TheCloud/PronounsPage
1
0
Fork 0
mirror of https://gitlab.com/PronounsPage/PronounsPage.git synced 2025-09-24 05:05:20 -04:00
Code Issues Packages Projects Releases Wiki Activity

(refactor) decouble pronunciation API from pronouns to allow more flexible use

Browse Source
This commit is contained in:
Valentyne Stigloher 2023-11-25 23:32:52 +01:00
parent fa5bb609ec
commit f82344eb0c
5 changed files with 117 additions and 59 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
components/Example.vue
View File

@ -7,7 +7,7 @@
<small v-if="link">
(<nuxt-link :to="'/' + pronoun.canonicalName"><Spelling escape :text="pronoun.canonicalName"/></nuxt-link>)
</small>
<template v-if="config.pronunciation.enabled && pronunciation && pronoun.pronounceable && example.pronounce(pronoun)">
<template v-if="config.pronunciation.enabled && pronunciation && pronoun.pronounceable && example.toPronunciationString(pronoun)">
<a v-for="(pLink, name) in pronunciationLinks"
class="mr-2"
dir="ltr"
@ -20,8 +20,6 @@
</template>
<script>
import { pronouns } from '../src/data';
export default {
props: {
example: { required: true },
@ -37,32 +35,12 @@
}
},
computed: {
pronounBase() {
const name = this.pronoun.name();
for (let key in pronouns) {
if (!pronouns.hasOwnProperty(key)) { continue; }
if (key === name) {
return key;
}
for (let alias of pronouns[key].aliases) {
if (alias === name) {
return key;
}
}
}
return null;
},
pronounToString() {
return this.pronounBase && pronouns[this.pronounBase].equals(this.pronoun) ? this.pronounBase : this.pronoun.toString();
},
pronunciationLinks() {
const justOne = Object.keys(this.config.pronunciation.voices).length === 1;
const links = {};
for (let country in this.config.pronunciation.voices) {
if (!this.config.pronunciation.voices.hasOwnProperty(country)) { continue; }
links[justOne ? '' : country] = `/api/pronounce/${country}/${this.pronounToString}?example=${encodeURIComponent(this.example.toString())}`;
links[justOne ? '' : country] = `/api/pronounce/${country}/${encodeURIComponent(this.example.toPronunciationString(this.pronoun))}`;
}
return links;
}

34
server/routes/pronounce.js
View File

@ -1,9 +1,6 @@
import { Router } from 'express';
import { loadTsv } from '../loader.js';
import { buildPronoun, parsePronouns } from '../../src/buildPronoun.js';
import { Example } from '../../src/classes.js';
import sha1 from 'sha1';
import { handleErrorAsync } from '../../src/helpers.js';
import { convertPronunciationStringToSsml, handleErrorAsync } from '../../src/helpers.js';
import awsConfig from '../aws.js';
import Polly from 'aws-sdk/clients/polly.js';
@ -11,28 +8,10 @@ import S3 from 'aws-sdk/clients/s3.js';
const router = Router();
router.get('/pronounce/:voice/:pronoun*', handleErrorAsync(async (req, res) => {
const pronounString = req.params.pronoun + req.params[0];
const pronoun = buildPronoun(
parsePronouns(loadTsv('pronouns/pronouns')),
pronounString,
);
router.get('/pronounce/:voice/*', handleErrorAsync(async (req, res) => {
const text = req.params[0];
if (!pronoun || !req.query.example) {
return res.status(404).json({error: 'Not found'});
}
let [singular, plural, isHonorific] = req.query.example.split('|');
const example = new Example(
Example.parse(singular),
Example.parse(plural || singular),
!!parseInt(isHonorific || '0'),
)
const text = example.pronounce(pronoun);
// quick length check to avoid abuse. remove SSML tags but keep both tag value and attributes
if (!text || text.replace(/<[^ ]+/g, '').replace('>', '').length > 256) {
if (!text || text.length > 256) {
return res.status(404).json({error: 'Not found'});
}
@ -44,7 +23,8 @@ router.get('/pronounce/:voice/:pronoun*', handleErrorAsync(async (req, res) => {
const s3 = new S3(awsConfig);
const polly = new Polly(awsConfig);
const key = `pronunciation/${global.config.locale}-${req.params.voice}/${pronounString}/${sha1(text)}.mp3`;
const ssml = convertPronunciationStringToSsml(text);
const key = `pronunciation/${global.config.locale}-${req.params.voice}/${sha1(ssml)}.mp3`;
try {
const s3getResponse = await s3.getObject({Key: key}).promise();
@ -52,7 +32,7 @@ router.get('/pronounce/:voice/:pronoun*', handleErrorAsync(async (req, res) => {
} catch {
const pollyResponse = await polly.synthesizeSpeech({
TextType: 'ssml',
Text: text,
Text: ssml,
OutputFormat: 'mp3',
LanguageCode: voice.language,
VoiceId: voice.voice,

