[GUI Plugin] Quality Check

[JayLaFunk]

Originally Posted by kiwidude

Hi Jay,

I guess the question is why you copy this all off for "future reference". When the quality check is completed, it "marks" the books that need attention, displayed using this search:
marked:epub_toc_too_small

So if you intend to edit the books to fix them, you can do so right from the search results window in calibre.

If instead you want to process them at a later date - well either just run the QualityCheck again at that time (since it is fairly quick). Or select all those books and via the bulk metadata dialog add a tag or set some custom column for you to search that way. However I would always just run the QC again.

Is there some particular reason why the above approaches do not work for you? I see what you mean about the log, the reason it is like that is just code laziness on my part (I would have to write a little extra to read the actual title/authors from the database and pump out to the log). If its a showstopper I can add it for a future release, but I'm curious as to why you feel using the log and manually typing in searches from it rather than just using the search results is better for you?

Thanks Kiwidude for the reply,

I suppose it's just a personal preference that I save it to a notepad file, I don't really want to add tags to the books and I know I can save the search in Calibre but I just open Calibre with the Notepad file side by side, scroll down to the first book in the list have a look at and try and fix the TOC, then cross it off the Notepad file, then hopefully after time the file will be blank as this will take me a bit of time, some of the books have bad TOC or none, I just thought that after doing a check with your plugin then viewing the log and option to save the info to the clipboard, would people need the file path in that info, just the book and resulting info in red would do me but as I say it's just my preference...

Many Thanks
Jay

[DoctorOhh]

Originally Posted by JayLaFunk

I suppose it's just a personal preference that I save it to a notepad file, I don't really want to add tags to the books

Everyone is entitled to their personal preferences, but adding tags to the books takes about 5 seconds.

Good Luck with your editing.

[Krogenar]

First off, this is a great plugin. Once I clean up my library (and stop adding to it) I plan on using this plugin to tweak my library even more. But I had a few quick questions, maybe suggestions.

Is there some way to make this plugin 'score' an ebook for quality, and then place that value in a custom column? My reasoning is that I sometimes add a book to the library and discover that a duplicate exists. This normally doesn't bother me; I allow the addition, and then try to determine which copy is 'better' -- more readable. Having a quality 'score' in a column would make this a lot faster.

Even better would be a way to have the plugin calculate the quality score before including the duplicate, so you could make the decision then and there. If the scores were close, you could add it and then manually compare. But if the score for the new book was terribly low compared to the existing copy, you could skip the addition entirely.

Or, maybe there's a much simpler way to do it that I'm missing. Any advice would be welcome. Again, thanks for the terrific PI.

[kiwidude]

@Krogenar - welcome to MobileRead.

I vaguely recall something similar being asked on these forums six months or so ago. The problem is that it is virtually impossible to define the "rules" for a "score" that has any meaning. Suggestions for any rules were conspicuously absent from your post

Personally I think that deciding which book to keep is an incredibly subjective process that only manual eyeballing can resolve. There are perhaps some "simple" rules that could be automated, but the amount of effort to do so in proportion to the amount of other factors that cannot makes it not worthwhile imho. For instance, some of the factors I use include:
- Have quotes been lost from a conversion due to an encoding issue (but defining what constitutes a quote is itself non-trivial)
- Emphasis present in text like italics (which not all books use)
- Paragraph structure (obvious PDF conversions, spacing missing near italics etc)
- Presence of images like maps, chapter headings etc

Then you have the issue of different formats etc - you might have a good quality LIT from a different source than the EPUB version. It all just gets too hard too darn quickly.

[Krogenar]

Originally Posted by kiwidude

@Krogenar - welcome to MobileRead.

I vaguely recall something similar being asked on these forums six months or so ago. The problem is that it is virtually impossible to define the "rules" for a "score" that has any meaning. Suggestions for any rules were conspicuously absent from your post

...

Then you have the issue of different formats etc - you might have a good quality LIT from a different source than the EPUB version. It all just gets too hard too darn quickly.

Ok, so there too many subjective metrics magnifying one another to be really feasible. You could make some system allowing the user to determine what constitutes the 'perfect' version of each format (no mismatched tags, has a cover, etc.) but that sounds like an awful lot of work.

[theducks]

Originally Posted by Krogenar

Ok, so there too many subjective metrics magnifying one another to be really feasible. You could make some system allowing the user to determine what constitutes the 'perfect' version of each format (no mismatched tags, has a cover, etc.) but that sounds like an awful lot of work.

