python - how do i take advantage of sqlite manifest typing / type affinity using sqlalchemy? -

I like the idea of ​​sqlite typing / type affinity:

Basically, if I set the affinity of a column as 'numerical', then it floats to store it as an integer or type, but still allows me to store the stars if I I think I think this is the best 'default' type for a column, when I get that data Unbelievable first time what I want to store it.

I go to:

metadata = metadata () new_table = table (table_name, metadata) =

but when I try some string values ​​and store them I like "ABC" in the table, sqlalchemy drops:

  file "[...] \ sqlalchemy \ processors.py", line 79, to_float wa C float ValueError: invalid for float verbatim: abc  

Boo, her. So, is there any way that I can explain Skylightmi to typing squilit? Probably Can I use sqlachemy.dialects.sqlite instead of sqlalchemy.types?

for bonus points [edit: I should be able to use the table through introspection / reflection So then some kind of work with meta.reflect () Security would be great! ; -)

OK, here's where I came with:

Define a custom column type, according to

documentation and some tests & amp; Error has given me this:

  class MyDuckType (sqlalchemy.types.TypeDecorator): "" "SQLalchemy custom column type, which is to handle scripts typing using 'numerical connection' Is designed for those who handle both intellectually numbers and strings "implication = sqlite.NUMERIC def bind_processor (self, quote): While typing # db, the methodology for type rape is any #E Not on SQL typing deef The results / processor (self, dialect, coltype): # function for typed shameless during reading #db, none # e pass value, which is stored by esclip, is already def processor_bind_prim (self, price, bid ): A change in a personal value before the store in #DN return value def. Process_result_value (self, price, bid): #db return value after getting the def copy (self) # personalized changes No change for: # Definitely not sure what this is for the return of MyDuc (s)  

The current sqlalchemy bid type in bind_processor gives to_float, which is why I was getting errors before for my bonus points: Manually column type in MyDuckType in my metadata. Settings () Code:

  def get_database_tables (engine): This is a bug.  meta.mettata () meta.reflect (bind = engine) table = table.values ​​in table.values ​​(): for col in tbl.c: col.type = MyDuckType ( Return table  

Any suggestions / correction seems to work for me?