@budhash/gomanize 1.0.0

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/README.md ADDED
@@ -0,0 +1,47 @@
1
+ # @budhash/gomanize
2
+
3
+ Romanize Hindi (Devanagari) into readable, diacritic-free Latin script — the
4
+ [gomanize](https://github.com/budhash/gomanize) engine (Go) compiled to
5
+ WebAssembly, with a thin JS/TS wrapper. It is the **same engine as the CLI**, so
6
+ output is identical; nothing is reimplemented.
7
+
8
+ Live demo: <https://budhash.com/gomanize>
9
+
10
+ ## Install
11
+
12
+ ```bash
13
+ npm i @budhash/gomanize
14
+ ```
15
+
16
+ ## Use
17
+
18
+ ```js
19
+ import { load } from "@budhash/gomanize";
20
+
21
+ const g = await load(); // instantiate once
22
+ g.translit("नमस्ते दुनिया"); // "namaste duniya"
23
+ g.translit("गाना", { longVowels: true }); // "gaanaa"
24
+ g.translit("मेरे सपनों की रानी", { rerank: true }); // "mere sapnon ki rani"
25
+ ```
26
+
27
+ CommonJS works too: `const { load } = require("@budhash/gomanize");`
28
+
29
+ ## Options
30
+
31
+ All optional, default `false`:
32
+
33
+ | Group | Option | Effect |
34
+ |-------|--------|--------|
35
+ | rule tweak | `longVowels` | write "aa" for every ā (गाना → gaanaa) |
36
+ | rule tweak | `simpleNasals` | करें → karen, not karein |
37
+ | rule tweak | `keepMedialSchwa` | disable CCV deletion (जनता → janata) |
38
+ | learned | `schwaModel` | decision-tree schwa classifier |
39
+ | learned | `lexicon` | prefer attested spellings |
40
+ | learned | `rerank` | re-rank candidates with a char n-gram model |
41
+
42
+ ## Notes
43
+
44
+ - `load()` is async (instantiates the wasm) and cached; `translit` is synchronous after.
45
+ - The `.wasm` is ~1.4 MB gzipped — it embeds the schwa model, lexicon, and n-gram data.
46
+ - Runs on **Node ≥ 18** and modern browsers/bundlers (the wasm + `wasm_exec.js` are resolved relative to the module).
47
+ - Romanization, not strict transliteration — see the [project README](https://github.com/budhash/gomanize). MIT licensed.
Binary file
@@ -0,0 +1,575 @@
1
+ // Copyright 2018 The Go Authors. All rights reserved.
2
+ // Use of this source code is governed by a BSD-style
3
+ // license that can be found in the LICENSE file.
4
+
5
+ "use strict";
6
+
7
+ (() => {
8
+ const enosys = () => {
9
+ const err = new Error("not implemented");
10
+ err.code = "ENOSYS";
11
+ return err;
12
+ };
13
+
14
+ if (!globalThis.fs) {
15
+ let outputBuf = "";
16
+ globalThis.fs = {
17
+ constants: { O_WRONLY: -1, O_RDWR: -1, O_CREAT: -1, O_TRUNC: -1, O_APPEND: -1, O_EXCL: -1, O_DIRECTORY: -1 }, // unused
18
+ writeSync(fd, buf) {
19
+ outputBuf += decoder.decode(buf);
20
+ const nl = outputBuf.lastIndexOf("\n");
21
+ if (nl != -1) {
22
+ console.log(outputBuf.substring(0, nl));
23
+ outputBuf = outputBuf.substring(nl + 1);
24
+ }
25
+ return buf.length;
26
+ },
27
+ write(fd, buf, offset, length, position, callback) {
28
+ if (offset !== 0 || length !== buf.length || position !== null) {
29
+ callback(enosys());
30
+ return;
31
+ }
32
+ const n = this.writeSync(fd, buf);
33
+ callback(null, n);
34
+ },
35
+ chmod(path, mode, callback) { callback(enosys()); },
36