Flightcrew (part of Sigil) will validate the document.
A valid document can still be butt ugly, contain spelling errors, mis-matched quotes...
The Quality check (EPUB section) already checks for a whole list of things. Some are 'quality' issues like cover size, missing jacket. Others are more Validation and kruft cleanup.

My books look different from some others. That makes the different, not wrong. Kiwidude outlined some common errors that would be wrong in any book.

[Thunder007]

The latest version of Calibre changed the way titles are sorted in other languages.

Change the algorithm that generates title sort strings to strip leading articles from both english and the current language set for the calibre user interface. In addition, in the edit metadata dialog, calibre will use the book's language when calculating the sort string. This behavior can be adjusted via Preferences->Tweaks.

Source: http://calibre-ebook.com/whats-new

Would it be possible to let your plug-in check the title sort in languages other than English?

[kiwidude]

Changes in this release:

• When performing a title sort check, take the first ebook language into account if any
• When checking for TOC < 3 items, display the title and authors rather than the path

[Thunder007]

Originally Posted by kiwidude

Changes in this release:

• When performing a title sort check, take the first ebook language into account if any
• When checking for TOC < 3 items, display the title and authors rather than the path

Thank you.

[JayLaFunk]

Originally Posted by kiwidude

Changes in this release:

• When checking for TOC < 3 items, display the title and authors rather than the path

[JoonieB]

Can you confirm that this option only works upto a series index of 999?

Unexpectedly my library had two books numbered 1400 and 1513 in a series ("Silhouette Intimate Moments series" on Goodreads).

Not sure it would be worth the effort of amending your code. There can't be many series with over 1,000 books! The highest series index in my library is only 23.

[kiwidude]

@JoonieB - indeed, any series which contains a book which has a series number > 999 will exclude the whole series from series checks. In theory I could make that threshold configurable for a future release.

However as you say it is a highly rare exceptional case, and the downside of producing thousands of lines of logging and performance impact of doing so means it isn't something that most would want. In my own library when those sort of nonsensical series numbers come up I just remove the series information completely. It is clearly just some made up "series" denoting an order of publication, with no meaningful relationship between the books. When there is a relationship between them (a mini series), Goodreads tends to list that instead. And as the metadata download plugin can only retrive information for one series even if there were multiple on the web page, you would have to manually assign and overwrite series names and values if you chose to prefer the high numbered "publication" series over the more meaningful content based ones.

So just delete the "Silhouette..." series name info and don't worry about it would be my advice

[kiwidude]

Changes in this release:

• Move all the metadata based checks under a submenu, and reorder menus
• Add a "Repeat last check" menu item to allow simply running the last check again
• Add "Exclude from check..." menu item, to allow excluding books from repeatedly showing you want exempt
• Add "View exclusions..." menu item, to allow viewing all excluded books per check and deleting if desired.
• Fix missing images from menu for CSS based items
• Remove configuration of the Author commas/no commas and titles with series menu items

The big ticket item in this release is the ability to add books to an exclusion list for a particular check, so they don't keep coming up again and again if you have a genuine reason. A prime example of usage for this is the "Check titlecase" check where there are always books that you would prefer to keep the published casing variation rather than what calibre's rules come up with. Other examples could be the author sorts and titles with series checks - but nearly every check in this plugin allows adding exemptions.

As you would hopefully expect, once you add such books to an exemption for that particular check, they will not appear the next time you run the check. I have also added a "Repeat last check" menu item for conveninece with re-running without digging back into the submenus.

I also took the opportunity to restructure the menus a bit for consistency. If you had keyboard shortcuts assigned to some of those moved menus they will need assigning again.

[dhume01]

Came over from this thread here.

It would be great in a future version of the plugin to have the ability to specify whether you want to search for books that either have:

• Titles in Upper case
• Titles in lower case
• Authors in Upper case
• Authors in lower case

Right now the search for title case feature in the plugin doesn't make it clear what's wrong with the particular book, and if you're doing a mass update, it's a bit harder to make sure the cleanup is being done correctly.

[kiwidude]

@dhume01 - I'll consider it as right now the plugin doesn't have any author casing checks at all. Though I'm not convinced anyone needs that level of granularity.

Why? For the same reasons as your comment above about "the existing title case feature doesn't make its clear what is wrong". That's because no-one cares about the why, they just care that it is wrong. That they can fix it by selecting those rows, bringing up the bulk metadata editor and ticking the "Change title to title case" option, or individually right clicking on a title while editing it. I don't believe anyone (outside of possibly yourself) really wants to have multiple menu items to subclass them in that way?