Searching for Calibre-Created covers?

[Dullahir]

Hey there, is there a search function to search for Calibre-Generated book covers or covers that have borders, so to speak?

[theducks]

Originally Posted by Dullahir

Hey there, is there a search function to search for Calibre-Generated book covers or covers that have borders, so to speak?

I don't think these are possible:

Once a cover has been created, it is just another Image file.

Borders come in 2 types:
1) is applied via the (Image) tag: (the border family of controls)

2) is an embedded 'edge' area in the Image

The Trim button in the Metadata edit window attempts to work some magic here, but is not always successful (conservatively )

[theducks]

A Calibre cover will have a consistent (H & V) size : mine come out at 350x400.
this is not a 100% indicator.

[theducks]

Kovid
Is there any reason the Cover Generator could not include the word Calibre in the JPEG images metadata?

[kovidgoyal]

Originally Posted by theducks

Kovid
Is there any reason the Cover Generator could not include the word Calibre in the JPEG images metadata?

In theory, sure. In practice, not so much. Someone would have to write code to insert/ediut metadata in jpeg files, then someone would have to ensure that every code path that could possibly resize the jpeg files preserves that metadata. In other words, patches welcome.

[theducks]

Originally Posted by kovidgoyal

In theory, sure. In practice, not so much. Someone would have to write code to insert/ediut metadata in jpeg files, then someone would have to ensure that every code path that could possibly resize the jpeg files preserves that metadata. In other words, patches welcome.

Other plugins modify a Calibre Generated (Default) cover ?

I was thinking Simple.
If Calibre generated the default cover: Simply Tag it (so it could be found later by a plugin/OS tool) that Only looks for that tag word.

All "Normal covers do not get a tag", even if Calibre resizes them.

[BetterRed]

Originally Posted by theducks

Other plugins modify a Calibre Generated (Default) cover ?

I was thinking Simple.
If Calibre generated the default cover: Simply Tag it (so it could be found later by a plugin/OS tool) that Only looks for that tag word.

All "Normal covers do not get a tag", even if Calibre resizes them.

@theducks - apologies if I'm misunderstanding you but...

It might be better to 'flag' the book in the database rather than 'tagging' the image. This would ensure that Calibre's integrity is maintained. The database is the repository for metadata, if one starts distributing it in different places then... what next

The Open With plugin allows one to open the cover in an image editor. For editing a cover, I use IrfanView, where my default settings in Save discards all the metadata - EXIF, IPTC, XMP & JPG Comment.

The database 'flag' could be a column with one of four values: None, Calibre Default, Generate PI and Custom. All the features of Calibre could be 'applied' to the column, sorting, selection etc, but maybe not editing.

BR

[theducks]

Originally Posted by BetterRed

@theducks - apologies if I'm misunderstanding you but...

It might be better to 'flag' the book in the database rather than 'tagging' the image. This would ensure that Calibre's integrity is maintained. The database is the repository for metadata, if one starts distributing it in different places then... what next

The Open With plugin allows one to open the cover in an image editor. For editing a cover, I use IrfanView, where my default settings in Save discards all the metadata - EXIF, IPTC, XMP & JPG Comment.

The database 'flag' could be a column with one of four values: None, Calibre Default, Generate PI and Custom. All the features of Calibre could be 'applied' to the column, sorting, selection etc, but maybe not editing.

BR

The advantage of doing it in the image (cover.jpeg) is it can be detected WITHOUT having the book in any particular Library.

[BetterRed]

Originally Posted by theducks

The advantage of doing it in the image (cover.jpeg) is it can be detected WITHOUT having the book in any particular Library.

What software would to be used to 'detect' a default cover on windows, os/x & linux ?

My suggestion for a 'cover type' column (None, Default, Generate, Custom), envisaged it being a standard column, not a custom column. So why would the book need to be in any particular library?

BR

[theducks]

Originally Posted by BetterRed

What software would to be used to 'detect' a default cover on windows, os/x & linux ?

My suggestion for a 'cover type' column (None, Default, Generate, Custom), envisaged it being a standard column, not a custom column. So why would the book need to be in any particular library?

BR

If the image was tagged (I did this using File: Properties):
Any Photo Search tool (Even Windows Search) :
file type: Pictures
Keywords: Calibre

[BetterRed]

But why should I have to wander off into the realm of image metadata in order to determine what books in my Calibre library don't have a 'real' cover.

