length = 0, O.push(h)), s.push(N.text), d.push(`[${h}]` + N.text);
|
|
60
60
|
}
|
|
61
|
-
return
|
|
62
|
-
`)),
|
|
63
|
-
`))), { segmentsResult: a, stringResult: r, indexStringResult:
|
|
61
|
+
return s.length > 0 && (r.push(s.join(`
|
|
62
|
+
`)), u.push(d.join(`
|
|
63
|
+
`))), { segmentsResult: a, stringResult: r, indexStringResult: u, indexResult: O };
|
|
64
64
|
}
|
|
65
65
|
function i2(n, e) {
|
|
66
66
|
if (n.length === 0)
|
|
@@ -68,10 +68,23 @@ function i2(n, e) {
|
|
|
68
68
|
const a = [];
|
|
69
69
|
let t = n[0].start, i = n[0].end;
|
|
70
70
|
for (let r = 1; r < n.length; r++) {
|
|
71
|
-
const { start:
|
|
72
|
-
|
|
71
|
+
const { start: u, end: s } = n[r];
|
|
72
|
+
u - i <= e.maxDistance ? i = Number(s) : (a.push({ start: t, end: i }), t = u, i = s);
|
|
73
73
|
}
|
|
74
|
-
|
|
74
|
+
if (a.push({ start: t, end: i }), e.padding) {
|
|
75
|
+
for (let s = 0; s < a.length; s++)
|
|
76
|
+
a[s].start = Math.max(0, a[s].start - e.padding), a[s].end += e.padding, a[s].end = Math.min(a[s].end, n[n.length - 1].end);
|
|
77
|
+
if (a.length === 1)
|
|
78
|
+
return a;
|
|
79
|
+
const r = [];
|
|
80
|
+
let u = a[0];
|
|
81
|
+
for (let s = 1; s < a.length; s++) {
|
|
82
|
+
const d = a[s];
|
|
83
|
+
d.start <= u.end ? u.end = Math.max(u.end, d.end) : (r.push(u), u = d);
|
|
84
|
+
}
|
|
85
|
+
return r.push(u), r;
|
|
86
|
+
}
|
|
87
|
+
return a;
|
|
75
88
|
}
|
|
76
89
|
const Z7 = /* @__PURE__ */ Object.freeze(/* @__PURE__ */ Object.defineProperty({
|
|
77
90
|
__proto__: null,
|
|
@@ -96,15 +109,15 @@ function t2(n) {
|
|
|
96
109
|
"i"
|
|
97
110
|
);
|
|
98
111
|
function a(t = "") {
|
|
99
|
-
return t.split(/[:.]/).map((i, r,
|
|
100
|
-
if (r ===
|
|
112
|
+
return t.split(/[:.]/).map((i, r, u) => {
|
|
113
|
+
if (r === u.length - 1) {
|
|
101
114
|
if (i.length === 1)
|
|
102
115
|
return "." + i + "00";
|
|
103
116
|
if (i.length === 2)
|
|
104
117
|
return "." + i + "0";
|
|
105
118
|
} else if (i.length === 1)
|
|
106
119
|
return (r === 0 ? "0" : ":0") + i;
|
|
107
|
-
return r === 0 ? i : r ===
|
|
120
|
+
return r === 0 ? i : r === u.length - 1 ? "." + i : ":" + i;
|
|
108
121
|
}).join("");
|
|
109
122
|
}
|
|
110
123
|
return `WEBVTT
|
|
@@ -147,12 +160,12 @@ function s2(n = "", e = { propName: {} }) {
|
|
|
147
160
|
const a = [], t = n.split(/(?:\r\n|\r|\n)/gm).filter((h) => h.length > 0), i = t.map((h, N) => t[N + 2] && /\d{2}:\d{2}:\d{2}[,\.]\d{2,3}\s*-->\s*\d{2}:\d{2}:\d{2}[,\.]\d{2,3}/.test(t[N + 2]) ? h + `\r
|
|
148
161
|
` : h).join(`
|
|
149
162
|
`).split(/(?:\r\n|\r|\n)/gm), r = o2(i) + 1;
|
|
150
|
-
let
|
|
163
|
+
let u = 0, s, d, O;
|
|
151
164
|
for (let h = 0; h < r; h++) {
|
|
152
|
-
for (O = {},
|
|
153
|
-
|
|
165
|
+
for (O = {}, d = [], h = r2(i, h), O.id = parseInt(i[h++], 10), s = i[h++].split(/[\t ]*-->[\t ]*/), O[e.propName.start || "start"] = e.timeInText ? s[0] : H(s[0]), u = s[1].indexOf(" "), u !== -1 && (s[1] = s[1].substr(0, u)), O[e.propName.end || "end"] = e.timeInText ? s[1] : H(s[1]); h < r && i[h]; )
|
|
166
|
+
d.push(i[h++]);
|
|
154
167
|
const N = e.propName.text || "text";
|
|
155
|
-
O[N] =
|
|
168
|
+
O[N] = d.join(e.keepMultiLine ? `
|
|
156
169
|
` : " ").replace(/\{(\\[\w]+\(?([\w\d]+,?)+\)?)+\}/gi, ""), O[N] = O[N].replace(/</g, "<").replace(/>/g, ">"), O[N] = O[N].replace(/<(\/?(font|b|u|i|s))((\s+(\w|\w[\w\-]*\w)(\s*=\s*(?:".*?"|'.*?'|[^'">\s]+))?)+\s*|\s*)(\/?)>/gi, "<$1$3$7>"), e.ignoreLineBreaks ? O[N] = O[N].replace(/\\N/gi, "") : O[N] = O[N].replace(/\\N/gi, "<br />"), a.push(O);
|
|
157
170
|
}
|
|
158
171
|
return a;
|
|
@@ -161,28 +174,28 @@ function l2(n, e) {
|
|
|
161
174
|
const a = /<\/?[^>]+(>|$)/g;
|
|
162
175
|
return new Promise((t, i) => {
|
|
163
176
|
let r = {};
|
|
164
|
-
const
|
|
165
|
-
let
|
|
166
|
-
const
|
|
177
|
+
const u = [];
|
|
178
|
+
let s = !1;
|
|
179
|
+
const d = n.split(`
|
|
167
180
|
`);
|
|
168
|
-
|
|
169
|
-
h.replace(a, "") === " " || h.replace(a, "") === "" || (h.indexOf("-->") !== -1 ? (
|
|
181
|
+
d.forEach((h, N) => {
|
|
182
|
+
h.replace(a, "") === " " || h.replace(a, "") === "" || (h.indexOf("-->") !== -1 ? (s = !0, typeof r.start == "number" && r.start >= 0 && u.push(h1(r)), r = {
|
|
170
183
|
start: L1(h.split("-->")[0].trimRight().split(" ").pop()),
|
|
171
184
|
end: L1(h.split("-->")[1].trimLeft().split(" ").shift()),
|
|
172
185
|
part: ""
|
|
173
|
-
}) :
|
|
174
|
-
${h}`, N ===
|
|
186
|
+
}) : s && (u.length !== 0 ? (u[u.length - 1].part.replace(a, ""), h.replace(a, ""), r.part.length === 0 ? r.part = h : r.part = `${r.part}
|
|
187
|
+
${h}`, N === d.length - 1 && u.push(h1(r))) : (r.part = h, u.push(h1(r)), r.part = "")));
|
|
175
188
|
}), r = [];
|
|
176
189
|
const O = /(<([0-9:.>]+)>)/ig;
|
|
177
|
-
|
|
178
|
-
const
|
|
190
|
+
u.forEach((h) => {
|
|
191
|
+
const B = h.part.split().map(function(c) {
|
|
179
192
|
return c.replace(O, function(k) {
|
|
180
193
|
return k.split("").reduce(function(v, y) {
|
|
181
194
|
return `==${k.replace("<", "").replace(">", "")}`;
|
|
182
195
|
}, 0);
|
|
183
196
|
});
|
|
184
197
|
})[0].replace(a, "").split(" "), g = [];
|
|
185
|
-
|
|
198
|
+
B.forEach(function(c) {
|
|
186
199
|
if (c.indexOf("==") > -1) {
|
|
187
200
|
const k = c.split("=="), v = k[0];
|
|
188
201
|
if (k[1], v == "" || v == "##")
|
|
@@ -197,7 +210,7 @@ ${h}`, N === L.length - 1 && B.push(h1(r))) : (r.part = h, B.push(h1(r)), r.part
|
|
|
197
210
|
time: void 0
|
|
198
211
|
});
|
|
199
212
|
}), h.words = g, h.part = h.part.replace(a, "");
|
|
200
|
-
}), t(
|
|
213
|
+
}), t(u);
|
|
201
214
|
});
|
|
202
215
|
}
|
|
203
216
|
function L1(n) {
|
|
@@ -242,41 +255,41 @@ function B2(n) {
|
|
|
242
255
|
}));
|
|
243
256
|
if (n.words.length > 0) {
|
|
244
257
|
const i = n.words;
|
|
245
|
-
let r = 0,
|
|
258
|
+
let r = 0, u = 0;
|
|
246
259
|
for (; r < t.length; ) {
|
|
247
|
-
const
|
|
248
|
-
let
|
|
249
|
-
for (let O =
|
|
260
|
+
const s = t[r];
|
|
261
|
+
let d = s.text;
|
|
262
|
+
for (let O = u; O < i.length; O++) {
|
|
250
263
|
const h = i[O];
|
|
251
|
-
if (h &&
|
|
252
|
-
|
|
264
|
+
if (h && d.startsWith(h.word))
|
|
265
|
+
s.children = s.children || [], s.children.push({
|
|
253
266
|
st: b(h.start),
|
|
254
267
|
et: b(h.end),
|
|
255
268
|
text: h.word
|
|
256
|
-
}),
|
|
257
|
-
else if (h && new RegExp(`^[\\s\\p{P}]*${h.word}`, "u").test(
|
|
258
|
-
|
|
259
|
-
const N =
|
|
269
|
+
}), d = d.substring(h.word.length);
|
|
270
|
+
else if (h && new RegExp(`^[\\s\\p{P}]*${h.word}`, "u").test(d)) {
|
|
271
|
+
s.children = s.children || [];
|
|
272
|
+
const N = d.match(new RegExp(`^[\\s\\p{P}]*${h.word}`, "u"))[0];
|
|
260
273
|
if (/^[,.?…,。?!!]/u.test(N)) {
|
|
261
274
|
const m = ((e = N.match(/^[,.?…,。?!!]+/u)) == null ? void 0 : e[0]) || "", M = N.substring(m.length);
|
|
262
|
-
if (
|
|
263
|
-
const
|
|
264
|
-
|
|
275
|
+
if (s.children.length > 0) {
|
|
276
|
+
const B = s.children[s.children.length - 1];
|
|
277
|
+
B.text += m;
|
|
265
278
|
}
|
|
266
|
-
|
|
279
|
+
s.children.push({
|
|
267
280
|
st: b(h.start),
|
|
268
281
|
et: b(h.end),
|
|
269
282
|
text: M
|
|
270
283
|
});
|
|
271
284
|
} else
|
|
272
|
-
|
|
285
|
+
s.children.push({
|
|
273
286
|
st: b(h.start),
|
|
274
287
|
et: b(h.end),
|
|
275
288
