Easy command line interface to automatically translate the text in your HTML file(s) to multiple languages and store the translation data in JSON files. Auto-detect changes and straightforward integration into your workflow. Documentation includes code snippets to integrate the translation data into your website project using vanilla JavaScript. Usage of a third party translation services, e.g. Google Translate API, is recommended but optional.
- Website-Text-Translator
The program will read and parse the HTML file(s) in the source folder. Upon first run unique and sequentially numbered text-id's are added to HTML elements that contain text nodes or elements that have translatable attributes values:
- text elements e.g.
<p>,<h1>,<span>, etc. - element attribute text values:
alt,title,placeholder - meta tag content:
description,keywords
Thus for each text node or attribute text value a key is generated and stored in the data-attribute of each element, for example:
<p data-text_id="[100]">Hello World!</p>
<a
href="#"
title="To the homepage"
data-text_id="[101]"
data-text_id__title="102"
>Home</a
>Next, the program will generate a base language JSON file containing the key-value pair for each text node found in your HTML. This file does not contain any translation. It is the reference file that stores all the text-id keys and the corresponding text values found in your HTML document(s).
For example, if your base language is English, the file text_data_en.json is going to be your base language file and it will look like this:
{
"100": "Hello World!",
"101": "Home",
"102": "To the homepage"
}As a last step — and if enabled, the program will fetch translation strings from the Google Translate API and store them in a separate JSON language file for each target language.
For example, for Spanish you get text_data_es.json:
{
"100": "¡Hola Mundo!",
"101": "Inicio",
"102": "A la página de inicio"
}For example, for Catalan you get text_data_ca.json:
{
"100": "Hola món!",
"101": "Inici",
"102": "A la pàgina d'inici"
}If you don't want to use the Google Translate API, you will have to generate these translations files yourself or use a different service.
Importing the JSON language data files in your website project requires a separate discussion, see How to load the translation files in your project.
Once your HTML document(s) have been processed for the first time, you can continue your development work and incorporate the translation process in your workflow by running the program again after you have added, removed or changed text elements in your document(s).
The program will parse the HTML files again and compare the changes your made to the existing key-value pairs in the base language file. Text from elements that have been removed in your HTML document will be deleted from the JSON language file(s). Elements that have been added will get a new text-id. Elements that have been changed will keep their existing text-id but the text value will be updated and a new translation string will be fetched from the Google Translate API.
- Base language: the language used for the text in your HTML document.
- Target language(s): the target language(s) to obtain translation strings for.
- Language file(s): JSON file(s) containing the key-value pairs for each language.
Run the script npx wtt config to start a series of command line prompts to set the essential configuration options like base language, target language(s), source folder, destination folder and backup functionality.
More advanced configuration options are available, see Additional configuration options for more information.
-
Google Application Credentials: to fetch translation strings from the Google Translate API you need to setup authentication for Google Cloud Services. Without, you will see an error message for an invalid API key. See Getting started with Google Cloud Authentication.
-
The program can run without the Google Translate API. Run the command
npx wtt start --translate-noor set thetranslateoption in the config file tofalse. In this case the program will only run the first part of the program: add text-id's to the HTML files and generate the base language JSON file. No target language translation files will be generated. -
The code of the program is modular and could be modified to use different third party translations services without too much trouble. However, this is not part of the current state of the program. Feel free to experiment and contribute Github: Website-text-translator. The only two modules that would need to change are
src/translate/index.jsandsrc/translate/googleTranslate.js. -
If your HTML document does not contain too much text and you don't want to use third party services, you can always manually add your own translations by creating your own JSON target language files that follow the same key-value pair structure as the base language JSON file generated by the program. Note that after each run of the program the base language reference file will always be up to date with the state of the HTML document(s). However, without a third party translation service enabled, you will need to manually update the translation data for each target language file.
By default the program will create a numbered backup of your HTML (and related JSON files) in the backup folder. You can disable the backup feature by using the configuration wizard (npx wtt config), a command line flag npx wtt start --backup-no or by setting backup: false in the configuration file.
