Dealing with large wrapped text

[Ausic]

Hi MR,

I am using Sigil to create an ePub. The source file is a PDF document. I converted the PDF to word so it makes copying the text to paste into my ePub file easier but the text are wrapped. The text are also wrapped when pasted into Sigil.

I then converted the PDF to text, the text wrap is still present. It will take me a long time to have to unwrap paragraph after paragraph of text. Is there a way to unwrap the text in bulk, for multiple paragraph at the same time?

[DiapDealer]

This has been a problem with pdf conversions forever. It is a terrible source format. It is always better to use whatever the source for creating the PDF was. There are people who will share their regular expressions for unwrapping bad pdf conversions, but nothing is going to be 100% effective. Manual editing and fixing will always be required.

Sigil has nothing inherent to help in this regard.

[JSWolf]

Originally Posted by Ausic

Hi MR,

I am using Sigil to create an ePub. The source file is a PDF document. I converted the PDF to word so it makes copying the text to paste into my ePub file easier but the text are wrapped. The text are also wrapped when pasted into Sigil.

I then converted the PDF to text, the text wrap is still present. It will take me a long time to have to unwrap paragraph after paragraph of text. Is there a way to unwrap the text in bulk, for multiple paragraph at the same time?


Attachment 226863

I've used AI to convert from PDF to ePub and it works pretty good. Better then most conversion programs.

[DiapDealer]

This is not a Sigil topic. This is an age-old PDF conversion issue.

[theducks]

I have 10 saved searches.
These are ALL really templates and get touched up to match the actual class="..."
Only the simple Join to lower do I ever use 'replace all' (and that is after checking what it does). For the rest, I use Find (this is what you also use to Skip without a replace)

Tip 1 is always 'Reformat: Multiple lines' before using the CSS cleans it adds the optional last ; before the }. And Prettify for the ALL HTML before using.

Tip 2 is to complete the basic cleans like join to lower before STEPPING thru the joins for honorific Be careful on the order you use things with Capitals and Periods

Mr.</p>
<p class="foo">Smith...

is a case that should be joined with a space and is NOT 2 paragraphs.

Here is a screenshot of my normal saved 'Joins' It is not even 75% of what you will see. Those will still need hand crafting.
Some of mine include support for both curly and straight quotes.

For de-hyphen, I start with Calibre's Editor:Spell check (with the filter set to -) as it offers a joined version as a choice (note: not all words with - are wrong. Never just remove blindly)

Sigil and Calibre are TOOLS. I use both as the ARE different on what and how. Neither is wrong or bad. They fit with the way I do stuff

HTH

[Sarmat89]

It's very, very easy. It's just what HTML was made for. Your paragraphs are separated with blank lines; add there a pair of tags </p><p> and you're set.

[DiapDealer]

Originally Posted by Sarmat89

It's very, very easy. It's just what HTML was made for. Your paragraphs are separated with blank lines; add there a pair of tags </p><p> and you're set.

I'm fairly certain that in this instance, it is an overabundance of </p><p> tags that is the problem. Unwrapping broken paragraphs involves removing a lot of extraneous </p><p> pairs, not adding more.

[DNSB]

Originally Posted by Sarmat89

It's very, very easy. It's just what HTML was made for. Your paragraphs are separated with blank lines; add there a pair of tags </p><p> and you're set.

The most common error I find in PDF conversions are lines that look like:

...this is a sample</p>
<p>line which continues in a new paragraph...

I use a collection of regexes to correct most of these errors but they are run manually with me going yes/no to each suggested change and not in a full automatic mode since that generally ends up making more problems than it solves.

Adding a <p></p> after each </p> would do nothing to help with this problem.

[Karellen]

@Ausic

Can you post the code for the text in your screenshot.

[BetterRed]

I've been doing this for 2-3 PDFs a week for several years.

• After opening the PDF in Word, I check the Styles, adjust any that don't make sense,
• Save a DOCX from Word,
• Open that DOCX in the [Sigil] DOCXImport plugin, which maps the Word Styles in the DOCX to an EPUB Style sheet. (see plugin & Mammoth doco)

The PDFs are from one of two sources. For each source I have a Word template, a Mammoth style map, and a Cascading style sheet.

All I do to the resultant EPUBs is to fix the metadata with Sigil's Metadata Edit tool.

BR

[JSWolf]

I've used AI to convert some PDF to ePub. The results are actually pretty good.

[Turtle91]

Originally Posted by JSWolf

I've used AI to convert some PDF to ePub. The results are actually pretty good.

Echo??

[Sarmat89]

Originally Posted by DNSB

Adding a <p></p> after each </p> would do nothing to help with this problem.

The right solution is preventing the paragraphs from taking place. Both pdftotext, pdftoxml, or LibreOffice allow for line-based production while still detecting paragraphs. Such plain text can be 'unwrapped' by any HTML viewer.