lipilekhika 1.0.0 → 1.0.2

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.
Files changed (80) hide show
  1. package/LICENCE +1 -1
  2. package/README.md +133 -128
  3. package/dist/cjs/{Assamese-D-JeQpHm.cjs → Assamese-DB4wYtnM.cjs} +1 -1
  4. package/dist/cjs/{Bengali-Cb8p6Ss6.cjs → Bengali-Cmp4ERCx.cjs} +1 -1
  5. package/dist/cjs/{Brahmi-WrTB6e7g.cjs → Brahmi-DXNNRCQo.cjs} +1 -1
  6. package/dist/cjs/{Devanagari-CubB6Ubb.cjs → Devanagari-C5Xu55TI.cjs} +1 -1
  7. package/dist/cjs/{Granth-C6Es6NQ-.cjs → Granth-DfIl5B1W.cjs} +1 -1
  8. package/dist/cjs/{Gujarati-ToSwR4ym.cjs → Gujarati-VhhVj3ND.cjs} +1 -1
  9. package/dist/cjs/{Gurumukhi-enKEt02s.cjs → Gurumukhi-B3KktUHR.cjs} +1 -1
  10. package/dist/cjs/{Kannada-BuxZGQgS.cjs → Kannada-Cim8EpsZ.cjs} +1 -1
  11. package/dist/cjs/{Malayalam-DrxDIRjl.cjs → Malayalam-ByvdN31f.cjs} +1 -1
  12. package/dist/cjs/{Modi-CYRCB8M0.cjs → Modi-BmaT2WSv.cjs} +1 -1
  13. package/dist/cjs/Normal-Dtdwej3_.cjs +1 -0
  14. package/dist/cjs/{Odia-DyI85meX.cjs → Odia-B2SrneR4.cjs} +1 -1
  15. package/dist/cjs/{Purna-Devanagari-72KHauTF.cjs → Purna-Devanagari-B6wi3Ozk.cjs} +1 -1
  16. package/dist/cjs/Romanized-CuNlC73n.cjs +1 -0
  17. package/dist/cjs/{Sharada-CsNaWIt6.cjs → Sharada-BhQ08FSG.cjs} +1 -1
  18. package/dist/cjs/Siddham-D4nvxdh7.cjs +1 -0
  19. package/dist/cjs/{Sinhala-Cu5sdgUY.cjs → Sinhala-BfF1GhfP.cjs} +1 -1
  20. package/dist/cjs/{Tamil-Extended-PyLgZJyX.cjs → Tamil-Extended-BWvMEAwT.cjs} +1 -1
  21. package/dist/cjs/{Tamil-CiqM4WXX.cjs → Tamil-lWn6n3eQ.cjs} +1 -1
  22. package/dist/cjs/{Telugu-CaCRkj-p.cjs → Telugu-zv0Kis3h.cjs} +1 -1
  23. package/dist/cjs/index.cjs +1 -1
  24. package/dist/cjs/script_normalization-CA-LgBRA.cjs +2 -0
  25. package/dist/cjs/typing.cjs +1 -1
  26. package/dist/esm/{Assamese-4WO2wzXg.js → Assamese-J9rLtNyc.js} +1 -1
  27. package/dist/esm/{Bengali-Dhlxv7Pr.js → Bengali-CzBJ2A7u.js} +1 -1
  28. package/dist/esm/{Brahmi-BvC4Luvf.js → Brahmi-BEpeQWDq.js} +1 -1
  29. package/dist/esm/{Devanagari-C4NmXCKb.js → Devanagari-BnMkKqfQ.js} +1 -1
  30. package/dist/esm/{Granth-B3US1gHe.js → Granth-Dm7b2ooD.js} +1 -1
  31. package/dist/esm/{Gujarati-DfQS2sSY.js → Gujarati-C-B0DC9F.js} +1 -1
  32. package/dist/esm/{Gurumukhi-dVJRMJNL.js → Gurumukhi-GmwtiirL.js} +1 -1
  33. package/dist/esm/{Kannada-ZB5P9u4T.js → Kannada-BYMa9Anc.js} +1 -1
  34. package/dist/esm/{Malayalam-BQROfOnO.js → Malayalam-BJnHbd1I.js} +1 -1
  35. package/dist/esm/{Modi-B16t9oPJ.js → Modi-D1HxQQAw.js} +1 -1
  36. package/dist/esm/Normal-9g1lSg7V.js +25 -0
  37. package/dist/esm/{Odia-hd6WA8ll.js → Odia-CkBeE4O8.js} +1 -1
  38. package/dist/esm/{Purna-Devanagari-CqkVL5zc.js → Purna-Devanagari-BqwAqNzf.js} +1 -1
  39. package/dist/esm/Romanized-BFCb5s8R.js +25 -0
  40. package/dist/esm/{Sharada-Bk4pUPeN.js → Sharada-CJqMtpHb.js} +1 -1
  41. package/dist/esm/Siddham-DxnhsFps.js +29 -0
  42. package/dist/esm/{Sinhala-C6t7GaP6.js → Sinhala-BbPpiimZ.js} +1 -1
  43. package/dist/esm/{Tamil-CR4VFQvV.js → Tamil-CyqLqDzd.js} +5 -5
  44. package/dist/esm/{Tamil-Extended-nmDMmVeX.js → Tamil-Extended-CXbdAVNJ.js} +1 -1
  45. package/dist/esm/{Telugu-BCg-EHjf.js → Telugu-o2swovuz.js} +1 -1
  46. package/dist/esm/index.mjs +3 -2
  47. package/dist/esm/script_normalization-HKw9dSJ0.js +704 -0
  48. package/dist/esm/typing.mjs +1 -1
  49. package/dist/index.d.ts +9 -31
  50. package/dist/lipilekhika.umd.js +2 -2
  51. package/dist/typing.d.ts +1 -1
  52. package/dist/umd_json/script_data/Assamese.json +1 -1
  53. package/dist/umd_json/script_data/Bengali.json +1 -1
  54. package/dist/umd_json/script_data/Brahmi.json +1 -1
  55. package/dist/umd_json/script_data/Devanagari.json +1 -1
  56. package/dist/umd_json/script_data/Granth.json +1 -1
  57. package/dist/umd_json/script_data/Gujarati.json +1 -1
  58. package/dist/umd_json/script_data/Gurumukhi.json +1 -1
  59. package/dist/umd_json/script_data/Kannada.json +1 -1
  60. package/dist/umd_json/script_data/Malayalam.json +1 -1
  61. package/dist/umd_json/script_data/Modi.json +1 -1
  62. package/dist/umd_json/script_data/Normal.json +1 -1
  63. package/dist/umd_json/script_data/Odia.json +1 -1
  64. package/dist/umd_json/script_data/Purna-Devanagari.json +1 -1
