echogarden 3.0.5 → 3.2.0
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/data/lexicons/heteronyms.en.json +2 -2
- package/data/lexicons/words.en.json +120 -0
- package/dist/alignment/SemanticTextAlignment.js +2 -2
- package/dist/alignment/SemanticTextAlignment.js.map +1 -1
- package/dist/api/Alignment.js +2 -2
- package/dist/api/Alignment.js.map +1 -1
- package/dist/api/Synthesis.js +2 -2
- package/dist/api/Synthesis.js.map +1 -1
- package/dist/encodings/Ascii.d.ts +1 -3
- package/dist/encodings/Ascii.d.ts.map +1 -1
- package/dist/encodings/Ascii.js +13 -9
- package/dist/encodings/Ascii.js.map +1 -1
- package/dist/encodings/Base64.d.ts.map +1 -1
- package/dist/encodings/Base64.js +9 -0
- package/dist/encodings/Base64.js.map +1 -1
- package/dist/encodings/Hex.d.ts.map +1 -1
- package/dist/encodings/Hex.js +11 -2
- package/dist/encodings/Hex.js.map +1 -1
- package/dist/encodings/LEB128.d.ts +1 -15
- package/dist/encodings/LEB128.d.ts.map +1 -1
- package/dist/encodings/LEB128.js +1 -247
- package/dist/encodings/LEB128.js.map +1 -1
- package/dist/encodings/Utf16.d.ts +1 -3
- package/dist/encodings/Utf16.d.ts.map +1 -1
- package/dist/encodings/Utf16.js +13 -9
- package/dist/encodings/Utf16.js.map +1 -1
- package/dist/encodings/Utf32.d.ts.map +1 -1
- package/dist/encodings/Utf32.js +6 -0
- package/dist/encodings/Utf32.js.map +1 -1
- package/dist/encodings/Utf8.d.ts +9 -5
- package/dist/encodings/Utf8.d.ts.map +1 -1
- package/dist/encodings/Utf8.js +36 -78
- package/dist/encodings/Utf8.js.map +1 -1
- package/dist/nlp/Segmentation.d.ts +2 -8
- package/dist/nlp/Segmentation.d.ts.map +1 -1
- package/dist/nlp/Segmentation.js +42 -19
- package/dist/nlp/Segmentation.js.map +1 -1
- package/dist/nlp/TextNormalizer.d.ts.map +1 -1
- package/dist/nlp/TextNormalizer.js +29 -14
- package/dist/nlp/TextNormalizer.js.map +1 -1
- package/dist/recognition/AmazonTranscribeSTT.js +3 -3
- package/dist/recognition/AmazonTranscribeSTT.js.map +1 -1
- package/dist/synthesis/AzureCognitiveServicesTTS.js +1 -1
- package/dist/synthesis/AzureCognitiveServicesTTS.js.map +1 -1
- package/dist/synthesis/EspeakTTS.d.ts.map +1 -1
- package/dist/synthesis/EspeakTTS.js +15 -8
- package/dist/synthesis/EspeakTTS.js.map +1 -1
- package/dist/synthesis/KokoroTTS.d.ts.map +1 -1
- package/dist/synthesis/KokoroTTS.js +0 -5
- package/dist/synthesis/KokoroTTS.js.map +1 -1
- package/dist/synthesis/MicrosoftEdgeTTS.js +1 -1
- package/dist/synthesis/MicrosoftEdgeTTS.js.map +1 -1
- package/dist/synthesis/VitsTTS.d.ts.map +1 -1
- package/dist/synthesis/VitsTTS.js +0 -5
- package/dist/synthesis/VitsTTS.js.map +1 -1
- package/dist/utilities/FileSystem.d.ts +12 -7
- package/dist/utilities/FileSystem.d.ts.map +1 -1
- package/dist/utilities/FileSystem.js +196 -79
- package/dist/utilities/FileSystem.js.map +1 -1
- package/dist/utilities/FileWriter.d.ts +1 -0
- package/dist/utilities/FileWriter.d.ts.map +1 -1
- package/dist/utilities/FileWriter.js +8 -1
- package/dist/utilities/FileWriter.js.map +1 -1
- package/dist/utilities/ObjectUtilities.d.ts.map +1 -1
- package/dist/utilities/ObjectUtilities.js +15 -13
- package/dist/utilities/ObjectUtilities.js.map +1 -1
- package/dist/utilities/PackageManager.js +2 -2
- package/dist/utilities/PackageManager.js.map +1 -1
- package/dist/utilities/PathUtilities.d.ts +1 -0
- package/dist/utilities/PathUtilities.d.ts.map +1 -1
- package/dist/utilities/PathUtilities.js +18 -0
- package/dist/utilities/PathUtilities.js.map +1 -1
- package/dist/utilities/StringUtilities.d.ts.map +1 -1
- package/dist/utilities/StringUtilities.js.map +1 -1
- package/dist/utilities/Timeline.js +2 -2
- package/dist/utilities/Timeline.js.map +1 -1
- package/dist/utilities/WikipediaReader.js +5 -5
- package/dist/utilities/WikipediaReader.js.map +1 -1
- package/package.json +13 -12
- package/src/alignment/SemanticTextAlignment.ts +2 -2
- package/src/api/Alignment.ts +2 -2
- package/src/api/Synthesis.ts +2 -2
- package/src/encodings/Ascii.ts +14 -10
- package/src/encodings/Base64.ts +9 -0
- package/src/encodings/Hex.ts +11 -2
- package/src/encodings/Utf16.ts +14 -10
- package/src/encodings/Utf32.ts +6 -0
- package/src/encodings/Utf8.ts +45 -82
- package/src/nlp/Segmentation.ts +53 -19
- package/src/nlp/TextNormalizer.ts +40 -16
- package/src/recognition/AmazonTranscribeSTT.ts +3 -3
- package/src/synthesis/AzureCognitiveServicesTTS.ts +1 -1
- package/src/synthesis/EspeakTTS.ts +17 -9
- package/src/synthesis/KokoroTTS.ts +0 -6
- package/src/synthesis/MicrosoftEdgeTTS.ts +1 -1
- package/src/synthesis/VitsTTS.ts +0 -6
- package/src/utilities/FileSystem.ts +224 -85
- package/src/utilities/FileWriter.ts +11 -1
- package/src/utilities/ObjectUtilities.ts +19 -15
- package/src/utilities/PackageManager.ts +2 -2
- package/src/utilities/PathUtilities.ts +19 -0
- package/src/utilities/StringUtilities.ts +0 -2
- package/src/utilities/Timeline.ts +2 -2
- package/src/utilities/WikipediaReader.ts +5 -5
- package/src/encodings/HtmlEscape.ts +0 -39
- package/src/encodings/LEB128.ts +0 -362
- package/src/utilities/StringBuilder.ts +0 -45
package/dist/encodings/Utf8.js
CHANGED
|
@@ -1,3 +1,6 @@
|
|
|
1
|
+
//////////////////////////////////////////////////////////////////////////////
|
|
2
|
+
// UTF-8 Encoding
|
|
3
|
+
//////////////////////////////////////////////////////////////////////////////
|
|
1
4
|
export function encodeUtf8(text) {
|
|
2
5
|
const textEncoder = new TextEncoder();
|
|
3
6
|
return textEncoder.encode(text);
|
|
@@ -7,91 +10,46 @@ export function encodeUtf8Into(text, outputArray) {
|
|
|
7
10
|
const result = textEncoder.encodeInto(text, outputArray);
|
|
8
11
|
return result;
|
|
9
12
|
}
|
|
10
|
-
export
|
|
13
|
+
export class ChunkedUtf8Encoder {
|
|
14
|
+
textEncoder = new TextEncoder();
|
|
15
|
+
pendingHighSurrogate = '';
|
|
16
|
+
writeChunk(stringChunk) {
|
|
17
|
+
if (this.pendingHighSurrogate !== '') {
|
|
18
|
+
stringChunk = this.pendingHighSurrogate + stringChunk;
|
|
19
|
+
this.pendingHighSurrogate = '';
|
|
20
|
+
}
|
|
21
|
+
const lastCodeUnit = stringChunk.charCodeAt(stringChunk.length - 1);
|
|
22
|
+
if (lastCodeUnit >= 0xD800 && lastCodeUnit <= 0xDBFF) {
|
|
23
|
+
this.pendingHighSurrogate = stringChunk[stringChunk.length - 1];
|
|
24
|
+
stringChunk = stringChunk.substring(0, stringChunk.length - 1);
|
|
25
|
+
}
|
|
26
|
+
return this.textEncoder.encode(stringChunk);
|
|
27
|
+
}
|
|
28
|
+
finalize() {
|
|
29
|
+
const result = this.textEncoder.encode(this.pendingHighSurrogate);
|
|
30
|
+
this.pendingHighSurrogate = '';
|
|
31
|
+
return result;
|
|
32
|
+
}
|
|
33
|
+
}
|
|
34
|
+
//////////////////////////////////////////////////////////////////////////////
|
|
35
|
+
// UTF-8 Decoding
|
|
36
|
+
//////////////////////////////////////////////////////////////////////////////
|
|
37
|
+
export function decodeUtf8(utf8Bytes) {
|
|
11
38
|
const maxChunkLength = 2 ** 24;
|
|
12
|
-
const
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
|
|
39
|
+
const chunkedUtf8Decoder = new ChunkedUtf8Decoder();
|
|
40
|
+
let resultString = '';
|
|
41
|
+
for (let offset = 0; offset < utf8Bytes.length; offset += maxChunkLength) {
|
|
42
|
+
const utf8Chunk = utf8Bytes.subarray(offset, offset + maxChunkLength);
|
|
43
|
+
const stringChunk = chunkedUtf8Decoder.writeChunk(utf8Chunk);
|
|
44
|
+
resultString += stringChunk;
|
|
16
45
|
}
|
|
17
|
-
return
|
|
46
|
+
return resultString;
|
|
18
47
|
}
|
|
19
48
|
export class ChunkedUtf8Decoder {
|
|
20
|
-
str = '';
|
|
21
49
|
textDecoder = new TextDecoder('utf-8');
|
|
22
50
|
writeChunk(chunk) {
|
|
23
51
|
const decodedChunk = this.textDecoder.decode(chunk, { stream: true });
|
|
24
|
-
|
|
25
|
-
}
|
|
26
|
-
toString() {
|
|
27
|
-
return this.str;
|
|
28
|
-
}
|
|
29
|
-
}
|
|
30
|
-
//////////////////////////////////////////////////////////////////////////////////////////////
|
|
31
|
-
// Pure JavaScript implementations
|
|
32
|
-
//////////////////////////////////////////////////////////////////////////////////////////////
|
|
33
|
-
function encodeUtf8Into_JS(str, outputArray) {
|
|
34
|
-
let readOffset = 0;
|
|
35
|
-
let writeOffset = 0;
|
|
36
|
-
while (readOffset < str.length) {
|
|
37
|
-
const charCode = str.codePointAt(readOffset++);
|
|
38
|
-
if (charCode <= 0x7f) {
|
|
39
|
-
outputArray[writeOffset++] = charCode;
|
|
40
|
-
}
|
|
41
|
-
else if (charCode <= 0x7ff) {
|
|
42
|
-
outputArray[writeOffset++] = 0xc0 | (charCode >>> 6);
|
|
43
|
-
outputArray[writeOffset++] = 0x80 | (charCode & 63);
|
|
44
|
-
}
|
|
45
|
-
else if (charCode <= 0xffff) {
|
|
46
|
-
outputArray[writeOffset++] = 0xe0 | (charCode >>> 12);
|
|
47
|
-
outputArray[writeOffset++] = 0x80 | ((charCode >>> 6) & 63);
|
|
48
|
-
outputArray[writeOffset++] = 0x80 | (charCode & 63);
|
|
49
|
-
}
|
|
50
|
-
else if (charCode <= 0x10ffff) {
|
|
51
|
-
outputArray[writeOffset++] = 0xf0 | (charCode >>> 18);
|
|
52
|
-
outputArray[writeOffset++] = 0x80 | ((charCode >>> 12) & 63);
|
|
53
|
-
outputArray[writeOffset++] = 0x80 | ((charCode >>> 6) & 63);
|
|
54
|
-
outputArray[writeOffset++] = 0x80 | (charCode & 63);
|
|
55
|
-
readOffset++;
|
|
56
|
-
}
|
|
57
|
-
}
|
|
58
|
-
