screenshot-to-code
screenshot-to-code copied to clipboard
abi
Support uploading pdfs
Can you share your use case? What is it that you're looking to convert?
Looking to support.
A pdf of text mixed with graphics just to text Each page should be converted to an image Then to html And then unify pdf
On Mon, 20 Nov 2023 at 16:26 Abi Raja @.***> wrote:
Can you share your use case? What is it that you're looking to convert?
Looking to support.
— Reply to this email directly, view it on GitHub https://github.com/abi/screenshot-to-code/issues/52#issuecomment-1819167370, or unsubscribe https://github.com/notifications/unsubscribe-auth/AWJQXWT2MJYSPQ3ASHI5BRTYFNSDBAVCNFSM6AAAAAA7S5IGIWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJZGE3DOMZXGA . You are receiving this because you authored the thread.Message ID: @.***>
Magazine for example. So when I do it manually it works very well Just screenshot per page, then I input to the tool, get HTML, ask "without pictures, one column", then I download the HTML, convert to PDF (with printing into PDF), and then unify the PDFs. Then I convert to EPUB and I have a magazine of PDF -> EPUB ready for Kindle.
On Mon, Nov 20, 2023 at 7:51 PM Abi Raja @.***> wrote:
@NaorFirefly https://github.com/NaorFirefly thanks. What kind of PDFs are they?
— Reply to this email directly, view it on GitHub https://github.com/abi/screenshot-to-code/issues/52#issuecomment-1819542091, or unsubscribe https://github.com/notifications/unsubscribe-auth/AWJQXWQ5D7OUCMX427T353DYFOKC3AVCNFSM6AAAAAA7S5IGIWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJZGU2DEMBZGE . You are receiving this because you were mentioned.Message ID: @.***>
Before supporting PDF upload, multi-image upload should be supported given that PDF typically contains more than one image.
@NaorFirefly that workflow makes sense. I'm not going to work on this but others are free to take this up. Should be relatively easy to add.
@clean99 makes sense. For PDF, there should be a bunch of JS libraries like pdf.js that should be able to convert PDF into a set of images.
Can you at least add the multi image upload? The other parts are easy
On Tue, 21 Nov 2023 at 4:30 Abi Raja @.***> wrote:
@NaorFirefly https://github.com/NaorFirefly that workflow makes sense. I'm not going to work on this but others are free to take this up. Should be relatively easy to add.
@clean99 https://github.com/clean99 makes sense. For PDF, there should be a bunch of JS libraries like pdf.js that should be able to convert PDF into a set of images.
— Reply to this email directly, view it on GitHub https://github.com/abi/screenshot-to-code/issues/52#issuecomment-1820115977, or unsubscribe https://github.com/notifications/unsubscribe-auth/AWJQXWXHZ65VEAB3EELVPHDYFQG5DAVCNFSM6AAAAAA7S5IGIWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGEYTKOJXG4 . You are receiving this because you were mentioned.Message ID: @.***>
@abi I want to work on this issue. Just needed clarifications on a couple of things.
Issue: Currently the application does not support pdf uploads.
Possible Solution: First of all, add support to upload pdfs which in turn will convert the each page to a separate image.
We can do this on frontend or backend. Let me know which one you'd prefer. I would suggest just giving multi image support to the frontend to begin. Then we might have to generate results for each image separately because of the limited context for openAI API.
Multi-image sounds great to begin with!! thank you
On Tue, Nov 21, 2023 at 6:23 AM Piyush Mishra @.***> wrote:
@abi https://github.com/abi I want to work on this issue. Just needed clarifications on a couple of things.
Issue: Currently the application does not support pdf uploads.
Possible Solution: First of all, add support to upload pdfs which in turn will convert the each page to a separate image.
We can do this on frontend or backend. Let me know which one you'd prefer. I would suggest just giving multi image support to the frontend to begin. Then we might have to generate results for each image separately because of the limited context for openAI API.
— Reply to this email directly, view it on GitHub https://github.com/abi/screenshot-to-code/issues/52#issuecomment-1820217949, or unsubscribe https://github.com/notifications/unsubscribe-auth/AWJQXWQKO3DKO4MO3K7MOQTYFQUCXAVCNFSM6AAAAAA7S5IGIWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGIYTOOJUHE . You are receiving this because you were mentioned.Message ID: @.***>
@PiyushMishra318 I would start with the simplest thing possible: accept multiple images and send them all in one request. Let's not worry about context for now. Input is 128K and output is 4K so input is really not a concern.