suggestion: tag groups should use Calibre tag hierarchy

[Terisa de morgan]

Thank you very much, eagerly awaiting this feature.

[Purple Lady]

This sounds great! I currently have a column that uses the pipe character instead of a period that I used for sub-collections on my Sony. I'll change that to a period for this.

[kaufman]

Originally Posted by chaley

It works by considering the first level of the hierarchy as a "first letter". Items that are not hierarchical (do not contain a period) show up under their normal first letter. Hierarchical items show up under the first level of the hierarchy with a period appended to distinguish it from an non-hierarchical entry. See the attached screen capture.

Love the idea.

If you have "Group by First Letter" turned off, do non-hierarchical item show up next to the top letter of hierarchical ones?

[comox]

Thanks for your responsiveness and effort. Much appreciated.

[chaley]

Originally Posted by kaufman

Love the idea.

If you have "Group by First Letter" turned off, do non-hierarchical item show up next to the top letter of hierarchical ones?

First, thank you for your email messages and ideas for progressing. They led to me "discovering" that I could use the first-letter changes made to support numeric ranges to support 1-level hierarchies.

Turning off first-letter organization makes the list flat, sorted in the usual fashion. For example, without first letters you will see the following items:

Fashion
Feverish
Fiction.Mystery
Fiction.Mystery.Police
Fiction.Mystery.Thriller
Fiction.Science Fiction
Fractals
Great literature

With first letters you will see

F
Fiction.
G

Tapping on Fiction. in first letter will show you

Fiction.
  Fiction.Mystery
  Fiction.Mystery.Police
  Fiction.Mystery.Thriller
  Fiction.Science Fiction

[chaley]

Originally Posted by comox

Thanks for your responsiveness and effort. Much appreciated.

You are welcome. I really do understand that this isn't solving the problem, but it is a step along the way.

I am looking for some way to deal with the word wrapping issue. One thing I am considering is turning all the dots into spaces so Android would have somewhere to wrap other than in the middle of the word. Unfortunately there are two problems:
1) A.B would become the same as A B. I suspect that CC would fall over dead if this happened.
2) I would need to process the tags in book details so that A.B actually searched for A B if clicked.

[kaufman]

Glad I could help a little.

That sounds fine to me. I just went and updated all my tags so that everything is part of a 2-level hierarchy.

[kaufman]

Originally Posted by chaley

BTW: I can't change the names of the subitems to remove the first level of the hierarchy because that would break clicking on tags in book details.

Originally Posted by chaley

2) I would need to process the tags in book details so that A.B actually searched for A B if clicked.

Is it possible that your solution for the second part will also be applicable to the first one?

I have to tell you that as soon as I read the first quote I started a little timer in my head for how long before you posted that you had figured out a way to do it.

[chaley]

Originally Posted by chaley

I am looking for some way to deal with the word wrapping issue. One thing I am considering is turning all the dots into spaces so Android would have somewhere to wrap other than in the middle of the word. Unfortunately there are two problems:
1) A.B would become the same as A B. I suspect that CC would fall over dead if this happened.
2) I would need to process the tags in book details so that A.B actually searched for A B if clicked.

Originally Posted by kaufman

Is it possible that your solution for the second part will also be applicable to the first one?

I have to tell you that as soon as I read the first quote I started a little timer in my head for how long before you posted that you had figured out a way to do it.

I have thought about this all day and don't have any reasonable solution.

Calibre ensures that there are no duplicate items (tags or otherwise) in a field for a book. As soon as one changes the value for a tag that guarantee is out the window, which means that CC would need to be prepared to find multiple items with the same value. Today it isn't prepared to do that. Of course I could use some kind of disambiguation such as adding the primary DB key to the item, but that is a) very ugly and b) hard to read. I have up-to-now rejected the idea. The base item value (the tag) is a better solution than that. Other disambiguation schemes have similar problems, changing the value of the item in question.

And in any event, I have already exceeded the time I am willing to spend on this feature, so it is staying like it is.

[kaufman]

Originally Posted by chaley

And in any event, I have already exceeded the time I am willing to spend on this feature, so it is staying like it is.

Its no problem, and this is already a great addition to the program as you have described it.

I'm just saying that you have a track record of doing things you originally said were too hard, so I won't be completely surprised if you come back in a bit with a solution.

[comox]

I have no Android programming experience but I wonder if you could write a custom function to override the wrap function in the standard Android library?

[chaley]

Originally Posted by comox

I have no Android programming experience but I wonder if you could write a custom function to override the wrap function in the standard Android library?

No, you can't write such a function. Wrapping is done in the heart of Android in the "native library". As far as I can find there is no way to replace that library short of making a custom version of the operating system.

However, there is a hack that works, sort of. If one accepts to do everything twice then it is possible to get the measurements with the first pass then show the correct info in the second pass. Took 12 hours to make this appear to work. I say "appear to" because I am not convinced that 100% of the cases are covered. In addition, because everything is done twice there may be situations where performance is affected. We will see.

[comox]

Thanks kindly. I am very impressed with the effort you are putting into this. Looking forward to testing it.

[comox]

The latest version is a big step forward. Thanks.

I may reorganize my tag hierarchy to remove a level so that CC works better with my library. But I will still have 1 or 2 more levels than CC supports.

I was thinking there might be another feature CC could add that would bring it very close to full support of hierarchical tags.

The idea is to extend the existing search feature so we have a second "tag filter" box just below the existing search box.

Any text entered into the tag filter would be compared against tags and only those books that match would be displayed.

For example, consider books with tags:

Non-Fiction.Biography
Non-Fiction.Food.Cookbook
Non-Fiction.Food.Cookbook.Baking
Non-Fiction.Food.Cookbook.Baking.Bread
Non-Fiction.Food.Cookbook.Baking.Pastry
Non-Fiction.Food.Cookbook.Ethnic
Non-Fiction.Food.Cookbook.Ethnic.Italian
Non-Fiction.Food.Cookbook.Ethnic.Japanese
Non-Fiction.Food.Preservation
Non-Fiction.Food.Preservation.Canning
Non-Fiction.Food.Preservation.Dehydration
Non-Fiction.Food.Preservation.Fermentation
Non-Fiction.History
Non-Fiction.Science
Non-Fiction.Science.Biology
Non-Fiction.Science.Physics

Entering the tag filter ...
"Non-Fiction.Food" would display all food books.
"Non-Fiction.Food.Cookbook" would display all cookbooks.
"Non-Fiction.Food.Cookbook.Baking" would display all baking books.
"Non-Fiction.Science" would display all science books.

We could then use the tag browser to navigate the filtered subset of books.

The existing search would apply another level of filter. So for example, tag filter "Non-Fiction.Food" with search "vegetarian" would display only food books with "vegetarian" in the title.

[chaley]

Originally Posted by comox

I was thinking there might be another feature CC could add that would bring it very close to full support of hierarchical tags.

...

This is an interesting idea. From your examples, it seems that you are discussing a "contains" search, and not a "starts with" search. It isn't clear if the search is partial. For example, do "Science" and "Fiction" both match "Science Fiction"? I don't want to get into regular expressions, but it might be OK to use beginning and end anchors ("^" and "$").

If implemented, it would work on all categories, not just tags. That would make it both faster and easier to build, because it could be done in the DB. This is important, because otherwise performance would suffer greatly.

One problem: real estate. I am not sure where one would put the box. Perhaps I could make the current search area do double duty. Needs thought.