  65. package/dist/umd_json/script_data/Romanized.json +1 -1
  66. package/dist/umd_json/script_data/Sharada.json +1 -1
  67. package/dist/umd_json/script_data/Siddham.json +1 -1
  68. package/dist/umd_json/script_data/Sinhala.json +1 -1
  69. package/dist/umd_json/script_data/Tamil-Extended.json +1 -1
  70. package/dist/umd_json/script_data/Tamil.json +1 -1
  71. package/dist/umd_json/script_data/Telugu.json +1 -1
  72. package/package.json +4 -2
  73. package/dist/cjs/Normal-BfIhy-Ts.cjs +0 -1
  74. package/dist/cjs/Romanized-Dy_qfDqa.cjs +0 -1
  75. package/dist/cjs/Siddham-Do9NoVoz.cjs +0 -1
  76. package/dist/cjs/script_normalization-BlG9S8Ex.cjs +0 -2
  77. package/dist/esm/Normal-Chi5ccip.js +0 -25
  78. package/dist/esm/Romanized-CQkrk7RV.js +0 -25
  79. package/dist/esm/Siddham-Cf9Jp5KF.js +0 -29
  80. package/dist/esm/script_normalization-D8Eu1NYy.js +0 -712
@@ -1,712 +0,0 @@
1
- function Q(n, e, l, t) {
2
- const { accessor: a, compareFn: s } = t ?? {}, c = a ?? ((i, r) => i[r]);
3
- let o = 0, h = e.length - 1, _ = -1;
4
- for (; o <= h; ) {
5
- const i = Math.floor((o + h) / 2), r = e[i], u = c(n, r);
6
- let m;
7
- s ? m = s(l, u) : m = l < u ? -1 : l > u ? 1 : 0, m === 0 ? (_ = r, h = i - 1) : m < 0 ? h = i - 1 : o = i + 1;
8
- }
9
- return _;
10
- }
11
- function Y(n, e, l) {
12
- const { accessor: t, compareFn: a } = l ?? {}, s = t ?? ((_, i) => _[i]);
13
- let c = 0, o = n.length - 1, h = -1;
14
- for (; c <= o; ) {
15
- const _ = Math.floor((c + o) / 2), i = s(n, _);
16
- let r;
17
- a ? r = a(e, i) : r = e < i ? -1 : e > i ? 1 : 0, r === 0 ? (h = _, o = _ - 1) : r < 0 ? o = _ - 1 : c = _ + 1;
18
- }
19
- return h;
20
- }
21
- const ve = (n, e, l) => {
22
- const t = n[e];
23
- return t ? typeof t == "function" ? t() : Promise.resolve(t) : new Promise((a, s) => {
24
- (typeof queueMicrotask == "function" ? queueMicrotask : setTimeout)(s.bind(null, /* @__PURE__ */ new Error("Unknown variable dynamic import: " + e + (e.split("/").length !== l ? ". Note that variables only represent file names one level deep." : ""))));
25
- });
26
- }, le = {}, ue = async (n) => (le[n] || (le[n] = ve(/* @__PURE__ */ Object.assign({ "../script_data/Assamese.json": () => import("./Assamese-4WO2wzXg.js"), "../script_data/Bengali.json": () => import("./Bengali-Dhlxv7Pr.js"), "../script_data/Brahmi.json": () => import("./Brahmi-BvC4Luvf.js"), "../script_data/Devanagari.json": () => import("./Devanagari-C4NmXCKb.js"), "../script_data/Granth.json": () => import("./Granth-B3US1gHe.js"), "../script_data/Gujarati.json": () => import("./Gujarati-DfQS2sSY.js"), "../script_data/Gurumukhi.json": () => import("./Gurumukhi-dVJRMJNL.js"), "../script_data/Kannada.json": () => import("./Kannada-ZB5P9u4T.js"), "../script_data/Malayalam.json": () => import("./Malayalam-BQROfOnO.js"), "../script_data/Modi.json": () => import("./Modi-B16t9oPJ.js"), "../script_data/Normal.json": () => import("./Normal-Chi5ccip.js"), "../script_data/Odia.json": () => import("./Odia-hd6WA8ll.js"), "../script_data/Purna-Devanagari.json": () => import("./Purna-Devanagari-CqkVL5zc.js"), "../script_data/Romanized.json": () => import("./Romanized-CQkrk7RV.js"), "../script_data/Sharada.json": () => import("./Sharada-Bk4pUPeN.js"), "../script_data/Siddham.json": () => import("./Siddham-Cf9Jp5KF.js"), "../script_data/Sinhala.json": () => import("./Sinhala-C6t7GaP6.js"), "../script_data/Tamil-Extended.json": () => import("./Tamil-Extended-nmDMmVeX.js"), "../script_data/Tamil.json": () => import("./Tamil-CR4VFQvV.js"), "../script_data/Telugu.json": () => import("./Telugu-BCg-EHjf.js") }), `../script_data/${n}.json`, 3).then(
27
- (e) => e.default
28
- )), le[n]), J = {
29
- "all_to_normal:replace_pancham_varga_varna_with_n": { from_script_type: "all", to_script_name: ["Normal"], check_in: "output", rules: [{ type: "replace_prev_krama_keys", check_in: "output", prev: [60], following: [52, 53, 54, 55, 56, 57, 58, 59], replace_with: [72] }, { type: "replace_prev_krama_keys", check_in: "output", prev: [67], following: [61, 62, 63, 64, 65, 66], replace_with: [72] }] },
30