return { read: str.length, written: writeOffset };
|
|
59
|
-
}
|
|
60
|
-
function decodeUtf8_JS(utf8Bytes) {
|
|
61
|
-
let decodedString = '';
|
|
62
|
-
let readOffset = 0;
|
|
63
|
-
while (readOffset < utf8Bytes.length) {
|
|
64
|
-
const leadByte = utf8Bytes[readOffset++];
|
|
65
|
-
let outputCodePoint;
|
|
66
|
-
if (leadByte >>> 7 === 0) {
|
|
67
|
-
outputCodePoint = leadByte;
|
|
68
|
-
}
|
|
69
|
-
else if (leadByte >>> 5 === 6) {
|
|
70
|
-
outputCodePoint =
|
|
71
|
-
(leadByte & 31) << 6 |
|
|
72
|
-
(utf8Bytes[readOffset++] & 63);
|
|
73
|
-
}
|
|
74
|
-
else if (leadByte >>> 4 === 14) {
|
|
75
|
-
outputCodePoint =
|
|
76
|
-
(leadByte & 15) << 12 |
|
|
77
|
-
(utf8Bytes[readOffset++] & 63) << 6 |
|
|
78
|
-
(utf8Bytes[readOffset++] & 63);
|
|
79
|
-
}
|
|
80
|
-
else if (leadByte >>> 3 === 30) {
|
|
81
|
-
outputCodePoint =
|
|
82
|
-
(leadByte & 7) << 18 |
|
|
83
|
-
(utf8Bytes[readOffset++] & 63) << 12 |
|
|
84
|
-
(utf8Bytes[readOffset++] & 63) << 6 |
|
|
85
|
-
(utf8Bytes[readOffset++] & 63);
|
|
86
|
-
}
|
|
87
|
-
else {
|
|
88
|
-
throw new Error(`Invalid UTF-8 stream: An invalid lead byte value encountered at position ${readOffset}`);
|
|
89
|
-
}
|
|
90
|
-
decodedString += String.fromCodePoint(outputCodePoint);
|
|
91
|
-
}
|
|
92
|
-
if (readOffset > utf8Bytes.length) {
|
|
93
|
-
throw new Error(`UTF-8 decoding failed. Byte sequence is truncated.`);
|
|
52
|
+
return decodedChunk;
|
|
94
53
|
}
|
|
95
|
-
return decodedString;
|
|
96
54
|
}
|
|
97
55
|
//# sourceMappingURL=Utf8.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"Utf8.js","sourceRoot":"","sources":["../../src/encodings/Utf8.ts"],"names":[],"mappings":"AAEA,MAAM,UAAU,UAAU,CAAC,IAAY;IACtC,MAAM,WAAW,GAAG,IAAI,WAAW,EAAE,CAAA;IAErC,OAAO,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC,CAAA;AAChC,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,IAAY,EAAE,WAAuB;IACnE,MAAM,WAAW,GAAG,IAAI,WAAW,EAAE,CAAA;IAErC,MAAM,MAAM,GAAG,WAAW,CAAC,UAAU,CAAC,IAAI,EAAE,WAAW,CAAC,CAAA;IAExD,OAAO,MAAM,CAAA;AACd,CAAC;AAED,MAAM,
|
|
1
|
+
{"version":3,"file":"Utf8.js","sourceRoot":"","sources":["../../src/encodings/Utf8.ts"],"names":[],"mappings":"AAEA,8EAA8E;AAC9E,iBAAiB;AACjB,8EAA8E;AAC9E,MAAM,UAAU,UAAU,CAAC,IAAY;IACtC,MAAM,WAAW,GAAG,IAAI,WAAW,EAAE,CAAA;IAErC,OAAO,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC,CAAA;AAChC,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,IAAY,EAAE,WAAuB;IACnE,MAAM,WAAW,GAAG,IAAI,WAAW,EAAE,CAAA;IAErC,MAAM,MAAM,GAAG,WAAW,CAAC,UAAU,CAAC,IAAI,EAAE,WAAW,CAAC,CAAA;IAExD,OAAO,MAAM,CAAA;AACd,CAAC;AAED,MAAM,OAAO,kBAAkB;IACb,WAAW,GAAG,IAAI,WAAW,EAAE,CAAA;IAExC,oBAAoB,GAAG,EAAE,CAAA;IAEjC,UAAU,CAAC,WAAmB;QAC7B,IAAI,IAAI,CAAC,oBAAoB,KAAK,EAAE,EAAE,CAAC;YACtC,WAAW,GAAG,IAAI,CAAC,oBAAoB,GAAG,WAAW,CAAA;YAErD,IAAI,CAAC,oBAAoB,GAAG,EAAE,CAAA;QAC/B,CAAC;QAED,MAAM,YAAY,GAAG,WAAW,CAAC,UAAU,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;QAEnE,IAAI,YAAY,IAAI,MAAM,IAAI,YAAY,IAAI,MAAM,EAAE,CAAC;YACtD,IAAI,CAAC,oBAAoB,GAAG,WAAW,CAAC,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;YAE/D,WAAW,GAAG,WAAW,CAAC,SAAS,CAAC,CAAC,EAAE,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;QAC/D,CAAC;QAED,OAAO,IAAI,CAAC,WAAW,CAAC,MAAM,CAAC,WAAW,CAAC,CAAA;IAC5C,CAAC;IAED,QAAQ;QACP,MAAM,MAAM,GAAG,IAAI,CAAC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAA;QAEjE,IAAI,CAAC,oBAAoB,GAAG,EAAE,CAAA;QAE9B,OAAO,MAAM,CAAA;IACd,CAAC;CACD;AAED,8EAA8E;AAC9E,iBAAiB;AACjB,8EAA8E;AAC9E,MAAM,UAAU,UAAU,CAAC,SAAqB;IAC/C,MAAM,cAAc,GAAG,CAAC,IAAI,EAAE,CAAA;IAE9B,MAAM,kBAAkB,GAAG,IAAI,kBAAkB,EAAE,CAAA;IAEnD,IAAI,YAAY,GAAG,EAAE,CAAA;IAErB,KAAK,IAAI,MAAM,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC,MAAM,EAAE,MAAM,IAAI,cAAc,EAAE,CAAC;QAC1E,MAAM,SAAS,GAAG,SAAS,CAAC,QAAQ,CAAC,MAAM,EAAE,MAAM,GAAG,cAAc,CAAC,CAAA;QACrE,MAAM,WAAW,GAAG,kBAAkB,CAAC,UAAU,CAAC,SAAS,CAAC,CAAA;QAE5D,YAAY,IAAI,WAAW,CAAA;IAC5B,CAAC;IAED,OAAO,YAAY,CAAA;AACpB,CAAC;AAED,MAAM,OAAO,kBAAkB;IACb,WAAW,GAAG,IAAI,WAAW,CAAC,OAAO,CAAC,CAAA;IAEvD,UAAU,CAAC,KAAiB;QAC3B,MAAM,YAAY,GAAG,IAAI,CAAC,WAAW,CAAC,MAAM,CAAC,KAAK,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,CAAA;QAErE,OAAO,YAAY,CAAA;IACpB,CAAC;CACD"}
|
|
@@ -1,19 +1,13 @@
|
|
|
1
1
|
import { ParagraphBreakType, WhitespaceProcessing } from '../api/Common.js';
|
|
2
2
|
import * as TextSegmentation from '@echogarden/text-segmentation';
|
|
3
|
-
export declare const wordCharacterRegExp: RegExp;
|
|
4
|
-
export declare const emojiSequenceRegExp: RegExp;
|
|
5
|
-
export declare const includesPunctuationRegExp: RegExp;
|
|
6
|
-
export declare const isAllPunctuationRegExp: RegExp;
|
|
7
|
-
export declare const phraseSeparators: string[];
|
|
8
|
-
export declare const symbolWords: string[];
|
|
9
3
|
export declare function isWordOrEmojiOrSymbolWord(str: string): boolean;
|
|
10
4
|
export declare function isWordOrEmoji(str: string): boolean;
|
|
11
5
|
export declare function isSymbolWord(str: string): boolean;
|
|
12
|
-
export declare function
|
|
6
|
+
export declare function includesWordCharacter(str: string): boolean;
|
|
13
7
|
export declare function includesPunctuation(str: string): boolean;
|
|
14
8
|
export declare function isAllPunctuation(str: string): boolean;
|
|
15
9
|
export declare function includesEmoji(str: string): boolean;
|
|
16
|
-
export declare function isAllWhitespace(str: string):
|
|
10
|
+
export declare function isAllWhitespace(str: string): boolean;
|
|
17
11
|
export declare function splitToParagraphs(text: string, paragraphBreakType: ParagraphBreakType, whitespaceProcessingMethod: WhitespaceProcessing): string[];
|
|
18
12
|
export declare function splitToLines(text: string): string[];
|
|
19
13
|
export declare function parseText(text: string, langCode: string): Promise<TextSegmentation.SegmentationResult>;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"Segmentation.d.ts","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,kBAAkB,CAAA;
|
|
1
|
+
{"version":3,"file":"Segmentation.d.ts","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,kBAAkB,CAAA;AAI3E,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AAmDjE,wBAAgB,yBAAyB,CAAC,GAAG,EAAE,MAAM,WAEpD;AAED,wBAAgB,aAAa,CAAC,GAAG,EAAE,MAAM,WAExC;AAED,wBAAgB,YAAY,CAAC,GAAG,EAAE,MAAM,WAEvC;AAED,wBAAgB,qBAAqB,CAAC,GAAG,EAAE,MAAM,WAEhD;AAED,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,WAE9C;AAED,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,MAAM,WAE3C;AAED,wBAAgB,aAAa,CAAC,GAAG,EAAE,MAAM,WAExC;AAED,wBAAgB,eAAe,CAAC,GAAG,EAAE,MAAM,WAE1C;AAKD,wBAAgB,iBAAiB,CAAC,IAAI,EAAE,MAAM,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,0BAA0B,EAAE,oBAAoB,YAoBvI;AAED,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,YAExC;AAED,wBAAsB,SAAS,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,gDAQ7D;AAED,wBAAsB,YAAY,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,gBAAgB,CAAC,YAAY,CAAC,CA8BzG;AAED,wBAAgB,yBAAyB,CAAC,IAAI,EAAE,MAAM,EAAE,0BAA0B,EAAE,oBAAoB,UAUvG;AAOD,wBAAsB,gBAAgB,CAAC,IAAI,EAAE,MAAM,EAAE,iBAAiB,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,iBAAiB,UAAO,EAAE,eAAe,UAAO,uBAoEjJ;AAED,wBAAsB,oCAAoC,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,uBA4BxF;AAED,qBAAa,QAAQ;IACpB,OAAO,EAAE,MAAM,EAAE,CAAK;IAEtB,QAAQ,CAAC,mBAAmB,QAAO;IAEnC,IAAI,MAAM,WAA+D;IAEzE,IAAI,IAAI,WAAuF;CAC/F;AAED,qBAAa,MAAM;IAClB,KAAK,EAAE,IAAI,EAAE,CAAK;IAElB,IAAI,MAAM,WAAyD;IAEnE,IAAI,IAAI,WAAiF;IAEzF,IAAI,QAAQ,qBAMX;IAED,IAAI,mBAAmB,YAA+E;CACtG;AAED,qBAAa,IAAI;IAChB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAA;IACrB,mBAAmB,EAAE,OAAO,CAAA;IAE5B,YAAY,IAAI,EAAE,MAAM,EAAE,mBAAmB,EAAE,OAAO,EAGrD;IAED,IAAI,uBAAuB,YAAmD;IAE9E,IAAI,YAAY,YAAqC;IAErD,IAAI,iBAAiB,YAEpB;IAED,IAAI,MAAM,WAA8B;CACxC;AAED,MAAM,MAAM,OAAO,GAAG,QAAQ,GAAG,MAAM,GAAG,IAAI,CAAA;AAE9C,qBAAa,QAAQ;IACpB,QAAQ,EAAE,OAAO,EAAE,CAAK;IAExB,IAAI,MAAM,WAAgE;IAE1E,IAAI,IAAI,WAA0F;IAElG,IAAI,OAAO,YAA8B;IAEzC,IAAI,UAAU,YAA2B;IAEzC,IAAI,WAAW,wBAMd;CACD"}
|
package/dist/nlp/Segmentation.js
CHANGED
|
@@ -1,43 +1,64 @@
|
|
|
1
1
|
import { sumArray, logToStderr } from '../utilities/Utilities.js';
|
|
2
2
|
import { getShortLanguageCode } from '../utilities/Locale.js';
|
|
3
|
-
import {
|
|
3
|
+
import { splitAndPreserveSeparators } from '../utilities/StringUtilities.js';
|
|
4
|
+
import { anyOf, buildRegExp, unicodeProperty, oneOrMore, possibly, codepoint, inputStart, inputEnd, whitespace } from 'regexp-composer';
|
|
4
5
|
import * as TextSegmentation from '@echogarden/text-segmentation';
|
|
5
6
|
import { splitChineseTextToWords_Jieba } from './ChineseSegmentation.js';
|
|
6
7
|
import { splitJapaneseTextToWords_Kuromoji } from './JapaneseSegmentation.js';
|
|
7
8
|
const log = logToStderr;
|
|
8
|
-
|
|
9
|
+
const includesWordCharacterPattern = anyOf(unicodeProperty('Letter'), unicodeProperty('Number'));
|
|
10
|
+
const includesWordCharacterRegExp = buildRegExp(includesWordCharacterPattern);
|
|
9
11
|
// See: https://mathiasbynens.be/notes/es-unicode-property-escapes
|
|
10
|
-
export const emojiSequenceRegExp = /\p{Emoji_Modifier_Base}\p{Emoji_Modifier}?|\p{Emoji_Presentation}|\p{Emoji}\uFE0F/u
|
|
11
|
-
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
|
|
12
|
+
