Basic Text Translation

[4mc]

I've read this long thread which has a noble request which I of course would support. viewtopic.php?p=189905#p189905

That said, I could cut a lot of wasted time with a basic function that would take highlighted text in the current pdf, within the limitations of Google translate - Auto detect language / 5,000 characters input.

JUSTIFICATION: I am spending a huge amount of time creating PDF's from Polish language music magazines at the moment. Some 50-60 jpgs per issue. Each pdf is OCR'd either by the scanning program or PDF X-Change Editor. Once OCR'd each magazine is scanned for about 30-keywords.

When a keyword is found, the paragraph and adjacent text is highlighted, copied and pasted manually into google translate and the translation checked to see if the content is relevant. I'd say 90% of the time it is not. It is a casual mention of a keyword or name rather than a meaningful discussion about it.

PROPOSED SOLUTION: Until the full ability to translate scanned text in one language, into another is available - highlighted text could be treated the same way it is now for Internet Search. The user would select the text to be translated, right click, select the translation engine, google by default, also possibly LLM's, paste the text into the input area and submit. Output would be displayed on the browser.

Here is a sample of input/output manual pasted into google translate and the same text pasted into a copilot chat preceded by "translate this".

{9EED57CC-DEB1-40EB-BCB9-69BA05F1D6C7}.png

{80D36489-D90C-4137-8A10-97E85735717A}.png

[Daniel - PDF-XChange]

Hello, 4mc

Unfortunately, as much as this seems a "basic" solution for a human to perform, implementing this in a usable way within PDF is extremely complex.

There is no communication from translation providers on which passage relates to where - so formatting is a complete lost cause. Various languages are going to have longer or shorter versions of numerous phrases, minor differences or any switching in dialec/language mid section would also pose problems and the largest issue of all.. Unlike many other formats, PDF is not a "flow based" format (it does not even have a concept of paragraphs - that is all emulated and "estimated" based on the relative proximity & positions of single line, often single character, text objects).

Making the text that is being translated, reflow between pages, extending the document if necessary, or otherwise making changes of other related content that could be important to the translation and section... These are far from easy, and in all honesty, prohibitively difficult to the point of making it nearly impossible at this point in time.
It may be something we can offer in the more distant future, but document translation of any kind, beyond the "send selection to an online translation tool" function, is not in the card today.

Kind regards,

[4mc]

Yes, and I kept reading and have done as suggested here viewtopic.php?p=184944#p184944

It works fine for Google Translate, as others have noted way before me

[Daniel - PDF-XChange]

[4mc]

It may be something we can offer in the more distant future, but document translation of any kind, beyond the "send selection to an online translation tool" function, is not in the card today.

Kind regards,

This is what I was asking for, it didn't occur to me there would be a thread that explained how to adapt Internet Search feature.

Another simple way that would be great for me is a function that export the current page to a JPG and upload it into copilot chat

{921D6CA5-12AE-4548-8FFD-607E897B87FD}.png

[Daniel - PDF-XChange]

Hello, 4mc

My apologies as well then, I read your post as indicating that you had already tried that feature and it wasn't quite what you needed!

If that is what you were looking for, I should certainly have passed along a link to that thread!
Sorry for that, and I am happy you found the solution you were looking for!

Kind regards,