- "brahmic_to_brahmic:replace_pancham_varga_varna_with_anusvAra": { from_script_type: "brahmic", to_script_type: "brahmic", check_in: "input", rules: [{ type: "replace_prev_krama_keys", check_in: "input", prev: [60, 49], following: [52, 53, 54, 55, 56, 57, 58, 59], replace_with: [45] }, { type: "replace_prev_krama_keys", check_in: "input", prev: [67, 49], following: [61, 62, 63, 64, 65, 66], replace_with: [45] }, { type: "replace_prev_krama_keys", check_in: "input", prev: [81, 49], following: [74, 75, 76, 77, 78, 79, 80], replace_with: [45] }, { type: "replace_prev_krama_keys", check_in: "input", prev: [72, 49], following: [68, 69, 70, 71], replace_with: [45] }, { type: "replace_prev_krama_keys", check_in: "input", prev: [88, 49], following: [82, 83, 85, 86, 87], replace_with: [45] }] },
31
- "all_to_sinhala:use_conjunct_enabling_halant": { from_script_type: "all", to_script_name: ["Sinhala"], check_in: "output", rules: [{ type: "direct_replace", check_in: "output", to_replace: [[49]], replace_text: "්‍", replace_with: [] }] },
32
- "all_to_normal:remove_virAma_and_double_virAma": { from_script_type: "all", to_script_name: ["Normal", "Romanized"], check_in: "output", rules: [{ type: "direct_replace", check_in: "output", to_replace: [[116], [115]], replace_with: [-1] }] },
33
- "all_to_normal:replace_avagraha_with_a": { from_script_type: "all", to_script_name: ["Normal", "Romanized"], check_in: "output", rules: [{ type: "direct_replace", check_in: "output", to_replace: [[48]], replace_with: [1] }] },
34
- "normal_to_all:use_typing_chars": { from_script_name: ["Normal"], to_script_type: "all", check_in: "output", rules: [] },
35
- "all_to_normal:preserve_specific_chars": { from_script_type: "all", to_script_name: ["Normal"], check_in: "output", rules: [] }
36
- }, xe = (n, e) => {
37
- const l = n.krama_text_arr[e]?.[0];
38
- return typeof l == "string" ? l : null;
39
- }, U = (n, e) => xe(n, e) ?? "", Z = (n, e) => Q(
40
- n.krama_text_arr,
41
- n.krama_text_arr_index,
42
- e,
43
- {
44
- accessor: (l, t) => l[t][0]
45
- }
46
- ), Ae = () => {
47
- let n = [];
48
- function e(i) {
49
- i && n.push(i);
50
- }
51
- function l(i) {
52
- for (const r of i) e(r);
53
- }
54
- function t() {
55
- return n.at(-1);
56
- }
57
- function a() {
58
- return t()?.at(-1);
59
- }
60
- function s() {
61
- const i = t();
62
- if (!i) return;
63
- const r = i.at(-1), u = i.slice(0, -1);
64
- return u ? n[n.length - 1] = u : n.pop(), r;
65
- }
66
- function c(i, r) {
67
- n.splice(-i, i, ...r.filter(Boolean));
68
- }
69
- function o(i, r) {
70
- const u = s();
71
- if (!u) {
72
- l(i), l(r);
73
- return;
74
- }
75
- l(i), e(u), l(r);
76
- }
77
- function h(i) {
78
- const r = n.at(i);
79
- return r ? {
80
- cp: i,
81
- ch: r,
82
- width: r.length
83
- } : null;
84
- }
85
- function _(i, r) {
86
- i < 0 && (i = n.length + i), !(i < 0 || i >= n.length) && (n[i] = r);
87
- }
88
- return {
89
- emit: e,
90
- emitPieces: l,
91
- lastPiece: t,
92
- lastChar: a,
93
- popLastChar: s,
94
- rewriteTailPieces: c,
95
- withLastCharMovedAfter: o,
96
- toString: () => n.join(""),
97
- rewriteAt: _,
98
- peekAt: h
99
- };
100
- }, we = (n) => {
101
- let e = [];
102
- function l() {
103
- e = [];
104
- }
105
- function t() {
106
- return e.length;
107
- }
108
- function a(u) {
109
- return e.at(u);
110
- }
111
- function s() {
112
- return e.at(-1);
113
- }
114
- function c() {
115
- return s()?.[0];
116
- }
117
- function o() {
118
- return s()?.[1]?.type;
119
- }
120
- function h(u) {
121
- return a(u)?.[1]?.type;
122
- }
123
- function _(u) {
124
- return a(u)?.[0];
125
- }
126
- function i(u) {
127
- return o() === u;
128
- }
129
- function r(u) {
130
- const m = u[0];
131
- m !== void 0 && m.length > 0 && e.push(u), e.length > n && e.shift();
132
- }
133
- return {
134
- clear: l,
135
- length: t,
136
- at: a,
137
- last: s,
138
- lastText: c,
139
- lastType: o,
140
- typeAt: h,
141
- textAt: _,
142
- isLastType: i,
143
- push: r
144
- };
145
- }, be = (n) => {
146
- let e = 0;
147
- function l(o) {
148
- const h = n.codePointAt(o);
149
- if (h === void 0) return null;
150
- const _ = String.fromCodePoint(h);
151
- return { cp: h, ch: _, width: _.length };
152
- }
153
- function t() {
154
- return l(e);
155
- }
156
- function a(o) {
157
- return l(e + o);
158
- }
159
- function s(o) {
160
- e += o;
161
- }
162
- function c(o, h) {
163
- return n.substring(o, h);
164
- }
165
- return {
166
- get pos() {
167
- return e;
168
- },
169
- peek: t,
170
- peekAt: l,
171
- peekAtOffsetUnits: a,
172
- advance: s,
173
- slice: c
174
- };
175
- }, ce = (n, e, l, t) => {
176
- for (let a = 0; a < l.length; a++) {
177
- const s = l[l.length - 1 - a], c = n(e - a);
178
- if (c === null) return { matched: !1, matchedLen: 0 };
179
- const o = Z(t, c.ch);
180
- if (o === -1 || o !== s)
181
- return { matched: !1, matchedLen: 0 };
182
- }
183
- return { matched: !0, matchedLen: l.length };
184
- }, se = (n, e) => n.map((l) => U(e, l)).filter(Boolean), Te = ["²", "³", "⁴"], S = (n) => !!n && Te.indexOf(
185
- n
186
- ) !== -1, je = ["॒", "॑", "᳚", "᳛"], oe = (n) => !!n && je.indexOf(n) !== -1, L = (n) => n === "Tamil-Extended", H = (n, e, l, t) => {
187
- if (e.length === 0) return;
188
- if (!t) {
189
- n.emitPieces(e);
190
- return;
191
- }
192
- const a = e[0] ?? "";
193
- if (a[0] === l) {
194
- const s = a.slice(1), c = [];
195
- s && c.push(s);
196
- for (let o = 1; o < e.length; o++) c.push(e[o]);
197
- n.withLastCharMovedAfter([l], c);
198
- } else
199
- n.withLastCharMovedAfter(e, []);
200
- }, pe = ["#an", "#sss", "#ss", "#s"], Se = ["↓", "↑↑↑", "↑↑", "↑"], Ne = (n, e) => {
201
- if (!n) return n;
202
- if (n.indexOf("x") !== -1 && (n = n.replaceAll("x", "kSh")), e === "Tamil-Extended")
203
- for (let l = 0; l < pe.length; l++) {
204
- const t = pe[l];
205
- n.indexOf(t) !== -1 && (n = n.replaceAll(t, Se[l]));
206
- }
207
- return n;
208
- }, Le = [" ", `
209
- `, "\r", " ", ",", "~", "!", "@", "?", "%"], Ee = 3;
210
- function j(n, e, l) {
211
- const {
212
- from_script_name: t,
213
- to_script_name: a,
214
- from_script_data: s,
215
- to_script_data: c,
216
- trans_options: o,
217
- result: h,
218
- prev_context: _,
219
- BRAHMIC_HALANT: i,
220
- BRAHMIC_NUQTA: r,
221
- typing_mode: u,
222
- include_inherent_vowel: m
223
- } = n, { next: d, last_extra_call: E } = l ?? {};
224
- let K = !1;
225
- if (
226
- // vyanjana, nuqta, svara
227
- (r && _.typeAt(-3) === "vyanjana" && _.textAt(-2) === r && _.typeAt(-1) === "mAtrA" || // or vyanjana, svara
228
- _.typeAt(-2) === "vyanjana" && _.typeAt(-1) === "mAtrA") && // to anya or null
229
- (!e || e[1]?.type === "anya") && _.clear(), s.script_type === "brahmic" && c.script_type === "other"
230
- )
231
- e[0] !== i && (!(L(t) && e[0] && e[0].length > 0) || e[0].charAt(0) !== i) && // (BRAHMIC_NUQTA ? item[0] !== BRAHMIC_NUQTA : true) &&
232
- (!r || e[0] !== r) && // ^ two special cases to ignore
233
- // vyanjana or vyanjana, nuqta
234
- (_.typeAt(-1) === "vyanjana" || r && _.typeAt(-2) === "vyanjana" && _.textAt(-1) === r) && // to anya or null
235
- (e[1]?.type !== "mAtrA" && e[0] !== i || e[1]?.type === "anya" || e[1] === null || e[1] === void 0) && h.emit(c.schwa_character);
236
- else if (s.script_type === "other" && c.script_type === "brahmic")
237
- if (_.typeAt(-1) === "vyanjana" && (e[1]?.type === "mAtrA" || e[1]?.type === "svara")) {
238
- const y = e[1].type === "svara" ? U(c, e[1].mAtrA_krama_ref?.[0] ?? -1) : e[0];
239
- L(a) && S(h.lastChar()) ? H(h, [y], c.halant, !0) : H(h, [y], c.halant, !1), K = !0;
240
- } else // default transliteration behavior (without schwa deletion)
241
- !m && _.typeAt(-1) === "vyanjana" && !(e[0] === i || e[1]?.type === "mAtrA") ? L(a) && S(h.lastChar()) ? H(h, [i], c.halant, !0) : (H(h, [i], c.halant, !1), a === "Sinhala" && o["all_to_sinhala:use_conjunct_enabling_halant"] && h.rewriteAt(-1, h.lastPiece() + "‍")) : (
242
- // if to include inherent vowel then check when to add that skipped halant
243
- m && e && e[1]?.type === "vyanjana" && (_.typeAt(-1) === "vyanjana" || r && _.typeAt(-2) === "vyanjana" && _.textAt(-1) === r) && (L(a) && S(h.lastChar()) ? H(h, [i], c.halant, !0) : (H(h, [i], c.halant, !1), a === "Sinhala" && o["all_to_sinhala:use_conjunct_enabling_halant"] && h.rewriteAt(-1, h.lastPiece() + "‍")))
244
- );
245
- let q = !1;
246
- return u && (d === void 0 || d.length === 0) && !E && // the case below is to enable typing of #an, #s (Vedic svara chihnas too)
247
- !(L(a) && S(h.lastChar())) && (q = !0, e[1]?.type === "vyanjana" && (q = !1), q && _.clear()), (!u || !E && !q) && _.push(e), K;
248
- }
249