//export const emojiSequenceRegExp = /\p{Emoji_Modifier_Base}\p{Emoji_Modifier}?|\p{Emoji_Presentation}|\p{Emoji}\uFE0F/u
|
|
13
|
+
const includesEmojiSequencePattern = anyOf([unicodeProperty('Emoji_Modifier_Base'), possibly(unicodeProperty('Emoji_Modifier'))], unicodeProperty('Emoji_Presentation'), [unicodeProperty('Emoji'), codepoint('FE0F')]);
|
|
14
|
+
const includesEmojiSequenceRegExp = buildRegExp(includesEmojiSequencePattern);
|
|
15
|
+
const symbolWordsList = [
|
|
16
|
+
'$', '€', '¢', '£', '¥', '©', '®', '™', '%', '&', '#', '~', '@', '+', '±', '÷',
|
|
17
|
+
'/', '\\', '^', '*', '×', '=', '≈', '¼', '½', '¾', '→', '≤', '≥', '∞'
|
|
18
|
+
];
|
|
19
|
+
const includesSymbolWordPattern = anyOf(...symbolWordsList);
|
|
20
|
+
const includesSymbolWordRegExp = buildRegExp(includesSymbolWordPattern);
|
|
21
|
+
const isAllSymbolWordsPattern = [inputStart, oneOrMore(includesSymbolWordPattern), inputEnd];
|
|
22
|
+
const isAllSymbolWordsRegExp = buildRegExp(isAllSymbolWordsPattern);
|
|
23
|
+
const includesWordCharacterOrEmojiPattern = anyOf(includesWordCharacterPattern, includesEmojiSequencePattern);
|
|
24
|
+
const includesWordCharacterOrEmojiRegExp = buildRegExp(includesWordCharacterOrEmojiPattern);
|
|
25
|
+
const includesWordCharacterOrEmojiOrIsAllSymbolWordPattern = anyOf(includesWordCharacterPattern, includesEmojiSequencePattern, isAllSymbolWordsPattern);
|
|
26
|
+
const includesWordCharacterOrEmojiOrIsAllSymbolWordRegExp = buildRegExp(includesWordCharacterOrEmojiOrIsAllSymbolWordPattern);
|
|
27
|
+
const includesPunctuationPattern = unicodeProperty('Punctuation');
|
|
28
|
+
const includesPunctuationRegExp = buildRegExp(includesPunctuationPattern);
|
|
29
|
+
const isAllPunctuationPattern = [inputStart, oneOrMore(includesPunctuationPattern), inputEnd];
|
|
30
|
+
const isAllPunctuationRegExp = buildRegExp(isAllPunctuationPattern);
|
|
31
|
+
const phraseSeparatorsList = [',', '、', ',', '،', ';', ';', ':', ':', '—'];
|
|
32
|
+
const includesPhraseSeparatorsPattern = anyOf(...phraseSeparatorsList);
|
|
33
|
+
const includesPhraseSeparatorsRegExp = buildRegExp(includesPhraseSeparatorsPattern);
|
|
34
|
+
const isAllWhitespacePattern = [inputStart, oneOrMore(whitespace), inputEnd];
|
|
35
|
+
const isAllWhitespaceRegExp = buildRegExp(isAllWhitespacePattern);
|
|
15
36
|
///////////////////////////////////////////////////////////////////////////////////////////////
|
|
16
37
|
// Predicates
|
|
17
38
|
///////////////////////////////////////////////////////////////////////////////////////////////
|
|
18
39
|
export function isWordOrEmojiOrSymbolWord(str) {
|
|
19
|
-
return
|
|
40
|
+
return includesWordCharacterOrEmojiOrIsAllSymbolWordRegExp.test(str.trim());
|
|
20
41
|
}
|
|
21
42
|
export function isWordOrEmoji(str) {
|
|
22
|
-
return
|
|
43
|
+
return includesWordCharacterOrEmojiRegExp.test(str.trim());
|
|
23
44
|
}
|
|
24
45
|
export function isSymbolWord(str) {
|
|
25
|
-
return
|
|
46
|
+
return isAllSymbolWordsRegExp.test(str.trim());
|
|
26
47
|
}
|
|
27
|
-
export function
|
|
28
|
-
return
|
|
48
|
+
export function includesWordCharacter(str) {
|
|
49
|
+
return includesWordCharacterRegExp.test(str.trim());
|
|
29
50
|
}
|
|
30
51
|
export function includesPunctuation(str) {
|
|
31
|
-
return includesPunctuationRegExp.test(str
|
|
52
|
+
return includesPunctuationRegExp.test(str.trim());
|
|
32
53
|
}
|
|
33
54
|
export function isAllPunctuation(str) {
|
|
34
|
-
return isAllPunctuationRegExp.test(str
|
|
55
|
+
return isAllPunctuationRegExp.test(str.trim());
|
|
35
56
|
}
|
|
36
57
|
export function includesEmoji(str) {
|
|
37
|
-
return
|
|
58
|
+
return includesEmojiSequenceRegExp.test(str.trim());
|
|
38
59
|
}
|
|
39
60
|
export function isAllWhitespace(str) {
|
|
40
|
-
return
|
|
61
|
+
return isAllWhitespaceRegExp.test(str);
|
|
41
62
|
}
|
|
42
63
|
///////////////////////////////////////////////////////////////////////////////////////////////
|
|
43
64
|
// Paragraph, line, sentence, phrase, and word segmentation
|
|
@@ -212,9 +233,11 @@ export class Word {
|
|
|
212
233
|
this.text = text;
|
|
213
234
|
this.isSentenceFinalizer = isSentenceFinalizer;
|
|
214
235
|
}
|
|
215
|
-
get containsOnlyPunctuation() { return !
|
|
216
|
-
get isSymbolWord() { return
|
|
217
|
-
get isPhraseSeperator() {
|
|
236
|
+
get containsOnlyPunctuation() { return !isWordOrEmojiOrSymbolWord(this.text); }
|
|
237
|
+
get isSymbolWord() { return isSymbolWord(this.text); }
|
|
238
|
+
get isPhraseSeperator() {
|
|
239
|
+
return this.containsOnlyPunctuation && includesPhraseSeparatorsRegExp.test(this.text);
|
|
240
|
+
}
|
|
218
241
|
get length() { return this.text.length; }
|
|
219
242
|
}
|
|
220
243
|
export class Fragment {
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"Segmentation.js","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACjE,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAE7D,OAAO,EAAE,aAAa,EAAE,0BAA0B,EAAE,MAAM,iCAAiC,CAAA;AAE3F,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,6BAA6B,EAAE,MAAM,0BAA0B,CAAA;AACxE,OAAO,EAAE,iCAAiC,EAAE,MAAM,2BAA2B,CAAA;AAE7E,MAAM,GAAG,GAAG,WAAW,CAAA;AAEvB,MAAM,CAAC,MAAM,mBAAmB,GAAG,yBAAyB,CAAA;AAE5D,kEAAkE;AAClE,MAAM,CAAC,MAAM,mBAAmB,GAAG,oFAAoF,CAAA;AAEvH,MAAM,CAAC,MAAM,yBAAyB,GAAG,oBAAoB,CAAA;AAC7D,MAAM,CAAC,MAAM,sBAAsB,GAAG,uBAAuB,CAAA;AAE7D,MAAM,CAAC,MAAM,gBAAgB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAC7E,MAAM,CAAC,MAAM,WAAW,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAE7K,+FAA+F;AAC/F,aAAa;AACb,+FAA+F;AAC/F,MAAM,UAAU,yBAAyB,CAAC,GAAW;IACpD,OAAO,aAAa,CAAC,GAAG,CAAC,IAAI,WAAW,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAA;AACvD,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,GAAW;IACxC,OAAO,MAAM,CAAC,GAAG,CAAC,IAAI,aAAa,CAAC,GAAG,CAAC,CAAA;AACzC,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,GAAW;IACvC,OAAO,WAAW,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AACzC,CAAC;AAED,MAAM,UAAU,MAAM,CAAC,GAAW;IACjC,OAAO,mBAAmB,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AAC7C,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,GAAW;IAC9C,OAAO,yBAAyB,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AACnD,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,GAAW;IAC3C,OAAO,sBAAsB,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AAChD,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,GAAW;IACxC,OAAO,mBAAmB,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AAC7C,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,GAAW;IAC1C,OAAO,GAAG,IAAI,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;AAChC,CAAC;AAED,+FAA+F;AAC/F,2DAA2D;AAC3D,+FAA+F;AAC/F,MAAM,UAAU,iBAAiB,CAAC,IAAY,EAAE,kBAAsC,EAAE,0BAAgD;IACvI,IAAI,UAAU,GAAa,EAAE,CAAA;IAE7B,IAAI,kBAAkB,KAAK,QAAQ,EAAE,CAAC;QACrC,UAAU,GAAG,0BAA0B,CAAC,IAAI,EAAE,WAAW,CAAC,CAAA;IAC3D,CAAC;SAAM,IAAI,kBAAkB,KAAK,QAAQ,EAAE,CAAC;QAC5C,UAAU,GAAG,0BAA0B,CAAC,IAAI,EAAE,kBAAkB,CAAC,CAAA;IAClE,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,kCAAkC,kBAAkB,GAAG,CAAC,CAAA;IACzE,CAAC;IAED,IAAI,UAAU,CAAC,MAAM,IAAI,CAAC,IAAI,eAAe,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QAC9D,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA;QAC7C,UAAU,CAAC,KAAK,EAAE,CAAA;IACnB,CAAC;IAED,UAAU,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,yBAAyB,CAAC,CAAC,EAAE,0BAA0B,CAAC,CAAC,CAAA;IAC1F,UAAU,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAEjD,OAAO,UAAU,CAAA;AAClB,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,IAAY;IACxC,OAAO,0BAA0B,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;AAClD,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,IAAY,EAAE,QAAgB;IAC7D,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,MAAM,YAAY,GAAG,MAAM,YAAY,CAAC,IAAI,EAAE,aAAa,CAAC,CAAA;IAE5D,MAAM,UAAU,GAAG,MAAM,gBAAgB,CAAC,mBAAmB,CAAC,YAAY,CAAC,CAAA;IAE3E,OAAO,UAAU,CAAA;AAClB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,IAAY,EAAE,QAAgB;IAChE,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,IAAI,aAAa,KAAK,IAAI,IAAI,aAAa,KAAK,KAAK,IAAI,aAAa,KAAK,IAAI,EAAE,CAAC;QACjF,IAAI,SAAS,GAAa,EAAE,CAAA;QAE5B,IAAI,aAAa,KAAK,IAAI,IAAI,aAAa,KAAK,KAAK,EAAE,CAAC;YACvD,SAAS,GAAG,MAAM,6BAA6B,CAAC,IAAI,CAAC,CAAA;QACtD,CAAC;aAAM,CAAC;YACP,SAAS,GAAG,MAAM,iCAAiC,CAAC,IAAI,CAAC,CAAA;QAC1D,CAAC;QAED,MAAM,YAAY,GAAG,IAAI,gBAAgB,CAAC,YAAY,EAAE,CAAA;QAExD,IAAI,MAAM,GAAG,CAAC,CAAA;QAEd,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;YAClC,MAAM,WAAW,GAAG,MAAM,CAAA;YAC1B,MAAM,SAAS,GAAG,WAAW,GAAG,QAAQ,CAAC,MAAM,CAAA;YAC/C,MAAM,aAAa,GAAG,CAAC,yBAAyB,CAAC,QAAQ,CAAC,CAAA;YAE1D,YAAY,CAAC,OAAO,CAAC,QAAQ,EAAE,WAAW,EAAE,aAAa,CAAC,CAAA;YAE1D,MAAM,GAAG,SAAS,CAAA;QACnB,CAAC;QAED,OAAO,YAAY,CAAA;IACpB,CAAC;SAAM,CAAC;QACP,OAAO,gBAAgB,CAAC,YAAY,CAAC,IAAI,EAAE,EAAE,QAAQ,EAAE,aAAa,EAAE,CAAC,CAAA;IACxE,CAAC;AACF,CAAC;AAED,MAAM,UAAU,yBAAyB,CAAC,IAAY,EAAE,0BAAgD;IACvG,IAAI,0BAA0B,KAAK,kBAAkB,EAAE,CAAC;QACvD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,UAAU,CAAC,WAAW,EAAE,GAAG,CAAC,CAAA;IAChD,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC,CAAA;IAC3C,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAA;IACZ,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,0CAA0C,0BAA0B,GAAG,CAAC,CAAA;IACzF,CAAC;AACF,CAAC;AAED,+FAA+F;AAC/F,wBAAwB;AACxB,EAAE;AACF,uEAAuE;AACvE,+FAA+F;AAC/F,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,IAAY,EAAE,iBAAyB,EAAE,QAAgB,EAAE,iBAAiB,GAAG,IAAI,EAAE,eAAe,GAAG,IAAI;IACjJ,MAAM,UAAU,GAAG,MAAM,oCAAoC,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IAE7E,MAAM,SAAS,GAAe,EAAE,CAAA;IAChC,IAAI,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;IAEpC,MAAM,oCAAoC,GAAG,GAAG,EAAE,CAAC,iBAAiB,GAAG,eAAe,CAAC,MAAM,CAAA;IAC7F,MAAM,yBAAyB,GAAG,GAAG,EAAE;QACtC,IAAI,eAAe,CAAC,UAAU,EAAE,CAAC;YAChC,SAAS,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;YAC/B,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;QACjC,CAAC;IACF,CAAC,CAAA;IAED,MAAM,mBAAmB,GAAG,CAAC,OAAgB,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,IAAI,oCAAoC,EAAE,CAAA;IAE1G,KAAK,MAAM,QAAQ,IAAI,UAAU,EAAE,CAAC;QACnC,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;YACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YACvC,SAAQ;QACT,CAAC;QAED,IAAI,iBAAiB,EAAE,CAAC;YACvB,yBAAyB,EAAE,CAAA;YAE3B,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;gBACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;gBACvC,SAAQ;YACT,CAAC;QACF,CAAC;QAED,KAAK,MAAM,MAAM,IAAI,QAAQ,CAAC,OAAO,EAAE,CAAC;YACvC,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;gBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;gBACrC,SAAQ;YACT,CAAC;YAGD,IAAI,eAAe,EAAE,CAAC;gBACrB,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;oBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;oBACrC,SAAQ;gBACT,CAAC;YACF,CAAC;YAED,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,KAAK,EAAE,CAAC;gBACjC,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,MAAM,IAAI,KAAK,CAAC,gCAAgC,IAAI,CAAC,MAAM,mDAAmD,iBAAiB,EAAE,CAAC,CAAA;YACnI,CAAC;QACF,CAAC;IACF,CAAC;IAED,yBAAyB,EAAE,CAAA;IAE3B,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,oCAAoC,CAAC,IAAY,EAAE,QAAgB;IACxF,MAAM,aAAa,GAAG,MAAM,SAAS,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IAErD,MAAM,SAAS,GAAe,EAAE,CAAA;IAEhC,KAAK,MAAM,aAAa,IAAI,aAAa,CAAC,SAAS,EAAE,CAAC;QACrD,MAAM,QAAQ,GAAG,IAAI,QAAQ,EAAE,CAAA;QAE/B,KAAK,MAAM,WAAW,IAAI,aAAa,CAAC,OAAO,EAAE,CAAC;YACjD,MAAM,MAAM,GAAG,IAAI,MAAM,EAAE,CAAA;YAE3B,KAAK,MAAM,SAAS,IAAI,WAAW,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBACnD,MAAM,mBAAmB,GAAG,SAAS,KAAK,aAAa,CAAC,KAAK,CAAC,SAAS,CAAA;gBAEvE,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,mBAAmB,CAAC,CAAA;gBAE1D,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YACxB,CAAC;YAED,IAAI,MAAM,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC7B,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YAC9B,CAAC;QACF,CAAC;QAED,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;IACzB,CAAC;IAED,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,OAAO,QAAQ;IACpB,OAAO,GAAa,EAAE,CAAA;IAEb,mBAAmB,GAAG,IAAI,CAAA;IAEnC,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEzE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,OAAO,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,MAAM,EAAE,EAAE,CAAC,MAAM,GAAG,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;CAC/F;AAED,MAAM,OAAO,MAAM;IAClB,KAAK,GAAW,EAAE,CAAA;IAElB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEnE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,KAAK,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,MAAM,GAAG,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAEzF,IAAI,QAAQ;QACX,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;YAC5B,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IACzC,CAAC;IAED,IAAI,mBAAmB,KAAK,OAAO,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,mBAAmB,CAAC,CAAC,CAAC,KAAK,CAAA,CAAC,CAAC;CACtG;AAED,MAAM,OAAO,IAAI;IACP,IAAI,CAAQ;IACrB,mBAAmB,CAAS;IAE5B,YAAY,IAAY,EAAE,mBAA4B;QACrD,IAAI,CAAC,IAAI,GAAG,IAAI,CAAA;QAChB,IAAI,CAAC,mBAAmB,GAAG,mBAAmB,CAAA;IAC/C,CAAC;IAED,IAAI,uBAAuB,KAAK,OAAO,CAAC,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,YAAY,CAAA,CAAC,CAAC;IAEnG,IAAI,YAAY,KAAK,OAAO,WAAW,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA,CAAC,CAAC;IAE7D,IAAI,iBAAiB,KAAK,OAAO,IAAI,CAAC,uBAAuB,IAAI,aAAa,CAAC,IAAI,CAAC,IAAI,EAAE,gBAAgB,CAAC,CAAA,CAAC,CAAC;IAE7G,IAAI,MAAM,KAAK,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,CAAA,CAAC,CAAC;CACxC;AAID,MAAM,OAAO,QAAQ;IACpB,QAAQ,GAAc,EAAE,CAAA;IAExB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAE1E,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,OAAO,EAAE,EAAE,CAAC,MAAM,GAAG,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAElG,IAAI,OAAO,KAAK,OAAO,IAAI,CAAC,MAAM,IAAI,CAAC,CAAA,CAAC,CAAC;IAEzC,IAAI,UAAU,KAAK,OAAO,CAAC,IAAI,CAAC,OAAO,CAAA,CAAC,CAAC;IAEzC,IAAI,WAAW;QACd,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YAClB,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAC/C,CAAC;CACD"}
|
|
1
|
+
{"version":3,"file":"Segmentation.js","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACjE,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAE7D,OAAO,EAAE,0BAA0B,EAAE,MAAM,iCAAiC,CAAA;AAC5E,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,eAAe,EAAE,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,UAAU,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAA;AAEvI,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,6BAA6B,EAAE,MAAM,0BAA0B,CAAA;AACxE,OAAO,EAAE,iCAAiC,EAAE,MAAM,2BAA2B,CAAA;AAE7E,MAAM,GAAG,GAAG,WAAW,CAAA;AAEvB,MAAM,4BAA4B,GAAG,KAAK,CAAC,eAAe,CAAC,QAAQ,CAAC,EAAE,eAAe,CAAC,QAAQ,CAAC,CAAC,CAAA;AAChG,MAAM,2BAA2B,GAAG,WAAW,CAAC,4BAA4B,CAAC,CAAA;AAE7E,kEAAkE;AAClE,yHAAyH;AACzH,MAAM,4BAA4B,GAAG,KAAK,CACzC,CAAC,eAAe,CAAC,qBAAqB,CAAC,EAAE,QAAQ,CAAC,eAAe,CAAC,gBAAgB,CAAC,CAAC,CAAC,EACrF,eAAe,CAAC,oBAAoB,CAAC,EACrC,CAAC,eAAe,CAAC,OAAO,CAAC,EAAE,SAAS,CAAC,MAAM,CAAC,CAAC,CAC7C,CAAA;AACD,MAAM,2BAA2B,GAAG,WAAW,CAAC,4BAA4B,CAAC,CAAA;AAE7E,MAAM,eAAe,GAAG;IACvB,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG;IAC9E,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG;CACrE,CAAA;AAED,MAAM,yBAAyB,GAAG,KAAK,CAAC,GAAG,eAAe,CAAC,CAAA;AAC3D,MAAM,wBAAwB,GAAG,WAAW,CAAC,yBAAyB,CAAC,CAAA;AAEvE,MAAM,uBAAuB,GAAG,CAAC,UAAU,EAAE,SAAS,CAAC,yBAAyB,CAAC,EAAE,QAAQ,CAAC,CAAA;AAC5F,MAAM,sBAAsB,GAAG,WAAW,CAAC,uBAAuB,CAAC,CAAA;AAEnE,MAAM,mCAAmC,GAAG,KAAK,CAAC,4BAA4B,EAAE,4BAA4B,CAAC,CAAA;AAC7G,MAAM,kCAAkC,GAAG,WAAW,CAAC,mCAAmC,CAAC,CAAA;AAE3F,MAAM,oDAAoD,GAAG,KAAK,CAAC,4BAA4B,EAAE,4BAA4B,EAAE,uBAAuB,CAAC,CAAA;AACvJ,MAAM,mDAAmD,GAAG,WAAW,CAAC,oDAAoD,CAAC,CAAA;AAE7H,MAAM,0BAA0B,GAAG,eAAe,CAAC,aAAa,CAAC,CAAA;AACjE,MAAM,yBAAyB,GAAG,WAAW,CAAC,0BAA0B,CAAC,CAAA;AAEzE,MAAM,uBAAuB,GAAG,CAAC,UAAU,EAAE,SAAS,CAAC,0BAA0B,CAAC,EAAE,QAAQ,CAAC,CAAA;AAC7F,MAAM,sBAAsB,GAAG,WAAW,CAAC,uBAAuB,CAAC,CAAA;AAEnE,MAAM,oBAAoB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAC1E,MAAM,+BAA+B,GAAG,KAAK,CAAC,GAAG,oBAAoB,CAAC,CAAA;AACtE,MAAM,8BAA8B,GAAG,WAAW,CAAC,+BAA+B,CAAC,CAAA;AAEnF,MAAM,sBAAsB,GAAG,CAAC,UAAU,EAAE,SAAS,CAAC,UAAU,CAAC,EAAE,QAAQ,CAAC,CAAA;AAC5E,MAAM,qBAAqB,GAAG,WAAW,CAAC,sBAAsB,CAAC,CAAA;AAEjE,+FAA+F;AAC/F,aAAa;AACb,+FAA+F;AAC/F,MAAM,UAAU,yBAAyB,CAAC,GAAW;IACpD,OAAO,mDAAmD,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC5E,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,GAAW;IACxC,OAAO,kCAAkC,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC3D,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,GAAW;IACvC,OAAO,sBAAsB,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC/C,CAAC;AAED,MAAM,UAAU,qBAAqB,CAAC,GAAW;IAChD,OAAO,2BAA2B,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AACpD,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,GAAW;IAC9C,OAAO,yBAAyB,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAClD,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,GAAW;IAC3C,OAAO,sBAAsB,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC/C,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,GAAW;IACxC,OAAO,2BAA2B,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AACpD,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,GAAW;IAC1C,OAAO,qBAAqB,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;AACvC,CAAC;AAED,+FAA+F;AAC/F,2DAA2D;AAC3D,+FAA+F;AAC/F,MAAM,UAAU,iBAAiB,CAAC,IAAY,EAAE,kBAAsC,EAAE,0BAAgD;IACvI,IAAI,UAAU,GAAa,EAAE,CAAA;IAE7B,IAAI,kBAAkB,KAAK,QAAQ,EAAE,CAAC;QACrC,UAAU,GAAG,0BAA0B,CAAC,IAAI,EAAE,WAAW,CAAC,CAAA;IAC3D,CAAC;SAAM,IAAI,kBAAkB,KAAK,QAAQ,EAAE,CAAC;QAC5C,UAAU,GAAG,0BAA0B,CAAC,IAAI,EAAE,kBAAkB,CAAC,CAAA;IAClE,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,kCAAkC,kBAAkB,GAAG,CAAC,CAAA;IACzE,CAAC;IAED,IAAI,UAAU,CAAC,MAAM,IAAI,CAAC,IAAI,eAAe,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QAC9D,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA;QAC7C,UAAU,CAAC,KAAK,EAAE,CAAA;IACnB,CAAC;IAED,UAAU,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,yBAAyB,CAAC,CAAC,EAAE,0BAA0B,CAAC,CAAC,CAAA;IAC1F,UAAU,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAEjD,OAAO,UAAU,CAAA;AAClB,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,IAAY;IACxC,OAAO,0BAA0B,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;AAClD,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,IAAY,EAAE,QAAgB;IAC7D,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,MAAM,YAAY,GAAG,MAAM,YAAY,CAAC,IAAI,EAAE,aAAa,CAAC,CAAA;IAE5D,MAAM,UAAU,GAAG,MAAM,gBAAgB,CAAC,mBAAmB,CAAC,YAAY,CAAC,CAAA;IAE3E,OAAO,UAAU,CAAA;AAClB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