Consider splitting UPC into main code and supplemental

[minomy13]

Is your feature request related to a problem? Please describe.
I'd like to search for UPCs. The problem is, that if the variant cover wasn't added yet, I can't even find the related standard cover.

Describe the solution you'd like
It would be great if main part of the UPC and the supplemental would be split so I can find the related standard cover / series.

Alternatives
An alternative would be allowing wildcards in search parameters, so I could do upc=759606210015*.

[bpepple]

Currently, only the main cover UPC is used in the search. Variant covers are stored in a separate model and not included, so that would need to be taken into consideration also.

[btxgit]

I apologize if this is old information , but I think the idea here is to split the UPC-A 12 digit code from the 5 digit supplemental code, which is usually 3 digits for the issue # + 2 digits for the "variant id" (which is often 1 digit for the printing and 1 digit for the variant #, but not always).

This is a topic I know something about, since one of my attempts at automatic comic identification caused me to educate myself in the dark arts of barcodes. I wrote a script to attempt to find a barcode on my digital copies of my comics, or pictures of the cover from my phone camera. At first, it worked incredibly well - it could identify not just a book, but also the specific variant in many cases. But when I started tested on a wider range of books, I started to notice weirdness, like why codes were 10, 12, 13, 14 or 17 digits, and why some books had the same barcode as others. And that answer required that I learn all about barcodes.

Barcodes were invented to make checkout quicker and more accurate. The idea being you had a barcode scanner + a price database, and you'd scan a code, which would be looked up in the database, and that price would be added to the customer's bill. Seems easy...

The main standard in use by US comics is UPC-A, which is a 12 digit code (11 digits + 1 check digit) with optional supplemental ids (2 or 5 digits, possibly others) Prior to 1987, comics used 12 digit UPCs if they used a barcode at all. This 12 digit code was broken into "company" (6 digits) and "product) (5 digits). For most products, this was more than enough - you could have 99,999 products under a single company. However, for periodicals like comics, this was a major problem since publishers would be given a company ID, but had to pay to register each code. And 99,999 unique codes would last a lot less time if you needed a new code for each new issue for each title. So, periodical publishers would register say 12 codes per product, and each month, they'd rotate the code, looping back to the original code every year. So, for books before 1987 or so, UPC codes were not enough to uniquely identify an issue. To add insult to injury, some publishers would recycle their codes across titles (like if a book stopped publishing, they would give a new title those codes) - therefore, the UPC wasn't enough to uniquely identify a title. They might've be usable to identify a publisher, but not an imprint, but I can't remember.

In 1987, Marvel started using UPC with a 2 digit supplemental code, which were used to hold the issue number. Now instead of rotating every year, they rotated every 100 issues, and only registered one UPC-A code for a specific series. In my experience, I didn't run into these more than few times, so I can't really say how effective it was/is.

In 1993, the 5 digit supplemental code was created, giving a 12+5=17 digit code. The 5 digits were usually for holding the issue # (3 digits) the printing # (1 digt) and the variant cover # (1 digit). I'm sure most of you are thinking "haha only 1 digit for the variant?) and you would be right to do so, although at that point in comics history, variants were much less of a frequent thing - in part because it would need to have a separate product code to have a different price. Now they could at least support 10 (or 9) different prices for the same issue #, and having to obtain a new product code every 1,000 issues was not really a problem for anyone. At this point, publishers had what they needed to assign a unique ID to every book they published. However, this was not always done for various reasons.,

Books UPC 12+5 is the most common one I described above. UPC 12 was used before that, but you needed a different ID for every issue, which ate through the 12 digits very quickly. So, publishers would get 12 codes for each of their titles, and would rotate the code every new issue. To add further insult to injury, many codes were reused on different titles, which ruins the ability to not only uniquely identify a specific comic by its UPC code, but also to uniquely identify a book by its series.

It wasn't until the early 2000's that (I think) Diamond told the publishers that if they wanted to distribute their comics thru Diamond, each book needed a unique ID. Finally, the comic industry understood how databases worked!

However, UPC was not the only barcode standard - there's a standard made specifically for books called ISBN. They are 10 or 13 digits (9 or 12 + a check code), and make more sense for books. However, they specifically don't support periodicals, so no comic book issues are allowed. So we can ignroe ISBN, right? Nope, because OGNs and collections (TPB, etc) ARE allowed. For the old 10 ISBN codes, you could convert them to ISBN-13 by prepending "978" to the ISBN-10. However, you then had to re-calculate the check digit, meaning you couldn't just scan for 978 + the isbn-10. They (ISO) have since added 979 as a new prefix for books, since 978 codes have all be consumed by now.

Why did they start with prefix 978? It's because they wanted to map into the Euro standard called "EAN-13", which is meant to be a general purpose international ID, where the first 3 digits are usually a country code. 978 and 979 were attributed to a fictitious country called "Bookland". So EAN-13 and ISBN-13 are completely compatible as long as ISBN doesn't allocate a number that maps to a different product under EAN-13.

So, in order to make use of the barcode, I first checked the length of the code. 17 digits usually meant a UPC+5 supplemental code, and I'd first search for the whole 17 digit code (where a perfect match would mean I'd get the publisher, comic title, issue # and the variant / printing. if that didn't match, I'd try for a 15 digit match, which would give me everything but the variant / printing. If that didn't match, I'd try for a 14 digit match to handle the marvel 1987 format.

If the code was 13 digits, I'd verify that the code started in 978 or 979. If it did, I knew it was an EAN-13 compatible code, and could search the EAN databases in addition to the ISBN databases. A perfect match would give you the publisher and title, which was generally all the info there was, since ISBN's weren't for periodicals.

If the code was 10 digits, I would search for that, and if no hit, I'd search for 978 then 979 + 10 digit UPC, each time re-calculating the check digit.

So what happeneed to this script I wrote? Well ... there were two pieces to it - one that did the barcode OCR, and another that harvested UPC / ISBNs from retailers, a necessary step for my tool to work on new issues since the public databases weren't updated too quickly. I didn't especially like the idea of my scripts being used by people to scour various web retailers' websites for the codes.

But the biggest problem is based on what I mentioned before: barcodes were meant to simplify the checkout experience. Around the time I was finishing up my script, digital retailers started to offer digital comics for sale in sync with the paper copies, but without barcodes. Why cover up part of the cover art with a barcode when its original purpose is no longer relevant? I am someone who lugged his longboxes of comics from apartment to apartment, and I had dreamed of digital comics for years. So my script went from being a godsend to being useless in just a few weeks! Oh well.

Eventually, I came up with other methods to track down that metadata, but that's off-topic.

Apologies for the length of this post. Hope it explains why barcodes are not a panacea, but can be helpful under the right circumstances.