feat(translator): preprocess some common mistranslated publishing terms (#1964)

apps/readest-app/src/hooks/useTranslator.ts

@@ -1,7 +1,8 @@
 import { useState, useCallback, useEffect } from 'react';
 import { useAuth } from '@/context/AuthContext';
 import { ErrorCodes, getTranslator, getTranslators, TranslatorName } from '@/services/translators';
-import { getFromCache, storeInCache, polish, UseTranslatorOptions } from '@/services/translators';
+import { getFromCache, storeInCache, UseTranslatorOptions } from '@/services/translators';
+import { polish, preprocess } from '@/services/translators';
 import { eventDispatcher } from '@/utils/event';
 import { useTranslation } from './useTranslation';
 
@@ -10,6 +11,7 @@ export function useTranslator({
   sourceLang = 'AUTO',
   targetLang = 'EN',
   enablePolishing = true,
+  enablePreprocessing = true,
 }: UseTranslatorOptions = {}) {
   const _ = useTranslation();
   const { token } = useAuth();
@@ -42,7 +44,7 @@ export function useTranslator({
       const sourceLanguage = options?.source || sourceLang;
       const targetLanguage = options?.target || targetLang;
       const useCache = options?.useCache ?? false;
-      const textsToTranslate = input;
+      const textsToTranslate = enablePreprocessing ? preprocess(input) : input;
 
       if (textsToTranslate.length === 0 || textsToTranslate.every((t) => !t?.trim())) {
         return textsToTranslate;

apps/readest-app/src/services/translators/index.ts

@@ -1,4 +1,5 @@
 export * from './types';
 export * from './cache';
 export * from './polish';
+export * from './preprocess';
 export * from './providers';

apps/readest-app/src/services/translators/preprocess.ts

@@ -0,0 +1,31 @@
+export type Preprocessor = (text: string) => string;
+export type TextSubstitutions = Record<string, string>;
+
+const defaultSubstitutions: TextSubstitutions = {
+  Cover: 'The Cover',
+  Dedication: 'Dedication Page',
+  Acknowledgements: 'The Acknowledgements',
+};
+
+export const basicSubstitute = (text: string, substitutions: TextSubstitutions): string => {
+  const allSubstitutions = { ...defaultSubstitutions, ...substitutions };
+
+  if (Object.keys(allSubstitutions).length === 0) {
+    return text;
+  }
+
+  if (allSubstitutions[text]) {
+    return allSubstitutions[text];
+  }
+
+  return text;
+};
+
+export function createPreprocessor(substitutions: TextSubstitutions = {}): Preprocessor {
+  return (text: string) => basicSubstitute(text, substitutions);
+}
+
+export function preprocess(texts: string[], substitutions: TextSubstitutions = {}): string[] {
+  const preprocessor = createPreprocessor(substitutions);
+  return texts.map(preprocessor);
+}

apps/readest-app/src/services/translators/types.ts

@@ -23,6 +23,7 @@ export interface UseTranslatorOptions {
   sourceLang?: string;
   targetLang?: string;
   enablePolishing?: boolean;
+  enablePreprocessing?: boolean;
 }
 
 export const ErrorCodes = {

apps/readest-app/src/utils/walk.ts

@@ -30,7 +30,11 @@ export const walkTextNodes = (root: HTMLElement, rejectTags: string[] = []): HTM
           if (node.nodeType === Node.TEXT_NODE && node.textContent?.trim()) {
             return true;
           }
-          if (node.nodeType === Node.ELEMENT_NODE && (node as HTMLElement).tagName === 'SPAN') {
+          if (
+            node.nodeType === Node.ELEMENT_NODE &&
+            (node as HTMLElement).tagName === 'SPAN' &&
+            node.textContent?.trim()
+          ) {
             return true;
           }
           return false;