Maybe I don't - I just opened a calibre library folder of 3330 books in xplorer2's Browse Flat mode, clicked its Pictures only autofilter, and eyeballed the covers in thumbnail view - the default and generated covers stand out like sore thumbs. Took 2 minutes.

The Browse Flat feature is available in the Free Xplorer2 Lite version.

BR

[theducks]

Originally Posted by BetterRed

But why should I have to wander off into the realm of image metadata in order to determine what books in my Calibre library don't have a 'real' cover.

Maybe I don't - I just opened a calibre library folder of 3330 books in xplorer2's Browse Flat mode, clicked its Pictures only autofilter, and eyeballed the covers in thumbnail view - the default and generated covers stand out like sore thumbs. Took 2 minutes.

The Browse Flat feature is available in the Free Xplorer2 Lite version.

BR

There are many reasons for NO 'real covers'
My example never had a cover.

BTW
I assumed you wanted a List to post-process, which is why the configured search plan. The cover Browser also works. Hold down the control key: Right or Left arrow quickly flips about 10 Covers

[BetterRed]

Originally Posted by theducks

There are many reasons for NO 'real covers'
My example never had a cover.

BTW
I assumed you wanted a List to post-process, which is why the configured search plan. The cover Browser also works. Hold down the control key: Right or Left arrow quickly flips about 10 Covers

Quality check will find me books with no cover - so I think that ones covered

My original thought was that the list should result from Calibre maintaining a standard column called say "Cover Type". I didn't like the idea of Calibre becoming entangled in maintaining image metadata, I still don't like that idea.

But, in Windows, I now find the xplorer2 Browse Flat idea the most appealing. Requires no changes to Calibre core functionality or Generate Cover PI to introduce a new standard column or to write data to image metadata.

Viewing the covers via x2's Browse Flat feature presents the files in a temporary area it calls Scrap (its on the heap, not in the file system) one can cull items from Scrap (doesn't touch the files) so that one is left with a list of file paths to books with covers that you want to 'do something about'.

I'm hoping this list could be fed into Import List to set a Yes/No custom column to true for the books in the list. I haven't used Import List so I don't know if that's possible 'out of the box' but I'm optimistic that it will or can be made to do so.

In x2's Scrap View window I'm seeing 98 covers on my 1920/1200 monitor - 33 key presses to browse 3330 covers.


More to come...

BR

[BetterRed]

Originally Posted by Dullahir

Hey there, is there a search function to search for Calibre-Generated book covers or covers that have borders, so to speak?

OK - on Windows 7 you can do this procedurally, as described below

Just as I was about to post this I had an idea

I think I found a better way, involving an old but highly functional duplicate image finder I'll have a look at it and report back.

1. Run both phases of Check Library and resolve discrepancies
2. Close down Calibre
3. Back up your library(ies)
4. You will be reading the Calibre Library folders, so you need to be careful not to add, remove or change anything in them

You need a File Manager called xPlorer2 - its just a better Windows Explorer, the Free LITE version is all you need. It has a feature called Browse Flat, which allows you to see the files in a directory tree (such as a Calibre Library) as a single list. IMO its the best of breed of Free File Managers - very stable. There are some very sophisticated alternatives but the one's that don't crash generally cost money. Also X2 is the one I know, and for this task it worked well.

• In xplorer2 navigate to the Library folder, and select File->Browse Flat
• A Window will open showing all the files in the folder and all of its subfolders etc
• Click the Autofilter button and select 'pictures'
• Now you will just have all the cover.jpgs, Click the Views button and select Thumbnails
• Now you can see all the covers, what you do is to cull the good covers by removing them from the list, you can use the Delete key to do this. DO NOT use Shift/Delete as that will remove the cover from the list and the disk. And the end of the cull you be left with the not so good (bad) covers,
• Now switch back to Details View and select all the files (Ctrl/A)
• SHIFT/Right click the selected list of books - you should see an item Copy as Path, select it - that will copy the paths to the bad covers to the clipboard (its a Window 7 hidden feature, you can use it from Windows Explorer - why is it hidden, because programmers love hiding things)
• Open a Text editor (I suggest Notepad++ because it has a vertical editing mode (it calls its column editing) which is 'handy' to have) and paste the clipboard into a new file - the lines will look like
  
  

  "E:\Calibre Libraries\Main\First Author\Some  Book or Other (440)\cover.jpg"
  "E:\Calibre Libraries\Main\First Author\Yet Another Book (3156)\cover.jpg"
  "E:\Calibre Libraries\Main\Second Author\And Another Tome (3155)\cover.jpg"