16
src/classes.js
View File

@ -1,4 +1,4 @@
import { buildDict, buildList, capitalise } from './helpers.js';
import { buildDict, buildList, capitalise, escapePronunciationString } from './helpers.js';
import MORPHEMES from '../data/pronouns/morphemes.js';
const config = process.env.CONFIG || global.config;
@ -55,7 +55,7 @@ export class Example {
}).join(''));
}
pronounce(pronoun) {
toPronunciationString(pronoun) {
let interchangable = false;
const buildPronunciation = m => {
@ -68,29 +68,29 @@ export class Example {
return pronunciation
? (pronunciation.startsWith('=')
? pronunciation.substring(1)
: `<phoneme alphabet="ipa" ph="${pronunciation}">${morpheme}</phoneme>`
: `/${pronunciation}/`
)
: ( config.pronunciation.ipa && morpheme
? morpheme.split('').map(
c => [' ', ',', '.', ':', ';', '–', '-'].includes(c)
? c
: `<phoneme alphabet="ipa" ph="${c}">${c}</phoneme>`
: `/${c}/`
).join('')
: morpheme
);
}
const ssml = '<speak>' + this.parts(pronoun).map(part => {
const pronunciationString = this.parts(pronoun).map(part => {
return part.variable
? buildPronunciation(part.str)
: part.str;
}).join('') + '</speak>';
: escapePronunciationString(part.str);
}).join('');
if (interchangable) {
return null;
}
return ssml;
return pronunciationString;
}
toString() {

43
src/helpers.js
View File

@ -250,6 +250,49 @@ const escapeChars = {
export const escapeHtml = (text) => text.replace(/[&<>"]/g, tag => escapeChars[tag] || tag);
export const escapePronunciationString = (text) => {
return text.replaceAll('\\', '\\\\')
.replaceAll('/', '\\/');
};
export const convertPronunciationStringToSsml = (pronunciationString) => {
const escapedString = escapeHtml(pronunciationString);
let ssml = '';
let escape = false;
let currentPhonemes = null;
for (const character of escapedString) {
if (escape) {
if (currentPhonemes === null) {
ssml += character;
} else {
currentPhonemes += character;
}
escape = false;
} else {
if (character === '\\') {
escape = true;
} else if (character == '/') {
if (currentPhonemes === null) {
currentPhonemes = '';
} else {
ssml += `<phoneme alphabet="ipa" ph="${currentPhonemes}"></phoneme>`;
currentPhonemes = null;
}
} else {
if (currentPhonemes === null) {
ssml += character;
} else {
currentPhonemes += character;
}
}
}
}
if (currentPhonemes !== null) {
ssml += `/${currentPhonemes}`;
}
return `<speak>${ssml}</speak>`;
};
export class ImmutableArray extends Array {
sorted(a, b) {
return new ImmutableArray(...[...this].sort(a, b));

57
test/helpers.test.js Normal file
View File

@ -0,0 +1,57 @@
import { describe, expect, test } from '@jest/globals';
import { convertPronunciationStringToSsml, escapePronunciationString } from '../src/helpers.js';
describe('when escaping pronunciation', () => {
test.each([
{
description: 'slashes get escaped',
text: 'w/o n/A',
pronunciationString: String.raw`w\/o n\/A`,
},
{
description: 'backslashes get escaped',
text: String.raw`\n is the symbol for a newline, \t for a tab`,
pronunciationString: String.raw`\\n is the symbol for a newline, \\t for a tab`,
}
])('$description', ({ text, pronunciationString }) => {
expect(escapePronunciationString(text)).toBe(pronunciationString);
});
});
describe('when converting pronunciation', () => {
test.each([
{
description: 'simple text is passed as-is',
pronunciationString: 'text',
ssml: '<speak>text</speak>',
},
{
description: 'slashes describe IPA phonemes',
pronunciationString: '/ðeɪ/',
ssml: '<speak><phoneme alphabet="ipa" ph="ðeɪ"></phoneme></speak>',
},
{
description: 'simple text and slashes can be combined',
pronunciationString: '/ðeɪ/ are',
ssml: '<speak><phoneme alphabet="ipa" ph="ðeɪ"></phoneme> are</speak>',
},
{
description: 'slashes can be escaped at front',
pronunciationString: String.raw`w\/o, n/A`,
ssml: '<speak>w/o, n/A</speak>',
},
{
description: 'slashes can be escaped at back',
pronunciationString: String.raw`w/o, n\/A`,
ssml: '<speak>w/o, n/A</speak>',
},
{
description: 'provided HTML is escaped',
pronunciationString: '<break time="1s"/>',
ssml: '<speak>&lt;break time=&quot;1s&quot;/&gt;</speak>',
},
])('$description', ({ pronunciationString, ssml }) => {
expect(convertPronunciationStringToSsml(pronunciationString)).toBe(ssml);
});
});