Errors when getting metadata

erfjr · 03-05-2013, 10:41 PM

I've discovered many corrupted entries in my new Calibre deployment. It appears there's an issue in clearing the previous entry's data when multiple books are selected for metadata retrieval and no entry is found for a book.

I found "C# 2010 for Programmers" in the "Beginning Visual C# 2010" folder, and upon further examination, I discovered invalid ID's (ISBN and Google) in the metadata. This book was one of several hundred I bulk-added and then submitted for updated metadata.

After some experimenting, I deleted the incorrect entry and dropped the original PDF into my auto-add folder. The book was imported with a title of {ISBN number}.pdf (the file name was the actual title) and "unknown" as the author (it appears Calibre reads the PDF attributes). [It would be very cool if there was a way to trigger the metadata search automatically in auto-add mode]. The metadata created in the add process was minimal--just the title. This suggests the autoload process is not introducing the corruption, nor is drag-and-drop.

I did CTRL+D and the no metadata was retrieved. Then I cleared the ISBN from the original PDF, re-added it, and the imported file name was picked up as the title.

I've been unable to duplicate the problem, probably because of adding other books and doing some search-and-replace cleanup on titles and authors. I'm continuing to see if I can find the steps to cause consistent failure.

kiwidude · 03-05-2013, 11:01 PM

It isn't a bug in calibre, it is a problem with using bulk metadata download. Use it at your own peril. Bear in mind that metadata download is at the mercy of the search results coming back from the websites. If you do a search for "Foo", and the website gives a book back called "Bar", calibre doesn't know that this wasn't actually the book you wanted. Bulk metadata download will choose the first result coming back from the search results, which can be (as you have found) occasionally for a completely and totally different book.

If on the other hand the book has the (correct) ISBN populated, then that gets used in preference to a title/author search of the website. And you are far more likely to get the metadata for the correct book back.

The only way to 100% guarantee the right results is to not use bulk metadata download and do them one by one, certainly that is what I do.

erfjr · 03-07-2013, 03:27 AM

Kiwidude, thanks for your feedback--it makes 100% sense.

03-05-2013, 10:41 PM	#1
erfjr Junior Member Posts: 4 Karma: 10 Join Date: Mar 2013 Device: Ipad3	Errors when getting metadata I've discovered many corrupted entries in my new Calibre deployment. It appears there's an issue in clearing the previous entry's data when multiple books are selected for metadata retrieval and no entry is found for a book. I found "C# 2010 for Programmers" in the "Beginning Visual C# 2010" folder, and upon further examination, I discovered invalid ID's (ISBN and Google) in the metadata. This book was one of several hundred I bulk-added and then submitted for updated metadata. After some experimenting, I deleted the incorrect entry and dropped the original PDF into my auto-add folder. The book was imported with a title of {ISBN number}.pdf (the file name was the actual title) and "unknown" as the author (it appears Calibre reads the PDF attributes). [It would be very cool if there was a way to trigger the metadata search automatically in auto-add mode]. The metadata created in the add process was minimal--just the title. This suggests the autoload process is not introducing the corruption, nor is drag-and-drop. I did CTRL+D and the no metadata was retrieved. Then I cleared the ISBN from the original PDF, re-added it, and the imported file name was picked up as the title. I've been unable to duplicate the problem, probably because of adding other books and doing some search-and-replace cleanup on titles and authors. I'm continuing to see if I can find the steps to cause consistent failure.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Calibre metadata.db contains errors... Calibre2opds cannot open file.	anamardoll	Related Tools	25	02-12-2018 08:33 PM
Search for metadata errors	Gary Young	Library Management	2	11-19-2012 08:56 PM
Flightcrew says Errors, but Sigil says no Errors…	Barcelona	Sigil	4	02-09-2012 07:13 AM
metadata from amazon errors	kevinrs	Calibre	1	05-09-2011 11:09 AM
metadata.db errors	christinerutter	Calibre	20	10-06-2009 12:23 PM

03-05-2013, 11:01 PM	#2
kiwidude Calibre Plugins Developer Posts: 4,728 Karma: 2197770 Join Date: Oct 2010 Location: Australia Device: Kindle Oasis	It isn't a bug in calibre, it is a problem with using bulk metadata download. Use it at your own peril. Bear in mind that metadata download is at the mercy of the search results coming back from the websites. If you do a search for "Foo", and the website gives a book back called "Bar", calibre doesn't know that this wasn't actually the book you wanted. Bulk metadata download will choose the first result coming back from the search results, which can be (as you have found) occasionally for a completely and totally different book. If on the other hand the book has the (correct) ISBN populated, then that gets used in preference to a title/author search of the website. And you are far more likely to get the metadata for the correct book back. The only way to 100% guarantee the right results is to not use bulk metadata download and do them one by one, certainly that is what I do.

03-07-2013, 03:27 AM	#3
erfjr Junior Member Posts: 4 Karma: 10 Join Date: Mar 2013 Device: Ipad3	Kiwidude, thanks for your feedback--it makes 100% sense.

Advert