Fix #37 (undefined in Show PatternSet) by lazy match

.github/workflows/haskell-ci.yml

@@ -8,9 +8,9 @@
 #
 # For more information, see https://github.com/haskell-CI/haskell-ci
 #
-# version: 0.15.20220525
+# version: 0.15.20220710
 #
-# REGENDATA ("0.15.20220525",["github","regex-tdfa.cabal"])
+# REGENDATA ("0.15.20220710",["github","regex-tdfa.cabal"])
 #
 name: Haskell-CI
 on:
@@ -32,14 +32,14 @@ jobs:
     strategy:
       matrix:
         include:
-          - compiler: ghc-9.4.0.20220501
+          - compiler: ghc-9.4.0.20220623
             compilerKind: ghc
-            compilerVersion: 9.4.0.20220501
+            compilerVersion: 9.4.0.20220623
             setup-method: ghcup
             allow-failure: true
-          - compiler: ghc-9.2.2
+          - compiler: ghc-9.2.3
             compilerKind: ghc
-            compilerVersion: 9.2.2
+            compilerVersion: 9.2.3
             setup-method: ghcup
             allow-failure: false
           - compiler: ghc-9.0.2
@@ -107,17 +107,17 @@ jobs:
             mkdir -p "$HOME/.ghcup/bin"
             curl -sL https://downloads.haskell.org/ghcup/0.1.17.8/x86_64-linux-ghcup-0.1.17.8 > "$HOME/.ghcup/bin/ghcup"
             chmod a+x "$HOME/.ghcup/bin/ghcup"
-            if $HEADHACKAGE; then "$HOME/.ghcup/bin/ghcup" config add-release-channel https://raw.githubusercontent.com/haskell/ghcup-metadata/master/ghcup-prereleases-0.0.7.yaml; fi
-            "$HOME/.ghcup/bin/ghcup" install ghc "$HCVER"
-            "$HOME/.ghcup/bin/ghcup" install cabal 3.6.2.0
+            "$HOME/.ghcup/bin/ghcup" config add-release-channel https://raw.githubusercontent.com/haskell/ghcup-metadata/master/ghcup-prereleases-0.0.7.yaml;
+            "$HOME/.ghcup/bin/ghcup" install ghc "$HCVER" || (cat "$HOME"/.ghcup/logs/*.* && false)
+            "$HOME/.ghcup/bin/ghcup" install cabal 3.6.2.0 || (cat "$HOME"/.ghcup/logs/*.* && false)
           else
             apt-add-repository -y 'ppa:hvr/ghc'
             apt-get update
             apt-get install -y "$HCNAME"
             mkdir -p "$HOME/.ghcup/bin"
             curl -sL https://downloads.haskell.org/ghcup/0.1.17.8/x86_64-linux-ghcup-0.1.17.8 > "$HOME/.ghcup/bin/ghcup"
             chmod a+x "$HOME/.ghcup/bin/ghcup"
-            "$HOME/.ghcup/bin/ghcup" install cabal 3.6.2.0
+            "$HOME/.ghcup/bin/ghcup" install cabal 3.6.2.0 || (cat "$HOME"/.ghcup/logs/*.* && false)
           fi
         env:
           HCKIND: ${{ matrix.compilerKind }}
@@ -186,6 +186,7 @@ jobs:
                         26021a13b401500c8eb2761ca95c61f2d625bfef951b939a8124ed12ecf07329
                         f76d08be13e9a61a377a85e2fb63f4c5435d40f8feb3e12eb05905edb8cdea89
              key-threshold: 3
+          active-repositories: hackage.haskell.org, head.hackage.ghc.haskell.org:override
           EOF
           fi
           cat >> $CABAL_CONFIG <<EOF

CHANGELOG.md

@@ -1,5 +1,12 @@
 For the package version policy (PVP), see  http://pvp.haskell.org/faq .
 
