Re: [nimh-dev] More stuff

["Jon Stewart" <stew1@xxxxxxxxxxx>]

> I can think of a pythony simple way to do this.  Say you specified a sequence
> "cur 8 9-11 foo" where foo is an existing sequence consisting of "7 10-12", a
> nd
> cur is 4.
> 
> 1. ['cur', 8, '9-11', 'foo']    <-- split by whitespace
> 2. [4, 8, '9-11', [7, '10-12']]  <-- parse tokens (foo, cur)
> 3. [4, 8, [9, 11], [7, [10, 12]]]  <-- parse ranges 
> 4. [4, 8, [9, 11], 7, [10, 12]]  <-- eliminate sublists
> 5. [4, 8, [9, 11], 7, 12]  <-- eliminate overlap
> 6. [4, 7, 8, [9, 11], 12] <--  sort


I think it will be easier if it's like this:

1. ['cur', 8, '9-11', 'foo']
2. [4, 8, '9-11', 7, '10-12'] # list remains flat
3. [8, '9-11', 7, '10-12'] --> [(4, 5)]
4. ['9-11', 7, '10-12'] --> [(4, 5), (8, 9)]
5. [7, '10-12'] --> [(4, 5), (8, 12)]
6. ['10-12'] --> [(4, 5), (7, 12)]
7. [] --> [(4, 5), (7, 13)]

Numeric tokens are popped off the first list and a second list of ranges 
is created.

But my brain has been warped by too many years of C++'s static 
type-checking, so my preference may be due to my inability to think about 
collections of different types.


> Then 5. would be the final form that was stored in sequence.seq.  No need to
> wrap up [7, 8, [9,11], 12] into [7,12]; it's plenty easy to deal with in its
> current form.  For step 3, you'd have to take a piecewise continuous range
> and split it up into continuous ranges; for instance, if your folder contains
> messages 1-400 and 402-500, and you specify 1-500, it should store the value
> [[1,400],[402,500]]


No, it should store [1, 500] (or [1, 501]). Checking for existence later 
on will keep the code cleaner.


Jon
-- 
Jon Stewart                                 Advanced Los Angeles C++
stew1@xxxxxxxxxxx                           http://www.alacpp.org
_______________________________________________
nimh-dev mailing list
nimh-dev@xxxxxxxxxxx
http://lists.ellipsis.cx/mailman/listinfo/nimh-dev