Calibre PDF to LRF losing line breaks

[kad032000]

I'm having trouble explaining this so here's an example:

Let's say this is an example pdf file. This is the first paragraph.

This is the second paragraph.

And here's a third!

Sometimes the resulting lrf will look like

Let's say this is an example pdf file. This is the first paragraph.
This is the second paragraph.
And here's a third!

(Which is what I want.) But other times it will look like

Let's say this is an example pdf file. This is the first paragraph. This is the second paragraph. And here's a third!

Any ideas?

[kad032000]

FYI, by sometimes, I mean that there are some sections of a PDF where the first example will occur, and some where the second will occur, not that it will happen differently if I try it multiple times.

[kovidgoyal]

that's because the PDF reflow engine tries to guess which line endings are "hard" and which are not. It doesn't always succeed.

[rhadin]

Kovid,

I've also noted that when Calibre adds a book to the library, it chops off the first letter in the filename. For example, if the book filename is

The Three Musketeers.lrf

and I ask Calibre to add it to the library, it adds

he Three Musketeers.lrf

I have found the easiest way to reslove the problem is to rename the file by adding a leading x (e.g., xThe Three Musketeers.lrf) before adding it to the Calibre library.

[kovidgoyal]

this is when adding LRF files or any kind of file?

[pilotbob]

Originally Posted by kovidgoyal

this is when adding LRF files or any kind of file?

I've noticed this too. Has happened with the last few TOR .prc books that I have added to the db. Or maybe it has happened with all of them.

BOb

[kovidgoyal]

there was a bug in the PRC metadata reading code which should now be fixed in 0.4.73

[rhadin]

Originally Posted by kovidgoyal

this is when adding LRF files or any kind of file?

Like Bob, I noticed it with the Tor .prc files. But I also noticed it with Baen .lrf files and with .lrf files I have downloaded from MobileRead.

[kovidgoyal]

Can you post a link to a mobileread lrf that causes this to happen, I just tested with some random LRFs and it works for me.

[kad032000]

FYI,

For the books I'm currently using, I've found that if I convert to html, the line breaks for new paragraphs occur immediately after the last character of the paragraph whereas the line breaks between two lines of text in the same paragraph have a space before the break. Thus you can do a simple replace all in a text editor of " <br>" to "".

Which makes sense from a typing perspective. When typing, you don't press enter at the end of a normal line (the editor automatically moves you there), so there's always a space before the next word. And when you want to create a new line, you don't insert a space then press enter, you just press enter. At least, that's what I do...

[kovidgoyal]

An interesting observation. But I doubt it would be much more reliable than the current heuristics over the set of all PDF files. I do have a complete rewrite of the PDF reflow engine in my queue, so I'll keep this in mind when I get to it.

[rhadin]

Originally Posted by kovidgoyal

Can you post a link to a mobileread lrf that causes this to happen, I just tested with some random LRFs and it works for me.

Kovid, the next time it occurs I will send you a link. It occurred with version .72 but since I upgraded to version .73, it hasn't occurred on the one file I downloaded from MobileRead.