+### 1.3.1.3
+
+_2022-07-14, Andreas Abel_
+
+- Fix an `undefined` in `Show PatternSet` [#37](https://github.com/haskell-hvr/regex-tdfa/issues/37))
+- Document POSIX character classes (e.g. `[[:digit:]]`)
+
 ### 1.3.1.2 Revision 1
 
 _2022-05-25, Andreas Abel_

README.md

@@ -55,7 +55,7 @@ a =~ b :: String  -- or ByteString, or Text...
 λ> "alexis-de-tocqueville" =~ "[a-z]+" :: String
 >>> "alexis"
 
-λ> "alexis-de-tocqueville" =~ "[0-9]+" :: String
+λ> "alexis-de-tocqueville" =~ "[[:digit:]]+" :: String
 >>> ""
 ```
 
@@ -103,7 +103,7 @@ getAllTextMatches (a =~ b) :: [String]
 λ> getAllTextMatches ("john anne yifan" =~ "[a-z]+") :: [String]
 >>> ["john","anne","yifan"]
 
-λ> getAllTextMatches ("0a0b0" =~ "0[a-z]0") :: [String]
+λ> getAllTextMatches ("0a0b0" =~ "0[[:lower:]]0") :: [String]
 >>> ["0a0"]
 ```
 Note that `"0b0"` is not included in the result since it overlaps with `"0a0"`.
@@ -120,6 +120,17 @@ featureful than some other regex engines you might be used to, such as PCRE.
 * `\b` — Match beginning or end of word
 * `\B` — Match neither beginning nor end of word
 
+While shorthands like `\d` (for digit) are not recognized, one can use the respective
+POSIX character class inside `[...]`.  E.g., `[[:digit:][:lower:]_]` is short for
+`[0-9a-z_]`.  The supported character classes are listed on
+[Wikipedia](https://en.wikipedia.org/w/index.php?title=Regular_expression&oldid=1095256273#Character_classes)
+and defined in module
+[`TNFA`](https://github.com/haskell-hvr/regex-tdfa/blob/95d47cb982d2cf636b2cb6260a866f9907341c45/lib/Text/Regex/TDFA/TNFA.hs#L804-L816).
+
+Please also consult a variant of this documentation which is part of the
+[Text.Regex.TDFA haddock](http://hackage.haskell.org/package/regex-tdfa/docs/Text-Regex-TDFA.html),
+and the original documentation at the [Haskell wiki](https://wiki.haskell.org/Regular_expressions#regex-tdfa).
+
 ### Less common stuff
 
 #### Get match indices
@@ -148,16 +159,6 @@ getAllSubmatches (a =~ b) :: [(Int, Int)]  -- (index, length)
 
 `regex-tdfa` does not provide find-and-replace.
 
-## The relevant links
-
-This documentation is also available in [Text.Regex.TDFA haddock](http://hackage.haskell.org/package/regex-tdfa-1.2.3.2/docs/Text-Regex-TDFA.html).
-
-This was also documented at the [Haskell wiki](https://wiki.haskell.org/Regular_expressions#regex-tdfa).  The original Darcs repository was at [code.haskell.org](http://code.haskell.org/regex-tdfa/).  When not updated, this was forked and maintained by Roman Cheplyaka as [regex-tdfa-rc](http://hackage.haskell.org/package/regex-tdfa-rc).
-
-Then the repository moved to <https://github.com/ChrisKuklewicz/regex-tdfa>, which was primarily maintained by [Artyom (neongreen)](https://github.com/neongreen).
-
-Finally, maintainership was passed on again and the repository moved to its current location at <https://github.com/haskell-hvr/regex-tdfa>.
-
 ## Avoiding backslashes
 
 If you find yourself writing a lot of regexes, take a look at
@@ -193,10 +194,23 @@ By building on this thesis and adding a few more optimizations, regex-tdfa match
 
 Regardless of performance, nearly every single OS and Libra for POSIX regular expressions has bugs in sub-matches.  This was detailed on the [Regex POSIX Haskell wiki page](https://wiki.haskell.org/Regex_Posix), and can be demonstrated with the [regex-posix-unittest](http://hackage.haskell.org/package/regex-posix-unittest) suite of checks.  Test [regex-tdfa-unittest](http://hackage.haskell.org/package/regex-tdfa-unittest) should show regex-tdfa passing these same checks.  I owe my understanding of the correct behvior and many of these unit tests to Glenn Fowler at AT&T ("An Interpretation of the POSIX regex Standard").
 
