March 2013 - Beginners

Re: [Haskell-beginners] Parsec lexeme parsers and whitespace/eol characters
by Brent Yorgey 06 Mar '13

06 Mar '13

On Mon, Mar 04, 2013 at 10:13:37AM +0000, Sean Cormican wrote: > Hi Brent, > > I have been trying to use Parsec because there is more documentation and > tutorials on using it. > > I am extremely limited as a functional programmer and just as bad as an > actual programmer. I am currently 18 days away from the due date for this > project and I have made little or no progress due to my lack of ability, > would src-exts be difficult to pick up and use? haskell-src-exts works extremely well, and it's not too hard to find your way around it using the Haddock documentation. Start here: http://hackage.haskell.org/packages/archive/haskell-src-exts/1.13.5/doc/htm… Use one of the functions listed on that page to parse the file, and then click on the types to take you to pages explaining what you get back. For example, if you use parseFile you can see that you get back a (ParseResult Module). Clicking on ParseResult, you can see that a (ParseResult Module) is either (ParseOk m) where m is a Module, or ParseFailed containing an error message. So you can pattern-match to find out which. Then clicking on Module, you can see all the stuff contained in a value of type Module. Keep clicking on types to learn what they are. The biggest difficulty is that since this is a parser for *all of* Haskell, there will be *lots* of stuff that you don't care about. But it might still be easier than trying to write your own parser; I am not sure. -Brent

1 0

parse C header files and transformation
by Ted Feng 04 Mar '13

04 Mar '13

Hello list, How to do some C header files parsing and automatic code transformation. like: parse this <sys/time.h> header file, and generate code for this API: int gettimeofday(struct timeval *tv, struct timezone *tz); generated code(automatically call the API, and dump the structure appropriately): void dump_time(void) { struct timeval tv = {0}; struct timezone tz = {0}; gettimeofday(&tv, &tz); printf("tv_sec: %d, tv_usec: %d\n", tv.tv_sec, tv.tv_usec); printf("tz_minuteswest: %d, tz_dsttime: %d\n", tz.tz_minuteswest, tz.tz_dsttime); } because I'm working on a SDK which has hundreds of API to set and get headware states. It'll will be very nice to write some haskell code to parse the header files and generate C code that can dump hardware state automatically. BR Ted

1 1

Help with some heavy lifting...
by emacstheviking 04 Mar '13

04 Mar '13

I now have a working USB application that can find, locate and switch LED-s on and off on my Hexwax board, for which I thank all those that helped me out last week. I am now trying to "Haskell-ise" my code i.e. make it less amateurish with respect to some of its inner workings in a constant drive for inner cleanliness and warm fuzziness etc. When attempting to find the device, I use the System.USB.getDevices function which returns me IO (Vector Device), a list of everything that's currently plugged in and available and I then use Data.Vector.filterM like so: *handleFor ctx (cVendor, cProd) = do usbDevs <- getDevices ctx matches <- V.filterM (isTarget (cVendor, cProd)) usbDevs case V.null matches of True -> return Nothing False -> return $ Just $ matches!* *isTarget :: (Word16, Word16) -> Device -> IO Bool isTarget (vid, pid) dev = do getDeviceDesc dev >>= \info -> return $ (deviceVendorId info, deviceProductId info) == (vid, pid) * but... that is not as efficient as it could be because I could have N devices and then I just throw all but the first. Tut tut. Could do better. If I knew how... well I kind of do but I can't figure it out by myself yet! In the Data.Vector there is "Data.Vector.find" which is *exactly* what I want with small dent in the bodywork, the predicate function is pure: *find :: (a -> Bool<http://hackage.haskell.org/packages/archive/base/4.6.0.0/doc/html/Data-Bool…>) -> Vector<http://hackage.haskell.org/packages/archive/vector/0.10.0.1/doc/html/Data-V…>a -> Maybe<http://hackage.haskell.org/packages/archive/base/4.6.0.0/doc/html/Data-Mayb…>a * So my question is, how do I make it work? I know (but don't yet feel comfortable with) about liftM and all that but in this case I can't see how and where it would work. I "know" (like Spiderman knows there is danger) that it's crying out for something and the return type is perfect too as it would just match. SO...how can I do it chaps? And as usual... .any comments, style notes, idiomatic pointers(!) etc. are always welcome. Thanks, Sean Charles.