|
text: N
|
|
276
289
|
});
|
|
277
|
-
|
|
290
|
+
d = d.substring(N.length);
|
|
278
291
|
} else {
|
|
279
|
-
|
|
292
|
+
u = O;
|
|
280
293
|
break;
|
|
281
294
|
}
|
|
282
295
|
}
|
|
@@ -311,15 +324,15 @@ const O1 = {
|
|
|
311
324
|
// 宽松模式下,如果VAD第一段的秒数小于该值,则设置为从头开始
|
|
312
325
|
};
|
|
313
326
|
function d2(n) {
|
|
314
|
-
let e, a, t, i, r,
|
|
315
|
-
const h = !!(n && n.vad && n.vad.length > 0), N = "[--end--]", m = "[--Single--]", M = "[--Full--]",
|
|
327
|
+
let e, a, t, i, r, u, s, d, O;
|
|
328
|
+
const h = !!(n && n.vad && n.vad.length > 0), N = "[--end--]", m = "[--Single--]", M = "[--Full--]", B = /\[(\d{2}:\d{2}:\d{2}\.\d{3}) --> (\d{2}:\d{2}:\d{2}\.\d{3})\]/;
|
|
316
329
|
let g = !1, c = !1, k = "";
|
|
317
330
|
return v(), { feed: P, reset: v, end: y };
|
|
318
331
|
function v() {
|
|
319
|
-
e = !0, a = "", t = !1, i = "", r = !1,
|
|
332
|
+
e = !0, a = "", t = !1, i = "", r = !1, u = void 0, s = 0, d = !1, O = 0;
|
|
320
333
|
}
|
|
321
334
|
function y() {
|
|
322
|
-
if (
|
|
335
|
+
if (d) {
|
|
323
336
|
console.log("whisper parser is already end");
|
|
324
337
|
return;
|
|
325
338
|
}
|
|
@@ -329,38 +342,38 @@ function d2(n) {
|
|
|
329
342
|
g || (k += Q), !g && !c && C1(k) && (c = !0), !g && k.length > 50 && (g = !0);
|
|
330
343
|
}
|
|
331
344
|
function P(Q) {
|
|
332
|
-
var
|
|
345
|
+
var l, L, o;
|
|
333
346
|
C(Q);
|
|
334
347
|
const R = [];
|
|
335
|
-
a += Q, e && ((
|
|
348
|
+
a += Q, e && ((l = n == null ? void 0 : n.onStart) == null || l.call(n, {
|
|
336
349
|
type: "event",
|
|
337
350
|
event: "start",
|
|
338
351
|
data: []
|
|
339
352
|
})), e && L2(a) && (a = a.slice(D0.length)), e = !1;
|
|
340
353
|
const F = a.length;
|
|
341
|
-
for (;
|
|
342
|
-
const D = a[
|
|
343
|
-
if (++
|
|
354
|
+
for (; s < F; ) {
|
|
355
|
+
const D = a[s];
|
|
356
|
+
if (++s, r && u && D !== "[" && (u.text += D), D === "[" && (t && i.length && u && (u.text += i), i = "", t = !0, r = !1), t) {
|
|
344
357
|
if (i += D, i === M && (i = ""), i === m && (i = "", h && O++), i === N) {
|
|
345
|
-
|
|
358
|
+
u && R.push(u), d = !0;
|
|
346
359
|
break;
|
|
347
360
|
}
|
|
348
|
-
if (
|
|
349
|
-
const A =
|
|
350
|
-
if (
|
|
361
|
+
if (B.test(i)) {
|
|
362
|
+
const A = B.exec(i), V = A == null ? void 0 : A[1], T = A == null ? void 0 : A[2];
|
|
363
|
+
if (u && R.push(u), V && T) {
|
|
351
364
|
if (h && (n != null && n.vad) && n.vad.length > 0) {
|
|
352
365
|
let X = 0;
|
|
353
366
|
X = n.vad[O].start - O1.padding[0] < 0 ? 0 : n.vad[O].start - O1.padding[0];
|
|
354
367
|
const Y = V === "00:00:00.000";
|
|
355
368
|
let r1 = b(H(V) + X);
|
|
356
|
-
O > 0 && Y && (r1 = b(H(V) + X + O1.padding[0])),
|
|
369
|
+
O > 0 && Y && (r1 = b(H(V) + X + O1.padding[0])), u = {
|
|
357
370
|
st: r1,
|
|
358
371
|
et: b(H(T) + X),
|
|
359
372
|
text: "",
|
|
360
373
|
children: []
|
|
361
374
|
};
|
|
362
375
|
} else
|
|
363
|
-
|
|
376
|
+
u = {
|
|
364
377
|
st: V,
|
|
365
378
|
et: T,
|
|
366
379
|
text: "",
|
|
@@ -371,11 +384,11 @@ function d2(n) {
|
|
|
371
384
|
}
|
|
372
385
|
}
|
|
373
386
|
}
|
|
374
|
-
R.length && ((
|
|
387
|
+
R.length && ((L = n == null ? void 0 : n.onParse) == null || L.call(n, {
|
|
375
388
|
type: "event",
|
|
376
389
|
event: "message",
|
|
377
390
|
data: R
|
|
378
|
-
})),
|
|
391
|
+
})), d && ((o = n == null ? void 0 : n.onEnd) == null || o.call(n, {
|
|
379
392
|
type: "event",
|
|
380
393
|
event: "end",
|
|
381
394
|
data: a
|
|
@@ -387,19 +400,19 @@ function L2(n) {
|
|
|
387
400
|
return D0.every((e, a) => n.charCodeAt(a) === e);
|
|
388
401
|
}
|
|
389
402
|
function h2(n) {
|
|
390
|
-
let e, a, t, i, r,
|
|
403
|
+
let e, a, t, i, r, u, s, d;
|
|
391
404
|
const O = /\[(\d+)\]/, h = "[--end--]";
|
|
392
405
|
return N(), { feed: M, reset: N, end: m };
|
|
393
406
|
function N() {
|
|
394
|
-
e = !0, a = "", t = 0,
|
|
407
|
+
e = !0, a = "", t = 0, d = !1, u = "", i = void 0, r = !1, s = !1;
|
|
395
408
|
}
|
|
396
409
|
function m() {
|
|
397
410
|
M(h);
|
|
398
411
|
}
|
|
399
|
-
function M(
|
|
412
|
+
function M(B) {
|
|
400
413
|
var k, v, y, C, P;
|
|
401
414
|
const g = [];
|
|
402
|
-
a +=
|
|
415
|
+
a += B, e && ((k = n == null ? void 0 : n.onStart) == null || k.call(n, {
|
|
403
416
|
type: "event",
|
|
404
417
|
event: "start",
|
|
405
418
|
data: []
|
|
@@ -407,25 +420,25 @@ function h2(n) {
|
|
|
407
420
|
const c = a.length;
|
|
408
421
|
for (; t < c; ) {
|
|
409
422
|
const Q = a[t];
|
|
410
|
-
if (++t,
|
|
423
|
+
if (++t, s && i && Q !== "[" && (i.text += Q, (v = n == null ? void 0 : n.onProgress) == null || v.call(n, {
|
|
411
424
|
type: "event",
|
|
412
425
|
event: "message",
|
|
413
426
|
data: [i]
|
|
414
|
-
})), Q === "[" && (r &&
|
|
427
|
+
})), Q === "[" && (r && u.length && i && (i.text += u, (y = n == null ? void 0 : n.onProgress) == null || y.call(n, {
|
|
415
428
|
type: "event",
|
|
416
429
|
event: "message",
|
|
417
430
|
data: [i]
|
|
418
|
-
})),
|
|
419
|
-
if (
|
|
420
|
-
i && g.push(i),
|
|
431
|
+
})), u = "", r = !0, s = !1), r) {
|
|
432
|
+
if (u += Q, u === h) {
|
|
433
|
+
i && g.push(i), d = !0;
|
|
421
434
|
break;
|
|
422
435
|
}
|
|
423
|
-
if (O.test(
|
|
424
|
-
const R = O.exec(
|
|
436
|
+
if (O.test(u)) {
|
|
437
|
+
const R = O.exec(u), F = R == null ? void 0 : R[1];
|
|
425
438
|
i && g.push(i), F && (i = {
|
|
426
439
|
index: Number(F),
|
|
427
440
|
text: ""
|
|
428
|
-
},
|
|
441
|
+
}, s = !0, r = !1, u = "");
|
|
429
442
|
}
|
|
430
443
|
}
|
|
431
444
|
}
|
|
@@ -433,7 +446,7 @@ function h2(n) {
|
|
|
433
446
|
type: "event",
|
|
434
447
|
event: "message",
|
|
435
448
|
data: g
|
|
436
|
-
})),
|
|
449
|
+
})), d && ((P = n == null ? void 0 : n.onEnd) == null || P.call(n, {
|
|
437
450
|
type: "event",
|
|
438
451
|
event: "end",
|
|
439
452
|
data: a
|
|
@@ -570,9 +583,9 @@ function V0(n, e, a) {
|
|
|
570
583
|
function u1(n, e, a, t) {
|
|
571
584
|
var i = !a;
|
|
572
585
|
a || (a = {});
|
|
573
|
-
for (var r = -1,
|
|
574
|
-
var
|
|
575
|
-
|
|
586
|
+
for (var r = -1, u = e.length; ++r < u; ) {
|
|
587
|
+
var s = e[r], d = void 0;
|
|
588
|
+
d === void 0 && (d = n[s]), i ? c0(a, s, d) : V0(a, s, d);
|
|
576
589
|
}
|
|
577
590
|
return a;
|
|
578
591
|
}
|
|
@@ -624,14 +637,14 @@ var R0 = typeof exports == "object" && exports && !exports.nodeType && exports,
|
|
|
624
637
|
}
|
|
625
638
|
}(), I1 = w && w.isTypedArray, X3 = I1 ? U1(I1) : T3, U3 = Object.prototype, p3 = U3.hasOwnProperty;
|
|
626
639
|
function X0(n, e) {
|
|
627
|
-
var a = R1(n), t = !a && n3(n), i = !a && !t && T0(n), r = !a && !t && !i && X3(n),
|
|
640
|
+
var a = R1(n), t = !a && n3(n), i = !a && !t && T0(n), r = !a && !t && !i && X3(n), u = a || t || i || r, s = u ? W2(n.length, String) : [], d = s.length;
|
|
628
641
|
for (var O in n)
|
|
629
|
-
(e || p3.call(n, O)) && !(
|
|
642
|
+
(e || p3.call(n, O)) && !(u && // Safari 9 has enumerable `arguments.length` in strict mode.
|
|
630
643
|
(O == "length" || // Node.js 0.10 has enumerable non-index properties on buffers.
|
|
631
644
|
i && (O == "offset" || O == "parent") || // PhantomJS 2 has enumerable non-index properties on typed arrays.
|
|
632
645
|
r && (O == "buffer" || O == "byteLength" || O == "byteOffset") || // Skip index properties.