,IAAY,EAAE,QAAgB;IAChE,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,IAAI,aAAa,KAAK,IAAI,IAAI,aAAa,KAAK,KAAK,IAAI,aAAa,KAAK,IAAI,EAAE,CAAC;QACjF,IAAI,SAAS,GAAa,EAAE,CAAA;QAE5B,IAAI,aAAa,KAAK,IAAI,IAAI,aAAa,KAAK,KAAK,EAAE,CAAC;YACvD,SAAS,GAAG,MAAM,6BAA6B,CAAC,IAAI,CAAC,CAAA;QACtD,CAAC;aAAM,CAAC;YACP,SAAS,GAAG,MAAM,iCAAiC,CAAC,IAAI,CAAC,CAAA;QAC1D,CAAC;QAED,MAAM,YAAY,GAAG,IAAI,gBAAgB,CAAC,YAAY,EAAE,CAAA;QAExD,IAAI,MAAM,GAAG,CAAC,CAAA;QAEd,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;YAClC,MAAM,WAAW,GAAG,MAAM,CAAA;YAC1B,MAAM,SAAS,GAAG,WAAW,GAAG,QAAQ,CAAC,MAAM,CAAA;YAC/C,MAAM,aAAa,GAAG,CAAC,yBAAyB,CAAC,QAAQ,CAAC,CAAA;YAE1D,YAAY,CAAC,OAAO,CAAC,QAAQ,EAAE,WAAW,EAAE,aAAa,CAAC,CAAA;YAE1D,MAAM,GAAG,SAAS,CAAA;QACnB,CAAC;QAED,OAAO,YAAY,CAAA;IACpB,CAAC;SAAM,CAAC;QACP,OAAO,gBAAgB,CAAC,YAAY,CAAC,IAAI,EAAE,EAAE,QAAQ,EAAE,aAAa,EAAE,CAAC,CAAA;IACxE,CAAC;AACF,CAAC;AAED,MAAM,UAAU,yBAAyB,CAAC,IAAY,EAAE,0BAAgD;IACvG,IAAI,0BAA0B,KAAK,kBAAkB,EAAE,CAAC;QACvD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,UAAU,CAAC,WAAW,EAAE,GAAG,CAAC,CAAA;IAChD,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC,CAAA;IAC3C,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAA;IACZ,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,0CAA0C,0BAA0B,GAAG,CAAC,CAAA;IACzF,CAAC;AACF,CAAC;AAED,+FAA+F;AAC/F,wBAAwB;AACxB,EAAE;AACF,uEAAuE;AACvE,+FAA+F;AAC/F,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,IAAY,EAAE,iBAAyB,EAAE,QAAgB,EAAE,iBAAiB,GAAG,IAAI,EAAE,eAAe,GAAG,IAAI;IACjJ,MAAM,UAAU,GAAG,MAAM,oCAAoC,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IAE7E,MAAM,SAAS,GAAe,EAAE,CAAA;IAChC,IAAI,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;IAEpC,MAAM,oCAAoC,GAAG,GAAG,EAAE,CAAC,iBAAiB,GAAG,eAAe,CAAC,MAAM,CAAA;IAC7F,MAAM,yBAAyB,GAAG,GAAG,EAAE;QACtC,IAAI,eAAe,CAAC,UAAU,EAAE,CAAC;YAChC,SAAS,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;YAC/B,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;QACjC,CAAC;IACF,CAAC,CAAA;IAED,MAAM,mBAAmB,GAAG,CAAC,OAAgB,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,IAAI,oCAAoC,EAAE,CAAA;IAE1G,KAAK,MAAM,QAAQ,IAAI,UAAU,EAAE,CAAC;QACnC,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;YACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YACvC,SAAQ;QACT,CAAC;QAED,IAAI,iBAAiB,EAAE,CAAC;YACvB,yBAAyB,EAAE,CAAA;YAE3B,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;gBACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;gBACvC,SAAQ;YACT,CAAC;QACF,CAAC;QAED,KAAK,MAAM,MAAM,IAAI,QAAQ,CAAC,OAAO,EAAE,CAAC;YACvC,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;gBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;gBACrC,SAAQ;YACT,CAAC;YAGD,IAAI,eAAe,EAAE,CAAC;gBACrB,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;oBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;oBACrC,SAAQ;gBACT,CAAC;YACF,CAAC;YAED,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,KAAK,EAAE,CAAC;gBACjC,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,MAAM,IAAI,KAAK,CAAC,gCAAgC,IAAI,CAAC,MAAM,mDAAmD,iBAAiB,EAAE,CAAC,CAAA;YACnI,CAAC;QACF,CAAC;IACF,CAAC;IAED,yBAAyB,EAAE,CAAA;IAE3B,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,oCAAoC,CAAC,IAAY,EAAE,QAAgB;IACxF,MAAM,aAAa,GAAG,MAAM,SAAS,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IAErD,MAAM,SAAS,GAAe,EAAE,CAAA;IAEhC,KAAK,MAAM,aAAa,IAAI,aAAa,CAAC,SAAS,EAAE,CAAC;QACrD,MAAM,QAAQ,GAAG,IAAI,QAAQ,EAAE,CAAA;QAE/B,KAAK,MAAM,WAAW,IAAI,aAAa,CAAC,OAAO,EAAE,CAAC;YACjD,MAAM,MAAM,GAAG,IAAI,MAAM,EAAE,CAAA;YAE3B,KAAK,MAAM,SAAS,IAAI,WAAW,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBACnD,MAAM,mBAAmB,GAAG,SAAS,KAAK,aAAa,CAAC,KAAK,CAAC,SAAS,CAAA;gBAEvE,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,mBAAmB,CAAC,CAAA;gBAE1D,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YACxB,CAAC;YAED,IAAI,MAAM,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC7B,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YAC9B,CAAC;QACF,CAAC;QAED,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;IACzB,CAAC;IAED,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,OAAO,QAAQ;IACpB,OAAO,GAAa,EAAE,CAAA;IAEb,mBAAmB,GAAG,IAAI,CAAA;IAEnC,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEzE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,OAAO,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,MAAM,EAAE,EAAE,CAAC,MAAM,GAAG,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;CAC/F;AAED,MAAM,OAAO,MAAM;IAClB,KAAK,GAAW,EAAE,CAAA;IAElB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEnE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,KAAK,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,MAAM,GAAG,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAEzF,IAAI,QAAQ;QACX,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;YAC5B,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IACzC,CAAC;IAED,IAAI,mBAAmB,KAAK,OAAO,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,mBAAmB,CAAC,CAAC,CAAC,KAAK,CAAA,CAAC,CAAC;CACtG;AAED,MAAM,OAAO,IAAI;IACP,IAAI,CAAQ;IACrB,mBAAmB,CAAS;IAE5B,YAAY,IAAY,EAAE,mBAA4B;QACrD,IAAI,CAAC,IAAI,GAAG,IAAI,CAAA;QAChB,IAAI,CAAC,mBAAmB,GAAG,mBAAmB,CAAA;IAC/C,CAAC;IAED,IAAI,uBAAuB,KAAK,OAAO,CAAC,yBAAyB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA,CAAC,CAAC;IAE9E,IAAI,YAAY,KAAK,OAAO,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA,CAAC,CAAC;IAErD,IAAI,iBAAiB;QACpB,OAAO,IAAI,CAAC,uBAAuB,IAAI,8BAA8B,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IACtF,CAAC;IAED,IAAI,MAAM,KAAK,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,CAAA,CAAC,CAAC;CACxC;AAID,MAAM,OAAO,QAAQ;IACpB,QAAQ,GAAc,EAAE,CAAA;IAExB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAE1E,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,OAAO,EAAE,EAAE,CAAC,MAAM,GAAG,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAElG,IAAI,OAAO,KAAK,OAAO,IAAI,CAAC,MAAM,IAAI,CAAC,CAAA,CAAC,CAAC;IAEzC,IAAI,UAAU,KAAK,OAAO,CAAC,IAAI,CAAC,OAAO,CAAA,CAAC,CAAC;IAEzC,IAAI,WAAW;QACd,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YAClB,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAC/C,CAAC;CACD"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"TextNormalizer.d.ts","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAIA,wBAAgB,+BAA+B,CAC9C,KAAK,EAAE,MAAM,EAAE,EACf,kBAAkB,EAAE,MAAM,EAAE,EAC5B,8BAA8B,EAAE,MAAM,EAAE,EACxC,QAAQ,EAAE,MAAM;;;
|
|
1
|
+
{"version":3,"file":"TextNormalizer.d.ts","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAIA,wBAAgB,+BAA+B,CAC9C,KAAK,EAAE,MAAM,EAAE,EACf,kBAAkB,EAAE,MAAM,EAAE,EAC5B,8BAA8B,EAAE,MAAM,EAAE,EACxC,QAAQ,EAAE,MAAM;;;EAgHhB;AAED,wBAAgB,4BAA4B,CAAC,UAAU,EAAE,MAAM,UAe9D;AAED,wBAAgB,8BAA8B,CAAC,YAAY,EAAE,MAAM,UAoBlE;AAED,wBAAgB,6BAA6B,CAAC,IAAI,EAAE,MAAM,UAEzD;AAED,eAAO,MAAM,6BAA6B,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAqDhE,CAAA"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
import { getShortLanguageCode } from '../utilities/Locale.js';
|
|
2
2
|
import { substituteCharactersUsingLookup } from '../utilities/StringUtilities.js';
|
|
3
|
-
import { anyOf, buildRegExp, charRange, inputEnd, inputStart, repeated, zeroOrMore } from 'regexp-composer';
|
|
3
|
+
import { anyOf, buildRegExp, charRange, inputEnd, inputStart, oneOrMore, repeated, unicodeProperty, zeroOrMore } from 'regexp-composer';
|
|
4
4
|
export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWhitespaceWordOriginalIndex, language) {
|
|
5
5
|
language = getShortLanguageCode(language);
|
|
6
6
|
if (language != 'en') {
|
|
@@ -15,9 +15,15 @@ export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWh
|
|
|
15
15
|
const nextNonWhitespaceWord = nextNonWhitespaceWords[0];
|
|
16
16
|
const originalWordIndex = nonWhitespaceWordOriginalIndex[wordIndex];
|
|
17
17
|
const isFollowedByWhitespace = words[originalWordIndex + 1]?.trim().length === 0;
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
|
|
18
|
+
const isSpecialCharacterBeforeYear = ['(', ',', '©'].includes(words[originalWordIndex]);
|
|
19
|
+
//const isWordPrecedingAYear = wordsPrecedingAYear.includes(lowerCaseWord)
|
|
20
|
+
const isWordPrecedingAYear = isAllLettersOrApostropheRegExp.test(lowerCaseWord) || isSpecialCharacterBeforeYear;
|
|
21
|
+
const followedByFourDigitYearPattern = fourDigitYearRegExp.test(nextNonWhitespaceWord);
|
|
22
|
+
const isWordPrecedingADecade = wordsPrecedingADecade.includes(lowerCaseWord);
|
|
23
|
+
const followedByFourDigitDecadePattern = fourDigitDecadeRegExp.test(nextNonWhitespaceWord);
|
|
24
|
+
if (isWordPrecedingAYear &&
|
|
25
|
+
(isFollowedByWhitespace || isSpecialCharacterBeforeYear) &&
|
|
26
|
+
followedByFourDigitYearPattern) {
|
|
21
27
|
// Normalize a four digit year pattern, e.g. 'in 1995'.
|
|
22
28
|
const normalizedString = normalizeFourDigitYearString(nextNonWhitespaceWord);
|
|
23
29
|
normalizedFragments.push(word);
|
|
@@ -26,9 +32,9 @@ export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWh
|
|
|
26
32
|
referenceFragments.push(nextNonWhitespaceWord);
|
|
27
33
|
wordIndex += 1;
|
|
28
34
|
}
|
|
29
|
-
else if (
|
|
35
|
+
else if (isWordPrecedingADecade &&
|
|
30
36
|
isFollowedByWhitespace &&
|
|
31
|
-
|
|
37
|
+
followedByFourDigitDecadePattern) {
|
|
32
38
|
// Normalize a four digit decade pattern, e.g. 'the 1980s'.