3 7

Re: [Haskell-beginners] Help with some heavy lifting...
by emacstheviking 04 Mar '13

04 Mar '13

For the record... by using Control.Monad.MissingM I have achieved two things... * more concise code * yet more stuff to learn! Here is what i ended up with and I am happy that for me at my current level, that this is clean enough whilst still readable! handleFor :: Ctx -> (Word16,Word16) -> IO (Maybe Device) handleFor ctx (cVendor, cProd) = do getDevices ctx >>= \devs -> M.findM (isTarget (cVendor, cProd)) $ V.toList devs handlesFor :: Ctx -> (Word16, Word16) -> IO (Vector Device) handlesFor ctx (vendorId, productId) = do getDevices ctx >>= \devs -> V.filterM (isTarget (vendorId, productId)) devs isTarget :: (Word16, Word16) -> Device -> IO Bool isTarget (vid, pid) dev = do info <- getDeviceDesc dev return $ (deviceVendorId info, deviceProductId info) == (vid, pid)* * If I feel braver later I may try to convert some of it to "point-free" style just to see if I can but to be honest I find that hard to read in all but the simplest and most obvious compositions. Once again a big thank-you to everybody for their time. Sean. On 4 March 2013 02:04, David McBride <toad3k(a)gmail.com> wrote: > Hoogle only indexes some small subset of libraries, while hayoo indexes > everything on hackage. Unfortunately it seems like peoples' first > inclination is to visit hoogle because hey, it's like google right? But in > fact hayoo is far better unless you know what you are looking for something > in the core libraries. > > I think there is some technical merit behind hoogle, like being able to > search for function prototypes, and I do know you can make your own hoogle > command line searcher to search libraries installed on your machine. But > usually you are searching for something you don't already have. > > An actual google search comes up with > http://hackage.haskell.org/packages/archive/MissingM/0.0.4/doc/html/Control… actually has findM proving my point that others have felt its absense > in the standard library. > > On Sun, Mar 3, 2013 at 5:24 PM, emacstheviking <objitsu(a)gmail.com> wrote: > >> David, >> >> Hoogle doesn't appear to have any matches for "findM" that I could find. >> Your code is pretty close to what I came up with this morning except yours >> is clever with monads and mine was just boring recursing through the list >> till I hit a match. This solution of yours looks like it is in the spirit >> of what I think I saw in my mind so I am going to study it very hard and >> understand it! >> >> Sometimes you just have to grind it out! >> Thanks. >> >> >> On 3 March 2013 19:46, David McBride <toad3k(a)gmail.com> wrote: >> >>> I would probably go (untested): >>> >>> ... >>> usbDevs <- ... >>> matches <- findM (isTarget foo) $ V.toList usbDevs >>> ... >>> where >>> findM :: Monad m => (a -> m Boolean) -> [a] -> m (Maybe a) >>> findM _ [] = return Nothing >>> findM f (x:xs) = do >>> b <- f x >>> return $ if b >>> then Just x >>> else findM f xs >>> >>> I can almost guarantee you there is a findM already out there somewhere >>> to use, but hayoo is down right now so I can't search for it. >>> >>> On Sun, Mar 3, 2013 at 1:28 PM, emacstheviking <objitsu(a)gmail.com>wrote: >>> >>>> I now have a working USB application that can find, locate and switch >>>> LED-s on and off on my Hexwax board, for which I thank all those that >>>> helped me out last week. >>>> >>>> I am now trying to "Haskell-ise" my code i.e. make it less amateurish >>>> with respect to some of its inner workings in a constant drive for inner >>>> cleanliness and warm fuzziness etc. >>>> >>>> When attempting to find the device, I use the System.USB.getDevices >>>> function which returns me IO (Vector Device), a list of everything that's >>>> currently plugged in and available and I then use Data.Vector.filterM like >>>> so: >>>> >>>> *handleFor ctx (cVendor, cProd) = do >>>> usbDevs <- getDevices ctx >>>> matches <- V.filterM (isTarget (cVendor, cProd)) usbDevs >>>> case V.null matches of >>>> True -> return Nothing >>>> False -> return $ Just $ matches!