|
|
633
|
-
z2(O,
|
|
634
|
-
return
|
|
646
|
+
z2(O, d))) && s.push(O);
|
|
647
|
+
return s;
|
|
635
648
|
}
|
|
636
649
|
function U0(n, e) {
|
|
637
650
|
return function(a) {
|
|
@@ -844,8 +857,8 @@ function g4(n, e) {
|
|
|
844
857
|
}
|
|
845
858
|
function c4(n, e) {
|
|
846
859
|
for (var a = -1, t = n == null ? 0 : n.length, i = 0, r = []; ++a < t; ) {
|
|
847
|
-
var
|
|
848
|
-
e(
|
|
860
|
+
var u = n[a];
|
|
861
|
+
e(u, a, n) && (r[i++] = u);
|
|
849
862
|
}
|
|
850
863
|
return r;
|
|
851
864
|
}
|
|
@@ -972,41 +985,41 @@ var t0 = w && w.isSet, m6 = t0 ? U1(t0) : O6, D6 = 1, M6 = 2, A6 = 4, Y0 = "[obj
|
|
|
972
985
|
U[Y0] = U[N6] = U[U6] = U[p6] = U[g6] = U[c6] = U[b6] = U[v6] = U[y6] = U[Z6] = U[f6] = U[F6] = U[k6] = U[j0] = U[Q6] = U[C6] = U[T6] = U[R6] = U[Y6] = U[S6] = U[j6] = U[E6] = !0;
|
|
973
986
|
U[P6] = U[S0] = U[X6] = !1;
|
|
974
987
|
function l1(n, e, a, t, i, r) {
|
|
975
|
-
var
|
|
976
|
-
if (
|
|
977
|
-
return
|
|
988
|
+
var u, s = e & D6, d = e & M6, O = e & A6;
|
|
989
|
+
if (u !== void 0)
|
|
990
|
+
return u;
|
|
978
991
|
if (!t1(n))
|
|
979
992
|
return n;
|
|
980
993
|
var h = R1(n);
|
|
981
994
|
if (h) {
|
|
982
|
-
if (
|
|
983
|
-
return E2(n,
|
|
995
|
+
if (u = f4(n), !s)
|
|
996
|
+
return E2(n, u);
|
|
984
997
|
} else {
|
|
985
998
|
var N = j(n), m = N == S0 || N == V6;
|
|
986
999
|
if (T0(n))
|
|
987
|
-
return g4(n,
|
|
1000
|
+
return g4(n, s);
|
|
988
1001
|
if (N == j0 || N == Y0 || m && !i) {
|
|
989
|
-
if (
|
|
990
|
-
return
|
|
1002
|
+
if (u = d || m ? {} : u6(n), !s)
|
|
1003
|
+
return d ? Q4(n, A4(u, n)) : F4(n, M4(u, n));
|
|
991
1004
|
} else {
|
|
992
1005
|
if (!U[N])
|
|
993
1006
|
return i ? n : {};
|
|
994
|
-
|
|
1007
|
+
u = l6(n, N, s);
|
|
995
1008
|
}
|
|
996
1009
|
}
|
|
997
1010
|
r || (r = new W());
|
|
998
1011
|
var M = r.get(n);
|
|
999
1012
|
if (M)
|
|
1000
1013
|
return M;
|
|
1001
|
-
r.set(n,
|
|
1002
|
-
|
|
1014
|
+
r.set(n, u), m6(n) ? n.forEach(function(c) {
|
|
1015
|
+
u.add(l1(c, e, a, c, n, r));
|
|
1003
1016
|
}) : L6(n) && n.forEach(function(c, k) {
|
|
1004
|
-
|
|
1017
|
+
u.set(k, l1(c, e, a, k, n, r));
|
|
1005
1018
|
});
|
|
1006
|
-
var
|
|
1019
|
+
var B = O ? d ? T4 : C4 : d ? b1 : p1, g = h ? void 0 : B(n);
|
|
1007
1020
|
return H2(g || n, function(c, k) {
|
|
1008
|
-
g && (k = c, c = n[k]), V0(
|
|
1009
|
-
}),
|
|
1021
|
+
g && (k = c, c = n[k]), V0(u, k, l1(c, e, a, k, n, r));
|
|
1022
|
+
}), u;
|
|
1010
1023
|
}
|
|
1011
1024
|
var H6 = 1, G6 = 4;
|
|
1012
1025
|
function E0(n) {
|
|
@@ -1047,27 +1060,27 @@ const G = {
|
|
|
1047
1060
|
function I6(n, e) {
|
|
1048
1061
|
let a = [];
|
|
1049
1062
|
const t = new H0({
|
|
1050
|
-
onHit: (
|
|
1051
|
-
a =
|
|
1063
|
+
onHit: (d) => {
|
|
1064
|
+
a = d;
|
|
1052
1065
|
}
|
|
1053
|
-
}), i = (
|
|
1054
|
-
const O = H(
|
|
1066
|
+
}), i = (d) => d.trim().length === 0, r = (d) => {
|
|
1067
|
+
const O = H(d.st), h = H(d.et);
|
|
1055
1068
|
return O >= h;
|
|
1056
|
-
},
|
|
1069
|
+
}, u = {
|
|
1057
1070
|
emt: [],
|
|
1058
1071
|
ep: [],
|
|
1059
1072
|
zf: [],
|
|
1060
1073
|
punc: [],
|
|
1061
1074
|
em: [],
|
|
1062
1075
|
space: []
|
|
1063
|
-
},
|
|
1064
|
-
return n.forEach((
|
|
1065
|
-
const h = G.punc.test(
|
|
1066
|
-
t.push(
|
|
1076
|
+
}, s = [];
|
|
1077
|
+
return n.forEach((d, O) => {
|
|
1078
|
+
const h = G.punc.test(d.text), N = G.em.test(d.text), m = G.space.test(d.text), M = i(d.text), B = r(d), g = G.ep.test(d.text);
|
|
1079
|
+
t.push(d), d.punc = h ? 1 : 0, h && u.punc.push(O), d.em = N ? 1 : 0, N && u.em.push(O), d.space = m ? 1 : 0, m && (e != null && e.space ? d.text = d.text.replace(G.space, " ").trim() : u.space.push(O)), d.emt = i(d.text) ? 1 : 0, M && (e != null && e.emt || u.emt.push(O)), d.zf = r(d) ? 1 : 0, B && (e != null && e.zf || u.zf.push(O)), d.ep = g ? 1 : 0, g && (e != null && e.ep ? d.text = d.text.replace(G.ep, "") : u.ep.push(O)), !(e != null && e.zf && d.zf) && (e != null && e.emt && d.emt || s.push(d));
|
|
1067
1080
|
}), t.end(), {
|
|
1068
|
-
result:
|
|
1081
|
+
result: u,
|
|
1069
1082
|
repeat: a,
|
|
1070
|
-
segments:
|
|
1083
|
+
segments: s
|
|
1071
1084
|
};
|
|
1072
1085
|
}
|
|
1073
1086
|
var z6 = { TINYLD_CONFIG: "normal" }, J6 = /[,.。,、#%&/\\+*¡!¿?[\]!?;:…„“«»”"“_–—~]/gi, K6 = /[0-9]/g, w6 = /[\uFF10-\uFF19]/g, q6 = /\s\s+/g, W6 = /’/gi, x6 = /[\u0300-\u036f]/g;
|
|
@@ -1100,20 +1113,20 @@ function o7(n) {
|
|
|
1100
1113
|
function w0(n, e, a = !0) {
|
|
1101
1114
|
let t = [], i = a ? " ".repeat(e - 1) + n + " ".repeat(e - 1) : n;
|
|
1102
1115
|
for (let r = 0; r < i.length - (e - 1); r++) {
|
|
1103
|
-
let
|
|
1104
|
-
for (let O = 0; O < e; O++) i[r + O] === " " ?
|
|
1105
|
-
let
|
|
1106
|
-
|
|
1116
|
+
let u = [], s = 0;
|
|
1117
|
+
for (let O = 0; O < e; O++) i[r + O] === " " ? s += 1 : s = 0, !(s > 1) && u.push(i[r + O]);
|
|
1118
|
+
let d = u.join("");
|
|
1119
|
+
d.trim().length > 0 && d.length === e && t.push(d);
|
|
1107
1120
|
}
|
|
1108
1121
|
return t;
|
|
1109
1122
|
}
|
|
1110
1123
|
function s7(n, e, a, t) {
|
|
1111
1124
|
for (let i of I0) {
|
|
1112
1125
|
let r = w0(n, i);
|
|
1113
|
-
for (let
|
|
1114
|
-
if (!a.has(
|
|
1115
|
-
let
|
|
1116
|
-
if (!(t.only.length > 0 && !t.only.includes(
|
|
1126
|
+
for (let u of r) {
|
|
1127
|
+
if (!a.has(u)) continue;
|
|
1128
|
+
let s = k1(e.uniques[u]);
|
|
1129
|
+
if (!(t.only.length > 0 && !t.only.includes(s))) return t.verbose && console.log(`[Pass 1] detectUniqueGrams ${i}-grams - match '${u}' to ${s}`), s;
|
|
1117
1130
|
}
|
|
1118
1131
|
}
|
|
1119
1132
|
return "";
|
|
@@ -1127,33 +1140,33 @@ function l7(n, e, a) {
|
|
|
1127
1140
|
let h = n7(O), N = e.multiples[h];
|
|
1128
1141
|
if (!N) continue;
|
|
1129
1142
|
let m = new Set(Object.keys(N)), M = [];
|
|
1130
|
-
for (let
|
|
1143
|
+
for (let B of r) m.has(B) && (t.set(B, (t.get(B) || 0) + N[B] * h.length / 4), M.push(`${B} = ${N[B] / 1024 * 100}%`));
|
|
1131
1144
|
a.verbose && M.length > 0 && console.log(`Gram '${h}'`, M);
|
|
1132
1145
|
}
|
|
1133
|
-
let
|
|
1134
|
-
|
|
1135
|
-
let
|
|
1136
|
-
return a.verbose && console.log("Result", n,
|
|
1146
|
+
let u = [...t.entries()];
|
|
1147
|
+
u.sort((O, h) => h[1] - O[1]);
|
|
1148
|
+
let s = Math.max(...u.map((O) => O[1])) || 1, d = u.slice(0, 8).map((O) => ({ lang: k1(O[0]), accuracy: 1 - o0((s - O[1]) / s), score: o0(O[1]) }));
|
|
1149
|
+
return a.verbose && console.log("Result", n, d), d;
|
|
1137
1150
|
}
|
|
1138
1151
|
function u7(n, e, a, t) {
|
|
1139
1152
|
let i = r7(n);
|
|
1140
1153
|
i = i.map((O) => $6(O)).filter((O) => !!O), i.sort((O, h) => h.length - O.length), i = i.slice(0, 7), e.verbose && console.log("Analize chunks", i);
|
|
1141
|
-
let r = 0,
|
|
1154
|
+
let r = 0, u = {};
|
|
1142
1155
|
for (let O of i) {
|
|
1143
1156
|
let h = s7(O, a, t, e);
|
|
1144
1157
|
if (h) {
|
|
1145
|
-
|
|
1158
|
+
u[h] = (u[h] || 0) + 1 * O.length, r += O.length;
|
|
1146
1159
|
continue;
|
|
1147
1160
|
}
|
|
1148
1161
|
let N = o7(O);
|
|
1149
1162
|
for (let m of N) l7(m, a, e).forEach((M) => {
|
|
1150
|
-
|
|
1163
|
+
u[M.lang] = (u[M.lang] || 0) + M.accuracy;
|
|
1151
1164
|
}), r += m.length;
|
|
1152
1165
|
}
|
|
1153
|
-
let
|
|
1154
|
-
|
|
1155
|
-
let
|
|
1156
|
-
return e.verbose && console.log("Merge Results",
|
|
1166
|
+
let s = Object.entries(u).filter((O) => O[1] > 0);
|
|
1167
|
+
s.sort((O, h) => h[1] - O[1]);
|
|
1168
|
+
let d = s.map((O) => ({ lang: O[0], accuracy: O[1] / r }));
|
|
1169
|
+
return e.verbose && console.log("Merge Results", d), d;
|
|
1157
1170
|
}
|
|
1158
1171
|
var l0 = K0, B7 = { uniques: Object.fromEntries(Object.entries(l0.uniques).map((n) => [n[0], s0[parseInt(n[1].toString(), 36)]])), multiples: Object.fromEntries(Object.entries(l0.multiples).map((n) => {
|
|
1159
1172
|
var a;
|
|
@@ -1331,8 +1344,8 @@ function M7() {
|
|
|
1331
1344
|
return ~n.indexOf(r);
|
|
1332
1345
|
}, Z.isTimeAbbreviation = function(i, r) {
|
|
1333
1346
|
if (i === "a.m." || i === "p.m.") {
|
|
1334
|
-
var
|
|
1335
|
-
if (
|
|
1347
|
+
var u = r.replace(/\W+/g, "").slice(-3).toLowerCase();
|
|
1348
|
+
if (u === "day")
|
|
1336
1349
|
return !0;
|
|
1337
1350
|
}
|
|
1338
1351
|
return !1;
|
|
@@ -1345,10 +1358,10 @@ function M7() {
|
|
|
1345
1358
|
if (r.length > 0) {
|
|
1346
1359
|
if (i < 5 && r[0].length < 6 && a(r[0]))
|
|
1347
1360
|
return !0;
|
|
1348
|
-
var
|
|
1349
|
-
return /[A-Z]/.test(
|
|
1361
|
+
var u = r.filter(function(s) {
|
|
1362
|
+
return /[A-Z]/.test(s.charAt(0));
|
|
1350
1363
|
});
|
|
1351
|
-
return
|
|
1364
|
+
return u.length >= 3;
|
|
1352
1365
|
}
|
|
1353
1366
|
return !1;
|
|
1354
1367
|
};
|
|
@@ -1362,8 +1375,8 @@ function M7() {
|
|
|
1362
1375
|
}, Z.isConcatenated = function(i) {
|
|
1363
1376
|
var r = 0;
|
|
1364
1377
|
if ((r = i.indexOf(".")) > -1 || (r = i.indexOf("!")) > -1 || (r = i.indexOf("?")) > -1) {
|
|
1365
|
-
var
|
|
1366
|
-
if (
|
|
1378
|
+
var u = i.charAt(r + 1);
|
|
1379
|
+
if (u.match(/[a-zA-Z].*/))
|
|
1367
1380
|
return [i.slice(0, r), i.slice(r + 1)];
|
|
1368
1381
|
}
|
|
1369
1382
|
return !1;
|
|
@@ -1375,11 +1388,11 @@ var L0;
|
|
|
1375
1388
|
function A7() {
|
|
1376
1389
|
if (L0) return M1;
|
|
1377
1390
|
L0 = 1;
|
|
1378
|
-
var n = m7(), e = D7(), a = M7(), t = " @~@ ", i = t.trim(), r = new RegExp("\\S", ""),
|
|
1379
|
-
return M1.sentences = function(
|
|
1380
|
-
if (!
