@storyteller-platform/align 0.1.56 → 0.1.58

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 (118) hide show
  1. package/dist/align/align.cjs +38 -19
  2. package/dist/align/align.d.cts +12 -1
  3. package/dist/align/align.d.ts +12 -1
  4. package/dist/align/align.js +38 -19
  5. package/dist/align/ctc/Aligner.cjs +800 -0
  6. package/dist/align/ctc/Aligner.d.cts +56 -0
  7. package/dist/align/ctc/Aligner.d.ts +56 -0
  8. package/dist/align/ctc/Aligner.js +789 -0
  9. package/dist/align/ctc/align.cjs +189 -0
  10. package/dist/align/ctc/align.d.cts +14 -0
  11. package/dist/align/ctc/align.d.ts +14 -0
  12. package/dist/align/ctc/align.js +132 -0
  13. package/dist/{process/mime.cjs → align/ctc/emissions.cjs} +7 -20
  14. package/dist/align/ctc/emissions.d.cts +6 -0
  15. package/dist/align/ctc/emissions.d.ts +6 -0
  16. package/dist/align/ctc/emissions.js +7 -0
  17. package/dist/align/ctc/forcedAlign.cjs +273 -0
  18. package/dist/align/ctc/forcedAlign.d.cts +21 -0
  19. package/dist/align/ctc/forcedAlign.d.ts +21 -0
  20. package/dist/align/ctc/forcedAlign.js +248 -0
  21. package/dist/align/ctc/greedyDecode.cjs +61 -0
  22. package/dist/align/ctc/greedyDecode.d.cts +15 -0
  23. package/dist/align/ctc/greedyDecode.d.ts +15 -0
  24. package/dist/align/ctc/greedyDecode.js +37 -0
  25. package/dist/align/ctc/mediaOverlay.cjs +248 -0
  26. package/dist/align/ctc/mediaOverlay.d.cts +16 -0
  27. package/dist/align/ctc/mediaOverlay.d.ts +16 -0
  28. package/dist/align/ctc/mediaOverlay.js +224 -0
  29. package/dist/align/ctc/report.cjs +66 -0
  30. package/dist/align/ctc/report.d.cts +79 -0
  31. package/dist/align/ctc/report.d.ts +79 -0
  32. package/dist/align/ctc/report.js +42 -0
  33. package/dist/align/ctc/search.cjs +437 -0
  34. package/dist/align/ctc/search.d.cts +46 -0
  35. package/dist/align/ctc/search.d.ts +46 -0
  36. package/dist/align/ctc/search.js +412 -0
  37. package/dist/align/getSentenceRanges.cjs +9 -3
  38. package/dist/align/getSentenceRanges.d.cts +1 -0
  39. package/dist/align/getSentenceRanges.d.ts +1 -0
  40. package/dist/align/getSentenceRanges.js +9 -3
  41. package/dist/align/parse.cjs +17 -4
  42. package/dist/align/parse.d.cts +8 -4
  43. package/dist/align/parse.d.ts +8 -4
  44. package/dist/align/parse.js +19 -4
  45. package/dist/align/search.cjs +76 -23
  46. package/dist/align/search.d.cts +8 -4
  47. package/dist/align/search.d.ts +8 -4
  48. package/dist/align/search.js +76 -24
  49. package/dist/align/slugify.cjs +3 -2
  50. package/dist/align/slugify.d.cts +1 -1
  51. package/dist/align/slugify.d.ts +1 -1
  52. package/dist/align/slugify.js +3 -2
  53. package/dist/cli/bin.cjs +204 -114
  54. package/dist/cli/bin.js +194 -103
  55. package/dist/common/ffmpeg.cjs +10 -5
  56. package/dist/common/ffmpeg.js +10 -5
  57. package/dist/common/logging.cjs +33 -1
  58. package/dist/common/logging.js +33 -1
  59. package/dist/emit/emit.cjs +105 -0
  60. package/dist/emit/emit.d.cts +15 -0
  61. package/dist/emit/emit.d.ts +15 -0
  62. package/dist/emit/emit.js +86 -0
  63. package/dist/emit/fs.cjs +227 -0
  64. package/dist/emit/fs.d.cts +22 -0
  65. package/dist/emit/fs.d.ts +22 -0
  66. package/dist/emit/fs.js +162 -0
  67. package/dist/emit/parse.cjs +68 -0
  68. package/dist/emit/parse.d.cts +26 -0
  69. package/dist/emit/parse.d.ts +26 -0
  70. package/dist/emit/parse.js +58 -0
  71. package/dist/index.d.cts +1 -1
  72. package/dist/index.d.ts +1 -1
  73. package/dist/markup/model.cjs +0 -16
  74. package/dist/markup/model.d.cts +1 -1
  75. package/dist/markup/model.d.ts +1 -1
  76. package/dist/markup/model.js +0 -12
  77. package/dist/markup/parseDom.cjs +13 -14
