From a5ad6a05d7c271489933f3fd48757388716c17ff Mon Sep 17 00:00:00 2001
From: Veronika Romashkina <vrom911@gmail.com>
Date: Fri, 14 Oct 2022 14:50:08 +0100
Subject: [PATCH] [#70] Tripped by comments before pragmas

Resolves #70
---
 CHANGELOG.md                   |  6 ++++++
 src/Extensions/Module.hs       | 17 ++++++++++-------
 test/Test/Extensions/Module.hs | 16 ++++++++++++++++
 3 files changed, 32 insertions(+), 7 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index b116fc4..907b78e 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -3,6 +3,12 @@
 `extensions` uses [PVP Versioning][1].
 The changelog is available [on GitHub][2].
 
+## Unreleased
+
+* [#70](https://github.com/kowainik/extensions/issues/70):
+  Parse empty lines and spaces in before comments and pragmas in the beginning
+  of the file.
+
 ## 0.0.0.1 — May 9, 2020 🎖️
 
 * Handle one more parsing case.
diff --git a/src/Extensions/Module.hs b/src/Extensions/Module.hs
index d727f77..f1475fd 100644
--- a/src/Extensions/Module.hs
+++ b/src/Extensions/Module.hs
@@ -31,7 +31,7 @@ import Data.List (nub)
 import Data.List.NonEmpty (NonEmpty (..))
 import System.Directory (doesFileExist)
 import Text.Parsec (alphaNum, between, char, eof, many, many1, manyTill, noneOf, oneOf, parse,
-                    sepBy1, try, (<|>))
+                    sepBy1, skipMany, try, (<|>))
 import Text.Parsec.ByteString (Parser)
 import Text.Parsec.Char (anyChar, endOfLine, letter, newline, space, spaces, string)
 import Text.Read (readMaybe)
@@ -106,9 +106,12 @@ It parses language pragmas or comments until end of file or the first line with
 the function/import/module name.
 -}
 extensionsP :: Parser [ParsedExtension]
-extensionsP = concat <$> manyTill
-    (try singleExtensionsP <|> try optionsGhcP <|> try commentP <|> try cppP)
-    (eof <|> (() <$ manyTill endOfLine letter))
+extensionsP = concat <$>
+    ( newLines *>
+      manyTill
+        (try singleExtensionsP <|> try optionsGhcP <|> try commentP <|> try cppP)
+        (eof <|> (() <$ manyTill endOfLine letter))
+    )
 
 {- | Single LANGUAGE pragma parser.
 
@@ -124,7 +127,7 @@ singleExtensionsP =
     languagePragmaP (commaSep (nonExtP *> extensionP <* nonExtP) <* spaces)
 
 nonExtP :: Parser ()
-nonExtP = () <$ many (try cppP <|> try commentP)
+nonExtP = skipMany (try cppP <|> try commentP)
 
 {- | Parses all known and unknown 'OnOffExtension's or 'SafeHaskellExtension's.
 -}
@@ -165,7 +168,7 @@ pragmaP pragmaNameP p = between
 
 -- | Comma separated parser. Newlines and spaces are allowed around comma.
 commaSep :: Parser a -> Parser [a]
-commaSep p = p `sepBy1` (try $ newLines *> char ',' <* newLines)
+commaSep p = p `sepBy1` try (newLines *> char ',' <* newLines)
 
 {- | Haskell comment parser.
 Supports both single-line comments:
@@ -211,4 +214,4 @@ cppP =
 
 -- | Any combination of spaces and newlines.
 newLines :: Parser ()
-newLines = () <$ many (space <|> endOfLine)
+newLines = skipMany (space <|> endOfLine)
diff --git a/test/Test/Extensions/Module.hs b/test/Test/Extensions/Module.hs
index 4fab2e2..3f57b65 100644
--- a/test/Test/Extensions/Module.hs
+++ b/test/Test/Extensions/Module.hs
@@ -176,6 +176,22 @@ multiLineCommentsSpec = describe "Parsing extensions with multi-line comments" $
     itShouldParse
         "{- Haskelll -} {-# {- extension -} LANGUAGE {- called -} LambdaCase {- and nothing else -} #-} {- matters -}"
         [LambdaCase]
+    itShouldParse (unlines
+        [ ""
+        , "{- |"
+        , "Hello"
+        , "-}"
+        , "{-# LANGUAGE CPP #-}"
+        ])
+        [Cpp]
+    itShouldParse (unlines
+        [ ""
+        , "-- | Hey"
+        , "-- Hello"
+        , "--"
+        , "module Abc"
+        ])
+        []
     itShouldParse (unlines
         [ "{-# LANGUAGE   TypeApplications  , LambdaCase   #-}"
         , "{- hello -}"