|
|
33
39
|
const normalizedString = normalizeFourDigitDecadeString(nextNonWhitespaceWord);
|
|
34
40
|
normalizedFragments.push(word);
|
|
@@ -37,7 +43,7 @@ export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWh
|
|
|
37
43
|
referenceFragments.push(nextNonWhitespaceWord);
|
|
38
44
|
wordIndex += 1;
|
|
39
45
|
}
|
|
40
|
-
else if (
|
|
46
|
+
else if (fourDigitYearRangeRegExp.test(words.slice(originalWordIndex, originalWordIndex + 3).join(''))) {
|
|
41
47
|
// Normalize a year range pattern, e.g. '1835-1896', ensure there are no spaces between words
|
|
42
48
|
normalizedFragments.push(normalizeFourDigitYearString(nonWhitespaceWords[wordIndex]));
|
|
43
49
|
referenceFragments.push(nonWhitespaceWords[wordIndex]);
|
|
@@ -64,7 +70,7 @@ export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWh
|
|
|
64
70
|
referenceFragments.push(referenceString);
|
|
65
71
|
}
|
|
66
72
|
}
|
|
67
|
-
else if (
|
|
73
|
+
else if (followingCurrencyRegExp.test(lowerCaseWord)) {
|
|
68
74
|
const currencyWord = currencySymbolsAsWords[currencySymbols.indexOf(lowerCaseWord[lowerCaseWord.length - 1])];
|
|
69
75
|
const normalizedString = `${word.substring(0, word.length - 1)} ${currencyWord}`;
|
|
70
76
|
normalizedFragments.push(normalizedString);
|
|
@@ -182,13 +188,13 @@ const wordsFollowingACurrency = [
|
|
|
182
188
|
'million', 'billion', 'trillion'
|
|
183
189
|
];
|
|
184
190
|
const arabicNumeralPattern = charRange('0', '9');
|
|
185
|
-
const
|
|
191
|
+
const isNumberPattern = [
|
|
186
192
|
inputStart,
|
|
187
193
|
arabicNumeralPattern,
|
|
188
194
|
zeroOrMore(anyOf(arabicNumeralPattern, ',', '.')),
|
|
189
195
|
inputEnd
|
|
190
196
|
];
|
|
191
|
-
const
|
|
197
|
+
const isNumberRegExp = buildRegExp(isNumberPattern);
|
|
192
198
|
const precedingCurrencyPattern = [
|
|
193
199
|
inputStart,
|
|
194
200
|
anyOf(...currencySymbols),
|
|
@@ -204,11 +210,20 @@ const followingCurrencyPattern = [
|
|
|
204
210
|
anyOf(...currencySymbols),
|
|
205
211
|
inputEnd
|
|
206
212
|
];
|
|
207
|
-
const
|
|
213
|
+
const followingCurrencyRegExp = buildRegExp(followingCurrencyPattern);
|
|
208
214
|
const fourDigitYearPattern = [inputStart, repeated(4, arabicNumeralPattern), inputEnd];
|
|
209
|
-
const
|
|
215
|
+
const fourDigitYearRegExp = buildRegExp(fourDigitYearPattern);
|
|
210
216
|
const fourDigitDecadePattern = [inputStart, repeated(3, arabicNumeralPattern), '0s', inputEnd];
|
|
211
|
-
const
|
|
217
|
+
const fourDigitDecadeRegExp = buildRegExp(fourDigitDecadePattern);
|
|
212
218
|
const fourDigitYearRangePattern = [inputStart, repeated(4, arabicNumeralPattern), anyOf('-', '–'), repeated(4, arabicNumeralPattern), inputEnd];
|
|
213
|
-
const
|
|
219
|
+
const fourDigitYearRangeRegExp = buildRegExp(fourDigitYearRangePattern);
|
|
220
|
+
const isAllLettersPattern = [inputStart, oneOrMore(unicodeProperty('Letter')), inputEnd];
|
|
221
|
+
const isAllLettersRegExp = buildRegExp(isAllLettersPattern);
|
|
222
|
+
const isAllLettersOrApostrophePattern = [
|
|
223
|
+
inputStart,
|
|
224
|
+
unicodeProperty('Letter'),
|
|
225
|
+
zeroOrMore(anyOf(unicodeProperty('Letter'), `'`)),
|
|
226
|
+
inputEnd,
|
|
227
|
+
];
|
|
228
|
+
const isAllLettersOrApostropheRegExp = buildRegExp(isAllLettersOrApostrophePattern);
|
|
214
229
|
//# sourceMappingURL=TextNormalizer.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"TextNormalizer.js","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAC7D,OAAO,EAAE,+BAA+B,EAAE,MAAM,iCAAiC,CAAA;AACjF,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,SAAS,EAAE,QAAQ,EAAE,UAAU,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAA;
|
|
1
|
+
{"version":3,"file":"TextNormalizer.js","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAC7D,OAAO,EAAE,+BAA+B,EAAE,MAAM,iCAAiC,CAAA;AACjF,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,SAAS,EAAE,QAAQ,EAAE,UAAU,EAAE,SAAS,EAAE,QAAQ,EAAE,eAAe,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAA;AAEvI,MAAM,UAAU,+BAA+B,CAC9C,KAAe,EACf,kBAA4B,EAC5B,8BAAwC,EACxC,QAAgB;IAEhB,QAAQ,GAAG,oBAAoB,CAAC,QAAQ,CAAC,CAAA;IAEzC,IAAI,QAAQ,IAAI,IAAI,EAAE,CAAC;QACtB,OAAO,EAAE,mBAAmB,EAAE,CAAC,GAAG,kBAAkB,CAAC,EAAE,kBAAkB,EAAE,CAAC,GAAG,kBAAkB,CAAC,EAAE,CAAA;IACrG,CAAC;IAED,MAAM,mBAAmB,GAAa,EAAE,CAAA;IACxC,MAAM,kBAAkB,GAAa,EAAE,CAAA;IAEvC,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,kBAAkB,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,CAAC;QAC5E,MAAM,IAAI,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAA;QAC1C,MAAM,aAAa,GAAG,IAAI,CAAC,WAAW,EAAE,CAAA;QAExC,MAAM,sBAAsB,GAAG,kBAAkB,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAA;QACtE,MAAM,qBAAqB,GAAG,sBAAsB,CAAC,CAAC,CAAC,CAAA;QAEvD,MAAM,iBAAiB,GAAG,8BAA8B,CAAC,SAAS,CAAC,CAAA;QAEnE,MAAM,sBAAsB,GAAG,KAAK,CAAC,iBAAiB,GAAG,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,CAAA;QAChF,MAAM,4BAA4B,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,iBAAiB,CAAC,CAAC,CAAA;QAEvF,0EAA0E;QAC1E,MAAM,oBAAoB,GACzB,8BAA8B,CAAC,IAAI,CAAC,aAAa,CAAC,IAAI,4BAA4B,CAAA;QAEnF,MAAM,8BAA8B,GAAG,mBAAmB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;QAEtF,MAAM,sBAAsB,GAAG,qBAAqB,CAAC,QAAQ,CAAC,aAAa,CAAC,CAAA;QAC5E,MAAM,gCAAgC,GAAG,qBAAqB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;QAE1F,IAAI,oBAAoB;YACvB,CAAC,sBAAsB,IAAI,4BAA4B,CAAC;YACxD,8BAA8B,EAAE,CAAC;YAEjC,uDAAuD;YACvD,MAAM,gBAAgB,GAAG,4BAA4B,CAAC,qBAAqB,CAAC,CAAA;YAE5E,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAE7B,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAC1C,kBAAkB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;YAE9C,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,IACN,sBAAsB;YACtB,sBAAsB;YACtB,gCAAgC,EAAE,CAAC;YAEnC,2DAA2D;YAE3D,MAAM,gBAAgB,GAAG,8BAA8B,CAAC,qBAAqB,CAAC,CAAA;YAE9E,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAE7B,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAC1C,kBAAkB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;YAE9C,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,IAAI,wBAAwB,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,iBAAiB,EAAE,iBAAiB,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC;YAC1G,6FAA6F;YAC7F,mBAAmB,CAAC,IAAI,CAAC,4BAA4B,CAAC,kBAAkB,CAAC,SAAS,CAAC,CAAC,CAAC,CAAA;YACrF,kBAAkB,CAAC,IAAI,CAAC,kBAAkB,CAAC,SAAS,CAAC,CAAC,CAAA;YAEtD,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,kBAAkB,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAA;YAE1D,mBAAmB,CAAC,IAAI,CAAC,4BAA4B,CAAC,kBAAkB,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;YACzF,kBAAkB,CAAC,IAAI,CAAC,kBAAkB,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAA;YAE1D,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,IAAI,8BAA8B,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,CAAC;YAC/D,8FAA8F;YAE9F,MAAM,YAAY,GAAG,sBAAsB,CAAC,eAAe,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;YAEtF,IAAI,uBAAuB,CAAC,QAAQ,CAAC,qBAAqB,EAAE,WAAW,EAAE,CAAC,EAAE,CAAC;gBAC5E,MAAM,gBAAgB,GAAG,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,qBAAqB,IAAI,YAAY,EAAE,CAAA;gBAExF,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;gBAE1C,MAAM,eAAe,GAAG,GAAG,IAAI,IAAI,qBAAqB,EAAE,CAAA;gBAC1D,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;gBAExC,SAAS,IAAI,CAAC,CAAA;YACf,CAAC;iBAAM,CAAC;gBACP,MAAM,gBAAgB,GAAG,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,YAAY,EAAE,CAAA;gBAE/D,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;gBAE1C,MAAM,eAAe,GAAG,IAAI,CAAA;gBAC5B,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;YACzC,CAAC;QACF,CAAC;aAAM,IAAI,uBAAuB,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,CAAC;YACxD,MAAM,YAAY,GAAG,sBAAsB,CAAC,eAAe,CAAC,OAAO,CAAC,aAAa,CAAC,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;YAE7G,MAAM,gBAAgB,GAAG,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,IAAI,YAAY,EAAE,CAAA;YAEhF,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAE1C,MAAM,eAAe,GAAG,IAAI,CAAA;YAC5B,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;QACzC,CAAC;aAAM,CAAC;YACP,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAC9B,CAAC;IACF,CAAC;IAED,OAAO,EAAE,mBAAmB,EAAE,kBAAkB,EAAE,CAAA;AACnD,CAAC;AAED,MAAM,UAAU,4BAA4B,CAAC,UAAkB;IAC9D,MAAM,mBAAmB,GAAG,UAAU,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAClE,MAAM,oBAAoB,GAAG,UAAU,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEnE,IAAI,gBAAwB,CAAA;IAE5B,IAAI,mBAAmB,IAAI,EAAE,IAAI,oBAAoB,IAAI,EAAE,