Important: If you do disable the backup feature, make sure you have a backup of your existing files elsewhere before running the program because they will be overwritten.
- Source folder: this is the base folder used to find HTML files. The program will look for HTML files recursively through all sub-folders. The default value is
"./", the root folder of your project. - Destination folder: this is the folder where you JSON translation files will be stored. This folder must be the same as the source folder or a sub-folder of the source folder. For example, if your source folder is
./srcthen the destination folder can be./srcor./src/translationsor similar.
Note that your HTML file(s) will be overwritten by the program because text-id's need to be added to the original HTML. Hence, the importance of the backup option discussed above.
It is recommended to run the program in 'dry mode' first. In dry-mode no files will be generated or overwritten. In the terminal you will get to see a list of all the HTML elements that are detected by the program. This will give you an idea of how many elements will be processed and how many text-id's will be added to the HTML files. If you have multiple HTML documents or lots of text, you may want to limit the terminal log output: run npx wtt start --dry-run or npx wtt start --dry-run --info-no.
To run the program, use the command: npx wtt start.
Now you should see the following files in your destination folder: a base language JSON file and a translation JSON file for each target language. All files are named with a prefix: text_data_. For example, text_data_en.json, text_data_nl.json, text_data_fr.json, etc.
The text node in your HTML file(s) will have been updated with text-id's and each text-id will have a corresponding translation string in the translation JSON file(s).
Also, a numbered backup folder with copies of your original HTML file(s) will have been created.
Once the program has done it's work and you have HTML elements with txt-id's and corresponding key-value pairs in JSON translation files you can continue your regular development work. When you're done adding, changing or removing HTML, run the program again. Changes will be detected and dealt with automatically.
However, take into account the consideration and recommendations in this section.
- Keep the backup functionality enabled or at least make sure you have your work stored in a separate Git commit or backed up somewhere else before you run the program.
- Simple text changes, addition or deletions for an HTML element with an existing text-id can be dealt with automatically by the program, for example:
<!-- your original document -->
<p data-text_id="[105]">This is an original sentence that will be changed.</p>
<p data-text_id="[106]">This will be removed by future development work.</p>
<p data-text_id="[107]">This is an original sentence that remains unchanged.</p>
<!-- your development work introduces the following changes: -->
<p data-text_id="[105]">This sentence has been changed.</p>
<p data-text_id="[107]">This is an original sentence that remains unchanged.</p>
<p>This is a brand new sentence.</p>
<!-- result after running the program again: -->
<p data-text_id="[105]">This sentence has been changed.</p>
<p data-text_id="[107]">This is an original sentence that remains unchanged.</p>
<p data-text_id="[108]">This is a brand new sentence.</p>In this case the program will have detected the changes automatically: the text change in the first sentence means that the key-value pair "[105]" in the JSON language file(s) will have been updated to the new string value. The element with the key "[106]" has been removed, which means that the corresponding key-value pair in the JSON languages file(s) will have been deleted as well. Lastly, the introduction of a new HTML text element leads to the addition of a new data attribute on the element and inside the JSON base language file you will find a new key-value pair: {"108": "This is a brand new sentence"}. In the target language files your will find corresponding translation strings fetched from the Google Translate API.
All nested elements are treated as separate text nodes by the program. This is perfectly fine when you run the program for the very first time.
The simple changes discussed above, however, are not the only changes that might occur and it's impossible to predict all possible permutations for an element with an existing text-id. For example, you may decide to add various span elements, an anchor tag and an <em> tag to a paragraph element with an existing text-id.
If you introduce such structural changes during your future development work it is highly recommended to remove the existing text-id for that element in your HTML.
Structural changes are changes that, for example, introduce additional HTML elements inside an existing element. Another example would be when you remove a span element from an existing paragraph. Every change that substantially changes the number or order of existing text nodes inside an element are considered structural changes that require removal of the existing text-id on the affected element(s) inside your HTML document.
By removing the existing text-id from the HTML the program will generate a new key for each text node in the correct order and the old key-value pairs will be purged.
