From 46fb91b44d2784829a5949e1faeaedaea59e76a8 Mon Sep 17 00:00:00 2001
From: Vidar Holen <spam@vidarholen.net>
Date: Sun, 22 Jan 2017 15:24:21 -0800
Subject: [PATCH] Manually decode input files as lenient UTF-8.

---
 shellcheck.1.md |  9 +++++++++
 shellcheck.hs   | 41 ++++++++++++++++++++++++++++++++++++++---
 2 files changed, 47 insertions(+), 3 deletions(-)

diff --git a/shellcheck.1.md b/shellcheck.1.md
index 8e76e3c..00ef455 100644
--- a/shellcheck.1.md
+++ b/shellcheck.1.md
@@ -175,6 +175,15 @@ ShellCheck uses the follow exit codes:
 + 3: ShellCheck was invoked with bad syntax (e.g. unknown flag).
 + 4: ShellCheck was invoked with bad options (e.g. unknown formatter).
 
+# LOCALE
+This version of ShellCheck is only available in English. All files are
+leniently decoded as UTF-8, with a fallback of ISO-8859-1 for invalid
+sequences. `LC_CTYPE` is respected for output, and defaults to UTF-8 for
+locales where encoding is unspecified (such as the `C` locale).
+
+Windows users seeing `commitBuffer: invalid argument (invalid character)`
+should set their terminal to use UTF-8 with `chcp 65001`.
+
 # AUTHOR
 ShellCheck is written and maintained by Vidar Holen.
 
diff --git a/shellcheck.hs b/shellcheck.hs
index 392d271..4a40ba2 100644
--- a/shellcheck.hs
+++ b/shellcheck.hs
@@ -31,6 +31,7 @@ import qualified ShellCheck.Formatter.TTY
 import Control.Exception
 import Control.Monad
 import Control.Monad.Except
+import Data.Bits
 import Data.Char
 import Data.Functor
 import Data.Either
@@ -288,14 +289,48 @@ ioInterface options files = do
         fallback path _ = return path
 
 inputFile file = do
-    contents <-
+    handle <-
             if file == "-"
-            then getContents
-            else readFile file
+            then return stdin
+            else openBinaryFile file ReadMode
+
+    hSetBinaryMode handle True
+    contents <- decodeString <$> hGetContents handle -- closes handle
 
     seq (length contents) $
         return contents
 
+-- Decode a char8 string into a utf8 string, with fallback on
+-- ISO-8859-1. This avoids depending on additional libraries.
+decodeString = decode
+  where
+    decode [] = []
+    decode (c:rest) | isAscii c = c : decode rest
+    decode (c:rest) =
+        let num = (fromIntegral $ ord c) :: Int
+            next = case num of
+                _ | num >= 0xFF -> Nothing
+                  | num >= 0xFE -> construct (num .&. 0x00) 6 rest
+                  | num >= 0xFC -> construct (num .&. 0x01) 5 rest
+                  | num >= 0xF8 -> construct (num .&. 0x03) 4 rest
+                  | num >= 0xF0 -> construct (num .&. 0x07) 3 rest
+                  | num >= 0xE0 -> construct (num .&. 0x0F) 2 rest
+                  | num >= 0xC0 -> construct (num .&. 0x1F) 1 rest
+                  | True -> Nothing
+        in
+            case next of
+                Just (n, remainder) -> chr n : decode remainder
+                Nothing -> c : decode rest
+
+    construct x 0 rest = return (x, rest)
+    construct x n (c:rest) =
+        let num = (fromIntegral $ ord c) :: Int in
+            if num >= 0x80 && num <= 0xBF
+            then construct ((x `shiftL` 6) .|. (num .&. 0x3f)) (n-1) rest
+            else Nothing
+    construct _ _ _ = Nothing
+
+
 verifyFiles files =
     when (null files) $ do
         printErr "No files specified.\n"