+ chown(path, uid, gid, callback) { callback(enosys()); },
37
+ close(fd, callback) { callback(enosys()); },
38
+ fchmod(fd, mode, callback) { callback(enosys()); },
39
+ fchown(fd, uid, gid, callback) { callback(enosys()); },
40
+ fstat(fd, callback) { callback(enosys()); },
41
+ fsync(fd, callback) { callback(null); },
42
+ ftruncate(fd, length, callback) { callback(enosys()); },
43
+ lchown(path, uid, gid, callback) { callback(enosys()); },
44
+ link(path, link, callback) { callback(enosys()); },
45
+ lstat(path, callback) { callback(enosys()); },
46
+ mkdir(path, perm, callback) { callback(enosys()); },
47
+ open(path, flags, mode, callback) { callback(enosys()); },
48
+ read(fd, buffer, offset, length, position, callback) { callback(enosys()); },
49
+ readdir(path, callback) { callback(enosys()); },
50
+ readlink(path, callback) { callback(enosys()); },
51
+ rename(from, to, callback) { callback(enosys()); },
52
+ rmdir(path, callback) { callback(enosys()); },
53
+ stat(path, callback) { callback(enosys()); },
54
+ symlink(path, link, callback) { callback(enosys()); },
55
+ truncate(path, length, callback) { callback(enosys()); },
56
+ unlink(path, callback) { callback(enosys()); },
57
+ utimes(path, atime, mtime, callback) { callback(enosys()); },
58
+ };
59
+ }
60
+
61
+ if (!globalThis.process) {
62
+ globalThis.process = {
63
+ getuid() { return -1; },
64
+ getgid() { return -1; },
65
+ geteuid() { return -1; },
66
+ getegid() { return -1; },
67
+ getgroups() { throw enosys(); },
68
+ pid: -1,
69
+ ppid: -1,
70
+ umask() { throw enosys(); },
71
+ cwd() { throw enosys(); },
72
+ chdir() { throw enosys(); },
73
+ }
74
+ }
75
+
76
+ if (!globalThis.path) {
77
+ globalThis.path = {
78
+ resolve(...pathSegments) {
79
+ return pathSegments.join("/");
80
+ }
81
+ }
82
+ }
83
+
84
+ if (!globalThis.crypto) {
85
+ throw new Error("globalThis.crypto is not available, polyfill required (crypto.getRandomValues only)");
86
+ }
87
+
88
+ if (!globalThis.performance) {
89
+ throw new Error("globalThis.performance is not available, polyfill required (performance.now only)");
90
+ }
91
+
92
+ if (!globalThis.TextEncoder) {
93
+ throw new Error("globalThis.TextEncoder is not available, polyfill required");
94
+ }
95
+
96
+ if (!globalThis.TextDecoder) {
97
+ throw new Error("globalThis.TextDecoder is not available, polyfill required");
98
+ }
99
+
100
+ const encoder = new TextEncoder("utf-8");
101
+ const decoder = new TextDecoder("utf-8");
102
+
103
+ globalThis.Go = class {
104
+ constructor() {
105
+ this.argv = ["js"];
106
+ this.env = {};
107
+ this.exit = (code) => {
108
+ if (code !== 0) {
109
+ console.warn("exit code:", code);
110
+ }
111
+ };
112
+ this._exitPromise = new Promise((resolve) => {
113
+ this._resolveExitPromise = resolve;
114
+ });
115
+ this._pendingEvent = null;
116
+ this._scheduledTimeouts = new Map();
117
+ this._nextCallbackTimeoutID = 1;
118
+
119
+ const setInt64 = (addr, v) => {
120
+ this.mem.setUint32(addr + 0, v, true);
121
+ this.mem.setUint32(addr + 4, Math.floor(v / 4294967296), true);
122
+ }
123
+
124
+ const setInt32 = (addr, v) => {
125
