Merge remote-tracking branch 'danr/Improve-Haskell-highlighter'

rc/base/haskell.kak

@@ -12,19 +12,61 @@ hook global BufCreate .*[.](hs) %{
 # ‾‾‾‾‾‾‾‾‾‾‾‾
 
 add-highlighter -group / regions -default code haskell \
-    string   %{(?<!')"} (?<!\\)(\\\\)*"      ''  \
+    string   '(?<!\'\\)(?<!\')"'         (?<!\\)(\\\\)*" ''   \
-    comment  (--[^>])   $                    ''  \
+    macro   ^\h*?\K#                     (?<!\\)\n       ''   \
+    pragma  \{-#                         \#-\}           \{-  \
     comment \{-                            -\}           \{-  \
-    macro   ^\h*?\K#    (?<!\\)\n            ''
+    comment --(?![!#$%&*+./<>?@\\\^|~=]) $               ''
 
 add-highlighter -group /haskell/string  fill string
 add-highlighter -group /haskell/comment fill comment
+add-highlighter -group /haskell/pragma  fill meta
 add-highlighter -group /haskell/macro   fill meta
 
-add-highlighter -group /haskell/code regex \b(import)\b 0:meta
+# Use (?<!['\w]) and (?!['\w]) instead of \b because ' is valid in identifiers
-add-highlighter -group /haskell/code regex \b(True|False)\b 0:value
+add-highlighter -group /haskell/code regex (?<!['\w])0x+[A-Fa-f0-9]+ 0:value
-add-highlighter -group /haskell/code regex \b(as|case|class|data|default|deriving|do|else|hiding|if|in|infix|infixl|infixr|instance|let|module|newtype|of|qualified|then|type|where)\b 0:keyword
+add-highlighter -group /haskell/code regex (?<!['\w])\d+([.]\d+)? 0:value
-add-highlighter -group /haskell/code regex \b(Int|Integer|Char|Bool|Float|Double|IO|Void|Addr|Array|String)\b 0:type
+add-highlighter -group /haskell/code regex (?<!['\w])(import|hiding|qualified|module)(?!['\w]) 0:keyword
+add-highlighter -group /haskell/code regex (?<!['\w])(import)(?!['\w])[^\n]+(?<!['\w])(as)(?!['\w]) 2:keyword
+add-highlighter -group /haskell/code regex (?<!['\w])(class|data|default|deriving|infix|infixl|infixr|instance|module|newtype|pattern|type|where)(?!['\w]) 0:keyword
+add-highlighter -group /haskell/code regex (?<!['\w])(case|do|else|if|in|let|mdo|of|proc|rec|then)(?!['\w]) 0:attribute
+
+# The complications below is because period has many uses:
+# As function composition operator (possibly without spaces) like "." and "f.g"
+# Hierarchical modules like "Data.Maybe"
+# Qualified imports like "Data.Maybe.Just", "Data.Maybe.maybe", "Control.Applicative.<$>"
+# Quantifier separator in "forall a . [a] -> [a]"
+# Enum comprehensions like "[1..]" and "[a..b]" (making ".." and "Module..." illegal)
+
+# matches uppercase identifiers:  Monad Control.Monad
+# not non-space separated dot:    Just.const
+add-highlighter -group /haskell/code regex \b(\u['\w]*\.)*\u['\w]*(?!['\w])(?![.\l]) 0:variable
+
+# matches infix identifier: `mod` `Apa._T'M`
+add-highlighter -group /haskell/code regex `\b(\u['\w]*\.)*[\w]['\w]*` 0:operator
+# matches imported operators: M.! M.. Control.Monad.>>
+# not operator keywords:      M... M.->
+add-highlighter -group /haskell/code regex \b\u['\w]*\.(?!([~=|:@\\]|<-|->|=>|\.\.|::)[^~<=>|:!?/.@$*&#%+\^\-\\])[~<=>|:!?/.@$*&#%+\^\-\\]+ 0:operator
+# matches dot: .
+# not possibly incomplete import:  a.
+# not other operators:             !. .!
+add-highlighter -group /haskell/code regex (?<![\w~<=>|:!?/.@$*&#%+\^\-\\])\.(?![~<=>|:!?/.@$*&#%+\^\-\\]) 0:operator
+# matches other operators: ... > < <= ^ <*> <$> etc
+# not dot: .
+# not operator keywords:  @ .. -> :: ~
+add-highlighter -group /haskell/code regex (?<![~<=>|:!?/.@$*&#%+\^\-\\])(?!([~=|:.@\\]|<-|->|=>|\.\.|::)[^~<=>|:!?/.@$*&#%+\^\-\\])[~<=>|:!?/.@$*&#%+\^\-\\]+ 0:operator
+
+# matches operator keywords: @ ->
+add-highlighter -group /haskell/code regex (?<![~<=>|:!?/.@$*&#%+\^\-\\])(@|~|<-|->|=>|::|=|:|[|])(?![~<=>|:!?/.@$*&#%+\^\-\\]) 1:keyword
+# matches: forall [..variables..] .
+# not the variables
+add-highlighter -group /haskell/code regex \b(forall)\b[^.\n]*?(\.) 1:keyword 2:keyword
+
+# matches 'x' '\\' '\'' '\n' '\0'
+# not incomplete literals: '\'
+# not valid identifiers:   w' _'
+add-highlighter -group /haskell/code regex (?<!\w)'([^\\]|[\\]['"\w\d\\])' 0:string
+# this has to come after operators so '-' etc is correct
 
 # Commands
 # ‾‾‾‾‾‾‾‾