|
|
1391
|
+
var n = m7(), e = D7(), a = M7(), t = " @~@ ", i = t.trim(), r = new RegExp("\\S", ""), u = new RegExp("\\n+|[-#=_+*]{4,}", "g"), s = new RegExp("\\S+|\\n", "g");
|
|
1392
|
+
return M1.sentences = function(d, O) {
|
|
1393
|
+
if (!d || typeof d != "string" || !d.length)
|
|
1381
1394
|
return [];
|
|
1382
|
-
if (!r.test(
|
|
1395
|
+
if (!r.test(d))
|
|
1383
1396
|
return [];
|
|
1384
1397
|
var h = {
|
|
1385
1398
|
newline_boundaries: !1,
|
|
@@ -1395,49 +1408,49 @@ function A7() {
|
|
|
1395
1408
|
else
|
|
1396
1409
|
for (var N in O)
|
|
1397
1410
|
h[N] = O[N];
|
|
1398
|
-
if (a.setAbbreviations(h.abbreviations), h.newline_boundaries && (
|
|
1411
|
+
if (a.setAbbreviations(h.abbreviations), h.newline_boundaries && (d = d.replace(u, t)), h.html_boundaries) {
|
|
1399
1412
|
var m = "(<br\\s*\\/?>|<\\/(" + h.html_boundaries_tags.join("|") + ")>)", M = new RegExp(m, "g");
|
|
1400
|
-
|
|
1413
|
+
d = d.replace(M, "$1" + t);
|
|
1401
1414
|
}
|
|
1402
|
-
(h.sanitize || h.allowed_tags) && (h.allowed_tags || (h.allowed_tags = [""]),
|
|
1403
|
-
var
|
|
1404
|
-
h.preserve_whitespace ? (g =
|
|
1405
|
-
return
|
|
1406
|
-
})) :
|
|
1415
|
+
(h.sanitize || h.allowed_tags) && (h.allowed_tags || (h.allowed_tags = [""]), d = n(d, { allowedTags: h.allowed_tags }));
|
|
1416
|
+
var B, g;
|
|
1417
|
+
h.preserve_whitespace ? (g = d.split(/(<br\s*\/?>|\S+|\n+)/), B = g.filter(function(F, l) {
|
|
1418
|
+
return l % 2;
|
|
1419
|
+
})) : B = d.trim().match(s);
|
|
1407
1420
|
var c = 0, k = 0, v = [], y = [], C = [];
|
|
1408
|
-
if (!
|
|
1421
|
+
if (!B || !B.length)
|
|
1409
1422
|
return [];
|
|
1410
|
-
for (var P = 0, Q =
|
|
1411
|
-
if (c++, C.push(
|
|
1412
|
-
(h.newline_boundaries || h.html_boundaries) &&
|
|
1423
|
+
for (var P = 0, Q = B.length; P < Q; P++) {
|
|
1424
|
+
if (c++, C.push(B[P]), ~B[P].indexOf(",") && (c = 0), a.isBoundaryChar(B[P]) || e.endsWithChar(B[P], "?!") || B[P] === i) {
|
|
1425
|
+
(h.newline_boundaries || h.html_boundaries) && B[P] === i && C.pop(), y.push(C), c = 0, C = [];
|
|
1413
1426
|
continue;
|
|
1414
1427
|
}
|
|
1415
|
-
if ((e.endsWithChar(
|
|
1428
|
+
if ((e.endsWithChar(B[P], '"') || e.endsWithChar(B[P], "”")) && (B[P] = B[P].slice(0, -1)), e.endsWithChar(B[P], ".")) {
|
|
1416
1429
|
if (P + 1 < Q) {
|
|
1417
|
-
if (
|
|
1430
|
+
if (B[P].length === 2 && isNaN(B[P].charAt(0)) || a.isCommonAbbreviation(B[P]))
|
|
1418
1431
|
continue;
|
|
1419
|
-
if (a.isSentenceStarter(
|
|
1420
|
-
if (a.isTimeAbbreviation(
|
|
1432
|
+
if (a.isSentenceStarter(B[P + 1])) {
|
|
1433
|
+
if (a.isTimeAbbreviation(B[P], B[P + 1]) || a.isNameAbbreviation(c, B.slice(P, 6)) || a.isNumber(B[P + 1]) && a.isCustomAbbreviation(B[P]))
|
|
1421
1434
|
continue;
|
|
1422
|
-
} else if (e.endsWith(
|
|
1435
|
+
} else if (e.endsWith(B[P], "..") || a.isDottedAbbreviation(B[P]) || a.isNameAbbreviation(c, B.slice(P, 5)))
|
|
1423
1436
|
continue;
|
|
1424
1437
|
}
|
|
1425
1438
|
y.push(C), C = [], c = 0;
|
|
1426
1439
|
continue;
|
|
1427
1440
|
}
|
|
1428
|
-
(k =
|
|
1441
|
+
(k = B[P].indexOf(".")) > -1 && (a.isNumber(B[P], k) || a.isDottedAbbreviation(B[P]) || a.isURL(B[P]) || a.isPhoneNr(B[P])) || (v = a.isConcatenated(B[P])) && (C.pop(), C.push(v[0]), y.push(C), C = [], c = 0, C.push(v[1]));
|
|
1429
1442
|
}
|
|
1430
1443
|
C.length && y.push(C), y = y.filter(function(F) {
|
|
1431
1444
|
return F.length > 0;
|
|
1432
1445
|
});
|
|
1433
|
-
var R = y.slice(1).reduce(function(F,
|
|
1434
|
-
var
|
|
1435
|
-
return
|
|
1446
|
+
var R = y.slice(1).reduce(function(F, l) {
|
|
1447
|
+
var L = F[F.length - 1];
|
|
1448
|
+
return L.length === 1 && /^.{1,2}[.]$/.test(L[0]) && !/[.]/.test(l[0]) ? (F.pop(), F.push(L.concat(l)), F) : (F.push(l), F);
|
|
1436
1449
|
}, [y[0]]);
|
|
1437
|
-
return R.map(function(F,
|
|
1450
|
+
return R.map(function(F, l) {
|
|
1438
1451
|
if (h.preserve_whitespace && !h.newline_boundaries && !h.html_boundaries) {
|
|
1439
|
-
var
|
|
1440
|
-
return
|
|
1452
|
+
var L = F.length * 2;
|
|
1453
|
+
return l === 0 && (L += 1), g.splice(0, L).join("");
|
|
1441
1454
|
}
|
|
1442
1455
|
return F.join(" ");
|
|
1443
1456
|
});
|
|
@@ -1455,26 +1468,26 @@ function c7() {
|
|
|
1455
1468
|
i.findCharIndex = function(m, M) {
|
|
1456
1469
|
if (M >= m.length)
|
|
1457
1470
|
return -1;
|
|
1458
|
-
if (!
|
|
1471
|
+
if (!d(m))
|
|
1459
1472
|
return M;
|
|
1460
|
-
for (var
|
|
1473
|
+
for (var B = O(), g = 0; B.exec(m) !== null && !(B.lastIndex > M); )
|
|
1461
1474
|
g++;
|
|
1462
1475
|
return g;
|
|
1463
1476
|
}, i.findByteIndex = function(m, M) {
|
|
1464
|
-
return M >= this.length(m) ? -1 :
|
|
1477
|
+
return M >= this.length(m) ? -1 : s(m, O(), M);
|
|
1465
1478
|
}, i.charAt = function(m, M) {
|
|
1466
|
-
var
|
|
1467
|
-
if (
|
|
1479
|
+
var B = this.findByteIndex(m, M);
|
|
1480
|
+
if (B < 0 || B >= m.length)
|
|
1468
1481
|
return "";
|
|
1469
|
-
var g = m.slice(
|
|
1482
|
+
var g = m.slice(B, B + 8), c = N.exec(g);
|
|
1470
1483
|
return c === null ? g[0] : c[0];
|
|
1471
1484
|
}, i.charCodeAt = function(m, M) {
|
|
1472
|
-
var
|
|
1473
|
-
if (
|
|
1485
|
+
var B = u(m, M);
|
|
1486
|
+
if (B < 0)
|
|
1474
1487
|
return NaN;
|
|
1475
|
-
var g = m.charCodeAt(
|
|
1488
|
+
var g = m.charCodeAt(B);
|
|
1476
1489
|
if (55296 <= g && g <= 56319) {
|
|
1477
|
-
var c = g, k = m.charCodeAt(
|
|
1490
|
+
var c = g, k = m.charCodeAt(B + 1);
|
|
1478
1491
|
return (c - 55296) * 1024 + (k - 56320) + 65536;
|
|
1479
1492
|
}
|
|
1480
1493
|
return g;
|
|
@@ -1483,92 +1496,92 @@ function c7() {
|
|
|
1483
1496
|
55296 + (m >> 10),
|
|
1484
1497
|
56320 + (m & 1023)
|
|
1485
1498
|
)) : String.fromCharCode(m);
|
|
1486
|
-
}, i.indexOf = function(m, M,
|
|
1487
|
-
(typeof
|
|
1488
|
-
var g = this.findByteIndex(m,
|
|
1499
|
+
}, i.indexOf = function(m, M, B) {
|
|
1500
|
+
(typeof B > "u" || B === null) && (B = 0);
|
|
1501
|
+
var g = this.findByteIndex(m, B), c = m.indexOf(M, g);
|
|
1489
1502
|
return c < 0 ? -1 : this.findCharIndex(m, c);
|
|
1490
|
-
}, i.lastIndexOf = function(m, M,
|
|
1503
|
+
}, i.lastIndexOf = function(m, M, B) {
|
|
1491
1504
|
var g;
|
|
1492
|
-
if (typeof
|
|
1505
|
+
if (typeof B > "u" || B === null)
|
|
1493
1506
|
g = m.lastIndexOf(M);
|
|
1494
1507
|
else {
|
|
1495
|
-
var c = this.findByteIndex(m,
|
|
1508
|
+
var c = this.findByteIndex(m, B);
|
|
1496
1509
|
g = m.lastIndexOf(M, c);
|
|
1497
1510
|
}
|
|
1498
1511
|
return g < 0 ? -1 : this.findCharIndex(m, g);
|
|
1499
|
-
}, i.slice = function(m, M,
|
|
1512
|
+
}, i.slice = function(m, M, B) {
|
|
1500
1513
|
var g = this.findByteIndex(m, M), c;
|
|
1501
|
-
return g < 0 && (g = m.length), typeof
|
|
1502
|
-
}, i.substr = function(m, M,
|
|
1503
|
-
return M < 0 && (M = this.length(m) + M), typeof
|
|
1514
|
+
return g < 0 && (g = m.length), typeof B > "u" || B === null ? c = m.length : (c = this.findByteIndex(m, B), c < 0 && (c = m.length)), m.slice(g, c);
|
|
1515
|
+
}, i.substr = function(m, M, B) {
|
|
1516
|
+
return M < 0 && (M = this.length(m) + M), typeof B > "u" || B === null ? this.slice(m, M) : this.slice(m, M, M + B);
|
|
1504
1517
|
}, i.substring = i.slice, i.length = function(m) {
|
|
1505
1518
|
return this.findCharIndex(m, m.length - 1) + 1;
|
|
1506
1519
|
}, i.stringToCodePoints = function(m) {
|
|
1507
|
-
for (var M = [],
|
|
1508
|
-
var g = this.charCodeAt(m,
|
|
1520
|
+
for (var M = [], B = 0; B < m.length; B++) {
|
|
1521
|
+
var g = this.charCodeAt(m, B);
|
|
1509
1522
|
if (!g)
|
|
1510
1523
|
break;
|
|
1511
1524
|
M.push(g);
|
|
1512
1525
|
}
|
|
1513
1526
|
return M;
|
|
1514
1527
|
}, i.codePointsToString = function(m) {
|
|
1515
|
-
for (var M = [],
|
|
1516
|
-
M.push(this.fromCharCode(m[
|
|
1528
|
+
for (var M = [], B = 0; B < m.length; B++)
|
|
1529
|
+
M.push(this.fromCharCode(m[B]));
|
|
1517
1530
|
return M.join("");
|
|
1518
1531
|
}, i.stringToBytes = function(m) {
|
|
1519
|
-
for (var M = [],
|
|
1520
|
-
for (var g = m.charCodeAt(
|
|
1532
|
+
for (var M = [], B = 0; B < m.length; B++) {
|
|
1533
|
+
for (var g = m.charCodeAt(B), c = []; g > 0; )
|
|
1521
1534
|
c.push(g & 255), g >>= 8;
|
|
1522
1535
|
c.length == 1 && c.push(0), M = M.concat(c.reverse());
|
|
1523
1536
|
}
|
|
1524
1537
|
return M;
|
|
1525
1538
|
}, i.bytesToString = function(m) {
|
|
1526
|
-
for (var M = [],
|
|
1527
|
-
var g = m[
|
|
1539
|
+
for (var M = [], B = 0; B < m.length; B += 2) {
|
|
1540
|
+
var g = m[B], c = m[B + 1], k = g << 8 | c;
|
|
1528
1541
|
M.push(String.fromCharCode(k));
|
|
1529
1542
|
}
|
|
1530
1543
|
return M.join("");
|
|
1531
1544
|
}, i.stringToCharArray = function(m) {
|
|
1532
|
-
var M = [],
|
|
1545
|
+
var M = [], B = O();
|
|
1533
1546
|
do {
|
|
1534
|
-
var g =
|
|
1547
|
+
var g = B.exec(m);
|
|
1535
1548
|
if (g === null)
|
|
1536
1549
|
break;
|
|
1537
1550
|
M.push(g[0]);
|
|
1538
1551
|
} while (g !== null);
|
|
1539
1552
|
return M;
|
|
1540
1553
|
};
|
|
1541
|
-
function
|
|
1542
|
-
return
|
|
1554
|
+
function u(m, M) {
|
|
1555
|
+
return s(m, new RegExp(h.source, "g"), M);
|
|
1543
1556
|
}
|
|
1544
|
-
function
|
|
1545
|
-
if (!