+ this.mem.setUint32(addr + 0, v, true);
126
+ }
127
+
128
+ const getInt64 = (addr) => {
129
+ const low = this.mem.getUint32(addr + 0, true);
130
+ const high = this.mem.getInt32(addr + 4, true);
131
+ return low + high * 4294967296;
132
+ }
133
+
134
+ const loadValue = (addr) => {
135
+ const f = this.mem.getFloat64(addr, true);
136
+ if (f === 0) {
137
+ return undefined;
138
+ }
139
+ if (!isNaN(f)) {
140
+ return f;
141
+ }
142
+
143
+ const id = this.mem.getUint32(addr, true);
144
+ return this._values[id];
145
+ }
146
+
147
+ const storeValue = (addr, v) => {
148
+ const nanHead = 0x7FF80000;
149
+
150
+ if (typeof v === "number" && v !== 0) {
151
+ if (isNaN(v)) {
152
+ this.mem.setUint32(addr + 4, nanHead, true);
153
+ this.mem.setUint32(addr, 0, true);
154
+ return;
155
+ }
156
+ this.mem.setFloat64(addr, v, true);
157
+ return;
158
+ }
159
+
160
+ if (v === undefined) {
161
+ this.mem.setFloat64(addr, 0, true);
162
+ return;
163
+ }
164
+
165
+ let id = this._ids.get(v);
166
+ if (id === undefined) {
167
+ id = this._idPool.pop();
168
+ if (id === undefined) {
169
+ id = this._values.length;
170
+ }
171
+ this._values[id] = v;
172
+ this._goRefCounts[id] = 0;
173
+ this._ids.set(v, id);
174
+ }
175
+ this._goRefCounts[id]++;
176
+ let typeFlag = 0;
177
+ switch (typeof v) {
178
+ case "object":
179
+ if (v !== null) {
180
+ typeFlag = 1;
181
+ }
182
+ break;
183
+ case "string":
184
+ typeFlag = 2;
185
+ break;
186
+ case "symbol":
187
+ typeFlag = 3;
188
+ break;
189
+ case "function":
190
+ typeFlag = 4;
191
+ break;
192
+ }
193
+ this.mem.setUint32(addr + 4, nanHead | typeFlag, true);
194
+ this.mem.setUint32(addr, id, true);
195
+ }
196
+
197
+ const loadSlice = (addr) => {
198
+ const array = getInt64(addr + 0);
199
+ const len = getInt64(addr + 8);
200
+ return new Uint8Array(this._inst.exports.mem.buffer, array, len);
201
+ }
202
+
203
+ const loadSliceOfValues = (addr) => {
204
+ const array = getInt64(addr + 0);
205
+ const len = getInt64(addr + 8);
206
+ const a = new Array(len);
207
+ for (let i = 0; i < len; i++) {
208
+ a[i] = loadValue(array + i * 8);
209
+ }
210
+ return a;
211
+ }
212
+
213
+ const loadString = (addr) => {
214
+ const saddr = getInt64(addr + 0);
215
+ const len = getInt64(addr + 8);
216
+ return decoder.decode(new DataView(this._inst.exports.mem.buffer, saddr, len));
217
+ }
218
+
219
+ const testCallExport = (a, b) => {
220
+ this._inst.exports.testExport0();
221
+ return this._inst.exports.testExport(a, b);
222
+ }
223
+
224
+ const timeOrigin = Date.now() - performance.now();
225
+ this.importObject = {
226
+ _gotest: {
227
+ add: (a, b) => a + b,
228
+ callExport: testCallExport,
229
+ },
230
+ gojs: {
231
+ // Go's SP does not change as long as no Go code is running. Some operations (e.g. calls, getters and setters)
232
+ // may synchronously trigger a Go event handler. This makes Go code get executed in the middle of the imported
233
+ // function. A goroutine can switch to a new stack if the current stack is too small (see morestack function).
234
+ // This changes the SP, thus we have to update the SP used by the imported function.