* >>>> >>>> *isTarget :: (Word16, Word16) -> Device -> IO Bool >>>> isTarget (vid, pid) dev = do >>>> getDeviceDesc dev >>= \info -> >>>> return $ (deviceVendorId info, deviceProductId info) == (vid, pid) >>>> * >>>> >>>> but... that is not as efficient as it could be because I could have N >>>> devices and then I just throw all but the first. Tut tut. Could do better. >>>> If I knew how... well I kind of do but I can't figure it out by myself yet! >>>> >>>> In the Data.Vector there is "Data.Vector.find" which is *exactly* what >>>> I want with small dent in the bodywork, the predicate function is pure: >>>> >>>> *find :: (a -> Bool<http://hackage.haskell.org/packages/archive/base/4.6.0.0/doc/html/Data-Bool…>) >>>> -> Vector<http://hackage.haskell.org/packages/archive/vector/0.10.0.1/doc/html/Data-V…>a -> >>>> Maybe<http://hackage.haskell.org/packages/archive/base/4.6.0.0/doc/html/Data-Mayb…>a >>>> * >>>> So my question is, how do I make it work? I know (but don't yet feel >>>> comfortable with) about liftM and all that but in this case I can't see how >>>> and where it would work. I "know" (like Spiderman knows there is danger) >>>> that it's crying out for something and the return type is perfect too as it >>>> would just match. >>>> >>>> SO...how can I do it chaps? >>>> >>>> And as usual... .any comments, style notes, idiomatic pointers(!) etc. >>>> are always welcome. >>>> >>>> Thanks, >>>> Sean Charles. >>>> >>>> >>>> _______________________________________________ >>>> Beginners mailing list >>>> Beginners(a)haskell.org >>>> http://www.haskell.org/mailman/listinfo/beginners >>>> >>>> >>> >>> _____________________________ >>> >>> >>> >>> -- | Internal helper to simply check that the product and >>> vendor >>> -- values in the current Device handle match those in (vid, >>> pid). >>> >>> __________________ >>> Beginners mailing list >>> Beginners(a)haskell.org >>> http://www.haskell.org/mailman/listinfo/beginners >>> >>> >> >> _______________________________________________ >> Beginners mailing list >> Beginners(a)haskell.org >> http://www.haskell.org/mailman/listinfo/beginners >> >> > > _______________________________________________ > Beginners mailing list > Beginners(a)haskell.org > http://www.haskell.org/mailman/listinfo/beginners > >

1 0

Parsec lexeme parsers and whitespace/eol characters
by Sean Cormican 02 Mar '13

02 Mar '13

Hi, I'm working with expressions in which I use parsec libraries lexeme parsers: identifier, integer, whiteSpace etc. I am attempting to parse haskell declarations and I am facing some difficulty when trying to delimit the declarations which span multiple lines using eol character for some reason ( I am led to believe is due to the removal of whitespace with lexemes ) data Express = Seq [Express] | ID String | Num Integer | BoolConst Bool | EmptyList String seqOfExpr8 = do list <- (sepBy1 expr8 eol) return $ if length list == 1 then head list else Seq list where Seq list is a data declaration for a list of expression8's The parser works correctly unless trying to parse a lexeme parser token such as an identifier. Upon reading a lexeme token the parser will stop and output all tokens up until and including that token but no tokens past it. Also when trying to separate declarations that span multiple lines, does anyone have an idea on how to achieve this? I have been thinking that the parser should ignore any end line characters that are followed by whitespace ( as whitespace is used in Haskell layout rules ) Any thoughts on how I might solve this would be greatly appreciated! Seán

2 1