|
|
1546
|
-
return
|
|
1557
|
+
function s(m, M, B) {
|
|
1558
|
+
if (!d(m))
|
|
1559
|
+
return B;
|
|
1547
1560
|
for (var g = 0, c = 0; ; ) {
|
|
1548
1561
|
var k = M.exec(m), v;
|
|
1549
|
-
for (k ? v = k.index : v = m.length; c <
|
|
1562
|
+
for (k ? v = k.index : v = m.length; c < B; ) {
|
|
1550
1563
|
if (g == v) {
|
|
1551
|
-
c <
|
|
1564
|
+
c < B && (c++, k ? g += k[0].length : g++);
|
|
1552
1565
|
break;
|
|
1553
1566
|
}
|
|
1554
1567
|
g++, c++;
|
|
1555
1568
|
}
|
|
1556
|
-
if (c ==
|
|
1569
|
+
if (c == B)
|
|
1557
1570
|
break;
|
|
1558
1571
|
if (g >= m.length || !k)
|
|
1559
1572
|
return -1;
|
|
1560
1573
|
}
|
|
1561
1574
|
return g;
|
|
1562
1575
|
}
|
|
1563
|
-
function
|
|
1576
|
+
function d(m) {
|
|
1564
1577
|
return N.test(m);
|
|
1565
1578
|
}
|
|
1566
1579
|
function O(m, M) {
|
|
1567
1580
|
m == null && (m = ["[^]"]), M == null && (M = "g");
|
|
1568
|
-
var
|
|
1581
|
+
var B = [];
|
|
1569
1582
|
return r.forEach(function(g) {
|
|
1570
|
-
|
|
1571
|
-
}),
|
|
1583
|
+
B.push(g.source);
|
|
1584
|
+
}), B.push(h.source), B = B.concat(m), new RegExp(B.join("|"), M);
|
|
1572
1585
|
}
|
|
1573
1586
|
var h = /[\uD800-\uDBFF][\uDC00-\uDFFF]/, N = O([], "");
|
|
1574
1587
|
}
|
|
@@ -1586,35 +1599,35 @@ function P7() {
|
|
|
1586
1599
|
Object.defineProperty(e, "__esModule", {
|
|
1587
1600
|
value: !0
|
|
1588
1601
|
});
|
|
1589
|
-
function t(
|
|
1590
|
-
if (Array.isArray(
|
|
1591
|
-
for (var
|
|
1592
|
-
o[
|
|
1602
|
+
function t(l) {
|
|
1603
|
+
if (Array.isArray(l)) {
|
|
1604
|
+
for (var L = 0, o = Array(l.length); L < l.length; L++)
|
|
1605
|
+
o[L] = l[L];
|
|
1593
1606
|
return o;
|
|
1594
1607
|
} else
|
|
1595
|
-
return Array.from(
|
|
1608
|
+
return Array.from(l);
|
|
1596
1609
|
}
|
|
1597
|
-
function i(
|
|
1598
|
-
if (!(
|
|
1610
|
+
function i(l, L) {
|
|
1611
|
+
if (!(l instanceof L))
|
|
1599
1612
|
throw new TypeError("Cannot call a class as a function");
|
|
1600
1613
|
}
|
|
1601
1614
|
var r = /* @__PURE__ */ function() {
|
|
1602
|
-
function
|
|
1615
|
+
function l(L, o) {
|
|
1603
1616
|
for (var D = 0; D < o.length; D++) {
|
|
1604
1617
|
var A = o[D];
|
|
1605
|
-
A.enumerable = A.enumerable || !1, A.configurable = !0, "value" in A && (A.writable = !0), Object.defineProperty(
|
|
1618
|
+
A.enumerable = A.enumerable || !1, A.configurable = !0, "value" in A && (A.writable = !0), Object.defineProperty(L, A.key, A);
|
|
1606
1619
|
}
|
|
1607
1620
|
}
|
|
1608
|
-
return function(
|
|
1609
|
-
return o &&
|
|
1621
|
+
return function(L, o, D) {
|
|
1622
|
+
return o && l(L.prototype, o), D && l(L, D), L;
|
|
1610
1623
|
};
|
|
1611
1624
|
}();
|
|
1612
1625
|
a == null && typeof window < "u" && window !== null && (a = window.UtfString);
|
|
1613
|
-
var
|
|
1614
|
-
function
|
|
1615
|
-
i(this,
|
|
1626
|
+
var u = e.BreakIterator = function() {
|
|
1627
|
+
function l(L) {
|
|
1628
|
+
i(this, l), this.suppressions = L;
|
|
1616
1629
|
}
|
|
1617
|
-
return r(
|
|
1630
|
+
return r(l, [{
|
|
1618
1631
|
key: "eachSentence",
|
|
1619
1632
|
value: function(o, D) {
|
|
1620
1633
|
var A = this.ruleSetFor("sentence");
|
|
@@ -1662,12 +1675,12 @@ function P7() {
|
|
|
1662
1675
|
value: function() {
|
|
1663
1676
|
return this.ruleSetCache === void 0 && (this.ruleSetCache = {}), this.ruleSetCache;
|
|
1664
1677
|
}
|
|
1665
|
-
}]),
|
|
1666
|
-
}(),
|
|
1667
|
-
function
|
|
1668
|
-
i(this,
|
|
1678
|
+
}]), l;
|
|
1679
|
+
}(), s = function() {
|
|
1680
|
+
function l(L) {
|
|
1681
|
+
i(this, l), this.values = L;
|
|
1669
1682
|
}
|
|
1670
|
-
return r(
|
|
1683
|
+
return r(l, [{
|
|
1671
1684
|
key: "get",
|
|
1672
1685
|
value: function(o) {
|
|
1673
1686
|
return this.find(o)[2];
|
|
@@ -1686,12 +1699,12 @@ function P7() {
|
|
|
1686
1699
|
}
|
|
1687
1700
|
return null;
|
|
1688
1701
|
}
|
|
1689
|
-
}]),
|
|
1690
|
-
}(),
|
|
1691
|
-
function
|
|
1692
|
-
i(this,
|
|
1702
|
+
}]), l;
|
|
1703
|
+
}(), d = function() {
|
|
1704
|
+
function l(L) {
|
|
1705
|
+
i(this, l), this.text = L, this.length = L.length, this.codepoints = a.stringToCodePoints(L), this.reset();
|
|
1693
1706
|
}
|
|
1694
|
-
return r(
|
|
1707
|
+
return r(l, [{
|
|
1695
1708
|
key: "advance",
|
|
1696
1709
|
value: function() {
|
|
1697
1710
|
for (var o = arguments.length > 0 && arguments[0] !== void 0 ? arguments[0] : 1, D = 0; D < o; D++) {
|
|
@@ -1729,43 +1742,43 @@ function P7() {
|
|
|
1729
1742
|
value: function(o, D) {
|
|
1730
1743
|
return a.codePointsToString(this.codepoints.slice(o, D));
|
|
1731
1744
|
}
|
|
1732
|
-
}]),
|
|
1745
|
+
}]), l;
|
|
1733
1746
|
}(), O = function() {
|
|
1734
|
-
function
|
|
1735
|
-
i(this,
|
|
1747
|
+
function l(L) {
|
|
1748
|
+
i(this, l), this.values = L;
|
|
1736
1749
|
}
|
|
1737
|
-
return r(
|
|
1750
|
+
return r(l, [{
|
|
1738
1751
|
key: "getCategoryCount",
|
|
1739
1752
|
value: function() {
|
|
1740
1753
|
return this.values.categoryCount;
|
|
1741
1754
|
}
|
|
1742
|
-
}]),
|
|
1755
|
+
}]), l;
|
|
1743
1756
|
}(), h = e.NullSuppressions = function() {
|
|
1744
|
-
function
|
|
1745
|
-
if (i(this,
|
|
1757
|
+
function l() {
|
|
1758
|
+
if (i(this, l), this.instance)
|
|
1746
1759
|
return this.instance;
|
|
1747
1760
|
this.instance = this;
|
|
1748
1761
|
}
|
|
1749
|
-
return r(
|
|
1762
|
+
return r(l, [{
|
|
1750
1763
|
key: "shouldBreak",
|
|
1751
1764
|
value: function(o) {
|
|
1752
1765
|
return !0;
|
|
1753
1766
|
}
|
|
1754
|
-
}]),
|
|
1767
|
+
}]), l;
|
|
1755
1768
|
}(), N = e.RuleSet = function() {
|
|
1756
|
-
r(
|
|
1769
|
+
r(l, null, [{
|
|
1757
1770
|
key: "create",
|
|
1758
1771
|
value: function(o, D) {
|
|
1759
|
-
return new
|
|
1772
|
+
return new l(k.getInstance(o), D);
|
|
1760
1773
|
}
|
|
1761
1774
|
}]);
|
|
1762
|
-
function
|
|
1763
|
-
i(this,
|
|
1775
|
+
function l(L, o) {
|
|
1776
|
+
i(this, l), this.stateMachine = L, this.suppressions = o || new h();
|
|
1764
1777
|
}
|
|
1765
|
-
return r(
|
|
1778
|
+
return r(l, [{
|
|
1766
1779
|
key: "eachBoundary",
|
|
1767
1780
|
value: function(o, D) {
|
|
1768
|
-
var A = new
|
|
1781
|
+
var A = new d(o);
|
|
1769
1782
|
for (this.stateMachine.boundaryType !== "line" && D(0); !A.isEos(); )
|
|
1770
1783
|
this.stateMachine.handleNext(A), (A.isEos() || this.suppressions.shouldBreak(A)) && D(A.actualPosition);
|
|
1771
1784
|
}
|
|
@@ -1774,7 +1787,7 @@ function P7() {
|
|
|
1774
1787
|
value: function() {
|
|
1775
1788
|