EAAE,CAAC;QAC7D,gBAAgB,GAAG,GAAG,mBAAmB,IAAI,oBAAoB,EAAE,CAAA;IACpE,CAAC;SAAM,IAAI,mBAAmB,IAAI,EAAE,IAAI,mBAAmB,GAAG,EAAE,IAAI,CAAC,IAAI,oBAAoB,GAAG,EAAE,EAAE,CAAC;QACpG,gBAAgB,GAAG,GAAG,mBAAmB,OAAO,oBAAoB,EAAE,CAAA;IACvE,CAAC;SAAM,CAAC;QACP,gBAAgB,GAAG,UAAU,CAAA;IAC9B,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,MAAM,UAAU,8BAA8B,CAAC,YAAoB;IAClE,MAAM,mBAAmB,GAAG,QAAQ,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAClE,MAAM,oBAAoB,GAAG,QAAQ,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEnE,IAAI,gBAAwB,CAAA;IAE5B,MAAM,uBAAuB,GAAG,mBAAmB,GAAG,EAAE,CAAA;IACxD,MAAM,iBAAiB,GAAG,mBAAmB,GAAG,EAAE,IAAI,CAAC,IAAI,oBAAoB,IAAI,CAAC,CAAA;IAEpF,IAAI,CAAC,uBAAuB,IAAI,CAAC,iBAAiB,EAAE,CAAC;QACpD,IAAI,oBAAoB,IAAI,CAAC,EAAE,CAAC;YAC/B,gBAAgB,GAAG,GAAG,mBAAmB,IAAI,oBAAoB,GAAG,CAAA;QACrE,CAAC;aAAM,CAAC;YACP,gBAAgB,GAAG,GAAG,mBAAmB,WAAW,CAAA;QACrD,CAAC;IACF,CAAC;SAAM,CAAC;QACP,gBAAgB,GAAG,YAAY,CAAA;IAChC,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,MAAM,UAAU,6BAA6B,CAAC,IAAY;IACzD,OAAO,+BAA+B,CAAC,IAAI,EAAE,6BAA6B,CAAC,CAAA;AAC5E,CAAC;AAED,MAAM,CAAC,MAAM,6BAA6B,GAA2B;IACpE,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IAET,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;CACR,CAAA;AAED,MAAM,mBAAmB,GAAG;IAC3B,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,IAAI,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO;IACpG,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM;IACjF,SAAS,EAAE,UAAU,EAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,WAAW,EAAE,SAAS,EAAE,UAAU,EAAE,UAAU;IACxH,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK;IAClF,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,QAAQ;CAC9C,CAAA;AAED,MAAM,qBAAqB,GAAG;IAC7B,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG;CACxC,CAAA;AAED,MAAM,eAAe,GAAG;IACvB,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG;CAC9E,CAAA;AAED,MAAM,sBAAsB,GAAG;IAC9B,SAAS,EAAE,KAAK,EAAE,OAAO,EAAE,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,OAAO,EAAE,SAAS,EAAE,SAAS;CAClJ,CAAA;AAED,MAAM,uBAAuB,GAAG;IAC/B,SAAS,EAAE,SAAS,EAAE,UAAU;CAChC,CAAA;AAED,MAAM,oBAAoB,GAAG,SAAS,CAAC,GAAG,EAAE,GAAG,CAAC,CAAA;AAEhD,MAAM,eAAe,GAAG;IACvB,UAAU;IACV,oBAAoB;IACpB,UAAU,CAAC,KAAK,CAAC,oBAAoB,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IACjD,QAAQ;CACR,CAAA;AAED,MAAM,cAAc,GAAG,WAAW,CAAC,eAAe,CAAC,CAAA;AAEnD,MAAM,wBAAwB,GAAG;IAChC,UAAU;IACV,KAAK,CAAC,GAAG,eAAe,CAAC;IACzB,oBAAoB;IACpB,UAAU,CAAC,KAAK,CAAC,oBAAoB,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IACjD,QAAQ;CACR,CAAA;AAED,MAAM,8BAA8B,GAAG,WAAW,CAAC,wBAAwB,CAAC,CAAA;AAE5E,MAAM,wBAAwB,GAAG;IAChC,UAAU;IACV,oBAAoB;IACpB,UAAU,CAAC,KAAK,CAAC,oBAAoB,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IACjD,KAAK,CAAC,GAAG,eAAe,CAAC;IACzB,QAAQ;CACR,CAAA;AAED,MAAM,uBAAuB,GAAG,WAAW,CAAC,wBAAwB,CAAC,CAAA;AAErE,MAAM,oBAAoB,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,QAAQ,CAAC,CAAA;AACtF,MAAM,mBAAmB,GAAG,WAAW,CAAC,oBAAoB,CAAC,CAAA;AAE7D,MAAM,sBAAsB,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,IAAI,EAAE,QAAQ,CAAC,CAAA;AAC9F,MAAM,qBAAqB,GAAG,WAAW,CAAC,sBAAsB,CAAC,CAAA;AAEjE,MAAM,yBAAyB,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,QAAQ,CAAC,CAAA;AAC/I,MAAM,wBAAwB,GAAG,WAAW,CAAC,yBAAyB,CAAC,CAAA;AAEvE,MAAM,mBAAmB,GAAG,CAAC,UAAU,EAAE,SAAS,CAAC,eAAe,CAAC,QAAQ,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAA;AACxF,MAAM,kBAAkB,GAAG,WAAW,CAAC,mBAAmB,CAAC,CAAA;AAE3D,MAAM,+BAA+B,GAAG;IACvC,UAAU;IAEV,eAAe,CAAC,QAAQ,CAAC;IACzB,UAAU,CAAC,KAAK,CAAC,eAAe,CAAC,QAAQ,CAAC,EAAE,GAAG,CAAC,CAAC;IAEjD,QAAQ;CACR,CAAA;AAED,MAAM,8BAA8B,GAAG,WAAW,CAAC,+BAA+B,CAAC,CAAA"}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
|
-
import { LanguageCode } from '@aws-sdk/client-transcribe-streaming';
|
|
2
|
-
import { wordCharacterRegExp } from '../nlp/Segmentation.js';
|
|
3
1
|
import * as FFMpegTranscoder from '../codecs/FFMpegTranscoder.js';
|
|
2
|
+
import { LanguageCode } from '@aws-sdk/client-transcribe-streaming';
|
|
3
|
+
import { isWordOrEmojiOrSymbolWord } from '../nlp/Segmentation.js';
|
|
4
4
|
import { Logger } from '../utilities/Logger.js';
|
|
5
5
|
export async function recgonize(rawAudio, languageCode, region, accessKeyId, secretAccessKey, callbacks) {
|
|
6
6
|
const flac16Khz16bitMonoAudio = await FFMpegTranscoder.encodeFromChannels(rawAudio, { format: 'flac', sampleRate: 16000, sampleFormat: 's16', channelCount: 1 }, callbacks);
|
|
@@ -71,7 +71,7 @@ export async function recgonize(rawAudio, languageCode, region, accessKeyId, sec
|
|
|
71
71
|
const timeline = [];
|
|
72
72
|
for (const event of events) {
|
|
73
73
|
const text = event.Content;
|
|
74
|
-
if (!
|
|
74
|
+
if (!isWordOrEmojiOrSymbolWord(text)) {
|
|
75
75
|
continue;
|
|
76
76
|
}
|
|
77
77
|
const startTime = event.StartTime;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"AmazonTranscribeSTT.js","sourceRoot":"","sources":["../../src/recognition/AmazonTranscribeSTT.ts"],"names":[],"mappings":"AAAA,OAAO,EAAQ,YAAY,EAAwC,MAAM,sCAAsC,CAAA;AAC/G,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"AmazonTranscribeSTT.js","sourceRoot":"","sources":["../../src/recognition/AmazonTranscribeSTT.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAQ,YAAY,EAAwC,MAAM,sCAAsC,CAAA;AAC/G,OAAO,EAAE,yBAAyB,EAAE,MAAM,wBAAwB,CAAA;AAClE,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAK/C,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,QAAkB,EAAE,YAAoB,EAAE,MAAc,EAAE,WAAmB,EAAE,eAAuB,EAAE,SAA+B;IACtK,MAAM,uBAAuB,GAAG,MAAM,gBAAgB,CAAC,kBAAkB,CAAC,QAAQ,EAAE,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,KAAK,EAAE,YAAY,EAAE,KAAK,EAAE,YAAY,EAAE,CAAC,EAAE,EAAE,SAAS,CAAC,CAAA;IAE3K,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAA;IAE7C,IAAI,CAAC,CAAC,YAAY,IAAI,YAAY,CAAC,EAAE,CAAC;QACrC,IAAI,oBAAoB,GAAuB,SAAS,CAAA;QAExD,IAAI,YAAY,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC/B,oBAAoB,GAAG,2BAA2B,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,YAAY,CAAC,CAAC,CAAA;QACjG,CAAC;QAED,IAAI,oBAAoB,EAAE,CAAC;YAC1B,MAAM,CAAC,GAAG,CAAC,wBAAwB,YAAY,+CAA+C,oBAAoB,GAAG,CAAC,CAAA;YAEtH,YAAY,GAAG,oBAAoB,CAAA;QACpC,CAAC;aAAM,CAAC;YACP,MAAM,IAAI,KAAK,CAAC,iBAAiB,YAAY,wCAAwC,CAAC,CAAA;QACvF,CAAC;IACF,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,sDAAsD,CAAC,CAAA;IAEpE,MAAM,sBAAsB,GAAG,MAAM,MAAM,CAAC,sCAAsC,CAAC,CAAA;IAEnF,MAAM,yBAAyB,GAAG,IAAI,sBAAsB,CAAC,yBAAyB,CAAC;QACtF,MAAM;QACN,WAAW,EAAE;YACZ,WAAW;YACX,eAAe;SACf;KACD,CAAC,CAAA;IAEF,MAAM,WAAW,GAAG,KAAK,SAAS,CAAC;QAClC,MAAM,SAAS,GAAG,CAAC,IAAI,EAAE,CAAA;QAEzB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,uBAAuB,CAAC,MAAM,EAAE,CAAC,IAAI,SAAS,EAAE,CAAC;YACpE,MAAM,KAAK,GAAG,uBAAuB,CAAC,QAAQ,CAAC,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,CAAA;YAEhE,MAAM,EAAE,UAAU,EAAE,EAAE,UAAU,EAAE,KAAK,EAAE,EAAE,CAAA;QAC5C,CAAC;IACF,CAAC,CAAA;IAED,MAAM,MAAM,GAAyC;QACpD,YAAY,EAAE,YAA4B;QAC1C,oBAAoB,EAAE,QAAQ,CAAC,UAAU;QACzC,aAAa,EAAE,MAAM;QACrB,WAAW,EAAE,WAAW,EAAE;KAC1B,CAAA;IAED,MAAM,CAAC,KAAK,CAAC,4CAA4C,CAAC,CAAA;IAE1D,MAAM,OAAO,GAAG,IAAI,sBAAsB,CAAC,+BAA+B,CAAC,MAAM,CAAC,CAAA;IAElF,MAAM,QAAQ,GAAG,MAAM,yBAAyB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAE9D,IAAI,UAAU,GAAG,EAAE,CAAA;IACnB,IAAI,MAAM,GAAW,EAAE,CAAA;IAEvB,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,QAAQ,CAAC,sBAAuB,EAAE,CAAC;QAC5D,IAAI,CAAC,KAAK,CAAC,eAAe,EAAE,CAAC;YAC5B,SAAQ;QACT,CAAC;QAED,MAAM,eAAe,GAAG,KAAK,CAAC,eAAe,CAAA;QAE7C,MAAM,OAAO,GAAG,eAAe,CAAC,UAAW,CAAC,OAAQ,CAAA;QAEpD,IAAI,OAAO,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;YACzB,SAAQ;QACT,CAAC;QAED,MAAM,WAAW,GAAG,OAAO,CAAC,CAAC,CAAC,CAAA;QAC9B,MAAM,YAAY,GAAG,WAAW,CAAC,YAAY,CAAA;QAE7C,IAAI,CAAC,YAAY,IAAI,YAAY,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;YAC/C,SAAQ;QACT,CAAC;QAED,MAAM,gBAAgB,GAAG,YAAY,CAAC,CAAC,CAAC,CAAA;QACxC,0CAA0C;QAE1C,IAAI,WAAW,CAAC,SAAS,KAAK,KAAK,EAAE,CAAC;YACrC,MAAM,GAAG,CAAC,GAAG,MAAM,EAAE,GAAG,gBAAgB,CAAC,KAAM,CAAC,CAAA;YAChD,UAAU,IAAI,GAAG,GAAG,gBAAgB,CAAC,UAAW,CAAA;QACjD,CAAC;IACF,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAA;IAE9B,UAAU,GAAG,UAAU,CAAC,OAAO,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IAElD,MAAM,QAAQ,GAAa,EAAE,CAAA;IAE7B,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC5B,MAAM,IAAI,GAAG,KAAK,CAAC,OAAQ,CAAA;QAE3B,IAAI,CAAC,yBAAyB,CAAC,IAAI,CAAC,EAAE,CAAC;YACtC,SAAQ;QACT,CAAC;QAED,MAAM,SAAS,GAAG,KAAK,CAAC,SAAU,CAAA;QAClC,MAAM,OAAO,GAAG,KAAK,CAAC,OAAQ,CAAA;QAC9B,MAAM,UAAU,GAAG,KAAK,CAAC,UAAU,CAAA;QAEnC,QAAQ,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,OAAO,GAAG,SAAS,CAAA;QAEjD,QAAQ,CAAC,IAAI,CACZ;YACC,IAAI,EAAE,MAAM;YACZ,IAAI;YACJ,SAAS;YACT,OAAO;YACP,UAAU;SACV,CAAC,CAAA;IACJ,CAAC;IAED,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,OAAO,EAAE,UAAU,EAAE,QAAQ,EAAE,CAAA;AAChC,CAAC;AAED,MAAM,CAAC,MAAM,2BAA2B,GAAa;IACpD,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;IACP,OAAO;CACP,CAAA"}
