metadata download - overwrite vs add to...

[cybmole]

a metadata search / download will overwrite most things - author, title, cover... but it adds to summary text rather than overwrites.

so it's easy to get long summaries with duplicate text. would it not be more logical to overwrite this data

e.g. I download metadata, decide I don't really care for the cover, or the summary, so I delete ISBN, and download again, this time choosing a different match from the screen of matches. ... summary gets added to....

I'm not a fan of tags & social data so dunno if they overwrite or accumulate.

it also seems possible to add metadata to 1st page of book twice, as the book progresses through various conversions., I have seen it several times, though I cannot as yet define the process which causes it to happen.

[DoctorOhh]

Originally Posted by cybmole

it also seems possible to add metadata to 1st page of book twice, as the book progresses through various conversions., I have seen it several times, though I cannot as yet define the process which causes it to happen.

This happens when doing a mobi to mobi conversion. If you already have a Book Jacket and leave the option to insert metadata then a second book Jacket will be inserted. Apparently the book jacket is inserted into the existing html in a mobi book.

This doesn't happen in ePubs because the book jacket is a separate html inside the epub and replaced each time a conversion is run.

[kiwidude]

Originally Posted by cybmole

a metadata search / download will overwrite most things - author, title, cover... but it adds to summary text rather than overwrites.

so it's easy to get long summaries with duplicate text. would it not be more logical to overwrite this data

e.g. I download metadata, decide I don't really care for the cover, or the summary, so I delete ISBN, and download again, this time choosing a different match from the screen of matches. ... summary gets added to....

I've had this debate with Starson17 in the past on a ticket in the bug tracker. I think he may have made a few minor tweaks to it but basically we had problems reliably replicating the "exact doubling up". Though like you I had records which showed the same result of that. In some cases I found the duplicates were not quite exact - it is possible that I had added a missing space to end of a sentence, but then downloading again says "you are not exact" and doubles up...

I'm in your camp - I don't want summaries "appended" to under any circumstance. However the existing behaviour is to append if the comments differ in any way. The justification given that people may want additional reviews. I think when Kovid's new metadata download stuff comes in as a user the intent is to have more control over what gets overwritten etc.

In the meantime I just patched my local source code so it never overwrites if I have an existing value

I'm not a fan of tags & social data so dunno if they overwrite or accumulate.

Tags accumulate. Everything else overwrites if it has a value from memory like series, rating etc. Some of this you can turn off if you don't want it at all in the configuration for the metadata download plugins. Personally I again "tweaked" the source code, such that it only puts in values if I don't have them and never appends.

[cybmole]

Originally Posted by dwanthny

This happens when doing a mobi to mobi conversion. If you already have a Book Jacket and leave the option to insert metadata then a second book Jacket will be inserted. Apparently the book jacket is inserted into the existing html in a mobi book.

This doesn't happen in ePubs because the book jacket is a separate html inside the epub and replaced each time a conversion is run.

yes, that could explain what I have been seeing , assuming it also happens with epub to mobi conversions ( I do not usually have any reason for doing mobi to mobi but I do a lot of epub to mobi

- is that a bug or a feature ?

it would be nice, someday to be able to set conversion prefs by type, so my default for epub output could be add metadata, & my default for mobi output could be do not add...

[DoctorOhh]

Originally Posted by cybmole

yes, that could explain what I have been seeing , assuming it also happens with epub to mobi conversions ( I do not usually have any reason for doing mobi to mobi but I do a lot of epub to mobi

If the ePub had the metadata page inserted prior to conversion to mobi and you added another I guess it might double up.

Originally Posted by cybmole

is that a bug or a feature ?

Yes, it is part of an expanding set of features that might bug some folks who don't pay attention.

Before calibre 0.6.50 ePub to ePub conversions would end up with duplicate book jackets too. Any book that the metadata page was inserted after that version can now be converted again without the book jacket being duplicated. In epubs it is easy enough to identify and replace but in a Mobi book jacket isn't its own separate html, thus you have to be more careful in the settings when converting books where the book jacket is already inserted.

Originally Posted by cybmole

it would be nice, someday to be able to set conversion prefs by type, so my default for epub output could be add metadata, & my default for mobi output could be do not add...

That idea definitely has merit.

[Starson17]

Originally Posted by kiwidude

I've had this debate with Starson17 in the past on a ticket in the bug tracker. I think he may have made a few minor tweaks to it but basically we had problems reliably replicating the "exact doubling up".

Yes - although we were talking about record merging as well as metadata fetching. I added the "don't overwrite author/title" option for metadata fetching, but finer control over metadata fetching overwrite is on the ToDo list IIRC.

I'm in your camp - I don't want summaries "appended" to under any circumstance.

I'm in the other camp. I use summaries for searching. I want them full of stuff to be found in searching, and often add several summaries from different records and different online metadata databases specifically to get more info in them.