Note: you don't have to do anything to the text-id's or text values in the JSON file(s). This will be handled by the program. Just make sure that your structurally changed HTML element(s) no longer has any lingering text-id's.
In the example below the paragraph has multiple text nodes and nested elements:
<!-- your original html after the first run of the program -->
<p data-text_id="[100,101]">
Hello, this is a paragraph that will undergo
<em data-text_id="[102]">structural</em> changes.
</p>The base language JSON file will looks like this:
{
"100": "Hello, this is a paragraph that will undergo ",
"101": " changes.",
"102": "structural"
}<!-- your development work introduces structural changes, so remove the text-id's: -->
<p>
Hello, this paragraph underwent <em>structural</em> changes and now contains a
<a href="#" title="To next section">section-link</a>.
</p>
<!-- final result after running the program again: -->
<p data-text_id="[110,111]">
Hello, this paragraph underwent
<em data-text_id="[112]">structural</em> changes and now contains a
<a
href="#"
title="To next section"
data-text-id="[113]"
data-text_id__title="114"
>section-link</a
>.
</p>The base language JSON file will contain the following key-value pairs:
{
"110": "Hello, this paragraph underwent ",
"111": " changes and now contains a ",
"112": "structural",
"113": "section-link",
"114": "To next section"
}- Note that leading and trailing single spaces will be respected by the program.
- Also note that text nodes that only contain punctuation characters like "." or "!" will be ignored by the program if they are the very last child text node. They are separate text nodes and don't have to be part of the translation process. When loading translation strings into your website project these punctuation text nodes will remain unaffected.
During your development work you may encounter a scenario whereby you want to duplicate an existing element that already has one or more text-id's. For example:
<!-- in header -->
<a
href="/"
title="To the homepage"
data-text_id="[100]"
data-text_id__title="101"
>Home</a
>
<!-- duplicate in footer -->
<a
href="/"
title="To the homepage"
data-text_id="[100]"
data-text_id__title="101"
>Home</a
>This perfectly acceptable. It also means that less translations strings have to be fetched from the third party translation service. If you have lots of duplicate text content you may want to share text-id's between elements.
However, be mindful because in future versions of your project you might make text changes to one element and forget about the fact that other elements share the same text-id, possibly leading to unexpected results.
Therefore, it is not recommended to share text-id's between elements. If you copy-paste an element, remove the text-id on the newly created element and let the program generate a new text-id for each additional element. You may have duplicate translation strings in your JSON language files but this way future changes in one element will not affect other elements.
On each run the program will compare the existing target language files with the array of target languages in your configuration file. If a new target language is detected, the program will fetch additional translation strings and generate an additional language JSON file.
Removing a target language is a two-step manual process. First, remove the target language identifier string from target language array in the configuration file. Then, delete the corresponding translation JSON file from your destination directory.
If you want to start all over with a clean slate, you can use the strip command.
The strip command will remove all text-id's from the text elements in your HTML file(s). To run the strip program, use the command npx wtt strip.
Note that running this command will make all your JSON language files, including the base language reference file, obsolete. The keys in those files will no longer be linked to the corresponding elements in your HTML file(s). There is no way to undo this action. Therefore you will be prompted for a backup. Also, once the your HTML file(s) are successfully cleaned up, you will be prompted to delete the now obsolete JSON language files.
Note that translations by the Google Translate API or any other service are not perfect. You should still manually review the translation strings and correct them if necessary. You should make your corrections directly in the respective JSON target language file(s) that are generated by the program. Your manual corrections in the JSON target language file(s) will be preserved, unless you make changes to the corresponding element in the HTML file.
Text nodes with only whitespace will be ignored. Also, the program will try to normalize the text content by removing unnecessary double spaces, tabs and new lines that may have been introduced by accident. Leading and trailing space are normalized to a single space. This works well for most common cases, but there may be exceptions.
If you do encounter issues with leading or trailing whitespace you have two options to try:
- try adding or removing spaces manually in the string value in the language JSON file(s).
- remove the text-id on the affected element(s), including the parent element and wrap the problematic text into an additional
<span>element.
If none of this suits your needs, open an issue on GitHub: Website Text Translator.