  78. package/dist/markup/parseDom.d.cts +2 -2
  79. package/dist/markup/parseDom.d.ts +2 -2
  80. package/dist/markup/parseDom.js +13 -18
  81. package/dist/markup/resolvedPos.d.cts +1 -1
  82. package/dist/markup/resolvedPos.d.ts +1 -1
  83. package/dist/markup/segmentation.cjs +3 -1
  84. package/dist/markup/segmentation.js +3 -1
  85. package/dist/markup/serializeDom.cjs +1 -1
  86. package/dist/markup/serializeDom.d.cts +1 -1
  87. package/dist/markup/serializeDom.d.ts +1 -1
  88. package/dist/markup/serializeDom.js +1 -1
  89. package/dist/markup/transform.cjs +5 -2
  90. package/dist/markup/transform.d.cts +1 -1
  91. package/dist/markup/transform.d.ts +1 -1
  92. package/dist/markup/transform.js +5 -4
  93. package/dist/{model-CZ2mMHJO.d.cts → model-CeOMDDgt.d.cts} +3 -11
  94. package/dist/{model-CZ2mMHJO.d.ts → model-CeOMDDgt.d.ts} +3 -11
  95. package/dist/process/processAudiobook.cjs +12 -23
  96. package/dist/process/processAudiobook.js +14 -30
  97. package/dist/process/ranges.cjs +2 -2
  98. package/dist/process/ranges.js +2 -2
  99. package/dist/readium/guidedNavigation.cjs +8 -8
  100. package/dist/readium/guidedNavigation.js +8 -8
  101. package/dist/readium/manifest.cjs +5 -4
  102. package/dist/readium/manifest.js +5 -4
  103. package/dist/snapshot/ctc/snapshot.cjs +261 -0
  104. package/dist/snapshot/ctc/snapshot.d.cts +6 -0
  105. package/dist/snapshot/ctc/snapshot.d.ts +6 -0
  106. package/dist/snapshot/ctc/snapshot.js +200 -0
  107. package/dist/snapshot/parse.cjs +29 -16
  108. package/dist/snapshot/parse.d.cts +14 -10
  109. package/dist/snapshot/parse.d.ts +14 -10
  110. package/dist/snapshot/parse.js +32 -17
  111. package/dist/transcribe/parse.d.cts +6 -6
  112. package/dist/transcribe/parse.d.ts +6 -6
  113. package/dist/transcribe/transcribe.cjs +4 -2
  114. package/dist/transcribe/transcribe.js +4 -2
  115. package/package.json +6 -4
  116. package/dist/process/mime.d.cts +0 -3
  117. package/dist/process/mime.d.ts +0 -3
  118. package/dist/process/mime.js +0 -24
@@ -0,0 +1,105 @@
1
+ "use strict";
2
+ var __defProp = Object.defineProperty;
3
+ var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
4
+ var __getOwnPropNames = Object.getOwnPropertyNames;
5
+ var __hasOwnProp = Object.prototype.hasOwnProperty;
6
+ var __export = (target, all) => {
7
+ for (var name in all)
8
+ __defProp(target, name, { get: all[name], enumerable: true });
9
+ };
10
+ var __copyProps = (to, from, except, desc) => {
11
+ if (from && typeof from === "object" || typeof from === "function") {
12
+ for (let key of __getOwnPropNames(from))
13
+ if (!__hasOwnProp.call(to, key) && key !== except)
14
+ __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
15
+ }
16
+ return to;
17
+ };
18
+ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
19
+ var emit_exports = {};
20
+ __export(emit_exports, {
21
+ emit: () => emit
22
+ });
23
+ module.exports = __toCommonJS(emit_exports);
24
+ var import_promises = require("node:fs/promises");
25
+ var import_node_path = require("node:path");
26
+ var import_itertools = require("itertools");
27
+ var import_ghost_story = require("@storyteller-platform/ghost-story");
28
+ var import_media_types = require("@storyteller-platform/media-types");
29
+ var import_fs = require("./fs.cjs");
30
+ async function emit(input, output, options) {
31
+ const controller = new AbortController();
32
+ const signal = AbortSignal.any([
33
+ ...options.signal ? [options.signal] : [],
34
+ controller.signal
35
+ ]);
36
+ await (0, import_promises.mkdir)(output, { recursive: true });
37
+ const allFiles = await (0, import_promises.readdir)(input, { recursive: true });
38