return this.stateMachine.boundaryType;
|
|
1776
1789
|
}
|
|
1777
|
-
}]),
|
|
1790
|
+
}]), l;
|
|
1778
1791
|
}();
|
|
1779
1792
|
N.grapheme = {
|
|
1780
1793
|
forwardTable: {
|
|
@@ -4567,29 +4580,29 @@ function P7() {
|
|
|
4567
4580
|
categoryCount: 29
|
|
4568
4581
|
}
|
|
4569
4582
|
};
|
|
4570
|
-
var m = function(
|
|
4583
|
+
var m = function(L, o, D) {
|
|
4571
4584
|
var A = [];
|
|
4572
|
-
return
|
|
4585
|
+
return L[o](D, function(V, T, X) {
|
|
4573
4586
|
A.push(V);
|
|
4574
4587
|
}), A;
|
|
4575
4588
|
};
|
|
4576
|
-
e.wordSplit = function(
|
|
4577
|
-
return m(new
|
|
4578
|
-
}, e.sentenceSplit = function(
|
|
4579
|
-
return m(new
|
|
4580
|
-
}, e.graphemeSplit = function(
|
|
4581
|
-
return m(new
|
|
4582
|
-
}, e.lineSplit = function(
|
|
4583
|
-
return m(new
|
|
4589
|
+
e.wordSplit = function(L, o) {
|
|
4590
|
+
return m(new u(o), "eachWord", L);
|
|
4591
|
+
}, e.sentenceSplit = function(L, o) {
|
|
4592
|
+
return m(new u(o), "eachSentence", L);
|
|
4593
|
+
}, e.graphemeSplit = function(L, o) {
|
|
4594
|
+
return m(new u(o), "eachGraphemeCluster", L);
|
|
4595
|
+
}, e.lineSplit = function(L, o) {
|
|
4596
|
+
return m(new u(o), "eachLine", L);
|
|
4584
4597
|
};
|
|
4585
|
-
var M = 1,
|
|
4586
|
-
r(
|
|
4598
|
+
var M = 1, B = 0, g = 4, c = 0, k = function() {
|
|
4599
|
+
r(l, null, [{
|
|
4587
4600
|
key: "getInstance",
|
|
4588
4601
|
value: function(o) {
|
|
4589
|
-
var D =
|
|
4602
|
+
var D = l.getCache();
|
|
4590
4603
|
if (D[o] === void 0) {
|
|
4591
4604
|
var A = N[o];
|
|
4592
|
-
D[o] = new
|
|
4605
|
+
D[o] = new l(o, new O(A.metadata), new y(A.forwardTable.table, A.forwardTable.flags), new s(A.categoryTable));
|
|
4593
4606
|
}
|
|
4594
4607
|
return D[o];
|
|
4595
4608
|
}
|
|
@@ -4599,14 +4612,14 @@ function P7() {
|
|
|
4599
4612
|
return this.cache === void 0 && (this.cache = {}), this.cache;
|
|
4600
4613
|
}
|
|
4601
4614
|
}]);
|
|
4602
|
-
function
|
|
4603
|
-
i(this,
|
|
4615
|
+
function l(L, o, D, A) {
|
|
4616
|
+
i(this, l), this.boundaryType = L, this.metadata = o, this.ftable = D, this.categoryTable = A;
|
|
4604
4617
|
}
|
|
4605
|
-
return r(
|
|
4618
|
+
return r(l, [{
|
|
4606
4619
|
key: "handleNext",
|
|
4607
4620
|
value: function(o) {
|
|
4608
4621
|
var D = o.actualPosition, A = o.actualPosition, V = M, T = this.getRowIndex(V), X = 3, Y = "run";
|
|
4609
|
-
for (this.ftable.isBofRequired() && (X = 2, Y = "start"); V !=
|
|
4622
|
+
for (this.ftable.isBofRequired() && (X = 2, Y = "start"); V != B; ) {
|
|
4610
4623
|
if (o.isEos()) {
|
|
4611
4624
|
if (Y === "stop")
|
|
4612
4625
|
break;
|
|
@@ -4623,12 +4636,12 @@ function P7() {
|
|
|
4623
4636
|
value: function(o) {
|
|
4624
4637
|
return o * (this.metadata.getCategoryCount() + 4);
|
|
4625
4638
|
}
|
|
4626
|
-
}]),
|
|
4639
|
+
}]), l;
|
|
4627
4640
|
}(), v = 2, y = function() {
|
|
4628
|
-
function
|
|
4629
|
-
i(this,
|
|
4641
|
+
function l(L, o) {
|
|
4642
|
+
i(this, l), this.values = L, this.flags = o;
|
|
4630
4643
|
}
|
|
4631
|
-
return r(
|
|
4644
|
+
return r(l, [{
|
|
4632
4645
|
key: "get",
|
|
4633
4646
|
value: function(o) {
|
|
4634
4647
|
return this.values[o];
|
|
@@ -4638,12 +4651,12 @@ function P7() {
|
|
|
4638
4651
|
value: function() {
|
|
4639
4652
|
return (this.flags & v) != 0;
|
|
4640
4653
|
}
|
|
4641
|
-
}]),
|
|
4654
|
+
}]), l;
|
|
4642
4655
|
}(), C = function() {
|
|
4643
|
-
function
|
|
4644
|
-
i(this,
|
|
4656
|
+
function l(L, o) {
|
|
4657
|
+
i(this, l), this.value = L, this.children = o || {};
|
|
4645
4658
|
}
|
|
4646
|
-
return r(
|
|
4659
|
+
return r(l, [{
|
|
4647
4660
|
key: "setChild",
|
|
4648
4661
|
value: function(o, D) {
|
|
4649
4662
|
this.children[o] = D;
|
|
@@ -4659,7 +4672,7 @@ function P7() {
|
|
|
4659
4672
|
var o = {};
|
|
4660
4673
|
for (var D in this.children)
|
|
4661
4674
|
o[D] = this.children[D].copy();
|
|
4662
|
-
return new
|
|
4675
|
+
return new l(this.value, o);
|
|
4663
4676
|
}
|
|
4664
4677
|
}, {
|
|
4665
4678
|
key: "forEach",
|
|
@@ -4673,12 +4686,12 @@ function P7() {
|
|
|
4673
4686
|
for (var A in this.children)
|
|
4674
4687
|
this.children[A]._forEach(o, [].concat(t(D), [A]));
|
|
4675
4688
|
}
|
|
4676
|
-
}]),
|
|
4689
|
+
}]), l;
|
|
4677
4690
|
}(), P = function() {
|
|
4678
|
-
function
|
|
4679
|
-
i(this,
|
|
4691
|
+
function l(L) {
|
|
4692
|
+
i(this, l), this.root = L || new C(), this.locked = !1;
|
|
4680
4693
|
}
|
|
4681
|
-
return r(
|
|
4694
|
+
return r(l, [{
|
|
4682
4695
|
key: "add",
|
|
4683
4696
|
value: function(o, D) {
|
|
4684
4697
|
if (this.locked)
|
|
@@ -4702,7 +4715,7 @@ function P7() {
|
|
|
4702
4715
|
}, {
|
|
4703
4716
|
key: "copy",
|
|
4704
4717
|
value: function() {
|
|
4705
|
-
return new
|
|
4718
|
+
return new l(this.root.copy());
|
|
4706
4719
|
}
|
|
4707
4720
|
}, {
|
|
4708
4721
|
key: "forEach",
|
|
@@ -4717,22 +4730,22 @@ function P7() {
|
|
|
4717
4730
|
D.add(A, V);
|
|
4718
4731
|
}), D;
|
|
4719
4732
|
}
|
|
4720
|
-
}]),
|
|
4733
|
+
}]), l;
|
|
4721
4734
|
}(), Q = e.suppressions = {}, R = e.Suppressions = function() {
|
|
4722
|
-
r(
|
|
4735
|
+
r(l, null, [{
|
|
4723
4736
|
key: "create",
|
|
4724
4737
|
value: function(o) {
|
|
4725
4738
|
var D = new P(), A = new P();
|
|
4726
4739
|
return o.forEach(function(V) {
|
|
4727
4740
|
var T = a.stringToCodePoints(V);
|
|
4728
4741
|
D.add(T, !0), A.add(T.reverse(), !0);
|
|
4729
|
-
}), new
|
|
4742
|
+
}), new l(D, A, o);
|
|
4730
4743
|
}
|
|
4731
4744
|
}]);
|
|
4732
|
-
function
|
|
4733
|
-
i(this,
|
|
4745
|
+
function l(L, o, D) {
|
|
4746
|
+
i(this, l), this.forwardTrie = L, this.backwardTrie = o, this.list = D;
|
|
4734
4747
|
}
|
|
4735
|
-
return r(
|
|
4748
|
+
return r(l, [{
|
|
4736
4749
|
key: "lock",
|
|
4737
4750
|
value: function() {
|
|
4738
4751
|
this.forwardTrie.lock(), this.backwardTrie.lock();
|
|
@@ -4740,7 +4753,7 @@ function P7() {
|
|
|
4740
4753
|
}, {
|
|
4741
4754
|
key: "merge",
|
|
4742
4755
|
value: function(o) {
|
|
4743
|
-
return new
|
|
4756
|
+
return new l(this.forwardTrie.merge(o.forwardTrie), this.backwardTrie.merge(o.backwardTrie), [].concat(t(this.list), t(o.list)));
|
|
4744
4757
|
}
|
|
4745
4758
|
}, {
|
|
4746
4759
|
key: "shouldBreak",
|
|
@@ -4772,39 +4785,39 @@ function P7() {
|
|
|
4772
4785
|
D++;
|
|
4773
4786
|
}
|
|
4774
4787
|
}
|
|
4775
|
-
}]),
|
|
4788
|
+
}]), l;
|
|
4776
4789
|
}(), F = e.customSuppressions = {
|
|
4777
4790
|
en: R.create(["Dr."])