|
|
@@ -3,7 +3,7 @@ import * as FFMpegTranscoder from '../codecs/FFMpegTranscoder.js';
|
|
|
3
3
|
import { Logger } from '../utilities/Logger.js';
|
|
4
4
|
import { getRawAudioDuration } from '../audio/AudioUtilities.js';
|
|
5
5
|
import { concatUint8Arrays } from '../utilities/Utilities.js';
|
|
6
|
-
import { escapeHtml } from '
|
|
6
|
+
import { escapeHtml } from 'standard-html-escaper';
|
|
7
7
|
export async function synthesize(text, subscriptionKey, serviceRegion, languageCode = 'en-US', voice = 'Microsoft Server Speech Text to Speech Voice (en-US, AvaNeural)', ssmlEnabled = false, ssmlPitchString = '+0Hz', ssmlRateString = '+0%', callbacks) {
|
|
8
8
|
return new Promise((resolve, reject) => {
|
|
9
9
|
const logger = new Logger(callbacks.logLevel);
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"AzureCognitiveServicesTTS.js","sourceRoot":"","sources":["../../src/synthesis/AzureCognitiveServicesTTS.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,SAAS,MAAM,wCAAwC,CAAA;AAEnE,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AAEjE,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAE/C,OAAO,EAAY,mBAAmB,EAAE,MAAM,4BAA4B,CAAA;AAC1E,OAAO,EAAE,iBAAiB,EAAE,MAAM,2BAA2B,CAAA;AAC7D,OAAO,EAAE,UAAU,EAAE,MAAM,
|
|
1
|
+
{"version":3,"file":"AzureCognitiveServicesTTS.js","sourceRoot":"","sources":["../../src/synthesis/AzureCognitiveServicesTTS.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,SAAS,MAAM,wCAAwC,CAAA;AAEnE,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AAEjE,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAE/C,OAAO,EAAY,mBAAmB,EAAE,MAAM,4BAA4B,CAAA;AAC1E,OAAO,EAAE,iBAAiB,EAAE,MAAM,2BAA2B,CAAA;AAC7D,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAGlD,MAAM,CAAC,KAAK,UAAU,UAAU,CAC/B,IAAY,EACZ,eAAuB,EACvB,aAAqB,EACrB,YAAY,GAAG,OAAO,EACtB,KAAK,GAAG,iEAAiE,EACzE,WAAW,GAAG,KAAK,EACnB,eAAe,GAAG,MAAM,EACxB,cAAc,GAAG,KAAK,EACtB,SAA6B;IAE7B,OAAO,IAAI,OAAO,CAA6C,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QAClF,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAA;QAE7C,MAAM,CAAC,KAAK,CAAC,iDAAiD,CAAC,CAAA;QAE/D,MAAM,YAAY,GAAG,SAAS,CAAC,YAAY,CAAC,gBAAgB,CAAC,eAAe,EAAE,aAAa,CAAC,CAAA;QAE5F,YAAY,CAAC,uBAAuB,GAAG,YAAY,CAAA;QACnD,YAAY,CAAC,wBAAwB,GAAG,KAAK,CAAA;QAC7C,YAAY,CAAC,2BAA2B,GAAG,SAAS,CAAC,2BAA2B,CAAC,qBAAqB,CAAA;QAEtG,MAAM,iBAAiB,GAAG,SAAS,CAAC,iBAAiB,CAAC,gBAAgB,EAAE,CAAA;QAExE,MAAM,WAAW,GAAG,SAAS,CAAC,WAAW,CAAC,gBAAgB,CAAC,iBAAiB,CAAC,CAAA;QAE7E,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,iBAAiB,CAAC,YAAY,EAAE,WAAW,CAAC,CAAA;QAE5E,MAAM,MAAM,GAAqD,EAAE,CAAA;QAEnE,SAAS,CAAC,YAAY,GAAG,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE;YAC1C,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QACnB,CAAC,CAAA;QAED,MAAM,QAAQ,GAAG,KAAK,EAAE,MAAuC,EAAE,EAAE;YAClE,IAAI,MAAM,CAAC,YAAY,IAAI,IAAI,EAAE,CAAC;gBACjC,MAAM,CAAC,MAAM,CAAC,YAAY,CAAC,CAAA;gBAC3B,OAAM;YACP,CAAC;YAED,IAAI,YAAwB,CAAA;YAE5B,IAAI,KAAK,EAAE,CAAC;gBACX,MAAM,UAAU,GAAG,CAAC,IAAI,EAAE,CAAA;gBAC1B,MAAM,OAAO,GAAiB,EAAE,CAAA;gBAEhC,OAAO,IAAI,EAAE,CAAC;oBACb,MAAM,MAAM,GAAG,IAAI,UAAU,CAAC,UAAU,CAAC,CAAA;oBACzC,MAAM,UAAU,GAAG,MAAM,iBAAiB,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;oBAE9D,IAAI,UAAU,IAAI,CAAC,EAAE,CAAC;wBACrB,iBAAiB,CAAC,KAAK,EAAE,CAAA;wBACzB,MAAK;oBACN,CAAC;oBAED,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,UAAU,CAAC,CAAC,CAAA;gBAC7C,CAAC;gBAED,YAAY,GAAG,iBAAiB,CAAC,OAAO,CAAC,CAAA;YAC1C,CAAC;iBAAM,CAAC;gBACP,YAAY,GAAG,IAAI,UAAU,CAAC,MAAM,CAAC,SAAS,CAAC,CAAA;YAChD,CAAC;YAED,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,MAAM,QAAQ,GAAG,MAAM,gBAAgB,CAAC,gBAAgB,CACvD,YAAY,EACZ,KAAK,EACL,CAAC,EACD,EAAE,WAAW,EAAE,SAAS,CAAC,WAAW,EAAE,QAAQ,EAAE,SAAS,EAAE,CAC3D,CAAA;YAED,MAAM,CAAC,KAAK,CAAC,uCAAuC,CAAC,CAAA;YAErD,MAAM,QAAQ,GAAG,wBAAwB,CAAC,MAAM,EAAE,mBAAmB,CAAC,QAAQ,CAAC,CAAC,CAAA;YAEhF,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,OAAO,CAAC,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,CAAC,CAAA;QAC1C,CAAC,CAAA;QAED,MAAM,OAAO,GAAG,CAAC,KAAa,EAAE,EAAE;YACjC,MAAM,CAAC,KAAK,CAAC,CAAA;QACd,CAAC,CAAA;QAED,IAAI,CAAC,WAAW,IAAI,eAAe,IAAI,KAAK,IAAI,cAAc,IAAI,MAAM,EAAE,CAAC;YAC1E,WAAW,GAAG,IAAI,CAAA;YAClB,IAAI,GAAG,UAAU,CAAC,IAAI,CAAC,CAAA;QACxB,CAAC;QAED,IAAI,WAAW,EAAE,CAAC;YACjB,IAAI;gBACH,oFAAoF;oBACpF,gBAAgB,KAAK,IAAI;oBACzB,mBAAmB,eAAe,WAAW,cAAc,IAAI;oBAC/D,IAAI;oBACJ,YAAY;oBACZ,UAAU;oBACV,UAAU,CAAA;YAEX,SAAS,CAAC,cAAc,CAAC,IAAI,EAAE,QAAQ,EAAE,OAAO,CAAC,CAAA;QAClD,CAAC;aAAM,CAAC;YACP,SAAS,CAAC,cAAc,CAAC,IAAI,EAAE,QAAQ,EAAE,OAAO,CAAC,CAAA;QAClD,CAAC;IACF,CAAC,CAAC,CAAA;AACH,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,eAAuB,EAAE,aAAqB;IAChF,MAAM,YAAY,GAAG,SAAS,CAAC,YAAY,CAAC,gBAAgB,CAAC,eAAe,EAAE,aAAa,CAAC,CAAA;IAE5F,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,iBAAiB,CAAC,YAAY,EAAE,SAAS,CAAC,CAAA;IAE1E,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,cAAc,EAAE,CAAA;IAE/C,OAAO,MAAM,CAAC,MAAM,CAAA;AACrB,CAAC;AAED,MAAM,UAAU,wBAAwB,CAAC,MAAa,EAAE,aAAqB;IAC5E,MAAM,QAAQ,GAAa,EAAE,CAAA;IAE7B,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC5B,MAAM,YAAY,GAAG,KAAK,CAAC,YAAY,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAA;QAEjF,IAAI,YAAY,IAAI,cAAc,EAAE,CAAC;YACpC,SAAQ;QACT,CAAC;QAED,MAAM,IAAI,GAAW,KAAK,CAAC,IAAI,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAA;QAC3E,MAAM,MAAM,GAAW,KAAK,CAAC,WAAW,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAA;QACxF,MAAM,QAAQ,GAAW,KAAK,CAAC,QAAQ,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAA;QAEtF,MAAM,SAAS,GAAG,MAAM,GAAG,QAAQ,CAAA;QACnC,MAAM,OAAO,GAAG,CAAC,MAAM,GAAG,QAAQ,CAAC,GAAG,QAAQ,CAAA;QAE9C,QAAQ,CAAC,IAAI,CAAC;YACb,IAAI,EAAE,MAAM;YACZ,IAAI;YACJ,SAAS;YACT,OAAO;SACP,CAAC,CAAA;IACH,CAAC;IAED,OAAO,QAAQ,CAAA;AAChB,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"EspeakTTS.d.ts","sourceRoot":"","sources":["../../src/synthesis/EspeakTTS.ts"],"names":[],"mappings":"AAGA,OAAO,EAAE,QAAQ,EAAoB,MAAM,4BAA4B,CAAA;AAIvE,OAAO,EAAE,OAAO,EAAkC,MAAM,mBAAmB,CAAA;AAE3E,OAAO,EAAE,QAAQ,EAAiB,MAAM,0BAA0B,CAAA;AAKlE,OAAO,
|
|
1
|
+
{"version":3,"file":"EspeakTTS.d.ts","sourceRoot":"","sources":["../../src/synthesis/EspeakTTS.ts"],"names":[],"mappings":"AAGA,OAAO,EAAE,QAAQ,EAAoB,MAAM,4BAA4B,CAAA;AAIvE,OAAO,EAAE,OAAO,EAAkC,MAAM,mBAAmB,CAAA;AAE3E,OAAO,EAAE,QAAQ,EAAiB,MAAM,0BAA0B,CAAA;AAKlE,OAAO,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,eAAe,CAAA;AAWtE,wBAAsB,uBAAuB,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,aAAa,EAAE,aAAa,EAAE,QAAQ,EAAE,OAAO,EAAE,YAAK,EAAE,SAAS,EAAE,kBAAkB;;;;;;;GAyNlK;AAED,wBAAsB,mBAAmB,CAAC,SAAS,EAAE,MAAM,EAAE,EAAE,aAAa,EAAE,aAAa,EAAE,SAAS,EAAE,kBAAkB;;cAUvG,QAAQ;YACV,WAAW,EAAE;GAoR7B;AAED,wBAAsB,UAAU,CAAC,IAAI,EAAE,MAAM,EAAE,aAAa,EAAE,aAAa,EAAE,SAAS,EAAE,kBAAkB;;;GAkDzG;AAED,wBAAgB,SAAS,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,UAQpD;AAED,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,UAAO,UAexE;AAGD,wBAAgB,QAAQ,CAAC,OAAO,EAAE,MAAM,QAQvC;AAED,wBAAgB,SAAS,CAAC,MAAM,EAAE,MAAM,OAIvC;AAED,wBAAsB,OAAO,CAAC,IAAI,EAAE,MAAM,gBAIzC;AAED,wBAAsB,QAAQ,CAAC,KAAK,EAAE,MAAM,gBAI3C;AAED,wBAAsB,aAAa,CAAC,UAAU,EAAE,MAAM,gBAIrD;AAED,wBAAgB,aAAa,IAAI,KAAK,CAErC;AAED,wBAAsB,UAAU,CAAC,SAAS,EAAE,kBAAkB;gBAIhD,MAAM;UACZ,MAAM;eACD;QACV,QAAQ,EAAE,MAAM,CAAC;QACjB,IAAI,EAAE,MAAM,CAAA;KACZ,EAAE;KAIJ;AAwBD,wBAAsB,+BAA+B,CAAC,SAAS,EAAE,kBAAkB,iBAElF;AAED,MAAM,MAAM,eAAe,GAAG,UAAU,GAAG,MAAM,GAAG,SAAS,GAAG,KAAK,GAAG,MAAM,GAAG,MAAM,GAAG,gBAAgB,GAAG,iBAAiB,GAAG,YAAY,CAAA;AAE7I,MAAM,WAAW,WAAW;IAC3B,cAAc,EAAE,MAAM,CAAA;IACtB,IAAI,EAAE,eAAe,CAAA;IACrB,aAAa,EAAE,MAAM,CAAA;IACrB,WAAW,EAAE,MAAM,CAAA;IACnB,EAAE,CAAC,EAAE,MAAM,GAAG,MAAM,CAAA;CACpB;AACD,MAAM,WAAW,aAAa;IAC7B,KAAK,EAAE,MAAM,CAAA;IACb,IAAI,EAAE,OAAO,CAAA;IACb,IAAI,EAAE,MAAM,CAAA;IACZ,KAAK,EAAE,MAAM,CAAA;IACb,UAAU,EAAE,MAAM,CAAA;IAClB,QAAQ,EAAE,OAAO,CAAA;IACjB,gBAAgB,EAAE,OAAO,CAAA;CACzB;AAED,eAAO,MAAM,oBAAoB,EAAE,aAQlC,CAAA;AAED,wBAAsB,4BAA4B,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,SAAU,iBAiBlF"}
|