- function he(n, e, l) {
250
- const { custom_rules: t, cursor: a, result: s, from_script_data: c, to_script_data: o } = n, h = e + l;
251
- for (let _ = 0; _ < t.length; _++) {
252
- if (t[_].use_replace === !0) continue;
253
- const i = t[_];
254
- if (i.type === "replace_prev_krama_keys") {
255
- if (i.check_in === "input") {
256
- const r = ce(
257
- a.peekAt,
258
- h,
259
- i.prev,
260
- c
261
- ), u = a.peekAt(e);
262
- if (r.matched && u !== null) {
263
- const m = u.ch, d = Z(c, m);
264
- if (d !== -1 && i.following.indexOf(d) !== -1) {
265
- const E = se(i.replace_with, o);
266
- s.rewriteTailPieces(r.matchedLen, E);
267
- }
268
- }
269
- } else if (i.check_in === "output") {
270
- const r = s.lastPiece();
271
- if (!r) continue;
272
- const u = Z(o, r);
273
- if (u !== -1 && i.following.indexOf(u) !== -1) {
274
- const m = ce(s.peekAt, -2, i.prev, o);
275
- if (m.matched) {
276
- const d = se(i.replace_with, o);
277
- s.rewriteTailPieces(m.matchedLen + 1, [
278
- ...d,
279
- r
280
- ]);
281
- }
282
- }
283
- }
284
- } else if (i.type === "direct_replace") {
285
- const r = i.check_in === "output" ? o : c;
286
- for (const u of i.to_replace) {
287
- const m = ce(s.peekAt, -1, u, r);
288
- if (!m.matched) continue;
289
- if (i.replace_text) {
290
- s.rewriteTailPieces(m.matchedLen, [i.replace_text]);
291
- break;
292
- }
293
- const d = se(i.replace_with, r);
294
- s.rewriteTailPieces(m.matchedLen, d);
295
- break;
296
- }
297
- }
298
- }
299
- }
300
- const Ce = (n, e, l) => {
301
- if (!l) return {};
302
- const t = {};
303
- for (const [a, s] of Object.entries(l ?? {})) {
304
- if (!(a in J)) continue;
305
- const c = J[a];
306
- (c.from_script_type === "all" && c.to_script_type === "all" || (c.from_script_type !== void 0 && (c.from_script_type === "all" || c.from_script_type === n.script_type) || c.from_script_name !== void 0 && c.from_script_name.includes(n.script_name)) && (c.to_script_type !== void 0 && (c.to_script_type === "all" || c.to_script_type === e.script_type) || c.to_script_name !== void 0 && c.to_script_name.includes(e.script_name))) && (t[a] = s);
307
- }
308
- return t;
309
- }, Oe = (n, e, l) => {
310
- const t = Ce(
311
- n,
312
- e,
313
- l
314
- ), a = Object.keys(t).flatMap(
315
- (s) => t[s] === !0 ? J[s].rules : []
316
- );
317
- return { trans_options: t, custom_rules: a };
318
- }, me = (n, e) => n.replace_with.map((l) => U(e, l)).join(""), fe = (n, e, l, t) => {
319
- if (l.length === 0) return n;
320
- for (const a of l)
321
- if (!(a.use_replace !== !0 || a.check_in !== t)) {
322
- if (a.type === "replace_prev_krama_keys") {
323
- const s = a.prev.map((c) => U(e, c)).join("");
324
- for (let c of a.following) {
325
- const o = U(e, c);
326
- if (!o) continue;
327
- const h = me(a, e) + o;
328
- n = n.replaceAll(s + o, h);
329
- }
330
- } else if (a.type === "direct_replace") {
331
- const s = a.to_replace.map(
332
- (c) => c.map((o) => U(e, o)).join("")
333
- );
334
- for (let c of s)
335
- n = n.replaceAll(c, me(a, e));
336
- }
337
- }
338
- return n;
339
- }, Pe = !0, Me = !1, De = (n, e, l, t, a, s, c, o) => {
340
- const h = o?.useNativeNumerals ?? Pe, _ = o?.typing_mode ?? !1, i = o?.includeInherentVowel ?? Me;
341
- if (_ && e !== "Normal")
342
- throw new Error("Typing mode is only supported with Normal script as the input");
343
- _ && (s["normal_to_all:use_typing_chars"] = !0), _ && e === "Normal" && (n = Ne(n, l)), n = fe(n, t, c, "input");
344
- const r = Ae();
345
- let u = 0;
346
- const m = be(n), d = we(Ee), E = t.script_type === "brahmic" && a.script_type === "other" || t.script_type === "other" && a.script_type === "brahmic" || _ && e === "Normal" && a.script_type === "other", K = t.script_type === "brahmic" && a.script_type === "other" ? t.nuqta ?? null : t.script_type === "other" && a.script_type === "brahmic" ? a.nuqta ?? null : null, q = t.script_type === "brahmic" && a.script_type === "other" ? t.halant : t.script_type === "other" && a.script_type === "brahmic" ? a.halant : null, y = {
347
- from_script_name: e,
348
- to_script_name: l,
349
- from_script_data: t,
350
- to_script_data: a,
351
- trans_options: s,
352
- custom_rules: c,
353
- cursor: m,
354
- result: r,
355
- prev_context: d,
356