+### Maintainance history
+
+The original Darcs repository was at [code.haskell.org](http://code.haskell.org/regex-tdfa/).
+For a while a fork was maintained by Roman Cheplyaka as
+[regex-tdfa-rc](http://hackage.haskell.org/package/regex-tdfa-rc).
+
+Then the repository moved to <https://github.com/ChrisKuklewicz/regex-tdfa>,
+which was primarily maintained by [Artyom (neongreen)](https://github.com/neongreen).
+
+Finally, maintainership was passed on again and the repository moved to its current location
+at <https://github.com/haskell-hvr/regex-tdfa>.
+
 ## Other related packages
 
-You can find several other related packages by searching for "tdfa" on [hackage](http://hackage.haskell.org/packages/search?terms=tdfa).
+Searching for "tdfa" on [hackage](http://hackage.haskell.org/packages/search?terms=tdfa)
+finds some related packages (unmaintained as of 2022-07-14).
 
 ## Document notes
 
-This was written 2016-04-30.
+This README was originally written 2016-04-30.

lib/Text/Regex/TDFA/CorePattern.hs

@@ -36,14 +36,17 @@ module Text.Regex.TDFA.CorePattern(Q(..),P(..),WhichTest(..),Wanted(..)
 
 import Control.Monad (liftM2, forM, replicateM)
 import Control.Monad.RWS (RWS, runRWS, ask, local, listens, tell, get, put)
+
 import Data.Array.IArray(Array,(!),accumArray,listArray)
+import Data.Either (partitionEithers, rights)
 import Data.List(sort)
 import Data.IntMap.EnumMap2(EnumMap)
 import qualified Data.IntMap.EnumMap2 as Map(singleton,null,assocs,keysSet)
 --import Data.Maybe(isNothing)
 import Data.IntSet.EnumSet2(EnumSet)
 import qualified Data.IntSet.EnumSet2 as Set(singleton,toList,isSubsetOf)
 import Data.Semigroup as Sem
+
 import Text.Regex.TDFA.Common {- all -}
 import Text.Regex.TDFA.Pattern(Pattern(..),starTrans)
 -- import Debug.Trace
@@ -278,18 +281,6 @@ makeGroupArray :: GroupIndex -> [GroupInfo] -> Array GroupIndex [GroupInfo]
 makeGroupArray maxGroupIndex groups = accumArray (\earlier later -> later:earlier) [] (1,maxGroupIndex) filler
     where filler = map (\gi -> (thisIndex gi,gi)) groups
 
-fromRight :: [Either Tag GroupInfo] -> [GroupInfo]
-fromRight [] = []
-fromRight ((Right x):xs) = x:fromRight xs
-fromRight ((Left _):xs) = fromRight xs
-
-partitionEither :: [Either Tag GroupInfo] -> ([Tag],[GroupInfo])
-partitionEither = helper id id where
-  helper :: ([Tag]->[Tag]) -> ([GroupInfo]->[GroupInfo]) -> [Either Tag GroupInfo] -> ([Tag],[GroupInfo])
-  helper ls rs [] = (ls [],rs [])
-  helper ls rs ((Right x):xs) = helper  ls      (rs.(x:)) xs
-  helper ls rs ((Left  x):xs) = helper (ls.(x:)) rs       xs
-
 -- Partial function: assumes starTrans has been run on the Pattern
 -- Note that the lazy dependency chain for this very zigzag:
 --   varies information is sent up the tree
@@ -305,7 +296,7 @@ patternToQ :: CompOption -> (Pattern,(GroupIndex,DoPa)) -> (Q,Array Tag OP,Array
 patternToQ compOpt (pOrig,(maxGroupIndex,_)) = (tnfa,aTags,aGroups) where
   (tnfa,(tag_dlist,nextTag),groups) = runRWS monad startReader startState
   aTags = listArray (0,pred nextTag) (tag_dlist [])
-  aGroups = makeGroupArray maxGroupIndex (fromRight groups)
+  aGroups = makeGroupArray maxGroupIndex (rights groups)
 