235
+
236
+ // func wasmExit(code int32)
237
+ "runtime.wasmExit": (sp) => {
238
+ sp >>>= 0;
239
+ const code = this.mem.getInt32(sp + 8, true);
240
+ this.exited = true;
241
+ delete this._inst;
242
+ delete this._values;
243
+ delete this._goRefCounts;
244
+ delete this._ids;
245
+ delete this._idPool;
246
+ this.exit(code);
247
+ },
248
+
249
+ // func wasmWrite(fd uintptr, p unsafe.Pointer, n int32)
250
+ "runtime.wasmWrite": (sp) => {
251
+ sp >>>= 0;
252
+ const fd = getInt64(sp + 8);
253
+ const p = getInt64(sp + 16);
254
+ const n = this.mem.getInt32(sp + 24, true);
255
+ fs.writeSync(fd, new Uint8Array(this._inst.exports.mem.buffer, p, n));
256
+ },
257
+
258
+ // func resetMemoryDataView()
259
+ "runtime.resetMemoryDataView": (sp) => {
260
+ sp >>>= 0;
261
+ this.mem = new DataView(this._inst.exports.mem.buffer);
262
+ },
263
+
264
+ // func nanotime1() int64
265
+ "runtime.nanotime1": (sp) => {
266
+ sp >>>= 0;
267
+ setInt64(sp + 8, (timeOrigin + performance.now()) * 1000000);
268
+ },
269
+
270
+ // func walltime() (sec int64, nsec int32)
271
+ "runtime.walltime": (sp) => {
272
+ sp >>>= 0;
273
+ const msec = (new Date).getTime();
274
+ setInt64(sp + 8, msec / 1000);
275
+ this.mem.setInt32(sp + 16, (msec % 1000) * 1000000, true);
276
+ },
277
+
278
+ // func scheduleTimeoutEvent(delay int64) int32
279
+ "runtime.scheduleTimeoutEvent": (sp) => {
280
+ sp >>>= 0;
281
+ const id = this._nextCallbackTimeoutID;
282
+ this._nextCallbackTimeoutID++;
283
+ this._scheduledTimeouts.set(id, setTimeout(
284
+ () => {
285
+ this._resume();
286
+ while (this._scheduledTimeouts.has(id)) {
287
+ // for some reason Go failed to register the timeout event, log and try again
288
+ // (temporary workaround for https://github.com/golang/go/issues/28975)
289
+ console.warn("scheduleTimeoutEvent: missed timeout event");
290
+ this._resume();
291
+ }
292
+ },
293
+ getInt64(sp + 8),
294
+ ));
295
+ this.mem.setInt32(sp + 16, id, true);
296
+ },
297
+
298
+ // func clearTimeoutEvent(id int32)
299
+ "runtime.clearTimeoutEvent": (sp) => {
300
+ sp >>>= 0;
301
+ const id = this.mem.getInt32(sp + 8, true);
302
+ clearTimeout(this._scheduledTimeouts.get(id));
303
+ this._scheduledTimeouts.delete(id);
304
+ },
305
+
306
+ // func getRandomData(r []byte)
307
+ "runtime.getRandomData": (sp) => {
308
+ sp >>>= 0;
309
+ crypto.getRandomValues(loadSlice(sp + 8));
310
+ },
311
+
312
+ // func finalizeRef(v ref)
313
+ "syscall/js.finalizeRef": (sp) => {
314
+ sp >>>= 0;
315
+ const id = this.mem.getUint32(sp + 8, true);
316
+ this._goRefCounts[id]--;
317
+ if (this._goRefCounts[id] === 0) {
318
+ const v = this._values[id];
319
+ this._values[id] = null;
320
+ this._ids.delete(v);
321
+ this._idPool.push(id);
322
+ }
323
+ },
324
+
325
+ // func stringVal(value string) ref
326
+ "syscall/js.stringVal": (sp) => {
327
+ sp >>>= 0;
328
+ storeValue(sp + 24, loadString(sp + 8));
329
+ },
330
+
331
+ // func valueGet(v ref, p string) ref
332
+ "syscall/js.valueGet": (sp) => {
333
+ sp >>>= 0;
334
+ const result = Reflect.get(loadValue(sp + 8), loadString(sp + 16));
335
+ sp = this._inst.exports.getsp() >>> 0; // see comment above
336
+ storeValue(sp + 32, result);
337
+ },
338
+
339
+ // func valueSet(v ref, p string, x ref)
340
+ "syscall/js.valueSet": (sp) => {
341
+ sp >>>= 0;
342
+ Reflect.set(loadValue(sp + 8), loadString(sp + 16), loadValue(sp + 32));
343
+ },
344
+
345
+ // func valueDelete(v ref, p string)
346
+ "syscall/js.valueDelete": (sp) => {
347
+ sp >>>= 0;
348
+ Reflect.deleteProperty(loadValue(sp + 8), loadString(sp + 16));
