Best format to extract text from speed vs accuracy

[Txomin]

Good folk.

I have been given the task of extracting the text out of thousands of ebooks. Some of these are available in several formats (epub, lit, mobi, pdf, etc).

For the purpose of extracting the text (unicode):

1. Which source format is the best to extract from?
2. Which source format would be fastest to extract from?

Preliminary experiments so far point to epub/mobi as the best and pdf as the worst in terms of accuracy.

Does anyone have any experience on this?

Thank you all in advance.

[theducks]

Originally Posted by Txomin

Good folk.

I have been given the task of extracting the text out of thousands of ebooks. Some of these are available in several formats (epub, lit, mobi, pdf, etc).

For the purpose of extracting the text (unicode):

1. Which source format is the best to extract from?
2. Which source format would be fastest to extract from?

Preliminary experiments so far point to epub/mobi as the best and pdf as the worst in terms of accuracy.

Does anyone have any experience on this?

Thank you all in advance.

Your experiment is pretty good.

I know you can open (explode) EPUB pretty easy (Just use Tweak: built in). HTML is also accessible.

If you HAVE Acrobat, the PDF might not be so bad .

[Txomin]

Originally Posted by theducks

Your experiment is pretty good.

I know you can open (explode) EPUB pretty easy (Just use Tweak: built in). HTML is also accessible.

If you HAVE Acrobat, the PDF might not be so bad .

Thank you for the quick reply and please forgive my ignorance.

What is Tweak? I've been playing with ebook-convert.

[theducks]

Originally Posted by Txomin

Thank you for the quick reply and please forgive my ignorance.

What is Tweak? I've been playing with ebook-convert.

highlight the book entry: Tap 'T'
This tool allows you to unpack the books pieces to allow (small) edits, then put them back together when done, maintaining the original structure. For Bigger edits( add/remove chapters..., Sigil is easier for the novice-intermediate).

[BetterRed]

Originally Posted by Txomin

Thank you for the quick reply and please forgive my ignorance.

What is Tweak? I've been playing with ebook-convert.

Converting to what format - TXT (i.e. a file with the extension txt) I assume

AFAIK what you see in the EPUB Viewer is what you'll get in TXT output file - but without any formatting/styling or images - the important settings are the TXT Output settings

Given that EPUB is Calibre's native format I would anticipate it might be faster.

If you don't have access to PDF editing software like Acrobat, Nitro etc to do the conversions, then you could try

1. use MobiCreator program to convert PDF's to PRC
2. use Calibre to convert the PRC to EPUB
3. use Sigil to tidy up the EPUB - needed on most books, the Quality Check plugin could also help in this regard
4. then use Calibre to convert the Sigil'd/QC'd EPUB to TXT.

But if you have a lot of PDF's you would probably save a lot of time if you had Acrobat (you'll need the full product) then you could create the TXT files direct - can't recall if Acrobat does bulk operations. Another possibility are the Nitro PDF tools, some people here have them.

I suggest you steer clear of the "Free PDF to ..." converters unless you get a specific recommendation - as in the case of MobiCreator.

BR

[JSWolf]

Calibre can convert to TXT. Just dump your eBooks into Calibre (not PDF) and batch convert to TXT. You can leave it running overnight. You don't have to care which is faster as it will just do it while you are not at the computer. I don't know the maximum you can queue at one time, but you could do it with Calibre.

[Txomin]

Thank you all for the answers and leads.