- BRAHMIC_NUQTA: K,
357
- BRAHMIC_HALANT: q,
358
- typing_mode: _,
359
- include_inherent_vowel: i
360
- }, ee = (s["normal_to_all:use_typing_chars"] || _) && e === "Normal" ? a.typing_text_to_krama_map : t.text_to_krama_map;
361
- let v = -1;
362
- for (; m.pos < n.length; ) {
363
- u = m.pos;
364
- const ne = m.peek();
365
- if (!ne) break;
366
- const V = ne.ch, z = ne.width, _e = z === 2 ? 1 : z;
367
- if (v !== -1 && u >= v) {
368
- v = -1, m.advance(z);
369
- continue;
370
- }
371
- if (Le.indexOf(V) !== -1) {
372
- m.advance(z), E && (j(y, [" ", null]), d.clear()), r.emit(V);
373
- continue;
374
- }
375
- if (V.match(/^\d$/) && !h) {
376
- r.emit(V), m.advance(z), j(y, [V, null]);
377
- continue;
378
- }
379
- if (s["all_to_normal:preserve_specific_chars"] && l === "Normal") {
380
- const x = Y(
381
- t.custom_script_chars_arr,
382
- V,
383
- {
384
- accessor: (A, T) => A[T][0]
385
- }
386
- );
387
- if (x !== -1) {
388
- const A = t.custom_script_chars_arr[x];
389
- j(y, [
390
- A[0],
391
- t.list[A[1] ?? -1] ?? null
392
- ]);
393
- const T = t.typing_text_to_krama_map[A[2] ?? -1]?.[0];
394
- r.emit(T ?? ""), m.advance(A[0].length);
395
- continue;
396
- }
397
- }
398
- let R = -1;
399
- {
400
- let x = 0, A = -1;
401
- const T = E && t.script_type === "other" && a.script_type === "brahmic" && (d.typeAt(-1) === "vyanjana" || K && d.typeAt(-2) === "vyanjana" && d.textAt(-1) === K);
402
- for (; ; ) {
403
- const G = m.peekAt(u + _e + x), $ = G?.ch ?? "";
404
- v !== -1 && $ && S($) && (x += G?.width ?? 0);
405
- const w = u + _e + x, f = (
406
- // usage example: க்⁴ரு² -> ghR
407
- v !== -1 ? (
408
- // in this we will ignore the current charcter and scan one ahead
409
- m.slice(u, v) + (w > v ? m.slice(v + 1, w) : "")
410
- ) : m.slice(u, w)
411
- ), N = Y(ee, f, {
412
- accessor: (b, C) => b[C][0]
413
- });
414
- if (N === -1) {
415
- R = -1;
416
- break;
417
- }
418
- const k = ee[N];
419
- if (T && k[1].krama && k[1].krama.length >= 1) {
420
- const b = k[1].krama, C = b[0], O = a.krama_text_arr[C]?.[1], re = a.list[O ?? -1]?.type;
421
- if (b.length === 1 && O !== null && O !== void 0 && (re === "svara" || re === "mAtrA"))
422
- A = N;
423
- else if (A !== -1) {
424
- R = A;
425
- break;
426
- }
427
- }
428
- if (k[1].next && k[1].next.length > 0) {
429
- const b = m.peekAt(w), C = b?.ch;
430
- if (L(e) && t.script_type === "brahmic") {
431
- const F = b ? m.peekAt(w + b.width) : null, O = F?.ch, M = (b && F ? m.peekAt(w + b.width + F.width) : null)?.ch;
432
- if (v === -1 && O !== void 0 && S(O) && k[1].next.indexOf(O) !== -1) {
433
- const B = Y(
434
- t.text_to_krama_map,
435
- f + O,
436
- {
437
- accessor: (P, g) => P[g][0]
438
- }
439
- ), D = Q(
440
- t.krama_text_arr,
441
- t.krama_text_arr_index,
442
- C ?? "",
443
- {
444
- accessor: (P, g) => P[g][0]
445
- }
446
- );
447
- if (B !== -1 && D !== -1) {
448
- R = B;
449
- const P = t.krama_text_arr[D], g = t.list[P[1] ?? -1]?.type;
450
- if (C === t.halant || g === "mAtrA") {
451
- v = w + (b?.width ?? 0);
452
- break;
453
- }
454
- }
455
- } else if (v === -1 && M !== void 0 && S(M) && k[1].next.indexOf(M) !== -1) {
456
- const B = Y(
457
- t.text_to_krama_map,
458
- f + M,
459
- {
460
- accessor: (g, I) => g[I][0]
461
- }
462
- ), D = Q(
463
- t.krama_text_arr,
464
- t.krama_text_arr_index,
465
- C ?? "",
466
- {
467
- accessor: (g, I) => g[I][0]
468
- }
469
- ), P = Q(
470
- t.krama_text_arr,
471
- t.krama_text_arr_index,
472
- O ?? "",
473
- {
474
- accessor: (g, I) => g[I][0]
475
- }
476
- );
477
- if (B !== -1 && D !== -1 && P !== -1) {
478
- const g = t.krama_text_arr[D], I = t.krama_text_arr[P];
479
- R = B;
480
- const ie = t.list[g[1] ?? -1]?.type, ge = t.list[I[1] ?? -1]?.type;
481
- if (ie === "mAtrA" && ge === "mAtrA") {
482
- v = w + (b?.width ?? 0) + (F?.width ?? 0);
483
- break;
484
- }
485
- }
486
- }
487
- if (v === -1 && C !== void 0 && O !== void 0 && oe(O) && M !== void 0 && S(M) && k[1].next.indexOf(M) !== -1) {
488
- const B = Q(
489
- t.krama_text_arr,
490
- t.krama_text_arr_index,
491
- C ?? "",
492
- { accessor: (D, P) => D[P][0] }
493
- );
494
- if (B !== -1) {
495
- const D = t.krama_text_arr[B];
496
- if (t.list[D[1] ?? -1]?.type === "mAtrA") {
497
- const g = Y(
498
