Custom comparison function with orderby in select()

[Ian W. Scott]

I want to sort the results of a select() based on a string field. Normally I can just use select(orderby=db.mytable.myfield), but the strings are not in Latin characters, so the default string sorting produces unacceptable results. 

I've figured out how to create a comparison function using the PyICU bindings to the ICU library. In a regular list sort this allows me to do something like this:

    from icu import Locale, Collator

    myloc = Locale('el')                                                

    col = Collator.createInstance(myloc)                     

    words = ['ἀγ', 'βλα', 'ὁμηρ']                                    

    sorted_words = sorted(words, cmp=col.compare) 

Is there any way to pass this kind of custom collating function to the select()?   

[Anthony]

The sorting via orderby is handled by the database, so you can't use Python code for that, as it must be executed by the database. However, the Rows object has a .sort() method, which does the sorting in Python after the records are returned by the database. See http://web2py.com/books/default/chapter/29/06/the-database-abstraction-layer#find--exclude--sort.

Note, depending on the size of the dataset, sorting in Python might be slower than having the database handle it.

Anthony

[Ian W. Scott]

A belated thanks, Anthony. I had forgotten about the rows.sort() method.