I've copied your zip-to-epub version to my PRST1. It displays just as well as my own, i.e. nothing wrong with it.
It's not surprising that your 2 epubs display slightly differently because the mobi-to-epub conversion has changed the markup in the epub's html file.
The zip-to-epub contains a single pair of <pre>...</pre> tags with the formatted TXT file in-between - very little change from the html file you imported into calibre.
The zip-to-mobi-to-epub has no <pre> tags but instead a single <p><tt>...</tt></p> paragraph with a hard linebreak (<br />) at the end of every line of text.
The biggest difference I see between your version and mine is:
- on page 1, the title page, where the order of the 2 blocks of text are reversed.
- the content page numbers are also positioned differently. On mine the page no. is on the right of the line, on yours it's on the left of the line
Neither appear to affect the readability of the epub.
I'm guessing it's due to longer lines wrapping onto 2 lines when creating the initial TXT from PDF in XPDF pdftotext. I see the current version of XPDF is 3.03 whereas my copy is 3.02. I don't know whether this accounts for the slight difference, or whether it's something to do with Mac vs Windows. You could try reading the pdftotext documentation to see if changing the runtime parameters gets you a better output file. In particular, it's worth considering the
-htmlmeta option (as well as the
-layout option) which, at a glance, seems to combine the following steps into a single step:
- create the TXT file
- wrap an html header/footer around it, including the <pre>...</pre> tags
- output the html file to disk
Code:
"C:\Program Files\XPDF\pdftotext.exe" -layout -nopgbrk -htmlmeta screenplay.pdf screenplay.html
Whichever method you use, you need to compare the XPDF output with the original pdf, to make sure you have created a good source, before complicating things by doing an epub conversion.