search for file size

[alexxx]

having received my new kindle, I bulk-converted my ebooks to MOBI format.
As expected, I had a lot of conversion errors, which I'm trying to track down and I'm slowly solving by small changes in the source files (often HTML).

My problem is how to find the problem ebooks, which can be defined as having MOBI file format with zero file size. Can I search for this in some way?

alessandro

[toddos]

While the column header for size says that the search field is "size", it doesn't seem to actually work. However you can search for mobi formatted files using this in the search box (or using the handy categories on the left, under Formats)

formats:"=MOBI"

Once you have the list of all books with MOBI-formatted files, you can sort by the Size column to put the 0-size files at the top (or bottom).

[chaley]

Note that 'size' is the sum of all the formats, not the size of the format searched for.

And yes, size isn't a valid search term. I could fix that, I suppose ...

[alexxx]

a fix for that would be useful...
an even better fix I guess would be not to create a new zero-size file when the conversion fails!

alessandro

[theducks]

Originally Posted by chaley

Note that 'size' is the sum of all the formats, not the size of the format searched for.

And yes, size isn't a valid search term. I could fix that, I suppose ...

The data table has the formats and uncompressed_size.
Having the filter report the current "view" size (currently continues to report overall size would be )

[chaley]

Originally Posted by alexxx

a fix for that would be useful...
an even better fix I guess would be not to create a new zero-size file when the conversion fails!

The second is out of my area.

You could use a command-line command to find what you are looking for. The command is 'find'. If you are on windows, you need to install it from http://gnuwin32.sourceforge.net/packages/findutils.htm. Other systems will have it as a native command.

Once installed, in a terminal/command box with the current directory at the top folder of the library (the one containing metadata.db), run the command

find -name "*.mobi" -size 0

If you are on windows, you might need to type

"\Program Files\GnuWin32\bin\find.exe" -name "*.mobi" -size 0

You will get a list of .mobi files with zero length. The path will tell you the author and title.

[chaley]

Originally Posted by theducks

The data table has the formats and uncompressed_size.
Having the filter report the current "view" size (currently continues to report overall size would be )

Reporting the size of items found isn't something I am interested in doing. Size is a column in the books table, kept up-to-date when formats are added or removed, so searching on that would be easy. Recomputing 'size' dynamically would be expensive, as well as being a right bother to implement.

[kovidgoyal]

Actually, IIRC, size is the max of the sizes of the various formats, not the sum. Other than that, I agree with chaley.

[chaley]

Kovid remembers correctly.

The select that builds the size field for the books view is
SELECT MAX(uncompressed_size) FROM data WHERE book=books.id

(edit: yes, I am using table & view interchangeably. Sloppy of me...)

[chaley]

Originally Posted by chaley

Note that 'size' is the sum of all the formats, not the size of the format searched for.

And yes, size isn't a valid search term. I could fix that, I suppose ...

Size has been added as a search term, along with multiplier suffixes K, M, and G. The multipliers have their normal computer meaning, 1024, 1024**2, and 1024**3.

Examples: size:>1.1M will find books with a format larger than 1.1 *1024*1024. size:<=1k finds those equal to smaller than 1 * 1024. The multipliers can be lower case.

If you want to avoid the use of 1024, type the numbers out. For example, size:<1000000 will find sizes smaller than 1 million.