349
+ },
350
+
351
+ // func valueIndex(v ref, i int) ref
352
+ "syscall/js.valueIndex": (sp) => {
353
+ sp >>>= 0;
354
+ storeValue(sp + 24, Reflect.get(loadValue(sp + 8), getInt64(sp + 16)));
355
+ },
356
+
357
+ // valueSetIndex(v ref, i int, x ref)
358
+ "syscall/js.valueSetIndex": (sp) => {
359
+ sp >>>= 0;
360
+ Reflect.set(loadValue(sp + 8), getInt64(sp + 16), loadValue(sp + 24));
361
+ },
362
+
363
+ // func valueCall(v ref, m string, args []ref) (ref, bool)
364
+ "syscall/js.valueCall": (sp) => {
365
+ sp >>>= 0;
366
+ try {
367
+ const v = loadValue(sp + 8);
368
+ const m = Reflect.get(v, loadString(sp + 16));
369
+ const args = loadSliceOfValues(sp + 32);
370
+ const result = Reflect.apply(m, v, args);
371
+ sp = this._inst.exports.getsp() >>> 0; // see comment above
372
+ storeValue(sp + 56, result);
373
+ this.mem.setUint8(sp + 64, 1);
374
+ } catch (err) {
375
+ sp = this._inst.exports.getsp() >>> 0; // see comment above
376
+ storeValue(sp + 56, err);
377
+ this.mem.setUint8(sp + 64, 0);
378
+ }
379
+ },
380
+
381
+ // func valueInvoke(v ref, args []ref) (ref, bool)
382
+ "syscall/js.valueInvoke": (sp) => {
383
+ sp >>>= 0;
384
+ try {
385
+ const v = loadValue(sp + 8);
386
+ const args = loadSliceOfValues(sp + 16);
387
+ const result = Reflect.apply(v, undefined, args);
388
+ sp = this._inst.exports.getsp() >>> 0; // see comment above
389
+ storeValue(sp + 40, result);
390
+ this.mem.setUint8(sp + 48, 1);
391
+ } catch (err) {
392
+ sp = this._inst.exports.getsp() >>> 0; // see comment above
393
+ storeValue(sp + 40, err);
394
+ this.mem.setUint8(sp + 48, 0);
395
+ }
396
+ },
397
+
398
+ // func valueNew(v ref, args []ref) (ref, bool)
399
+ "syscall/js.valueNew": (sp) => {
400
+ sp >>>= 0;
401
+ try {
402
+ const v = loadValue(sp + 8);
403
+ const args = loadSliceOfValues(sp + 16);
404
+ const result = Reflect.construct(v, args);
405
+ sp = this._inst.exports.getsp() >>> 0; // see comment above
406
+ storeValue(sp + 40, result);
407
+ this.mem.setUint8(sp + 48, 1);
408
+ } catch (err) {
409
+ sp = this._inst.exports.getsp() >>> 0; // see comment above
410
+ storeValue(sp + 40, err);
411
+ this.mem.setUint8(sp + 48, 0);
412
+ }
413
+ },
414
+
415
+ // func valueLength(v ref) int
416
+ "syscall/js.valueLength": (sp) => {
417
+ sp >>>= 0;
418
+ setInt64(sp + 16, parseInt(loadValue(sp + 8).length));
419
+ },
420
+
421
+ // valuePrepareString(v ref) (ref, int)
422
+ "syscall/js.valuePrepareString": (sp) => {
423
+ sp >>>= 0;
424
+ const str = encoder.encode(String(loadValue(sp + 8)));
425
+ storeValue(sp + 16, str);
426
+ setInt64(sp + 24, str.length);
427
+ },
428
+
429
+ // valueLoadString(v ref, b []byte)
430
+ "syscall/js.valueLoadString": (sp) => {
431
+ sp >>>= 0;
432
+ const str = loadValue(sp + 8);
433
+ loadSlice(sp + 16).set(str);
434
+ },
435
+
436
+ // func valueInstanceOf(v ref, t ref) bool
437
+ "syscall/js.valueInstanceOf": (sp) => {
438
+ sp >>>= 0;
439
+ this.mem.setUint8(sp + 24, (loadValue(sp + 8) instanceof loadValue(sp + 16)) ? 1 : 0);
440
+ },
441
+
442
+ // func copyBytesToGo(dst []byte, src ref) (int, bool)
443
+ "syscall/js.copyBytesToGo": (sp) => {
444
+ sp >>>= 0;
445
+ const dst = loadSlice(sp + 8);
446
+ const src = loadValue(sp + 32);
447
+ if (!(src instanceof Uint8Array || src instanceof Uint8ClampedArray)) {
448
+ this.mem.setUint8(sp + 48, 0);
449
+ return;
450
+ }
451
+ const toCopy = src.subarray(0, dst.length);
452
+ dst.set(toCopy);
453
+ setInt64(sp + 40, toCopy.length);
454
+ this.mem.setUint8(sp + 48, 1);
455
+ },
456
+
457
+ // func copyBytesToJS(dst ref, src []byte) (int, bool)