|
|
4778
4791
|
};
|
|
4779
4792
|
Q.de = function() {
|
|
4780
|
-
var
|
|
4781
|
-
return F.de && (
|
|
4793
|
+
var l = R.create(["Port.", "Prot.", "Proz.", "Pfd.", "P.", "Phys.", "PIN.", "Alt.", "Allg.", "Ausg.", "Ausschl.", "Aug.", "Akt.", "Abs.", "Abt.", "Abw.", "A.", "Adr.", "Adj.", "App.", "Apr.", "Art.", "Di.", "Dipl.-Ing.", "Dipl.-Kfm.", "Dir.", "Dim.", "Dez.", "D.", "Dr.", "Dtzd.", "Dat.", "Do.", "Ges.", "Gesch.", "Gebr.", "Gem.", "Geograph.", "G.", "Ggf.", "frz.", "entspr.", "erw.", "evtl.", "ev.", "eigtl.", "einschl.", "e.h.", "e.Wz.", "e.V.", "exkl.", "ehem.", "Frl.", "Fr.", "Folg.", "Forts. f.", "Fa.", "Fam.", "Fn.", "Feb.", "F.", "Inh.", "Ing.", "Int.", "Inc.", "I.", "k.u.k.", "k. u. k.", "kgl.", "kath.", "kfm.", "Ca.", "Chr.", "Chin.", "C.", "Co.", "J.", "Jahrh.", "Jan.", "Jr.", "So.", "Sept.", "Sep.", "Sek.", "Sa.", "Spezif.", "St.", "Std.", "Str.", "StR.", "S.A.", "Schr.", "i.B.", "i.R.", "i.J.", "i.A.", "i.H.", "i.V.", "inkl.", "s.", "schwäb.", "schweiz.", "schles.", "sog.", "südd.", "o.B.", "o.A.", "o.g.", "od.", "am.", "amtl.", "a.a.O.", "a.a.S.", "a.Rh.", "a.M.", "a.D.", "u.", "z.", "zzgl.", "Kl.", "Kap.", "Kfm.", "Konv.", "Kop.", "Obj.", "Op.", "O.", "Okt.", "Msp.", "Mo.", "Mod.", "Mrd.", "M.", "Max.", "Mind.", "Min.", "Mi.", "Mio.", "U.", "Urspr.", "Ursprüngl.", "R.", "Rd.", "Rel.", "Reg.", "Rep.", "Einh.", "Erf.", "Evtl.", "Vgl.", "Verf.", "W.", "Wg.", "B.", "Bibl.", "Bed.", "Best.", "Ben.", "Ber.", "n.u.Z.", "n. Chr.", "näml.", "Hr.", "Hrn.", "Hrsg.", "Hbf.", "Hptst.", "Ztr.", "Z.", "d.Ä.", "d.J.", "b.", "beil.", "v.R.w.", "v.T.", "v. Chr.", "v. u. Z.", "v. H.", "v.u.Z.", "v.Chr.", "v.H.", "verh.", "verw.", "vgl.", "möbl.", "m.W.", "m.E.", "mtl.", "tägl.", "gest.", "gesch.", "gek.", "gedr.", "geb.", "ggf.", "ggfs.", "Nov.", "N.", "Nr.", "p.Adr.", "r.", "röm.", "jew.", "jhrl.", "Qu.", "hpts.", "Ltd.", "L.", "led.", "Test.", "Temp.", "Tel.", "Tägl.", "Trans.", "T.", "Y.", "österr."]);
|
|
4794
|
+
return F.de && (l = l.merge(F.de)), l.lock(), l;
|
|
4782
4795
|
}(), Q.en = function() {
|
|
4783
|
-
var
|
|
4784
|
-
return F.en && (
|
|
4796
|
+
var l = R.create(["L.", "Long.", "Link.", "Lib.", "Lt.", "Lev.", "Alt.", "All.", "Approx.", "Act.", "Aug.", "Abs.", "A.", "Adj.", "Adv.", "AD.", "AB.", "AA.", "As.", "E.", "Ex.", "Exec.", "Est.", "Ed.", "Etc.", "O.", "Ok.", "Org.", "Or.", "OK.", "Op.", "On.", "Maj.", "Mart.", "Mar.", "Misc.", "Min.", "MR.", "M.", "Mrs.", "Mr.", "Md.", "Mt.", "Mgr.", "Ms.", "Mb.", "P.O.", "P.M.", "P.V.", "Prof.", "Pro.", "Ph.D.", "Phys.", "PC.", "Pvt.", "PP.", "J.D.", "J.K.", "J.B.", "Jam.", "Jan.", "Job.", "Joe.", "Jun.", "Card.", "Cap.", "Capt.", "Cont.", "Conn.", "Col.", "Comm.", "C.O.D.", "C.F.", "Dec.", "Def.", "Dept.", "DC.", "D.", "Do.", "Diff.", "Sept.", "Sep.", "S.", "Sgt.", "Sq.", "Hat.", "Hz.", "Hon.B.A.", "G.", "Gb.", "Go.", "By.", "B.", "Var.", "VS.", "N.V.", "N.Y.", "Num.", "Nov.", "Nr.", "Up.", "U.", "Fn.", "Feb.", "Fri.", "F.", "Z.", "Kb.", "K.", "I.", "In.", "Id.", "Is.", "exec.", "X.", "vs.", "R.L.", "R.T.", "Rev.", "Rep.", "Yr.", "pp.", "To.", "T.", "Q.", "a.m."]);
|
|
4797
|
+
return F.en && (l = l.merge(F.en)), l.lock(), l;
|
|
4785
4798
|
}(), Q.es = function() {
|
|
4786
|
-
var
|
|
4787
|
-
return F.es && (
|
|
4799
|
+
var l = R.create(["Rdos.", "Rdo.", "RR.HH.", "Rvdos.", "Rvdo.", "Rvdmos.", "Rvdmo.", "RAM.", "Rol.", "R.D.", "R.U.", "Rev.", "Reg.", "JJ.OO.", "Sres.", "Srtas.", "Srta.", "Sra.", "Sras.", "Sr.", "SA.", "SS.MM.", "SS.AA.", "Sol.", "Seg.", "Sta.", "Sto.", "Sdad.", "fig.", "figs.", "feb.", "fund.", "may.", "mar.", "mié.", "mss.", "ms.", "mtro.", "oct.", "cap.", "col.", "cf.", "cfr.", "c/c.", "doc.", "dom.", "dic.", "dtor.", "dptos.", "dpto.", "depto.", "deptos.", "d. C.", "Excmo.", "Excmos.", "Excma.", "Excmas.", "Exc.", "Exmo.", "Exmos.", "Exma.", "Exmas.", "Emm.", "Em.", "Trab.", "Kit.", "Korn.", "K.", "Inc.", "Ilmas.", "Ilma.", "Ilmos.", "Ilmo.", "Id.", "FF.CC.", "Fabric.", "Fr.", "DC.", "Da.", "Dr.", "Drs.", "Dra.", "Desv.", "Desc.", "D.", "Dña.", "Dª.", "Dto.", "ago.", "abr.", "a. C.", "a. e. c.", "afma.", "afmas.", "afmo.", "afmos.", "trad.", "ed.", "ene.", "ej.", "e. c.", "e.g.", "jul.", "jue.", "jun.", "Comm.", "Col.", "Corp.", "C.", "Cía.", "CA.", "vid.", "vie.", "v.gr.", "vs.", "ss.", "sáb.", "sras.", "sres.", "srs.", "sept.", "ntra.", "ntro.", "nov.", "prov.", "prof.", "p.ej.", "licdo.", "licda.", "lun.", "bol.", "bco.", "Vda.", "Var.", "O.M.", "Ldo.", "Lda.", "Ltda.", "Ltd.", "Ltdo.", "Lcda.", "Lcdo.", "Lic.", "L.", "U.S.", "hnos.", "MM.", "Mons.", "M.", "Mr.", "Mrs.", "A.C.", "Av.", "Avda.", "Ant.", "All.", "AA.", "H.P.", "PP.", "Bien."]);
|
|
4800
|
+
return F.es && (l = l.merge(F.es)), l.lock(), l;
|
|
4788
4801
|
}(), Q.fr = function() {
|
|
4789
|
-
var
|
|
4790
|
-
return F.fr && (
|
|
4802
|
+
var l = R.create(["aux.", "adr.", "anc.", "anon.", "ann.", "acoust.", "avr.", "av. J.-C.", "append.", "ap. J.-C.", "config.", "collab.", "coll.", "cam.", "categ.", "M.", "MM.", "Mart.", "dim.", "dir.", "dict.", "doc.", "dest.", "déc.", "imprim.", "imm.", "indus.", "illustr.", "ill.", "oct.", "syst.", "symb.", "synth.", "sept.", "sam.", "bull.", "broch.", "P.", "Prof.", "hôp.", "juil.", "jeu.", "janv.", "C.", "Comm.", "voit.", "ven.", "Jr.", "graph.", "gouv.", "éd.", "édit.", "éval.", "équiv.", "fig.", "févr.", "niv.", "nov.", "quart.", "réf.", "DC.", "Desc.", "D.", "L.", "All.", "U.", "S.M.A.R.T.", "S.A.", "encycl.", "exempl.", "mer.", "mar.", "lun.", "Inc.", "Var.", "W.", "Op.", "trav. publ."]);
|
|
4803
|
+
return F.fr && (l = l.merge(F.fr)), l.lock(), l;
|
|
4791
4804
|
}(), Q.it = function() {
|
|
4792
|
-
var
|
|
4793
|
-
return F.it && (
|
|
4805
|
+
var l = R.create(["N.B.", "N.d.A.", "N.d.T.", "N.d.E.", "div.", "d.p.R.", "d.C.", "dott.", "dr.", "a.C.", "arch.", "ag.", "avv.", "all.", "fig.", "c.c.p.", "Cfr.", "C.P.", "vol.", "ver.", "Geom.", "O.d.G.", "S.p.A.", "S.M.A.R.T.", "S.A.R.", "Sig.", "rag.", "Mod.", "pag.", "p.", "tav.", "tab.", "DC.", "D.", "mitt.", "Ing.", "int.", "on.", "L.", "Ltd.", "Liv.", "U.S.", "sez."]);
|
|
4806
|
+
return F.it && (l = l.merge(F.it)), l.lock(), l;
|
|
4794
4807
|
}(), Q.pt = function() {
|
|
4795
|
-
var
|
|
4796
|
-
return F.pt && (
|
|
4808
|
+
var l = R.create(["psicol.", "port.", "pol.", "p.", "pág.", "pal.", "profa.", "prof.", "pron.", "prod.", "próx.", "pq.", "pp.", "fig.", "fisiol.", "filos.", "fil.", "ff.", "fev.", "fem.", "fot.", "fís.", "fasc.", "fac.", "fl.", "fr.", "f.", "compl.", "comp.", "com.", "col.", "cont.", "contab.", "cap.", "caps.", "cat.", "cf.", "créd.", "cel.", "cx.", "círc.", "cód.", "rep.", "res.", "relat.", "rel.", "ref.", "rementente", "r.s.v.p.", "rod.", "doc.", "dipl.", "dir.", "div.", "dez.", "des.", "desc.", "déb.", "d.C.", "astron.", "ago.", "agric.", "arquit.", "arqueol.", "arit.", "apt.", "ap.", "apart.", "abr.", "abrev.", "adm.", "anat.", "aux.", "autom.", "aer.", "a.C.", "a.m.", "eletrôn.", "eletr.", "elem.", "ed.", "educ.", "end.", "eng.", "esp.", "ex.", "geom.", "gên.", "gram.", "mov.", "matem.", "mat.", "mai.", "mar.", "méd.", "mús.", "máq.", "trad.", "transp.", "tecnol.", "tec.", "tel.", "Exmo.", "Exma.", "E.", "Est.", "univ.", "R.", "D.C", "Dir.", "Drs.", "Dra.", "Dras.", "Dr.", "l.", "liter.", "lit.", "lin.", "lat.", "long.", "P.M.", "Pe.", "Ph.D.", "jur.", "jul.", "jun.", "jan.", "neol.", "nov.", "náut.", "índ.", "Ltda.", "Ltd.", "ingl.", "ind.", "Mar.", "símb.", "sup.", "séc.", "set.", "sociol.", "V.T.", "vs.", "vol.", "S.", "Sta.", "Sto.", "N.Sra.", "N.T.", "out.", "odontol.", "odont.", "org.", "organiz.", "obs.", "A.C.", "A.M", "Av.", "Alm.", "álg.", "hist.", "Fr.", "Ilmo.", "Ilma.", "quím.", "W.C.", "bras.", "biogr.", "Jr."]);
|
|
4809
|
+
return F.pt && (l = l.merge(F.pt)), l.lock(), l;
|
|
4797
4810
|
}(), Q.ru = function() {
|
|
4798
|
-
var
|
|
4799
|
-
return F.ru && (
|
|
4811
|
+
var l = R.create(["руб.", "янв.", "до н. э.", "дек.", "сент.", "тел.", "тыс.", "февр.", "нояб.", "н. э.", "н.э.", "апр.", "авг.", "окт.", "отд.", "проф.", "кв.", "ул."]);
|
|
4812
|
+
return F.ru && (l = l.merge(F.ru)), l.lock(), l;