+ const filenames = allFiles.toSorted().filter((f) => import_media_types.MediaType.fromPath(f)?.kind === "audio");
39
+ if (!filenames.length) {
40
+ throw new Error(
41
+ `Failed to transcribe audio: found no audio files in ${input}`
42
+ );
43
+ }
44
+ const engine = "wav2vec2";
45
+ const model = options.model ?? "onnx-community/mms-300m-1130-forced-aligner-ONNX";
46
+ const emissions = [];
47
+ function aborted() {
48
+ return signal.aborted;
49
+ }
50
+ const timing = (0, import_ghost_story.createAggregator)();
51
+ timing.setMetadata("engine", engine);
52
+ timing.setMetadata("model", model);
53
+ for (const [i, filename] of (0, import_itertools.enumerate)(filenames)) {
54
+ let onFileProgress2 = function(progress) {
55
+ const updatedProgress = (i + progress) / filenames.length;
56
+ options.logger?.info(`Progress: ${Math.floor(updatedProgress * 100)}%`);
57
+ options.onProgress?.(updatedProgress);
58
+ };
59
+ var onFileProgress = onFileProgress2;
60
+ if (aborted()) throw new Error("Aborted");
61
+ const filepath = (0, import_node_path.join)(input, filename);
62
+ const emissionFilepath = (0, import_node_path.join)(
63
+ output,
64
+ `${(0, import_node_path.basename)(filename, (0, import_node_path.extname)(filename))}.json`
65
+ );
66
+ try {
67
+ await (0, import_promises.readFile)(emissionFilepath, {
68
+ encoding: "utf-8",
69
+ signal
70
+ });
71
+ options.logger?.info(`Found existing transcription for ${filepath}`);
72
+ emissions.push(emissionFilepath);
73
+ options.onProgress?.((emissions.length + 1) / filenames.length);
74
+ continue;
75
+ } catch {
76
+ }
77
+ if (aborted()) throw new Error("Aborted");
78
+ const result = await (0, import_ghost_story.emit)(filepath, {
79
+ ...options,
80
+ signal,
81
+ engine,
82
+ options: {
83
+ model,
84
+ ...options.device && { device: options.device },
85
+ ...options.dtype && { dtype: options.dtype },
86
+ onProgress: onFileProgress2
87
+ }
88
+ });
89
+ options.logger?.info(
90
+ (0, import_ghost_story.formatSingleReport)(
91
+ result.timing,
92
+ `Transcription Timing Report for ${filepath}`
93
+ )
94
+ );
95
+ timing.add(result.timing);
96
+ await (0, import_fs.writeEmissionsFile)(emissionFilepath, result, { signal });
97
+ emissions.push(emissionFilepath);
98
+ options.onProgress?.((i + 1) / filenames.length);
99
+ }
100
+ return timing;
101
+ }
102
+ // Annotate the CommonJS export names for ESM import in node:
103
+ 0 && (module.exports = {
104
+ emit
105
+ });
@@ -0,0 +1,15 @@
1
+ import { Logger } from 'pino';
2
+ import { TimingAggregator } from '@storyteller-platform/ghost-story';
3
+ import { Wav2Vec2Model, Wav2Vec2Device, Wav2Vec2DataType } from '@storyteller-platform/ghost-story/constants';
4
+
5
+ interface EmitOptions {
6
+ onProgress?: ((progress: number) => void) | null | undefined;
7
+ signal?: AbortSignal | null | undefined;
8
+ model?: Wav2Vec2Model | null | undefined;
9
+ device?: Wav2Vec2Device | null | undefined;
10
+ dtype?: Wav2Vec2DataType | null | undefined;
11
+ logger?: Logger | null | undefined;
12
+ }
13
+ declare function emit(input: string, output: string, options: EmitOptions): Promise<TimingAggregator>;
14
+
15
+ export { type EmitOptions, emit };
@@ -0,0 +1,15 @@
1
+ import { Logger } from 'pino';
2
+ import { TimingAggregator } from '@storyteller-platform/ghost-story';
3
+ import { Wav2Vec2Model, Wav2Vec2Device, Wav2Vec2DataType } from '@storyteller-platform/ghost-story/constants';
4
+
5
+ interface EmitOptions {
6
+ onProgress?: ((progress: number) => void) | null | undefined;
7
+ signal?: AbortSignal | null | undefined;
8
+ model?: Wav2Vec2Model | null | undefined;
9
+ device?: Wav2Vec2Device | null | undefined;
10
+ dtype?: Wav2Vec2DataType | null | undefined;