- t.text_to_krama_map,
499
- f + M,
500
- { accessor: (I, ie) => I[ie][0] }
501
- );
502
- if (g !== -1) {
503
- R = g, v = w + (b?.width ?? 0) + (F?.width ?? 0);
504
- break;
505
- }
506
- }
507
- }
508
- }
509
- }
510
- if (C !== void 0 && k[1].next.indexOf(C) !== -1) {
511
- x += b?.width ?? 0;
512
- continue;
513
- }
514
- }
515
- R = N;
516
- break;
517
- }
518
- }
519
- const p = R !== -1 ? ee[R] : null;
520
- if (p !== null) {
521
- const x = v !== -1 && p[0].length > 1 && t.list[t.krama_text_arr[p[1].krama?.[0] ?? -1]?.[1] ?? -1]?.type === "vyanjana" && S(p[0].at(-1) ?? void 0) ? 1 : 0, A = p[0].length - x;
522
- if (m.advance(A), u = m.pos, s["normal_to_all:use_typing_chars"] && "custom_back_ref" in p[1] && p[1].custom_back_ref !== void 0 && p[1].custom_back_ref !== null) {
523
- const T = a.custom_script_chars_arr[p[1].custom_back_ref] ?? null;
524
- J !== null && (r.emit(T[0]), j(
525
- y,
526
- [p[0], a.list[T[1] ?? -1] ?? null],
527
- {
528
- next: p[1].next ?? void 0
529
- }
530
- ));
531
- continue;
532
- }
533
- if (p[1].krama !== null && p[1].krama !== void 0 && p[1].krama.some((T) => T !== -1)) {
534
- const G = p[1].krama.map(
535
- (f) => U(a, f)
536
- // revert to the original character if the krama index is not found
537
- ), $ = G.join("");
538
- let w = !1;
539
- if (E)
540
- if (t.script_type === "brahmic" && a.script_type === "other") {
541
- let f = null;
542
- if (!s["normal_to_all:use_typing_chars"] && p[1].fallback_list_ref !== void 0 && p[1].fallback_list_ref !== null)
543
- f = t.list[p[1].fallback_list_ref];
544
- else if (!p[1].krama || p[1].krama.length === 0)
545
- f = null;
546
- else {
547
- const N = p[1].krama.map(
548
- (k) => t.list[t.krama_text_arr[k][1] ?? -1]
549
- );
550
- L(e) && N.some((k) => k?.type === "mAtrA") && N.some((k) => k?.type === "vyanjana") ? f = { ...N[0], type: "anya" } : L(e) && N.length > 1 && N.some((k) => k == null) ? f = N.at(-1) : f = N[0];
551
- }
552
- w = j(y, [p[0], f]);
553
- } else if (a.script_type === "brahmic" && t.script_type === "other") {
554
- let f = null;
555
- p[1].fallback_list_ref !== void 0 && p[1].fallback_list_ref !== null ? f = a.list[p[1].fallback_list_ref] : f = p[1].krama && p[1].krama.length > 0 ? a.list[a.krama_text_arr[p[1].krama[0]]?.[1] ?? -1] ?? null : null, _ && e === "Normal" ? w = j(y, [p[0], f], {
556
- next: p[1].next ?? void 0
557
- }) : w = j(y, [p[0], f]);
558
- } else _ && e === "Normal" && a.script_type === "other" && (w = j(y, [p[0], null], {
559
- next: p[1].next ?? void 0
560
- }));
561
- if (!w)
562
- if (a.script_type === "brahmic" && L(l) && (a.list[p[1].krama?.at(-1) ?? -1]?.type === "mAtrA" || $ === a.halant) && S(r.lastChar()))
563
- H(
564
- r,
565
- G,
566
- a.halant,
567
- !0
568
- );
569
- else if (L(l) && oe(G.at(-1)?.at(-1) ?? "") && S(r.lastChar())) {
570
- const f = r.popLastChar();
571
- r.emitPieces(G), r.emit(f ?? "");
572
- } else
573
- r.emitPieces(G);
574
- he(y, u, -A);
575
- continue;
576
- } else if (p[1].krama !== null && p[1].krama !== void 0 && p[1].krama.some((T) => T === -1)) {
577
- r.emit(p[0]), _ && j(y, [p[0], null], {
578
- next: p[1].next ?? void 0
579
- });
580
- continue;
581
- }
582
- } else
583
- m.advance(z), u = m.pos;
584
- const W = p === null ? V : p[0], X = Z(t, W);
585
- if (X === -1) {
586
- E && (j(y, [W, null]), d.clear()), r.emit(W);
587
- continue;
588
- }
589
- let ae = !1;
590
- if (E && (t.script_type === "brahmic" ? ae = j(y, [
591
- W,
592
- t.list[t.krama_text_arr[X][1] ?? -1]
593
- ]) : a.script_type === "brahmic" && (ae = j(y, [
594
- W,
595
- a.list[a.krama_text_arr[X][1] ?? -1]
596
- ]))), !ae) {
597
- const x = U(a, X);
598
- if (a.script_type === "brahmic" && L(l) && (a.list[a.krama_text_arr[X][1] ?? -1]?.type === "mAtrA" || x === a.halant) && S(r.lastChar()))
599
- H(r, [x], a.halant, !0);
600
- else if (L(l) && oe(x) && S(r.lastChar())) {
601
- const A = r.popLastChar();
602
- r.emit(x), r.emit(A ?? "");
603
- } else
604
- r.emit(x);
605
- }
606
- he(y, u, -z);
607
- }
608
- E && j(y, [void 0, null], { last_extra_call: !0 });
609
- let te = r.toString();
610
- return te = fe(te, a, c, "output"), {
611
- output: te,
612
- /** Can be used to manage context while using the typing feature.