By running the npx wtt config command a config file named wtt.config.json is created in the root folder of your NPM project. The program will automatically detect and use this file. You can edit this file to manually configure various additional options.
Use the languages object to specify the base language and the target languages. The base language is the language of the text in your HTML file(s). The target languages are the languages you want to translate to. The program will generate a JSON file for each target language. Only two-letter language codes are supported. Example:
{
"languages": {
"base": "en",
"targets": ["nl", "fr", "de"]
}
}The program can be run with optional command line flags that change the behavior of the program. You can disable backups npx wtt start --backup-no; disable the third party translation service npx wtt start --translate-no; and run the program in dry-run mode npx wtt start --dry-run. In dry mode no files are written.
By default you will see log messages in the terminal about the progress of the program. If you want to see less log messages in the terminal, use the flag npx wtt start --info-no or set the logLevel key to an empty string "" in the config file.
"mode": {
"dryRun": false,
"backup": true,
"translate": true,
"logLevel": "verbose"
},The program will ignore HTML elements unless they are included in the selectors array in the (default) config file. For example, <div> elements are ignored by default as it is considered bad practice to have text lingering inside a container element. Wrap your text in a paragraph or span element.
Other exclusions by default:
- elements with the HTML attribute
translate="no", - elements and text nodes that are empty or contain only whitespace,
- elements and text nodes that only contain a single special character like
&,.?!etc.
Custom exclusions: You can tell the program to ignore HTML elements by class or id. Add the class or id to the respective exclude arrays in the config file.
The most common HTML elements are included by default. No custom config required. However, do review the list of selectors and add or remove element selectors according to your needs.
The default list of HTML elements that will be processed:
[
"head meta[name=description]",
"head meta[name=keywords]",
"head title",
"h1",
"h2",
"h3",
"h4",
"h5",
"h6",
"p",
"span",
"a",
"strong",
"em",
"small",
"cite",
"abbr",
"img",
"picture",
"input",
"button",
];To add additional elements or exclude HTML elements that are included by default, use a string value that is a valid selector for the JavaScript querySelectorAll() method, see MDN: Document: querySelectorAll():
{
"elements": {
"addSelector": ["div", "custom-tag"],
"exclude": {
"defaultSelectorToExclude": ["button", "head title", "abbr"],
"classToExclude": ["yourClass"],
"idToExclude": ["yourId"]
}
}
}The default text-id is for text nodes is data-txt_id. Modifiers are used for attribute text values: data-txt_id__alt, data-txt__title, data-txt__placeholder, etc. You can change these values in the config file. However, be careful: the data prefix is required for the text-id to be part of the HTML dataset attribute and the __ modifiers are used in the program to identify the type of text. If you change the structure of these identifiers, the program may no longer work as expected or throw an error that you will have to debug yourself.
The text-id's are numbered in sequence. The default offset is 100, thus the first key will start at "100". If you want to start counting at a different value, set the keyCountOffset value in the config file:
{
"keyCountOffset": 100
}Your website project may contain JavaScript modules that append elements to the DOM dynamically. These elements cannot be accessed by this program because they are not part of the HTML markup. However, you can include them in the automated translation process by adding custom keys to the configuration file and inside the base language JSON file. This is a three step process:
In your JavaScript module that creates the HTML element, add data attribute(s) with the name of the relevant text-id's. Example usage:
const anchor = document.createElement("a");
anchor.href = "#";
anchor.textContent = "Next section";
anchor.dataset.text_id = "[110]"; // [<number>]
anchor.title = "Go to the next page section";
anchor.dataset.text_id__title = "111"; // "<number>"
document.body.appendChild(anchor);Next, go to your base language JSON file. If your base language is set to 'en', this will be 'text_data_en' in the destination folder. In this example the last existing key-value pair has a key number of "109". Add your custom key-value pair(s) after that:
{
"100": "Hello world",
...
...
"109": "Last sentence.",
"110": "Next section",
"111": "Go to the next page section",
}The last step is to add these keys to the customKeys array in the configuration file:
{
"customKeys": ["110", "111"]
}On the next run of the program these custom keys will be picked up by the program and translation strings will be fetched from the third party translation API and stored in the respective target language JSON file(s).
