Handle multiline comments in Package (ipkg)

...just like in Source (idr).

CHANGELOG_NEXT.md

@@ -49,6 +49,8 @@ This CHANGELOG describes the merged but unreleased changes. Please see [CHANGELO
 
 * Support for macOS PowerPC added.
 
+* Multiline comments `{- text -}` are now supported in ipkg files.
+
 ### Language changes
 
 * Autobind and Typebind modifier on operators allow the user to

CONTRIBUTORS

@@ -82,6 +82,7 @@ Wen Kokke
 Wind Wong
 Zoe Stafford
 Claudio Etterli
+Boyd Stephen Smith Jr.
 
 Apologies to anyone we've missed - let us know and we'll correct it (or just
 send a PR with the correction).

src/Parser/Lexer/Common.idr

@@ -14,6 +14,56 @@ comment
   <+> many (is '-') <+> reject (is '}')  -- not a closing delimiter
   <+> many (isNot '\n')                  -- till the end of line
 
+mutual
+  ||| The mutually defined functions represent different states in a
+  ||| small automaton.
+  ||| `toEndComment` is the default state and it will munch through
+  ||| the input until we detect a special character (a dash, an
+  ||| opening brace, or a double quote) and then switch to the
+  ||| appropriate state.
+  toEndComment : (k : Nat) -> Recognise (k /= 0)
+  toEndComment Z = empty
+  toEndComment (S k)
+               = some (pred (\c => c /= '-' && c /= '{' && c /= '"' && c /= '\''))
+                        <+> (eof <|> toEndComment (S k))
+             <|> is '{' <+> singleBrace k
+             <|> is '-' <+> singleDash k
+             <|> (charLit <|> pred (== '\'')) <+> toEndComment (S k)
+             <|> stringLit <+> toEndComment (S k)
+
+  ||| After reading a single brace, we may either finish reading an
+  ||| opening delimiter or ignore it (e.g. it could be an implicit
+  ||| binder).
+  singleBrace : (k : Nat) -> Lexer
+  singleBrace k
+     =  is '-' <+> many (is '-')              -- opening delimiter
+               <+> (eof <|> singleDash (S k)) -- `singleDash` handles the {----} special case
+                                              -- `eof` handles the file ending with {---
+    <|> toEndComment (S k)                    -- not a valid comment
+
+  ||| After reading a single dash, we may either find another one,
+  ||| meaning we may have started reading a line comment, or find
+  ||| a closing brace meaning we have found a closing delimiter.
+  singleDash : (k : Nat) -> Lexer
+  singleDash k
+     =  is '-' <+> doubleDash k    -- comment or closing delimiter
+    <|> is '}' <+> toEndComment k  -- closing delimiter
+    <|> toEndComment (S k)         -- not a valid comment
+
+  ||| After reading a double dash, we are potentially reading a line
+  ||| comment unless the series of uninterrupted dashes is ended with
+  ||| a closing brace in which case it is a closing delimiter.
+  doubleDash : (k : Nat) -> Lexer
+  doubleDash k = with Prelude.(::)
+      many (is '-') <+> choice            -- absorb all dashes
+        [ is '}' <+> toEndComment k                      -- closing delimiter
+        , many (isNot '\n') <+> toEndComment (S k)       -- line comment
+        ]
+
+export
+blockComment : Lexer
+blockComment = is '{' <+> is '-' <+> many (is '-') <+> (eof <|> toEndComment 1)
+
 -- Identifier Lexer
 -- There are multiple variants.
 

src/Parser/Lexer/Package.idr

@@ -99,6 +99,7 @@ andop = is '&' <+> is '&'
 rawTokens : TokenMap Token
 rawTokens =
   [ (comment, Comment . drop 2)
+  , (blockComment, Comment . shrink 2)
   , (namespacedIdent, uncurry DotSepIdent . mkNamespacedIdent)
   , (identAllowDashes, DotSepIdent Nothing)
   , (separator, const Separator)

src/Parser/Lexer/Source.idr

@@ -140,55 +140,6 @@ Pretty Void Token where
   pretty (MagicDebugInfo di) = pretty (show di)
   pretty (Unrecognised x) = pretty "Unrecognised" <++> pretty x
 
-mutual
-  ||| The mutually defined functions represent different states in a
-  ||| small automaton.
-  ||| `toEndComment` is the default state and it will munch through
-  ||| the input until we detect a special character (a dash, an
-  ||| opening brace, or a double quote) and then switch to the
-  ||| appropriate state.
-  toEndComment : (k : Nat) -> Recognise (k /= 0)
-  toEndComment Z = empty
-  toEndComment (S k)
-               = some (pred (\c => c /= '-' && c /= '{' && c /= '"' && c /= '\''))
-                        <+> (eof <|> toEndComment (S k))
-             <|> is '{' <+> singleBrace k
-             <|> is '-' <+> singleDash k
-             <|> (charLit <|> pred (== '\'')) <+> toEndComment (S k)
-             <|> stringLit <+> toEndComment (S k)
-
-  ||| After reading a single brace, we may either finish reading an
-  ||| opening delimiter or ignore it (e.g. it could be an implicit
-  ||| binder).
-  singleBrace : (k : Nat) -> Lexer
-  singleBrace k
-     =  is '-' <+> many (is '-')              -- opening delimiter
-               <+> (eof <|> singleDash (S k)) -- `singleDash` handles the {----} special case
-                                              -- `eof` handles the file ending with {---
-    <|> toEndComment (S k)                    -- not a valid comment
-
-  ||| After reading a single dash, we may either find another one,
-  ||| meaning we may have started reading a line comment, or find
-  ||| a closing brace meaning we have found a closing delimiter.
-  singleDash : (k : Nat) -> Lexer
-  singleDash k
-     =  is '-' <+> doubleDash k    -- comment or closing delimiter
-    <|> is '}' <+> toEndComment k  -- closing delimiter
-    <|> toEndComment (S k)         -- not a valid comment
-
-  ||| After reading a double dash, we are potentially reading a line
-  ||| comment unless the series of uninterrupted dashes is ended with
-  ||| a closing brace in which case it is a closing delimiter.
-  doubleDash : (k : Nat) -> Lexer
-  doubleDash k = with Prelude.(::)
-      many (is '-') <+> choice            -- absorb all dashes
-        [ is '}' <+> toEndComment k                      -- closing delimiter
-        , many (isNot '\n') <+> toEndComment (S k)       -- line comment
-        ]
-
-blockComment : Lexer
-blockComment = is '{' <+> is '-' <+> many (is '-') <+> (eof <|> toEndComment 1)
-
 docComment : Lexer
 docComment = is '|' <+> is '|' <+> is '|' <+> many (isNot '\n')
 

tests/idris2/pkg/multiline-comments/Main.idr

@@ -0,0 +1,4 @@
+module Main
+
+main : IO ()
+main = putStrLn "CouCou!"

tests/idris2/pkg/multiline-comments/expected

@@ -0,0 +1,2 @@
+1/1: Building Main (Main.idr)
+Now compiling the executable: test

tests/idris2/pkg/multiline-comments/run

@@ -0,0 +1,4 @@
+. ../../../testutils.sh
+
+# Previously reported: Uncaught error: Error: Can't recognise token.
+idris2 --build test.ipkg

tests/idris2/pkg/multiline-comments/test.ipkg

@@ -0,0 +1,12 @@
+package test
+{-
+A test of
+multiline
+comments.
+-}
+version = 0.1.0
+authors = "Boyd Stephen Smith Jr."
+main = Main
+executable = "test"
+{- "." is the top-level package directory -}
+sourcedir = "."