Ok folks, here's the latest. This has one notable omission - the new "Manage exemptions for book" dialog that Charles and I were posting about earlier. I know what needs to be done, just haven't had time and wanted some feedback on a huge number of other changes/additions.
This version will need
Calibre 0.7.55 (released today).
Main changes:- The new UI dialog I posted a screenshot of above, giving you the option to view the results one at a time or all at once
- Viewing all results at once turns on highlighting, sorting and search restriction
- Result groups are sorted by their "fuzzy key"
- Display on screen is sorted by title within marked group - wording just for Charles
- Show all exemptions applies a search restriction to just those records
- Marking a group or groups as exemptions has a confirmation dialog where the details pane shows you all the exemptions it will add
- Likewise the remove exemptions dialog shows you the restrictions that would be removed based on your selection
- Added a 'Clear duplicate results' menu item for exiting either duplicate groups or duplicate exemptions
- Placeholder menu for the 'Manage exemptions for this book' view
- You can show exemptions then hit next result and get returned to the appropriate duplicate groups view
- All menu items enabled/disabled based on state/selections
- Remembers any search restriction/highlighting mode you had before you started your search for duplicates and restores it when finished
- Various other bug-fixes, keyboard shortcut placeholders, some icons etc
Definitely been an interesting plugin to work on - albeit utterly consuming my week. Once I get the manage screen sorted then other than any further suggestions that come up I think it is close to done. I would like to release it as a standalone plugin for a week or two before looking at merging it into Calibre - both to let it get thrashed a bit and to give me a break.
Other outstanding possible ideas for it I have had:- Adding an additional "fuzzier" algorithm. Particularly for the author side, maybe something that checks only surname. So you could catch "S Meyer" versus "S.L. Meyer" versus "Stephanie Meyer" etc. which "Similar author" does not. I don't know what to call it though - "similar title, author surname" perhaps.
- In a similar vein there could be a fuzzier title option. One that would strip off stuff inside brackets to get rid of things like (Omnibus), (2010) etc. Maybe also anything after hyphens to catch where the filename regex brought series info into the title.
- There is a thought now in my mind that maybe the user should be allowed to choose the title and author algorithms independently. So they could have any permutation they like (e.g. exact title/fuzziest author, fuzziest title/exact author etc). Sounds fine until you wonder how the heck you fit ISBN into that
- Enhancing the descriptive text for each algorithm to include some examples on the RHS of books that would be matched versus ones that would not.
- Perhaps a star rating system for the algorithm to give a visual ranking of how strong the matching logic is. Though really a ranking system to be truly useful has to be on the match itself and we aren't going there.
Enjoy, look forward to your feedback as always...