• You need to edit the file to get this result
  
  

  First Author,Some  Book or Other
  First Author,Yet Another Book
  Second Author,And Another Tome

• Save the file somewhere (like the desktop) as a csv file - eg badcovers.csv
• Go into Calibre , you might want to Add a Custom column to your library, especially if you have a lot of 'bad covers', I suggest adding a Yes/No column called badcover (Bad Cover).
• You need the Import List plugin, install it if you don't already have it.
• I used the CSV option, so I'll assume its being used, select and and open the CSV File you put on the desktop - badcovers.csv. There's clip board/Reg Ex version if you prefer that method
• IMPORTANT At the bottom of the CSV panel, you'll see Title as 1 and Author as 2, flip them i.e. make Title as 2 and Author as 1.
• Click Next and the plug in will match the books in the library to those in the list - its what Kiwidude calls a 'Fuzzy" match, so there may be mismatches.
• At the top of the RHS list is a counter like 21 of 27 matches, you need to scroll though the list to find and resolve the mismatches. All should be resolvable because the library was the source of the CSV
• Now Click though the next panel and you will have the list of books with 'bad covers' in marked:imported_list - if its a short list you may be able to fix the bad covers now
• if its a long list then this is where the Bad Cover custom column can be used, select all the books and do a bulk edit to set the Bad Cover column to Yes (True). Now you can fix up the bad covers over time

It sounds a lot of work, and the culling of the covers is tedious - took me about 40 mins to cull about 2,800 'good covers' from a library of 3,330 books. However the rest of it is straightforward and not too much work - although I admit that I was using tools & techniques with which I am very familiar, so...

Zabkat xplorer2 Lite can be downloaded here http://zabkat.com/alldown.htm
Notepad++ can be downloaded here http://notepad-plus-plus.org/download/v6.3.2.html
Import List is here https://www.mobileread.com/forums/sho...d.php?t=187831

BR

[BetterRed]

Yep, there's always a another way, this method is faster and less tedious but because it uses numerical analysis it may not be 100% accurate - i.e. it might miss some default covers and identify real covers as defaults - but its less prone to human error than the eyeball culling approach.

This method replaces the 40 minutes of eyeball culling with < 4 minutes of numerical analysing

You don't need xplorer2, what you need is Prismatic Software's Dup Detector - which you can get from here, http://prismaticsoftware.com/dupdete...pdetector.html Its been around since Win 95 so it looks a bit clunky but its better than many similar products that cost $50 and more. Its free and it comes with a good quality CHM help file.

Its a fairly sophisticated duplicate image finder that looks at the characteristics of the actual image rather than file properties or metadata contents, and it can compare a bunch of images against a single reference - such as a sample default cover - which is just what we need.

But be careful - it has the ability to delete images

• Copy a default cover onto your desktop
  
• Start Dup Detector
• Select the Compare single image to primary data Method
  
• In Get Data click Build for Primary data file
• Navigate to the Calibre Library folder in which you're interested
• Make sure you check the Include all subfolders checkbox
• Click OK - it will now analyse all the cover.jpgs, shouldn't take long because covers are small; my 3330 books took a couple of minutes
• Click Open for the Single image selection and open the cover you coped to the desktop,
  
• In Find Dups you probably need to adjust some settings, these worked for me
• Match Method Color (euclidean), the next 3 checkboxes were off and I had no Match restrictions,
• I set Dup range to 88% to 100&
• In Options menu
  select the Sort High to Low option
  select Manual Delete
  do not select Deletes are permanent
  do not select any of the move options
  go into the Automatic & Semi-auto delete setup and turn it OFF
  
• Click Find Dups - warning this screen may loose most of the Back and Next buttons off the right edge of the - but you should not need them,
• Scroll down to the last pair of matches if its not a default cover then you may need to wind the from percentage up a notch, if it is a default cover then you may need to wind the from percentage down a notch
  
• Once your satisfied your not going to get any closer then select the menu item View->SaveAs log file, I suggest you save it to your desktop

Now you need to edit the file into a CSV as in the previous post, the Dup Detector log is a bit verbose, however Notepad++ has a simple macro recorder that you can use to good effect to remove the superfluous lines etc, but feel free to use vi and sed if you're minded to do so.

Then you do the same thing as above with the Import List plug in etc.

BR