Need a regex for importing books

[flinkdeldinky]

Imported a bunch of books into Calibre the normal way. Calibre got the metadata from most book files okay (they're pdf files) but in many cases it pretty much fubar'd alot of files. My idea is to clear out all the fubar'd files from Calibre and re-import them using a regex.

Unfortunately I'm not e regex guy and I found no useful examples to help me out. I only got a little bit of the way in figuring out a regex.

The file names are formatted as such.

isbn.publisher.title.date.pdf

Just to make things interesting all words are ended with a period. Publisher (always the same three words) and title (variable number of words) and date (month in 3 letters style then year in four digits).

Examples:
012345678X.This.Is.Publisher.This.is.a.Title.Apr.2 007.pdf
876543210x.This.Is.Publisher.A.Different.Title.Tha t.is.Longer.Jan.1997.pdf

This is the best regex I could get and it only gets isbn correct:

(?P<isbn>[0-9]+[A-Za-z])\.(?P<publisher>[A-Za-z]+\.[A-Za-z]+\.[A-Za-z]+)

When run on the second example:

isbn = 876543210x
publisher = This.Is.Publisher
and for some reason
title = 876543210x.This.Is.Publisher.A.Different.Title.Tha t.is.Longer.Jan.1997

I have no idea how to remove the periods from publisher. No idea how to get variable length titles. No idea how to get the dates.

Anybody got good grep out there?

[Starson17]

Originally Posted by flinkdeldinky

Anybody got good grep out there?

Try this:

(?P<isbn>\d+\w)\.(?P<publisher>\w+\.\w+\.\w+)\.(?P<title>.*)\.(?P<published>\w\w\w\.\d+)

It assumes three character month abbreviations. It doesn't remove periods, except between fields. It assumes three word publisher names.

[flinkdeldinky]

Thanks Starson17. Works as advertised. As you said id doesn't remove the periods on the title or publisher. Is that just not possible on file import? Is it possible on a bulk metadata search and replace?

My ultimate idea is t get all the info I can out of the file name and then let calibre do an internet lookup and hope it finds matches.

Again, thanks for the regex. Going to try and figure it out. I'm pretty impressed you could do the varaible title lengths. Not to surprised about the periods though as that's less of a grep and more of an edit. Hopefully the bulk metadata search and replace can do it.

[Starson17]

Originally Posted by flinkdeldinky

Thanks Starson17. Works as advertised.

You're welcome.

As you said id doesn't remove the periods on the title or publisher. Is that just not possible on file import?

No, it's not possible. File import can only get the character string in the filename, it's not designed to change those characters.

Is it possible on a bulk metadata search and replace?

Yes, do it that way.

I'm pretty impressed you could do the varaible title lengths.

Titles always vary - the trick is to find things on either side of it to match. In your case, I relied on the year digits and three characters for the month abbreviation. Those were stripped off the end, and everything in front was stripped off to match isbn, etc. What was left was title.

[flinkdeldinky]

Starson17, just wanted to let you know that everything went very well. I'm truly amazed. Your regex worked flawlessly and correcting the period problem with the Publisher and Title fields in the bulk metadata search and replace was super easy. Then I just did a bulk internet metadata search and everything went perfect.

Just want to thank you one last time. I really appreciate your effort.

[Starson17]

Originally Posted by flinkdeldinky

Starson17, I really appreciate your effort.

.

[sdspieg]

Hmmm - I still miss a few steps here. I have a similar issue - my titles contain the ISBN number, followed by a period and then publisher (followed by a period - and btw in my case the publisher can have from one to four words), Title (again any combination of words, each word followed by a period) and Date (followed by a period).
All I'd like to do is to extract the ISBN number from the tile and copy it to the identifier/isbn field - so that I can then do an automatic bulk download of metadata and covers.
Can somebody please explain how to do that? Here's what I tried: Edit Metadata|Search and replace; search mode=regex, search field=title, search for=??[sthg like your regex expression I presume]; replace with [how do I JUST get the ACTUAL isbn# here]; destination field identifier|isbn...
Thanks a bunch in advance!

-Stephan

[Serpentine]

Originally Posted by sdspieg

Hmmm - I still miss a few steps here...

Provide examples of the titles/filenames/whatever you want to capture from.

Makes life a whole lot easier.

[salines]

Hi,
I also have some problems with a regex for importing books.

My books look like this:
"author - title.epub"
"author - series xx - title.epub"

That work fine IF author NOT looks like: "Brain, Master-Mind"

The "-" within the author splits the "Master-Mind" and makes the "Mind..." to the series.

BTW: I always use " - " as separator.

Please help me. I tried it alone - but....no success....

[theducks]

Originally Posted by salines

Hi,
I also have some problems with a regex for importing books.

My books look like this:
"author - title.epub"
"author - series xx - title.epub"

That work fine IF author NOT looks like: "Brain, Master-Mind"

The "-" within the author splits the "Master-Mind" and makes the "Mind..." to the series.

BTW: I always use " - " as separator.

Please help me. I tried it alone - but....no success....

do you use just -
or do you use <space>-<space>

there is a difference (and it avoids hyphenated words )

[salines]

Originally Posted by theducks

do you use just -
or do you use <space>-<space>

there is a difference (and it avoids hyphenated words )

I'm using <space>-<space>.

[Serpentine]

It would help to have your base regex - it may have something special you use elsewhere.
If not, try this :

(?P<author>.+?) - (?:(?P<series>.+?) (?:(?P<series_index>\d+(?:\.\d+)?) - )?)?(?P<title>.+)

or with space as whitespace:
(?P<author>.+?)\s-\s(?:(?P<series>.+?)\s(?:(?P<series_index>\d+(?:\.\d+)?)\s-\s)?)?(?P<title>.+)

[salines]

or with space as whitespace:
(?P<author>.+?)\s-\s(??P<series>.+?)\s(??P<series_index>\d+(?:\. \d+)?)\s-\s)?)?(?P<title>.+)

"Fielding, Joy-ebbes - serie 01 - Tanz, Püppchen, tanz.pdf"
Works fine. Thank you!

But for
"Fielding, Joy-ebbes - Tanz, Püppchen, tanz.pdf"
it doesn't work.
author is ok: "Fielding, Joy-ebbes"
-> Series is here "Tanz,"
title is: "Püppchen, tanz!

Other question:
Should I switch the used regex if I add book for series and none series?
How can I switch the used regex for adding books fast?

[sdspieg]

Here are some examples:
01505798756X.Silly Press.The.Strange.Professional.Title.Jul.1985
043165591X.Wharton.School.Publishing.The.Delight.o f.Very.Silly.Titles.Hidden.Sep.2006

Thanks!

[theducks]

Originally Posted by sdspieg

Here are some examples:
01505798756X.Silly Press.The.Strange.Professional.Title.Jul.1985
043165591X.Wharton.School.Publishing.The.Delight.o f.Very.Silly.Titles.Hidden.Sep.2006

Thanks!

No reasonable way to determine which words belong together with (spaces) and which are new fields.