458
+ "syscall/js.copyBytesToJS": (sp) => {
459
+ sp >>>= 0;
460
+ const dst = loadValue(sp + 8);
461
+ const src = loadSlice(sp + 16);
462
+ if (!(dst instanceof Uint8Array || dst instanceof Uint8ClampedArray)) {
463
+ this.mem.setUint8(sp + 48, 0);
464
+ return;
465
+ }
466
+ const toCopy = src.subarray(0, dst.length);
467
+ dst.set(toCopy);
468
+ setInt64(sp + 40, toCopy.length);
469
+ this.mem.setUint8(sp + 48, 1);
470
+ },
471
+
472
+ "debug": (value) => {
473
+ console.log(value);
474
+ },
475
+ }
476
+ };
477
+ }
478
+
479
+ async run(instance) {
480
+ if (!(instance instanceof WebAssembly.Instance)) {
481
+ throw new Error("Go.run: WebAssembly.Instance expected");
482
+ }
483
+ this._inst = instance;
484
+ this.mem = new DataView(this._inst.exports.mem.buffer);
485
+ this._values = [ // JS values that Go currently has references to, indexed by reference id
486
+ NaN,
487
+ 0,
488
+ null,
489
+ true,
490
+ false,
491
+ globalThis,
492
+ this,
493
+ ];
494
+ this._goRefCounts = new Array(this._values.length).fill(Infinity); // number of references that Go has to a JS value, indexed by reference id
495
+ this._ids = new Map([ // mapping from JS values to reference ids
496
+ [0, 1],
497
+ [null, 2],
498
+ [true, 3],
499
+ [false, 4],
500
+ [globalThis, 5],
501
+ [this, 6],
502
+ ]);
503
+ this._idPool = []; // unused ids that have been garbage collected
504
+ this.exited = false; // whether the Go program has exited
505
+
506
+ // Pass command line arguments and environment variables to WebAssembly by writing them to the linear memory.
507
+ let offset = 4096;
508
+
509
+ const strPtr = (str) => {
510
+ const ptr = offset;
511
+ const bytes = encoder.encode(str + "\0");
512
+ new Uint8Array(this.mem.buffer, offset, bytes.length).set(bytes);
513
+ offset += bytes.length;
514
+ if (offset % 8 !== 0) {
515
+ offset += 8 - (offset % 8);
516
+ }
517
+ return ptr;
518
+ };
519
+
520
+ const argc = this.argv.length;
521
+
522
+ const argvPtrs = [];
523
+ this.argv.forEach((arg) => {
524
+ argvPtrs.push(strPtr(arg));
525
+ });
526
+ argvPtrs.push(0);
527
+
528
+ const keys = Object.keys(this.env).sort();
529
+ keys.forEach((key) => {
530
+ argvPtrs.push(strPtr(`${key}=${this.env[key]}`));
531
+ });
532
+ argvPtrs.push(0);
533
+
534
+ const argv = offset;
535
+ argvPtrs.forEach((ptr) => {
536
+ this.mem.setUint32(offset, ptr, true);
537
+ this.mem.setUint32(offset + 4, 0, true);
538
+ offset += 8;
539
+ });
540
+
541
+ // The linker guarantees global data starts from at least wasmMinDataAddr.
542
+ // Keep in sync with cmd/link/internal/ld/data.go:wasmMinDataAddr.
543
+ const wasmMinDataAddr = 4096 + 8192;
544
+ if (offset >= wasmMinDataAddr) {
545
+ throw new Error("total length of command line and environment variables exceeds limit");
546
+ }
547
+
548
+ this._inst.exports.run(argc, argv);
549
+ if (this.exited) {
550
+ this._resolveExitPromise();
551
+ }
552
+ await this._exitPromise;
553
+ }
554
+
555
+ _resume() {
556
+ if (this.exited) {
557
+ throw new Error("Go program has already exited");
558
+ }
559
+ this._inst.exports.resume();
560
+ if (this.exited) {
561
+ this._resolveExitPromise();
562
+ }
563
+ }
564
+
565
+ _makeFuncWrapper(id) {
566
+ const go = this;
567
+ return function () {
568
+ const event = { id: id, this: this, args: arguments };
569
+ go._pendingEvent = event;
570
+ go._resume();
571
+ return event.result;
572
+ };
573
+ }
574
+ }
575
+ })();
package/index.cjs ADDED
@@ -0,0 +1,31 @@
1
+ // @budhash/gomanize — CommonJS loader (Node). Same engine as index.mjs.
