Webpage Full Translation
If you think that multiple webpage selections are cumbersome, you can also use the webpage full translation feature to translate the entire text of the webpage at once.
The webpage full translation can be opened in the following ways:
- Right-click anywhere on the webpage and select 划词翻译 > Webpage Full Translation.
- After opening the translation panel, click the "文 A" icon in the upper left corner.
- After configuring the shortcut keys on the settings page, use the shortcut keys to open.
After opening, a control panel will pop up in the lower right corner of the webpage, where you can modify the translation service, switch between bilingual/original/translated display, etc.
Users who apply for keys by themselves should pay attention: the webpage you are browsing may contain a large number of characters, please pay attention to the consumption of characters! Some users have reported that using the full text translation function on dozens of webpages consumed two hundred thousand characters. Suggestions:
- When using the webpage full translation function, do not use the translation service you applied for a key by yourself.
- If you must use the translation service you applied for a key by yourself, then:
- Do not use it on webpages with many dynamic functions (such as Youtube, Twitter), only use it in document-type webpages that are basically only text.
- During use, go to the official platform of the translation service to see how many characters have been consumed, to avoid inadvertently exceeding the free quota and incurring charges.
Why can the full text translation function consume a large number of characters?
Modern webpages are very complex:
- Translation services will count punctuation marks, spaces, line breaks, etc., and general webpages contain a large number of these characters.
- Some content is initially invisible and only appears under specific conditions, such as a large amount of text that appears only after clicking a button, such as privacy agreements/user agreements, and this part of the content may also be translated.
- Some content is loaded infinitely dynamically, such as loading new content when scrolling to the bottom, and loading again when scrolling to the bottom, and so on. Each time there is new content, the full text translation function will trigger the translation.
Although 划词翻译 has done optimizations as much as possible (such as removing leading and trailing spaces, translating only the text in the visible area, etc.), it is impossible to cover the usage of all webpages in the world, and these situations will cause you to consume characters in large and continuous quantities without realizing it!
Frequently Asked Questions
Some text is not translated
Please make sure you have upgraded 划词翻译 to version 8.5.0 or higher. 划词翻译 v8.5.0 uses a brand new text detection algorithm, and theoretically there should be no more missing text.
The translation result is different from when selecting text
First of all, it should be noted that the translation service will adjust the translation result according to the context. For example, there are three sentences A, B, and C. When you combine these three sentences for translation, sentences A and C are the context of B, so the translation result of B may be different from when you translate it alone.
A real case
When translating the following two sentences in DeepL:
This project does not include a GUI application.
If you need one, please see the 3rdParty documentation.
The translation result is:
这个项目不包括一个GUI应用程序。
如果你需要，请看第三方文档。
But if you translate the first sentence alone, the translation result is:
这个项目不包括GUI应用。
Compared with the previous result, the words "一个" and "程序" are missing.
In this example, since the translated text is very short, the difference seems small, but in a real webpage full translation scenario, the more words, the greater the difference in translation results.
And the webpage full translation is to concatenate the text of multiple paragraphs into one sentence and send it to the translation service for translation, which means that in the case of webpage full translation, each paragraph has context, while when selecting a single paragraph, it is only the text of this paragraph that is sent to the translation service, without context.
The translation does not retain the links in the original text
This can be divided into two cases.
Case 1: The link is surrounded by text
If the link is surrounded by text, then the link is indeed not included in the translation of this paragraph in the translated text, which is expected—because 划词翻译 translates the entire paragraph containing this link as a whole, which can get the best translation result; if the text is split to retain the link, the translation result will be inaccurate.
For example, "There is a link in this sentence, and the link is flanked by text." When translated as a whole, the translation is "There is a link in this sentence, and the link is flanked by text.", but if it is split to retain the link, the translation will be "In this sentence there is a Link, there is text on both sides of the link.".
Case 2: The link is not surrounded by text
Links that are not surrounded by text (such as the news titles in Tencent News) can also be clicked in the translated text. If you encounter a link that is not surrounded by text and cannot be clicked in the translated mode, please report the issue to me, and I will make the adaptation.
Incorrect translation results for webpages with mixed languages
If the webpage contains only two languages, you can try changing the source language to one of the languages to resolve the issue of the translation result not being in Chinese. For example, in a webpage with a mix of Chinese and English, you can try the following steps:
- Upgrade 划词翻译 to v8.7.7 or higher.
- After enabling full text translation, change the source language from "Auto-detect" to "English".
Click to see how to change the source language for full text translation
If the webpage is composed of three or more languages, then this problem is unsolvable. The main reason for this problem is that the translation service does not support translating text containing multiple languages. If you have a better solution, please let me know.
Why can't 划词翻译 solve this problem?
First of all, it should be noted that translation services do not support translating text containing multiple languages. Taking Google Translate as an example:
- Open https://translate.google.com
- Set the source language to "Detect language" and the target language to "Chinese (Simplified)"
- Enter
I love you. 我爱你。in the text box
After that, you will see that Google Translate recognizes this text as Chinese and the translation result is still I love you. 我爱你。, and it does not translate "I love you." into Chinese as we would like, as shown in the figure below:
The principle of webpage full translation is to merge all the text on the webpage into one paragraph and send it to the translation service, which leads to the problem mentioned above. To solve this problem, the text cannot be merged, and it needs to be translated sentence by sentence, but this will require hundreds or thousands of translation service calls for a webpage, and these translation services do not allow frequent calls in a short period of time, and when the number of calls is high, a 429 error will occur.
Therefore, in 划词翻译, you can only try to switch the source language, but it still cannot fundamentally solve this problem. If you have a better solution, please let me know.
Invalid in the built-in PDF reader
Due to the special structure of PDF files, the webpage full translation function is indeed invalid in the built-in PDF reader.
When translating a text (.txt) file, the translation result is incomplete or there is no translation result
When translating a text file, 划词翻译 uses "two or more consecutive line breaks" as a marker to split the entire text file into multiple paragraphs. In other words, if your text file does not have two consecutive line breaks (such as this file), the entire text will be sent to the translation service for translation as a single paragraph. Since translation services generally limit the maximum number of characters submitted at a time to 4000, and if your text exceeds this number, it will result in an incomplete translation result or no translation result.
It is recommended that you process the text first, add multiple line breaks between different paragraphs, and then perform webpage full translation.