11
+ logger?: Logger | null | undefined;
12
+ }
13
+ declare function emit(input: string, output: string, options: EmitOptions): Promise<TimingAggregator>;
14
+
15
+ export { type EmitOptions, emit };
@@ -0,0 +1,86 @@
1
+ import "../chunk-BIEQXUOY.js";
2
+ import { mkdir, readFile, readdir } from "node:fs/promises";
3
+ import { basename, extname, join } from "node:path";
4
+ import { enumerate } from "itertools";
5
+ import {
6
+ createAggregator,
7
+ emit as emitFile,
8
+ formatSingleReport
9
+ } from "@storyteller-platform/ghost-story";
10
+ import { MediaType } from "@storyteller-platform/media-types";
11
+ import { writeEmissionsFile } from "./fs.js";
12
+ async function emit(input, output, options) {
13
+ const controller = new AbortController();
14
+ const signal = AbortSignal.any([
15
+ ...options.signal ? [options.signal] : [],
16
+ controller.signal
17
+ ]);
18
+ await mkdir(output, { recursive: true });
19
+ const allFiles = await readdir(input, { recursive: true });
20
+ const filenames = allFiles.toSorted().filter((f) => MediaType.fromPath(f)?.kind === "audio");
21
+ if (!filenames.length) {
22
+ throw new Error(
23
+ `Failed to transcribe audio: found no audio files in ${input}`
24
+ );
25
+ }
26
+ const engine = "wav2vec2";
27
+ const model = options.model ?? "onnx-community/mms-300m-1130-forced-aligner-ONNX";
28
+ const emissions = [];
29
+ function aborted() {
30
+ return signal.aborted;
31
+ }
32
+ const timing = createAggregator();
33
+ timing.setMetadata("engine", engine);
34
+ timing.setMetadata("model", model);
35
+ for (const [i, filename] of enumerate(filenames)) {
36
+ let onFileProgress2 = function(progress) {
37
+ const updatedProgress = (i + progress) / filenames.length;
38
+ options.logger?.info(`Progress: ${Math.floor(updatedProgress * 100)}%`);
39
+ options.onProgress?.(updatedProgress);
40
+ };
41
+ var onFileProgress = onFileProgress2;
42
+ if (aborted()) throw new Error("Aborted");
43
+ const filepath = join(input, filename);
44
+ const emissionFilepath = join(
45
+ output,
46
+ `${basename(filename, extname(filename))}.json`
47
+ );
48
+ try {
49
+ await readFile(emissionFilepath, {
50
+ encoding: "utf-8",
51
+ signal
52
+ });
53
+ options.logger?.info(`Found existing transcription for ${filepath}`);
54
+ emissions.push(emissionFilepath);
55
+ options.onProgress?.((emissions.length + 1) / filenames.length);
56
+ continue;
57
+ } catch {
58
+ }
59
+ if (aborted()) throw new Error("Aborted");
60
+ const result = await emitFile(filepath, {
61
+ ...options,
62
+ signal,
63
+ engine,
64
+ options: {
65
+ model,
66
+ ...options.device && { device: options.device },
67
+ ...options.dtype && { dtype: options.dtype },
68
+ onProgress: onFileProgress2
69
+ }
70
+ });
71
+ options.logger?.info(
72
+ formatSingleReport(
73
+ result.timing,
74
+ `Transcription Timing Report for ${filepath}`
75
+ )
76
+ );
77
+ timing.add(result.timing);
78
+ await writeEmissionsFile(emissionFilepath, result, { signal });
79
+ emissions.push(emissionFilepath);
80
+ options.onProgress?.((i + 1) / filenames.length);
81
+ }
82
+ return timing;
83
+ }
84
+ export {
85
+ emit
86
+ };
@@ -0,0 +1,227 @@
1
+ "use strict";
2
+ var __defProp = Object.defineProperty;
3
+ var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
4
+ var __getOwnPropNames = Object.getOwnPropertyNames;
5
+ var __hasOwnProp = Object.prototype.hasOwnProperty;
6
+ var __knownSymbol = (name, symbol) => (symbol = Symbol[name]) ? symbol : Symbol.for("Symbol." + name);
7
+ var __typeError = (msg) => {
8
+ throw TypeError(msg);
9
+ };
10
+ var __export = (target, all) => {
11
+ for (var name in all)
12
+ __defProp(target, name, { get: all[name], enumerable: true });