To remove a custom key, simply remove the key string from the configuration file array and run the program again. The program will detect the key as being obsolete and all corresponding key-value pairs in your base language JSON file and all target language JSON files will be purged.
By default the point of reference is the text content of the HTML file. This means that when the key-value pair inside the JSON base language file is compared to the text content of the corresponding HTML element and a change has been detected, the JSON file is updated with the value of the text content of the corresponding HTML element.
If you change the textUpdateDirection value to jsonToHtml, the update direction is reversed. This means that you should make the text changes in the JSON language file first. When you run the program, the HTML text content will be updated with the changes you made in the corresponding key-value pair in the JSON base language file.
This feature may be useful for simple and quick text updates. However, this is experimental.Use with caution. Make sure you have a backup.
{
"textUpdateDirection": "jsonToHtml"
}Your project will need a script that loads the translation strings from the respective JSON translation files. There are many ways to do this but the code snippets below cover the basics.
Create a separate module or simply copy the following module code in your project's main script, e.g. app.js.
// import language data, adjust the path to your folder structure
import txt_data_en from "../data/txt_data_en.json";
import txt_data_nl from "../data/txt_data_nl.json";
import txt_data_es from "../data/txt_data_es.json";
import txt_data_ca from "../data/txt_data_ca.json";
const languageData = {
nl: txt_data_nl,
en: txt_data_en,
es: txt_data_es,
ca: txt_data_ca,
};
/*
You will need a separate function to detect the browser language, user preference set in cookies or localStorage or code that handles click events in a language selection menu. This is beyond the scope of this tutorial. However, at the bottom you will find a simple drop-down menu you can use to for testing purposes.
Your handler functions should return a string value that corresponds to the languages you chose to support. In this example: "en", "nl", "es", "ca".
You pass that value to the setTextContent function below.
*/
function setTextContent(currentLanguage) {
const currentLangData = languageData[currentLanguage];
const textElems = document.querySelectorAll("[data-txt_id]");
setTextElems(textElems, currentLangData);
setAttributeTxt(currentLangData, "title");
setAttributeTxt(currentLangData, "alt");
setAttributeTxt(currentLangData, "placeholder");
setAttributeTxt(currentLangData, "meta");
}
function setTextElems(textElems, langData) {
textElems.forEach((elem) => {
const idArr = JSON.parse(elem.dataset.txt_id);
if (elem.childNodes.length === 0) {
elem.textContent = langData[idArr[0]];
} else {
const textNodes = Array.from(elem.childNodes).filter(
(node) => node.nodeType === 3 && node.textContent.trim().length
);
for (let i = 0; i < idArr.length; i++) {
textNodes[i].textContent = langData[idArr[i]];
}
}
});
}
function setAttributeTxt(langData, target) {
const elems = document.querySelectorAll(`[data-txt_id__${target}]`);
let name = target;
if (target === "meta") name = "content";
elems.forEach((elem) => {
elem.setAttribute(name, langData[elem.dataset[`txt_id__${target}`]]);
});
}
// example of simple language menu
function addLanguageMenu() {
const languages = {
nl: "Dutch",
en: "English",
es: "Spanish",
ca: "Catalan",
};
// create language menu
const labelElement = document.createElement("label");
labelElement.textContent = "Select Language: ";
labelElement.htmlFor = "selectLanguage";
const selectElement = document.createElement("select");
selectElement.id = "selectLanguage";
for (const langCode in languages) {
const optionElement = document.createElement("option");
optionElement.value = langCode;
optionElement.text = languages[langCode];
selectElement.appendChild(optionElement);
}
// Add event listener to handle language selection
selectElement.addEventListener("change", function (event) {
const selectedLanguage = event.target.value;
setTextContent(selectedLanguage);
});
document.body.appendChild(labelElement);
document.body.appendChild(selectElement);
}
// on page load default language is "nl", language menu for changing between languages
setTextContent("nl");
addLanguageMenu();