2
+ "use strict";
3
+
4
+ let _instance;
5
+
6
+ async function load(opts = {}) {
7
+ if (_instance) return _instance;
8
+ const fs = require("node:fs");
9
+ const path = require("node:path");
10
+ const execPath = opts.execURL || path.join(__dirname, "dist", "wasm_exec.js");
11
+ const wasmPath = opts.wasmURL || path.join(__dirname, "dist", "gomanize.wasm");
12
+ if (typeof globalThis.Go === "undefined") {
13
+ const vm = require("node:vm");
14
+ vm.runInThisContext(fs.readFileSync(execPath, "utf8"));
15
+ }
16
+ const bytes = fs.readFileSync(wasmPath);
17
+ const go = new globalThis.Go();
18
+ const { instance } = await WebAssembly.instantiate(bytes, go.importObject);
19
+ go.run(instance);
20
+ const fn = globalThis.gomanizeTranslit;
21
+ if (typeof fn !== "function") throw new Error("gomanize: wasm did not initialize");
22
+ _instance = {
23
+ translit(text, options) {
24
+ return fn(text == null ? "" : String(text), options || {});
25
+ },
26
+ };
27
+ return _instance;
28
+ }
29
+
30
+ module.exports = { load };
31
+ module.exports.default = module.exports;
package/index.d.ts ADDED
@@ -0,0 +1,54 @@
1
+ /**
2
+ * @budhash/gomanize — the gomanize engine (Go, compiled to WebAssembly) with a
3
+ * thin JS wrapper. Romanizes Devanagari (Hindi) into readable, diacritic-free
4
+ * Latin script. This is the same engine as the CLI, so output is identical.
5
+ */
6
+
7
+ /** Transliteration options. All default to `false` (the rule-engine baseline). */
8
+ export interface Options {
9
+ /** Rule tweak: write "aa" for every ā (गाना → gaanaa). */
10
+ longVowels?: boolean;
11
+ /** Rule tweak: simplified nasal endings (करें → karen, not karein). */
12
+ simpleNasals?: boolean;
13
+ /** Rule tweak: disable medial (CCV) schwa deletion (जनता → janata). */
14
+ keepMedialSchwa?: boolean;
15
+ /** Learned model: decision-tree schwa classifier for inherent-schwa decisions. */
16
+ schwaModel?: boolean;
17
+ /** Learned model: prefer attested spellings from the embedded lexicon. */
18
+ lexicon?: boolean;
19
+ /** Learned model: re-rank candidates with a character n-gram model. */
20
+ rerank?: boolean;
21
+ }
22
+
23
+ export interface Gomanize {
24
+ /**
25
+ * Romanize Devanagari text into Latin script. Whitespace and punctuation are
26
+ * preserved; words are romanized in place.
27
+ */
28
+ translit(text: string, options?: Options): string;
29
+ }
30
+
31
+ /** Where to load the engine assets from. Defaults resolve next to the module. */
32
+ export interface LoadOptions {
33
+ /** URL/path to gomanize.wasm. */
34
+ wasmURL?: string | URL;
35
+ /** URL/path to wasm_exec.js (ignored if a host page already defines `Go`). */
36
+ execURL?: string | URL;
37
+ }
38
+
39
+ /**
40
+ * Load and instantiate the WebAssembly engine. The result is cached, so calling
41
+ * `load()` again returns the same instance. `translit` is synchronous once
42
+ * loaded.
43
+ *
44
+ * ```js
45
+ * import { load } from "@budhash/gomanize";
46
+ * const g = await load();
47
+ * g.translit("नमस्ते दुनिया"); // "namaste duniya"
48
+ * g.translit("गाना", { longVowels: true }); // "gaanaa"
49
+ * ```
50
+ */
51
+ export function load(opts?: LoadOptions): Promise<Gomanize>;
52
+
53
+ declare const _default: { load: typeof load };
54
+ export default _default;
package/index.mjs ADDED
@@ -0,0 +1,59 @@
1
+ // @budhash/gomanize — ESM loader for the gomanize WebAssembly engine.
