pdf to ePub

[qsipl]

Hi

I need to convert a PDF to EPUB. Guide on this for a tool that converts PDF TO EPUB with styles matching the PDF?

[bobinatcat]

Calibre. Easy as.

[alanHd]

I have better luck using Mobipocket creator.

[HarryT]

Please ask questions in the correct format. Moved to the "Workshop" forum.

[willus]

@qsipl -- Welcome to MR. See this thread. Lots of good tips on PDF conversions.

[Toxaris]

Originally Posted by qsipl

Hi

I need to convert a PDF to EPUB. Guide on this for a tool that converts PDF TO EPUB with styles matching the PDF?

That tool does not exits at the moment. Apparently it is a holy grail to many. PDF is your worst format to start from.

[mrmikel]

First off there are at least two different types of epubs. One consists of nothing but images. To convert it you need to optically recognize it with an appropriate program. The most effective ones are far from free.

The second type has the text already in it. You can use any one of number of programs to extract it. Mobipocket Creator can do so.

The quality of the output is dependent on how much care the creator used on it. Some are just crude OCR output with many errors, but good enough to search by, sort of, which is why many are like that.

PDFs created from text originally can have great text, but since they are not required to store it linearly like in the original, chunks can be misplaced, the images overlayed in the original but not in the deconstruction.

Hence, PDF is the worst format to start from. But it is all many of us have.

[dgatwood]

Originally Posted by Toxaris

That tool does not exits at the moment. Apparently it is a holy grail to many. PDF is your worst format to start from.

You're inclined to understatement. Essentially, the question is approximately like asking, "How can I zoom in and enhance like they do on CSI" or "how can I copy the text of my research paper from a photograph of the screen," and for precisely the same reason—you can't readily extract information that isn't there.

As mrmikel pointed out, a PDF file basically consists of... at best, a series of strings, or at worst, a series of individual glyphs, along with font information and the location where each glyph or string should be drawn on the page. You don't have paragraphs, and you may or may not even have entire lines. This is why copy and paste from a PDF is notoriously error-prone.

One of the most hilarious examples of PDF's inadequacy that I've seen involved Apple's developer documentation PDFs from a few years back. In some PDF readers (notably, Apple's Preview prior to about OS X v10.8), depending on how you selected text, you would sometimes select the words, but not the spaces between them. You can probably imagine how much fun that was.

Worse, depending on how the PDF was created, there's no guarantee that it contains the mapping information needed to convert glyph IDs back into a Unicode code points. If it doesn't, then copying text from the PDF could return nothing, random garbage, or anything in between. So in that case, the question is more like asking how to retrieve your research paper from a photo of a Microsoft Word BSOD....