Fix EditorConfig section priority, glob depth matching, and root=true search stop

Co-authored-by: Freed-Wu <32936898+Freed-Wu@users.noreply.github.com>

Author: Copilot

shellcheck.hs

@@ -597,11 +597,12 @@ ioInterface options files = do
             | otherwise = dir ++ "/"
 
         collectDirConfigs dir = do
-            let next = takeDirectory dir
-            rest <- if next /= dir
+            current <- readConfig (dir </> ".editorconfig")
+            let isRoot = maybe False (isEditorConfigRoot . snd) current
+                next = takeDirectory dir
+            rest <- if next /= dir && not isRoot
                     then collectDirConfigs next
                     else return []
-            current <- readConfig (dir </> ".editorconfig")
             return $ maybeToList current ++ rest
 
         readGlobalEditorConfig = do

src/ShellCheck/EditorConfig.hs

@@ -24,7 +24,7 @@
 -- of sections whose glob matches the file being checked are extracted,
 -- and turned into the same "key=value" directive syntax that is used in
 -- .shellcheckrc files.
-module ShellCheck.EditorConfig (parseEditorConfig, globToRegexString, runTests) where
+module ShellCheck.EditorConfig (parseEditorConfig, isEditorConfigRoot, globToRegexString, runTests) where
 
 import Data.Char
 import Data.List
@@ -38,10 +38,20 @@ import Test.QuickCheck
 -- file being checked, return the shellcheck directives (as a
 -- "key=value\n" delimited blob, suitable for feeding into the same
 -- parser as .shellcheckrc) found in matching sections.
+--
+-- As per the EditorConfig spec, files are read top to bottom and
+-- properties from later sections override those from earlier ones
+-- (for the same key), so on conflicts the last matching section wins.
 parseEditorConfig :: String -> FilePath -> String
 parseEditorConfig contents name =
-    unlines . concatMap sectionDirectives $ sections
+    unlines . map render . lastWins . concatMap sectionDirectives $ sections
   where
+    render (key, value) = key ++ "=" ++ value
+
+    -- Keep only the last occurrence of each key, preserving the
+    -- relative order of the remaining (first-seen) entries.
+    lastWins = reverse . nubBy (\a b -> fst a == fst b) . reverse
+
     ls = lines contents
     sections = splitSections ls
 
@@ -71,24 +81,48 @@ parseEditorConfig contents name =
                 let key' = trim key
                     value' = trim value
                 in if "shellcheck." `isPrefixOf` key'
-                    then Just (drop (length "shellcheck.") key' ++ "=" ++ value')
+                    then Just (drop (length "shellcheck.") key', value')
                     else Nothing
             _ -> Nothing
 
     stripComment = takeWhile (\c -> c /= '#' && c /= ';')
 
+-- Does the top-level (pre-section) part of an EditorConfig file
+-- declare "root = true"? Per the spec, this stops the search for
+-- further EditorConfig files in parent directories.
+isEditorConfigRoot :: String -> Bool
+isEditorConfigRoot contents =
+    any isRootTrue . takeWhile (not . isSectionHeader) $ lines contents
+  where
+    isSectionHeader l =
+        case trim (stripComment l) of
+            ('[':cs@(_:_)) -> last cs == ']'
+            _ -> False
+
+    isRootTrue l =
+        case break (== '=') (trim (stripComment l)) of
+            (key, '=':value) ->
+                map toLower (trim key) == "root" && map toLower (trim value) == "true"
+            _ -> False
+
+    stripComment = takeWhile (\c -> c /= '#' && c /= ';')
+
 trim :: String -> String
 trim = dropWhileEnd isSpace . dropWhile isSpace
 
--- Does the (basename of the) file match the given EditorConfig glob?
+-- Does the (relative path of the) file match the given EditorConfig glob?
 matchesGlob :: String -> FilePath -> Bool
 matchesGlob pattern name =
     name `matches` mkRegex (globToRegexString pattern)
 
 -- Translate an EditorConfig glob pattern into an anchored regex string.
+-- Per the spec, patterns without a path separator are matched against
+-- the file at any depth (as if prefixed with "**/").
 globToRegexString :: String -> String
-globToRegexString pattern = "^" ++ go pattern ++ "$"
+globToRegexString pattern = "^" ++ prefix ++ go pattern ++ "$"
   where
+    prefix = if '/' `elem` pattern then "" else "(.*/)?"
+
     go [] = ""
     go ('*':'*':rest) = ".*" ++ go rest
     go ('*':rest) = "[^/]*" ++ go rest
@@ -129,6 +163,13 @@ prop_globNoMatch = not $ matchesGlob "*.ebuild" "foo.txt"
 prop_globQuestion = matchesGlob "foo?.sh" "food.sh"
 prop_globClass = matchesGlob "foo[0-9].sh" "foo1.sh"
 prop_globClassNeg = not $ matchesGlob "foo[!0-9].sh" "foo1.sh"
+-- Patterns without a path separator should match at any depth.
+prop_globAnyDepth = matchesGlob "*.sh" "sub/dir/foo.sh"
+prop_globAnyDepthPlain = matchesGlob "foo" "sub/foo"
+-- Patterns with a path separator are only matched against the full
+-- relative path.
+prop_globWithSlashNoMatch = not $ matchesGlob "sub/*.sh" "other/foo.sh"
+prop_globWithSlashMatch = matchesGlob "sub/*.sh" "sub/foo.sh"
 
 prop_parseEditorConfig1 =
     parseEditorConfig "[*.{ebuild,eclass}]\nshellcheck.shell=bash\nshellcheck.disable=SC2034\n" "foo.ebuild"
@@ -140,6 +181,15 @@ prop_parseEditorConfig3 =
 prop_parseEditorConfig4 =
     parseEditorConfig "root = true\n[*.sh]\nindent_style = space\nshellcheck.shell=bash\n" "foo.sh"
         == "shell=bash\n"
+-- A later, more specific section overrides an earlier, more general
+-- one for the same key.
+prop_parseEditorConfig5 =
+    parseEditorConfig "[*]\nshellcheck.shell=sh\n\n[foo]\nshellcheck.shell=bash\n" "foo"
+        == "shell=bash\n"
+-- Non-conflicting keys from earlier and later sections are all kept.
+prop_parseEditorConfig6 =
+    parseEditorConfig "[*]\nshellcheck.shell=sh\n\n[foo]\nshellcheck.disable=SC2034\n" "foo"
+        == "shell=sh\ndisable=SC2034\n"
 
 return []
 runTests = $quickCheckAll