Link of all my books with invalid ISBNs

While auditing my catalog recently, I noticed that a handful of my books have invalid ISBNs. I think a lot of this had to do with my early days of cataloging where I didn't know that the ISBNs printed on spines sometimes had extra numbers at the end. Or I just mistyped or something.

Anyway, it'd be really helpful if there were ISBN-specific catalog search modes much like the ones where you can search just for "Amazon-provided cover" or "No cover" that you'll find under the Statistics/Memes->Book covers.

I'd like:

Books with no ISBN
Books with invalid ISBNs

Pretty self-explanatory, I think.

I realize that the first one is probably super-easy, but the second one may be difficult due to database issues (you probably don't have a column for "is this ISBN valid" already). At the very least, you could at least find books that don't have ISBNs that are 10 or 13 characters long without doing too much CPU-intensive math.

Well, sorting by ISBN will bring up the no-ISBN and the too-many-digit ISBN pretty quickly, so as a temporary workaround it's not terrible.

Yeah, that's not a terrible workaround, though it's also not that great in some cases. For example, here's the "sorted" display for some of mine:

271182781X Guide du Musée d'Orsay, édition anglaise
2854951295 Gardens of Versailles
440-08029-195 Slaughterhouse-Five
9036615011 The Complete Encyclopedia of Greek Mythology
978-0-230-70872 Orbus
9780061161704 Unseen Academicals
9780061833106 The Discworld Graphic Novels

Notice the 11 digit one sandwiched in between two 10 digit ones (440). Unfortunately, it also keeps the hyphens (which apparently I used sometimes when looking them up). In other words, it's pretty messy. But better than nothing.

Also the invalid ISBNs feature could catch the 10 and 13 digit ones that were typed wrong (The last digit of the ISBN is a check digit, so the computer can actually tell if it's a valid ISBN or not -- Of course it can't tell if you got a good ISBN on the wrong book, sigh) Since LT is so dependent on ISBN, I've always wondered why they didn't make it easier to correct invalid ISBNs, so I'd be for this addition. (wouldn't use the no ISBN part though. Way over half my collection legitimately do not have ISBNs)

the invalid ISBNs feature could catch the 10 and 13 digit ones that were typed wrong

If LT was made the understand the "logic" of ISBN numbers. I suppose that could be done though.

Since LT is so dependent on ISBN

Oh dear, I hope not. Too much ISBN info is, to put it plainly, total crap. Publishers using ISBN more than once, etc., etc. etc. Never, but NEVER rely on ISBN alone.

#4 by aulsmith> Also the invalid ISBNs feature could catch the 10 and 13 digit ones that were typed wrong (The last digit of the ISBN is a check digit, so the computer can actually tell if it's a valid ISBN or not

Yeah, that's what I was suggesting in my original post. :) I was just hedging because LT may not store the validity of the ISBN anywhere in the db, so it might drain the CPU/db to try to calculate that for every row of your catalog on demand.

Of course it can't tell if you got a good ISBN on the wrong book, sigh

Ah, but I had some other ideas on that that I left out. An extra pony would be "Books with ISBNs not matching any other editions". This would show you all the works that have ISBNs, but none of the ISBNs match yours.

wouldn't use the no ISBN part though. Way over half my collection legitimately do not have ISBNs

Similar to the above, there could be a query that shows your books with missing ISBNs when the work has an ISBN. But I suppose it wouldn't be that helpful due to book club editions usually not having ISBNs (got a lot of those). Would be nice to have a checkbox in edit book for "This book has no ISBN". That way we can tell it "yes, I left it out on purpose."

#5 by BarkingMatt> Oh dear, I hope not. Too much ISBN info is, to put it plainly, total crap. Publishers using ISBN more than once, etc., etc. etc. Never, but NEVER rely on ISBN alone.

But it is. A large part of the cover logic, for example, is highly ISBN dependent. Plus the way it rolls up editions is author+title+ISBN. If your book has no ISBN, it gets rolled up with any other book with no ISBN with the same author and title, even if they are two different editions. A lot of the quicklinks are also ISBN-dependent.

6: I'd be for those ponies too! (But I hear what you're saying about the database.)

Perhaps LT could reserve 0000000000000 or just 0 for this book has no ISBN. It's easier than coding a checkbox and adding a database field.

#8 by jjwilson61> True. But if feels a bit kludgy and isn't something that most users would think of doing without reading Talk or digging into the help.

A better compromise might be to code the checkbox, but make it store "0" behind the scenes, rather than in a separate field. Though this may all be moot, because of indexing. I don't know a lot about the guts of mysql indexing, but it might be far more efficient to use a separate field.

>6

I would like to have the "This book has no ISBN" checkbox too.

Most of my books don't have any ISBN - but that's probably just me being an old fart.

More power to you guys.

#3 brightcopy,

Looking at your list,

271182781X Guide du Musée d'Orsay, édition anglaise
2854951295 Gardens of Versailles
440-08029-195 Slaughterhouse-Five
9036615011 The Complete Encyclopedia of Greek Mythology
978-0-230-70872 Orbus
9780061161704 Unseen Academicals
9780061833106 The Discworld Graphic Novels

it would appear that some of the ones with dashes are lacking either the checksum (Slaughterhouse Five, Orbus) or also the language code (0 or 1 for Englishe for Slaughterhouse Five). Since LT cannot interpret this as a valid ISBN, it has left it as entered with the dashes. Hence, sorting by ISBN has revealed outliers.

I think that 2 is the language code for French so your first example could be missing that or a digit in the middle.

Assuming that it is not possible to gain enough traction to add this as an LT feature, there are a couple of approaches you could try to work with this. One would be to export to tab-delimited and open the file in a spreadsheet like Excel. Sort by ISBN there and many will jump out. You can add columns with formulas to count the number of characters in the field or search for hyphen characters. You can even find or create a formula to either check the checksum or convert to/from ISBN-10/ISBN-13.

Note that the checksum number is different for any given book between these two systems but otherwise the numbers follow a pattern so long as the 978 prefix is in use.

978{ISBN-10 without checksum digit}{new checksum}

Of course the corrections would have to be made in LT.

Once you have some formulas that work, you may be able to put these into a Greasemonkey script to help mark suspicious ISBNs.

I know well the situation of most books in one's collection not having them. The SBN and ISBN were introduced in 1967 and soon after became common features printed in/on books. Older books usually had ISBNs added if they were reprinted after that point. It's a little like looking at a 5-digit zip code for a US address and realizing that there was a certain point where they began to appear vs. earlier methods.

James

Thanks, James. I'm actually playing with the API so I have them all programmatically. I might crank them through some javascript code I have for ISBN validation. I don't think I have that many and I'm just waiting to pull the books and correct them.

Since the books in my library whose ISBNs are missing or wrong gets the following text at the bottom of the book details page:

"Citation Requires valid ISBN."

I've always assumed that librarything already had the ability to determine whether an ISBN was valid or not.
Getting a list of them, is another thing..

#14 by amberwitch> Oh, it does. You may also notice that it shows both the ISBN13 and ISBN10 higher on the page only if the ISBN entered is valid (and convertible). So yeah, the question is when it applies it. It may only apply it when that specific page is brought up for only one book.

One thing that's interesting that I just noticed. It seems for the purposes of the editions page, invalid ISBNs are treated as no ISBN. I looked at the edition page before and after fixing one and it shifted one out a blank ISBN edition line and into the fixed ISBN. Makes sense, but worth noting.