13
+ };
14
+ var __copyProps = (to, from, except, desc) => {
15
+ if (from && typeof from === "object" || typeof from === "function") {
16
+ for (let key of __getOwnPropNames(from))
17
+ if (!__hasOwnProp.call(to, key) && key !== except)
18
+ __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
19
+ }
20
+ return to;
21
+ };
22
+ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
23
+ var __using = (stack, value, async) => {
24
+ if (value != null) {
25
+ if (typeof value !== "object" && typeof value !== "function") __typeError("Object expected");
26
+ var dispose, inner;
27
+ if (async) dispose = value[__knownSymbol("asyncDispose")];
28
+ if (dispose === void 0) {
29
+ dispose = value[__knownSymbol("dispose")];
30
+ if (async) inner = dispose;
31
+ }
32
+ if (typeof dispose !== "function") __typeError("Object not disposable");
33
+ if (inner) dispose = function() {
34
+ try {
35
+ inner.call(this);
36
+ } catch (e) {
37
+ return Promise.reject(e);
38
+ }
39
+ };
40
+ stack.push([async, dispose, value]);
41
+ } else if (async) {
42
+ stack.push([async]);
43
+ }
44
+ return value;
45
+ };
46
+ var __callDispose = (stack, error, hasError) => {
47
+ var E = typeof SuppressedError === "function" ? SuppressedError : function(e, s, m, _) {
48
+ return _ = Error(m), _.name = "SuppressedError", _.error = e, _.suppressed = s, _;
49
+ };
50
+ var fail = (e) => error = hasError ? new E(e, error, "An error was suppressed during disposal") : (hasError = true, e);
51
+ var next = (it) => {
52
+ while (it = stack.pop()) {
53
+ try {
54
+ var result = it[1] && it[1].call(it[2]);
55
+ if (it[0]) return Promise.resolve(result).then(next, (e) => (fail(e), next()));
56
+ } catch (e) {
57
+ fail(e);
58
+ }
59
+ }
60
+ if (hasError) throw error;
61
+ };
62
+ return next();
63
+ };
64
+ var fs_exports = {};
65
+ __export(fs_exports, {
66
+ EmissionsReader: () => EmissionsReader,
67
+ readEmissionsFile: () => readEmissionsFile,
68
+ readEmissionsFileHead: () => readEmissionsFileHead,
69
+ writeEmissionsFile: () => writeEmissionsFile
70
+ });
71
+ module.exports = __toCommonJS(fs_exports);
72
+ var import_promises = require("node:fs/promises");
73
+ var import_node_path = require("node:path");
74
+ var import_ghost_story = require("@storyteller-platform/ghost-story");
75
+ async function writeEmissionsFile(path, data, options = {}) {
76
+ const encoded = (0, import_ghost_story.encodeEmissions)(data);
77
+ await (0, import_promises.writeFile)(path, encoded, options);
78
+ }
79
+ async function readEmissionsFile(path, framesFrom, framesTo) {
80
+ var _stack = [];
81
+ try {
82
+ const file = __using(_stack, await (0, import_promises.open)(path, "r"), true);
83
+ async function getBytes(bytesFrom, bytesTo) {
84
+ const length = bytesTo - bytesFrom;
85
+ const buffer = new Uint8Array(length);
86
+ let read = 0;
87
+ while (read < length) {
88
+ const { bytesRead } = await file.read(
89
+ buffer,
90
+ read,
91
+ length - read,
92
+ bytesFrom + read
93
+ );
94
+ if (bytesRead === 0) break;
95
+ read += bytesRead;
96
+ }
97
+ return buffer;
98
+ }
99
+ return await (0, import_ghost_story.decodeEmissions)(getBytes, framesFrom, framesTo);
100
+ } catch (_) {
101
+ var _error = _, _hasError = true;
102
+ } finally {
103
+ var _promise = __callDispose(_stack, _error, _hasError);
104
+ _promise && await _promise;
105
+ }
106
+ }
107
+ async function readEmissionsFileHead(path) {
108
+ return await readEmissionsFile(path, 0, 0);
109
+ }
110
+ async function packJsonEmissionsFile(path) {
111
+ const directory = (0, import_node_path.dirname)(path);
112
+ const filebase = (0, import_node_path.basename)(path, (0, import_node_path.extname)(path));