613
- * If this is 0, the external context can be cleared
614
- */
615
- context_length: d.length()
616
- };
617
- }, ze = async (n, e, l, t, a) => {
618
- const [s, c] = await Promise.all([
619
- ue(e),
620
- ue(l)
621
- ]), { trans_options: o, custom_rules: h } = Oe(
622
- s,
623
- c,
624
- t
625
- );
626
- return De(
627
- n,
628
- e,
629
- l,
630
- s,
631
- c,
632
- o,
633
- h,
634
- a
635
- );
636
- }, Ie = { Devanagari: 1, Telugu: 2, Tamil: 3, "Tamil-Extended": 4, Bengali: 5, Kannada: 6, Gujarati: 7, Malayalam: 8, Odia: 9, Sinhala: 10, Normal: 11, Romanized: 12, Gurumukhi: 13, Assamese: 14, "Purna-Devanagari": 15, Brahmi: 16, Granth: 17, Modi: 18, Sharada: 20, Siddham: 21 }, Re = { English: 1, Sanskrit: 2, Hindi: 3, Telugu: 4, Tamil: 5, Bengali: 6, Kannada: 7, Gujarati: 8, Malayalam: 9, Odia: 10, Sinhala: 11, Marathi: 12, Nepali: 13, Punjabi: 14, Assamese: 15 }, Ge = {
637
- script_list: Ie,
638
- lang_list: Re
639
- }, { script_list: Be, lang_list: He } = Ge, ke = Object.keys(Be), de = Object.keys(He), Ke = Array.from(
640
- /* @__PURE__ */ new Set([...de, ...ke])
641
- ), Ue = {
642
- Bengali: "Bengali",
643
- Gujarati: "Gujarati",
644
- Hindi: "Devanagari",
645
- Kannada: "Kannada",
646
- Malayalam: "Malayalam",
647
- Odia: "Odia",
648
- Sinhala: "Sinhala",
649
- Tamil: "Tamil",
650
- Telugu: "Telugu",
651
- English: "Normal",
652
- Sanskrit: "Devanagari",
653
- Marathi: "Devanagari",
654
- Nepali: "Devanagari",
655
- Punjabi: "Gurumukhi",
656
- Assamese: "Assamese"
657
- }, ye = {
658
- dev: "Devanagari",
659
- te: "Telugu",
660
- tel: "Telugu",
661
- tam: "Tamil",
662
- "tam-ex": "Tamil-Extended",
663
- ben: "Bengali",
664
- be: "Bengali",
665
- ka: "Kannada",
666
- kan: "Kannada",
667
- gu: "Gujarati",
668
- guj: "Gujarati",
669
- mal: "Malayalam",
670
- or: "Odia",
671
- od: "Odia",
672
- oriya: "Odia",
673
- si: "Sinhala",
674
- sinh: "Sinhala",
675
- sin: "Sinhala",
676
- en: "Normal",
677
- eng: "Normal",
678
- la: "Normal",
679
- lat: "Normal",
680
- nor: "Normal",
681
- norm: "Normal",
682
- rom: "Romanized",
683
- gur: "Gurumukhi",
684
- as: "Assamese",
685
- sa: "Devanagari",
686
- san: "Devanagari",
687
- hin: "Devanagari",
688
- hi: "Devanagari",
689
- mar: "Devanagari",
690
- ne: "Devanagari",
691
- nep: "Devanagari",
692
- pun: "Gurumukhi"
693
- };
694
- function Ve(n) {
695
- return n.toLowerCase().replace(/(^|-)([a-z])/g, (e, l, t) => l + t.toUpperCase());
696
- }
697
- const qe = (n) => {
698
- const e = Ve(n);
699
- return ke.includes(e) ? e : de.includes(e) ? Ue[e] : n.toLocaleLowerCase() in ye ? ye[n.toLocaleLowerCase()] : null;
700
- };
701
- export {
702
- Ke as A,
703
- de as L,
704
- ke as S,
705
- ue as a,
706
- Ce as b,
707
- J as c,
708
- De as d,
709
- qe as g,
710
- Oe as r,
711
- ze as t
712
- };