Good catch. This is a minor change of backward compatibility (the '%' was
supposed to be in encoder) but we agreed this is an experimental feature
and you are suggesting a good improvement compatible with the example in
the manual.
I implemented it in trunk with some minor changes:
if isinstance(fieldtype, SQLCustomType):
value = fieldtype.encoder(obj)
if fieldtype.native in ('string','text'):
return self.adapt(value)
return value
On Tuesday, 5 June 2012 01:23:58 UTC-5, Alexander Shashkevych wrote:
>
> I tried another field name, but result the same. I think dal has
> incomplete handling of custom types. I analyzed dal code and found that
> constructed query for my example was:
>
> INSERT INTO posts(content,html,title) VALUES
> (\xd0\x94\xd0\xb2\xd0\xb0,\xd0\xa2\xd1\x80\xd0\xb8,\xd0\x9e\xd0\xb4\xd0\xb8\xd0\xbd);
>
> As you can see values are not enclosed by quotes. This is incorrect as for
> me. I rewrote my custom type:
>
> utext = SQLCustomType(
> type="text",
> native="text",
> encoder = lambda x: "'%s'" % x.replace("'", "''").encode("utf-8"),
> decoder = lambda x: x.decode("utf-8")
> )
>
> And it works after such update, but example from the web2py book wouldn't
> work also:
>
> compressed = SQLCustomType(
> type ='text',
> native='text',
> encoder =(lambda x: zlib.compress(x or '')),
> decoder = (lambda x: zlib.decompress(x))
> )
>
> Further analysis shown that DAL doesn't check SQLCustomType.type while
> constructing query and simply calls SQLCustomType.encoder(). But it should
> check the type of SQLCustomType and enclose it with quotes in case of
> string/text type. I updated BaseAdapter.represent() with following code:
>
> ...
> if isinstance(fieldtype, SQLCustomType):
> if fieldtype.type == "string" or fieldtype.type == "text": # added
> return self.adapt(fieldtype.encoder(obj)) # added
> return fieldtype.encoder(obj)
> ...
>
> Now my custom type doesn't require code that adds quotes - now this does
> DAL. =)
>
> Could you comment how my fix fits DAL design?
>