113
+ const emissionsName = (0, import_node_path.join)(directory, `${filebase}.emissions`);
114
+ const jsonData = await (0, import_promises.readFile)(path, { encoding: "utf-8" });
115
+ const parsed = JSON.parse(jsonData);
116
+ await writeEmissionsFile(emissionsName, {
117
+ ...parsed,
118
+ vocab: new Map(Object.entries(parsed.vocab)),
119
+ numFrames: parsed.emissions.length
120
+ });
121
+ await (0, import_promises.rm)(path);
122
+ return emissionsName;
123
+ }
124
+ class EmissionsReader {
125
+ constructor(numFrames = 0, frameMap, vocab, blankId, timings) {
126
+ this.numFrames = numFrames;
127
+ this.frameMap = frameMap;
128
+ this.vocab = vocab;
129
+ this.blankId = blankId;
130
+ this.timings = timings;
131
+ }
132
+ static async getFilepaths(fileOrDirectory) {
133
+ if ((0, import_node_path.extname)(fileOrDirectory) === ".emissions" || (0, import_node_path.extname)(fileOrDirectory) === ".json") {
134
+ return [fileOrDirectory];
135
+ }
136
+ const allFiles = await (0, import_promises.readdir)(fileOrDirectory, { recursive: true });
137
+ return allFiles.filter((f) => (0, import_node_path.extname)(f) === ".emissions" || (0, import_node_path.extname)(f) === ".json").toSorted().map((f) => (0, import_node_path.join)(fileOrDirectory, f));
138
+ }
139
+ static async from(fileOrDirectory) {
140
+ let vocab = null;
141
+ let blankId = null;
142
+ const timings = [];
143
+ const frameMap = [];
144
+ let numFrames = 0;
145
+ const emissionsFiles = await EmissionsReader.getFilepaths(fileOrDirectory);
146
+ let f = 0;
147
+ for (let path of emissionsFiles) {
148
+ if ((0, import_node_path.extname)(path) === ".json") {
149
+ path = await packJsonEmissionsFile(path);
150
+ }
151
+ const {
152
+ numFrames: numFrames2,
153
+ vocab: v,
154
+ blankIndex: b,
155
+ emissions: _,
156
+ ...emissionTimings
157
+ } = await readEmissionsFileHead(path);
158
+ vocab ??= v;
159
+ blankId ??= b;
160
+ timings.push(emissionTimings);
161
+ frameMap.push([path, f, f + numFrames2]);
162
+ f += numFrames2;
163
+ }
164
+ numFrames = f;
165
+ if (vocab === null || blankId === null) {
166
+ throw new Error("Failed to initialize EmissionsReader");
167
+ }
168
+ const reader = new EmissionsReader(
169
+ numFrames,
170
+ frameMap,
171
+ vocab,
172
+ blankId,
173
+ timings
174
+ );
175
+ return reader;
176
+ }
177
+ async *frames() {
178
+ let fileIndex = 0;
179
+ let chunkFrom = 0;
180
+ let frameIndex = 0;
181
+ while (frameIndex < this.numFrames) {
182
+ const [path, framesStart, framesEnd] = this.frameMap[fileIndex];
183
+ const chunkTo = Math.min(chunkFrom + 1e3, framesEnd - framesStart);
184
+ const { emissions: chunk } = await readEmissionsFile(
185
+ path,
186
+ chunkFrom,
187
+ chunkTo
188
+ );
189
+ for (const e of chunk) {
190
+ yield e;
191
+ }
192
+ frameIndex += chunkTo - chunkFrom;
193
+ if (chunkTo < chunkFrom + 1e3) {
194
+ fileIndex++;
195
+ chunkFrom = 0;
196
+ } else {
197
+ chunkFrom += 1e3;
198
+ }
199
+ }
200
+ }
201
+ async readFrames(framesFrom, framesTo) {
202
+ const startFile = this.frameMap.findIndex(([, , end]) => end > framesFrom);
203
+ if (startFile === -1) return [];
204
+ let endFile = this.frameMap.findLastIndex(([, start]) => start < framesTo);
205
+ if (endFile === -1) endFile = this.frameMap.length - 1;
206
+ const frames = [];
207
+ for (let i = startFile; i <= endFile; i++) {
208
+ const [path, from, to] = this.frameMap[i];
209
+ const { emissions } = await readEmissionsFile(
210
+ path,
211
+ Math.max(framesFrom, from) - from,
212
+ Math.min(to, framesTo) - from
213
+ );
214
+ for (const e of emissions) {
215
+ frames.push(e);
216
+ }
217
+ }
218
+ return frames;
219
+ }
220
+ }
221