|
|
4800
4813
|
}(), Q.tr = function() {
|
|
4801
|
-
var
|
|
4802
|
-
return F.tr && (
|
|
4814
|
+
var l = R.create(["age.", "agm.", "agy.", "Alb.", "Alm.", "anat.", "ant.", "Apt.", "Ar.", "ark.", "Arş.", "As.", "Asb.", "astr.", "astrol.", "Atğm.", "atm.", "Av.", "b.", "bağ.", "Bçvş.", "bit.", "biy.", "bk.", "bl.", "Bl.", "Bnb.", "bot.", "Böl.", "bs.", "Bşk.", "Bul.", "Bulg.", "çev.", "Çvş.", "db.", "dk.", "Doç.", "doğ.", "Dr.", "drl.", "Dz.", "dzl.", "Ecz.", "ed.", "ekon.", "Ens.", "f.", "Fak.", "fel.", "fiz.", "fizy.", "Gen.", "geom.", "gn.", "Gnkur.", "Gön.", "gr.", "hay.", "haz.", "hek.", "hlk.", "Hs.", "huk.", "Hv.", "Hz.", "İng.", "is.", "İsp.", "işl.", "İt.", "İz.", "Jap.", "jeol.", "K.", "kim.", "koor.", "Kor.", "Kora.", "Korg.", "kr.", "krş.", "Kuv.", "Lat.", "Mac.", "Mah.", "man.", "mat.", "mec.", "Mim.", "mim.", "min.", "Müh.", "Mür.", "müz.", "No.", "Nö.", "Okt.", "Onb.", "Or.", "Ora.", "Ord.", "Org.", "Ort.", "Osm.", "öl.", "ör.", "öz.", "ped.", "Port.", "Prof.", "psikol.", "Rum.", "Rus.", "s.", "sa.", "Sb.", "sf.", "Sl.", "Sn.", "sn.", "snt.", "Sok.", "sos.", "sp.", "Srp.", "Şb.", "T.", "T.C.", "tar.", "Tb.", "tek.", "tel.", "telg.", "Tğm.", "tiy.", "tlks.", "tls.", "Top.", "Tug.", "Tuğa.", "Tuğg.", "Tüm.", "Tüma.", "Tümg.", "Uzm.", "Üçvş.", "ünl.", "Ütğm.", "vb.", "vd.", "Vet.", "vs.", "Y.", "Yay.", "Yb.", "Yd.", "Yrd.", "Yun.", "yy.", "Yzb.", "zf.", "zm.", "zool."]);
|
|
4815
|
+
return F.tr && (l = l.merge(F.tr)), l.lock(), l;
|
|
4803
4816
|
}(), Q.all = R.create(function() {
|
|
4804
|
-
var
|
|
4805
|
-
for (var
|
|
4806
|
-
|
|
4807
|
-
return
|
|
4817
|
+
var l = [];
|
|
4818
|
+
for (var L in Q)
|
|
4819
|
+
l = l.concat(Q[L].list);
|
|
4820
|
+
return l;
|
|
4808
4821
|
}());
|
|
4809
4822
|
});
|
|
4810
4823
|
}(s1)), s1;
|
|
@@ -4842,11 +4855,11 @@ function k7(n = [], e) {
|
|
|
4842
4855
|
return n.length - 1;
|
|
4843
4856
|
}
|
|
4844
4857
|
function Q7(n, e, a) {
|
|
4845
|
-
var
|
|
4858
|
+
var u;
|
|
4846
4859
|
let t, i, r = -1;
|
|
4847
|
-
if ((
|
|
4848
|
-
for (let
|
|
4849
|
-
r + n.children[
|
|
4860
|
+
if ((u = n.children) != null && u.length) {
|
|
4861
|
+
for (let s = 0; s < n.children.length; s++)
|
|
4862
|
+
r + n.children[s].text.length < e || r >= e && r <= a && (t === void 0 && (t = s), i = s), r += n.children[s].text.length;
|
|
4850
4863
|
return { start: t, end: i };
|
|
4851
4864
|
}
|
|
4852
4865
|
}
|
|
@@ -4895,12 +4908,12 @@ function m0(n, e = 0, a = 2) {
|
|
|
4895
4908
|
}
|
|
4896
4909
|
function R7(n) {
|
|
4897
4910
|
const a = (n == null ? void 0 : n.sentenceLength) || 75;
|
|
4898
|
-
let t, i, r,
|
|
4911
|
+
let t, i, r, u, s, d;
|
|
4899
4912
|
const O = /[.,?!\],。!?…]{1}/, h = /[.?!\]。!?…]{1}/;
|
|
4900
|
-
let N = !1, m = !1, M = "",
|
|
4913
|
+
let N = !1, m = !1, M = "", B = null;
|
|
4901
4914
|
return g(), { feed: y, reset: g, end: c };
|
|
4902
4915
|
function g() {
|
|
4903
|
-
i = [], r = 0,
|
|
4916
|
+
i = [], r = 0, u = !1, s = null, d = [];
|
|
4904
4917
|
}
|
|
4905
4918
|
function c() {
|
|
4906
4919
|
y([{
|
|
@@ -4910,28 +4923,28 @@ function R7(n) {
|
|
|
4910
4923
|
}]);
|
|
4911
4924
|
}
|
|
4912
4925
|
function k(C) {
|
|
4913
|
-
N || (M += C.map((P) => P.text).join("")), !N && !m && C1(M) && (m = !0), M.length > 50 && (N = !0), !
|
|
4926
|
+
N || (M += C.map((P) => P.text).join("")), !N && !m && C1(M) && (m = !0), M.length > 50 && (N = !0), !B && N && (B = W0(M), console.log("detect language", M, B), console.log(B));
|
|
4914
4927
|
}
|
|
4915
4928
|
function v() {
|
|
4916
4929
|
var C, P;
|
|
4917
|
-
|
|
4930
|
+
d.length && ((C = n == null ? void 0 : n.repeatString) != null && C.length && d.forEach((Q) => {
|
|
4918
4931
|
var R;
|
|
4919
4932
|
(R = n.repeatString) == null || R.forEach((F) => {
|
|
4920
|
-
const
|
|
4921
|
-
if (
|
|
4922
|
-
console.log("===================================="), console.log(Q.text), console.log(
|
|
4923
|
-
let
|
|
4924
|
-
|
|
4933
|
+
const l = F7(Q.text, F);
|
|
4934
|
+
if (l.length) {
|
|
4935
|
+
console.log("===================================="), console.log(Q.text), console.log(l);
|
|
4936
|
+
let L = 0;
|
|
4937
|
+
l.forEach((o, D) => {
|
|
4925
4938
|
var A;
|
|
4926
4939
|
if (o.start === 0) {
|
|
4927
4940
|
const V = k7(Q.children, o.match);
|
|
4928
|
-
|
|
4941
|
+
L += m0(Q, 0, V + 1), console.log(L);
|
|
4929
4942
|
}
|
|
4930
4943
|
if (o.start > 0) {
|
|
4931
|
-
const V = D > 0 ? o.start -
|
|
4944
|
+
const V = D > 0 ? o.start - l[D - 1].end + L : o.start, T = D > 0 ? o.end - l[D - 1].end + L : o.end;
|
|
4932
4945
|
console.log(V, T, "=============================");
|
|
4933
4946
|
const X = Q7(Q, V, T);
|
|
4934
|
-
console.log((A = Q.children) == null ? void 0 : A.map((Y, r1) => Y.text + "--" + Y.text.length + " " + r1).join("||")), console.log(X), X && X.start !== void 0 && X.end !== void 0 && (
|
|
4947
|
+
console.log((A = Q.children) == null ? void 0 : A.map((Y, r1) => Y.text + "--" + Y.text.length + " " + r1).join("||")), console.log(X), X && X.start !== void 0 && X.end !== void 0 && (L += m0(Q, X.start, X.end - X.start + 1), console.log(L), console.log(Q.text));
|
|
4935
4948
|
}
|
|
4936
4949
|
}), console.log("----------"), console.log(Q.text), console.log("----------");
|
|
4937
4950
|
}
|
|
@@ -4939,11 +4952,11 @@ function R7(n) {
|
|
|
4939
4952
|
}), (P = n == null ? void 0 : n.onParse) == null || P.call(n, {
|
|
4940
4953
|
type: "event",
|
|
4941
4954
|
event: "message",
|
|
4942
|
-
data:
|
|
4955
|
+
data: d
|
|
4943
4956
|
}));
|
|
4944
4957
|
}
|
|
4945
4958
|
function y(C) {
|
|
4946
|
-
var Q, R, F,
|
|
4959
|
+
var Q, R, F, l;
|
|
4947
4960
|
k(C), i.push(...C), t && ((Q = n == null ? void 0 : n.onStart) == null || Q.call(n, {
|
|
4948
4961
|
type: "event",
|
|
4949
4962
|
event: "start",
|
|
@@ -4951,27 +4964,27 @@ function R7(n) {
|
|
|
4951
4964
|
})), t = !1;
|
|
4952
4965
|
const P = i.length;
|
|
4953
4966
|
for (; r < P; ) {
|
|
4954
|
-
const
|
|
4955
|
-
if (++r,
|
|
4956
|
-
|
|
4967
|
+
const L = i[r];
|
|
4968
|
+
if (++r, L.punc = O.test(L.text) ? 1 : 0, L.em = h.test(L.text) ? 1 : 0, L.text === "[--end--]") {
|
|
4969
|
+
s && d.push(s), u = !0;
|
|
4957
4970
|
break;
|
|
4958
4971
|
}
|
|
4959
|
-
if (
|
|
4960
|
-
...
|
|
4972
|
+
if (s ? (s.et = L.et, s.text = s.text + L.text, s.em = s.em || L.em, s.punc = s.punc || L.punc, (R = s.children) == null || R.push({ ...L })) : s = {
|
|
4973
|
+
...L,
|
|
4961
4974
|
children: [{
|
|
4962
|
-
...
|
|
4975
|
+
...L
|
|
4963
4976
|
}]
|
|
4964
|
-
},
|
|
4965
|
-
const o =
|
|
4977
|
+
}, L.punc || u && s) {
|
|
4978
|
+
const o = d[d.length - 1];
|
|
4966
4979
|
let D = !0;
|
|
4967
|
-
o && !o.em && (o.text +
|
|
4968
|
-
} else !
|
|
4969
|
-
if (
|
|
4970
|
-
const o =
|
|
4971
|
-
v(),
|
|
4980
|
+
o && !o.em && (o.text + s.text).length <= a ? D = !1 : o && o.em && L.em && (D = Q1(o.text + s.text).length !== 1), D ? d.push(E0(s)) : (o.et = s.et, o.text = o.text + s.text, o.em = s.em || o.em, o.punc = s.punc || o.punc, (F = o.children) == null || F.push(...s.children || [])), s = null;
|
|
4981
|
+
} else !L.punc && !s.punc && s.text.length > a && B && Q1(s.text, B.language).length === 1 && (B.language === "zh" || B.language === "ja" || B.language);
|
|
4982
|
+
if (d.length > 0) {
|
|
4983
|
+
const o = d.pop();
|
|
4984
|
+
v(), d = [o];
|
|
4972
4985
|
}
|
|
4973
4986
|
}
|
|
4974
|
-
|
|
4987
|
+
u && (v(), (l = n == null ? void 0 : n.onEnd) == null || l.call(n, {
|
|
4975
4988
|
type: "event",
|
|
4976
4989
|
event: "end",
|
|
4977
4990
|
data: i
|
|
@@ -5038,10 +5051,10 @@ class v7 {
|
|
|
5038
5051
|
else {
|
|
5039
5052
|
if (!(i instanceof Object))
|
|
5040
5053
|
break;
|
|
5041
|
-
let
|
|
5042
|
-
for (let
|
|
5043
|
-
i[t[
|
|
5044
|
-
|
|
5054
|
+
let u, s;
|
|
5055
|
+
for (let d = r; d < t.length; d++)
|
|
5056
|
+
i[t[d]] = {}, [u, s] = [i, t[d]], i = i[t[d]];
|
|
5057
|
+
u && s && (u[s] = { [this.delimit]: a || "" });
|
|
5045
5058
|
break;
|
|
5046
5059
|
}
|
|
5047
5060
|
r === t.length - 1 && (i[this.delimit] = a || "");
|