2
+ // Works in Node (reads the .wasm from disk) and in browsers/bundlers (fetches
3
+ // it, resolved relative to this module via import.meta.url). No top-level Node
4
+ // imports, so bundling for the browser stays clean.
5
+
6
+ let _instance;
7
+
8
+ const isNode = () =>
9
+ typeof process !== "undefined" && !!(process.versions && process.versions.node);
10
+
11
+ // wasm_exec.js is a plain script (not a module) shipped by the Go toolchain; it
12
+ // defines globalThis.Go. Evaluate it once in the global scope. If a host page
13
+ // already loaded it (e.g. via a <script> tag), this is a no-op.
14
+ async function ensureGo(url) {
15
+ if (typeof globalThis.Go !== "undefined") return;
16
+ if (isNode()) {
17
+ const { readFile } = await import("node:fs/promises");
18
+ const vm = await import("node:vm");
19
+ vm.runInThisContext(await readFile(url, "utf8"));
20
+ } else {
21
+ // eslint-disable-next-line no-eval
22
+ (0, eval)(await (await fetch(url)).text());
23
+ }
24
+ }
25
+
26
+ async function wasmBytes(url) {
27
+ if (isNode()) {
28
+ const { readFile } = await import("node:fs/promises");
29
+ return readFile(url);
30
+ }
31
+ return (await fetch(url)).arrayBuffer();
32
+ }
33
+
34
+ /**
35
+ * @param {{ wasmURL?: string|URL, execURL?: string|URL }} [opts]
36
+ * Override where the .wasm and wasm_exec.js are loaded from. Defaults resolve
37
+ * to ./dist/* next to this module (npm layout).
38
+ */
39
+ export async function load(opts = {}) {
40
+ if (_instance) return _instance;
41
+ const execURL = opts.execURL ?? new URL("./dist/wasm_exec.js", import.meta.url);
42
+ const wasmURL = opts.wasmURL ?? new URL("./dist/gomanize.wasm", import.meta.url);
43
+ await ensureGo(execURL);
44
+ const go = new globalThis.Go();
45
+ const { instance } = await WebAssembly.instantiate(await wasmBytes(wasmURL), go.importObject);
46
+ // main() sets globalThis.gomanizeTranslit and then blocks, keeping the
47
+ // exported function callable for the lifetime of the instance.
48
+ go.run(instance);
49
+ const fn = globalThis.gomanizeTranslit;
50
+ if (typeof fn !== "function") throw new Error("gomanize: wasm did not initialize");
51
+ _instance = {
52
+ translit(text, options) {
53
+ return fn(text == null ? "" : String(text), options || {});
54
+ },
55
+ };
56
+ return _instance;
57
+ }
58
+
59
+ export default { load };
package/package.json ADDED
@@ -0,0 +1,41 @@
1
+ {
2
+ "name": "@budhash/gomanize",
3
+ "version": "1.0.0",
4
+ "description": "Romanize Hindi (Devanagari) into readable, diacritic-free Latin script — the gomanize engine compiled to WebAssembly.",
5
+ "type": "module",
6
+ "exports": {
7
+ ".": {
8
+ "types": "./index.d.ts",
9
+ "import": "./index.mjs",
10
+ "require": "./index.cjs"
11
+ }
12
+ },
13
+ "types": "./index.d.ts",
14
+ "files": [
15
+ "index.mjs",
16
+ "index.cjs",
17
+ "index.d.ts",
18
+ "dist/gomanize.wasm",
19
+ "dist/wasm_exec.js",
20
+ "README.md"
21
+ ],
22
+ "engines": { "node": ">=18" },
23
+ "sideEffects": false,
24
+ "keywords": [
25
+ "hindi",
26
+ "devanagari",
27
+ "romanization",
28
+ "transliteration",
29
+ "romanize",
30
+ "transliterate",
31
+ "wasm",
32
+ "webassembly",
33
+ "nlp",
34
+ "indic"
35
+ ],
36
+ "license": "MIT",
37
+ "author": "budhash",
38
+ "homepage": "https://budhash.com/gomanize",
39
+ "repository": { "type": "git", "url": "git+https://github.com/budhash/gomanize.git" },
40
+ "bugs": { "url": "https://github.com/budhash/gomanize/issues" }
41
+ }