+ // Annotate the CommonJS export names for ESM import in node:
222
+ 0 && (module.exports = {
223
+ EmissionsReader,
224
+ readEmissionsFile,
225
+ readEmissionsFileHead,
226
+ writeEmissionsFile
227
+ });
@@ -0,0 +1,22 @@
1
+ import { EmissionData } from '@storyteller-platform/ghost-story';
2
+ import { EmissionTimings } from '../align/ctc/emissions.cjs';
3
+
4
+ declare function writeEmissionsFile(path: string, data: EmissionData, options?: {
5
+ signal?: AbortSignal;
6
+ }): Promise<void>;
7
+ declare function readEmissionsFile(path: string, framesFrom?: number, framesTo?: number): Promise<EmissionData>;
8
+ declare function readEmissionsFileHead(path: string): Promise<EmissionData>;
9
+ declare class EmissionsReader {
10
+ numFrames: number;
11
+ frameMap: [string, number, number][];
12
+ vocab: Map<string, number>;
13
+ blankId: number;
14
+ timings: EmissionTimings[];
15
+ private constructor();
16
+ private static getFilepaths;
17
+ static from(fileOrDirectory: string): Promise<EmissionsReader>;
18
+ frames(): AsyncGenerator<number[], void, unknown>;
19
+ readFrames(framesFrom: number, framesTo: number): Promise<number[][]>;
20
+ }
21
+
22
+ export { EmissionsReader, readEmissionsFile, readEmissionsFileHead, writeEmissionsFile };
@@ -0,0 +1,22 @@
1
+ import { EmissionData } from '@storyteller-platform/ghost-story';
2
+ import { EmissionTimings } from '../align/ctc/emissions.js';
3
+
4
+ declare function writeEmissionsFile(path: string, data: EmissionData, options?: {
5
+ signal?: AbortSignal;
6
+ }): Promise<void>;
7
+ declare function readEmissionsFile(path: string, framesFrom?: number, framesTo?: number): Promise<EmissionData>;
8
+ declare function readEmissionsFileHead(path: string): Promise<EmissionData>;
9
+ declare class EmissionsReader {
10
+ numFrames: number;
11
+ frameMap: [string, number, number][];
12
+ vocab: Map<string, number>;
13
+ blankId: number;
14
+ timings: EmissionTimings[];
15
+ private constructor();
16
+ private static getFilepaths;
17
+ static from(fileOrDirectory: string): Promise<EmissionsReader>;
18
+ frames(): AsyncGenerator<number[], void, unknown>;
19
+ readFrames(framesFrom: number, framesTo: number): Promise<number[][]>;
20
+ }
21
+
22
+ export { EmissionsReader, readEmissionsFile, readEmissionsFileHead, writeEmissionsFile };
@@ -0,0 +1,162 @@
1
+ import {
2
+ __callDispose,
3
+ __using
4
+ } from "../chunk-BIEQXUOY.js";
5
+ import { open, readFile, readdir, rm, writeFile } from "node:fs/promises";
6
+ import { basename, dirname, extname, join } from "node:path";
7
+ import {
8
+ decodeEmissions,
9
+ encodeEmissions
10
+ } from "@storyteller-platform/ghost-story";
11
+ async function writeEmissionsFile(path, data, options = {}) {
12
+ const encoded = encodeEmissions(data);
13
+ await writeFile(path, encoded, options);
14
+ }
15
+ async function readEmissionsFile(path, framesFrom, framesTo) {
16
+ var _stack = [];
17
+ try {
18
+ const file = __using(_stack, await open(path, "r"), true);
19
+ async function getBytes(bytesFrom, bytesTo) {
20
+ const length = bytesTo - bytesFrom;
21
+ const buffer = new Uint8Array(length);
22
+ let read = 0;
23
+ while (read < length) {
24
+ const { bytesRead } = await file.read(
25
+ buffer,
26
+ read,
27
+ length - read,
28
+ bytesFrom + read
29
+ );
30
+ if (bytesRead === 0) break;
31
+ read += bytesRead;
32
+ }
33
+ return buffer;
34
+ }
35
+ return await decodeEmissions(getBytes, framesFrom, framesTo);
36
+ } catch (_) {
37
+ var _error = _, _hasError = true;
38
+ } finally {
39
+ var _promise = __callDispose(_stack, _error, _hasError);
40
+ _promise && await _promise;
41
+ }
42
+ }
43
+ async function readEmissionsFileHead(path) {
44