   -- implicitly inside a PGroup 0 converted into a GroupInfo 0 undefined 0 1
   monad = go (starTrans pOrig) (Advice 0) (Advice 1)
@@ -354,8 +345,7 @@ patternToQ compOpt (pOrig,(maxGroupIndex,_)) = (tnfa,aTags,aGroups) where
   -- withOrbit uses MonadWriter(listens to makeOrbit/Left), collects
   -- children at all depths
   withOrbit :: PM a -> PM (a,[Tag])
-  withOrbit = listens childStars
-    where childStars x = let (ts,_) = partitionEither x in ts
+  withOrbit = listens $ fst . partitionEithers
 
   {-# INLINE makeGroup #-}
   -- makeGroup usesMonadWriter(tell/Right)
@@ -380,7 +370,7 @@ patternToQ compOpt (pOrig,(maxGroupIndex,_)) = (tnfa,aTags,aGroups) where
   withParent :: GroupIndex -> PM a -> PM (a,[Tag])
   withParent this = local (const (Just this)) . listens childGroupInfo
     where childGroupInfo x =
-            let (_,gs) = partitionEither x
+            let gs = snd $ partitionEithers x
                 children :: [GroupIndex]
                 children = norep . sort . map thisIndex
                            -- filter to get only immediate children (efficiency)

lib/Text/Regex/TDFA/Pattern.hs

@@ -1,6 +1,9 @@
+{-# OPTIONS_GHC -fno-warn-incomplete-uni-patterns #-}
+
 -- | This "Text.Regex.TDFA.Pattern" module provides the 'Pattern' data
 -- type and its subtypes.  This 'Pattern' type is used to represent
 -- the parsed form of a Regular Expression.
+
 module Text.Regex.TDFA.Pattern
     (Pattern(..)
     ,PatternSet(..)
@@ -105,9 +108,9 @@ instance Show PatternSet where
     in shows charSpec
        . showsPrec i scc' . showsPrec i sce' . showsPrec i sec'
        . if '-' `elem` special then showChar '-' else id
-    where byRange xAll@(x:xs) | length xAll <=3 = xAll
-                              | otherwise = groupRange x 1 xs
-          byRange _ = undefined
+    where byRange xAll@(~(x:xs))
+            | length xAll <=3 = xAll
+            | otherwise       = groupRange x 1 xs
           groupRange x n (y:ys) = if (fromEnum y)-(fromEnum x) == n then groupRange x (succ n) ys
                                   else (if n <=3 then take n [x..]
                                         else x:'-':(toEnum (pred n+fromEnum x)):[]) ++ groupRange y 1 ys

regex-tdfa.cabal

@@ -1,7 +1,6 @@
 cabal-version:          1.12
 name:                   regex-tdfa
-version:                1.3.1.2
-x-revision:             1
+version:                1.3.1.3
 
 build-Type:             Simple
 license:                BSD3
@@ -27,7 +26,7 @@ extra-source-files:
 
 tested-with:
   GHC == 9.4.1
-  GHC == 9.2.2
+  GHC == 9.2.3
   GHC == 9.0.2
   GHC == 8.10.7
   GHC == 8.8.4
@@ -47,7 +46,7 @@ source-repository head
 source-repository this
   type:                git
   location:            https://github.com/haskell-hvr/regex-tdfa.git
-  tag:                 v1.3.1.2-r1
+  tag:                 v1.3.1.3
 
 flag force-O2
   default: False