+ return await readEmissionsFile(path, 0, 0);
45
+ }
46
+ async function packJsonEmissionsFile(path) {
47
+ const directory = dirname(path);
48
+ const filebase = basename(path, extname(path));
49
+ const emissionsName = join(directory, `${filebase}.emissions`);
50
+ const jsonData = await readFile(path, { encoding: "utf-8" });
51
+ const parsed = JSON.parse(jsonData);
52
+ await writeEmissionsFile(emissionsName, {
53
+ ...parsed,
54
+ vocab: new Map(Object.entries(parsed.vocab)),
55
+ numFrames: parsed.emissions.length
56
+ });
57
+ await rm(path);
58
+ return emissionsName;
59
+ }
60
+ class EmissionsReader {
61
+ constructor(numFrames = 0, frameMap, vocab, blankId, timings) {
62
+ this.numFrames = numFrames;
63
+ this.frameMap = frameMap;
64
+ this.vocab = vocab;
65
+ this.blankId = blankId;
66
+ this.timings = timings;
67
+ }
68
+ static async getFilepaths(fileOrDirectory) {
69
+ if (extname(fileOrDirectory) === ".emissions" || extname(fileOrDirectory) === ".json") {
70
+ return [fileOrDirectory];
71
+ }
72
+ const allFiles = await readdir(fileOrDirectory, { recursive: true });
73
+ return allFiles.filter((f) => extname(f) === ".emissions" || extname(f) === ".json").toSorted().map((f) => join(fileOrDirectory, f));
74
+ }
75
+ static async from(fileOrDirectory) {
76
+ let vocab = null;
77
+ let blankId = null;
78
+ const timings = [];
79
+ const frameMap = [];
80
+ let numFrames = 0;
81
+ const emissionsFiles = await EmissionsReader.getFilepaths(fileOrDirectory);
82
+ let f = 0;
83
+ for (let path of emissionsFiles) {
84
+ if (extname(path) === ".json") {
85
+ path = await packJsonEmissionsFile(path);
86
+ }
87
+ const {
88
+ numFrames: numFrames2,
89
+ vocab: v,
90
+ blankIndex: b,
91
+ emissions: _,
92
+ ...emissionTimings
93
+ } = await readEmissionsFileHead(path);
94
+ vocab ??= v;
95
+ blankId ??= b;
96
+ timings.push(emissionTimings);
97
+ frameMap.push([path, f, f + numFrames2]);
98
+ f += numFrames2;
99
+ }
100
+ numFrames = f;
101
+ if (vocab === null || blankId === null) {
102
+ throw new Error("Failed to initialize EmissionsReader");
103
+ }
104
+ const reader = new EmissionsReader(
105
+ numFrames,
106
+ frameMap,
107
+ vocab,
108
+ blankId,
109
+ timings
110
+ );
111
+ return reader;
112
+ }
113
+ async *frames() {
114
+ let fileIndex = 0;
115
+ let chunkFrom = 0;
116
+ let frameIndex = 0;
117
+ while (frameIndex < this.numFrames) {
118
+ const [path, framesStart, framesEnd] = this.frameMap[fileIndex];
119
+ const chunkTo = Math.min(chunkFrom + 1e3, framesEnd - framesStart);
120
+ const { emissions: chunk } = await readEmissionsFile(
121
+ path,
122
+ chunkFrom,
123
+ chunkTo
124
+ );
125
+ for (const e of chunk) {
126
+ yield e;
127
+ }
128
+ frameIndex += chunkTo - chunkFrom;
129
+ if (chunkTo < chunkFrom + 1e3) {
130
+ fileIndex++;
131
+ chunkFrom = 0;
132
+ } else {
133
+ chunkFrom += 1e3;
134
+ }
135
+ }
136
+ }
137
+ async readFrames(framesFrom, framesTo) {
138
+ const startFile = this.frameMap.findIndex(([, , end]) => end > framesFrom);
139
+ if (startFile === -1) return [];
140
+ let endFile = this.frameMap.findLastIndex(([, start]) => start < framesTo);
141
+ if (endFile === -1) endFile = this.frameMap.length - 1;
142
+ const frames = [];
143
+ for (let i = startFile; i <= endFile; i++) {
144
+ const [path, from, to] = this.frameMap[i];
145
+ const { emissions } = await readEmissionsFile(
146
+ path,
147
+ Math.max(framesFrom, from) - from,
148
+ Math.min(to, framesTo) - from
149
+ );
150
+ for (const e of emissions) {
151
+ frames.push(e);
152
+ }
153
+ }
154
+ return frames;
155
+ }
156
+ }
157
+ export {
158
+ EmissionsReader,
159
+ readEmissionsFile,
160
+ readEmissionsFileHead,
161
+ writeEmissionsFile
162
+ };