From df2adeb3169fe8e813117da7aad2e6030ecb5da6 Mon Sep 17 00:00:00 2001
From: "D. Bohdan" <dbohdan@dbohdan.com>
Date: Mon, 20 May 2024 12:11:27 +0000
Subject: [PATCH] feat!: throw an error on UTF-16 or UTF-32 BOM

BREAKING CHANGE: UTF-16 and UTF-32 files that have the byte-order mark
are no longer passed through with `-p`/`--pass-through`.

v0.16.0
---
 README.md   |  7 +++++--
 VERSION     |  2 +-
 initool.sml | 42 ++++++++++++++++++++++++++++++++++++------
 3 files changed, 42 insertions(+), 9 deletions(-)

diff --git a/README.md b/README.md
index abced9f..a046426 100644
--- a/README.md
+++ b/README.md
@@ -220,8 +220,11 @@ Commands act on all of them at the same time.
 ### Text encodings
 
 Initool is encoding-naive and assumes one character is one byte.
-It correctly processes UTF-8-encoded files when given UTF-8 command-line arguments but can't open files in UTF-16 or UTF-32.
-On Windows, it will receive the command-line arguments in the encoding for your system's language for non-Unicode programs (e.g., [Windows-1252](https://en.wikipedia.org/wiki/Windows-1252)),
+It correctly processes UTF-8-encoded files when given UTF-8 command-line arguments.
+It exits with an encoding error if it detects the UTF-16 or UTF-32 [BOM](https://en.wikipedia.org/wiki/Byte_order_mark).
+Trying to open a UTF-16 or UTF-32 file without the BOM results in an "invalid line" error because initool fails to parse it.
+
+On Windows, initool will receive the command-line arguments in the encoding for your system's language for non-Unicode programs (e.g., [Windows-1252](https://en.wikipedia.org/wiki/Windows-1252)),
 which limits what you can do with UTF-8-encoded files.
 
 
diff --git a/VERSION b/VERSION
index a551051..04a373e 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-0.15.0
+0.16.0
diff --git a/initool.sml b/initool.sml
index 8639939..5489cb7 100644
--- a/initool.sml
+++ b/initool.sml
@@ -3,9 +3,29 @@
  * License: MIT
  *)
 
-type options = {ignoreCase: bool, passThrough: bool}
+exception Encoding of string
 
-fun idOptions (opts: options) : Id.options = {ignoreCase = #ignoreCase opts}
+val unsupportedEncoding = "unsupported encoding: "
+
+fun checkWrongEncoding (lines: string list) =
+  let
+    val _ =
+      case lines of
+        [first] =>
+          (case map Char.ord (String.explode first) of
+             0x00 :: 0x00 :: 0xFE :: 0xFF :: _ =>
+               raise Encoding (unsupportedEncoding ^ "UTF-32 BE")
+           | 0xFF :: 0xFE :: 0x00 :: 0x00 :: _ =>
+               raise Encoding (unsupportedEncoding ^ "UTF-32 LE")
+           | 0xFE :: 0xFF :: _ =>
+               raise Encoding (unsupportedEncoding ^ "UTF-16 BE")
+           | 0xFF :: 0xFE :: _ =>
+               raise Encoding (unsupportedEncoding ^ "UTF-16 LE")
+           | _ => ())
+      | _ => ()
+  in
+    lines
+  end
 
 fun readLines (filename: string) : string list =
   let
@@ -44,7 +64,8 @@ datatype result = Output of string | FailureOutput of string | Error of string
 
 fun processFileCustom quiet passThrough successFn filterFn filename =
   let
-    val parsed = Ini.parse passThrough (readLines filename)
+    val parsed =
+      ((Ini.parse passThrough) o checkWrongEncoding o readLines) filename
     val filtered = filterFn parsed
     val success = successFn (parsed, filtered)
     val output = if quiet then "" else Ini.stringify filtered
@@ -100,13 +121,17 @@ fun helpCommand [] = Output allUsage
       Error (invalidUsage ^ (formatArgs (cmd :: rest)) ^ "\n" ^ usage ^ cmd)
 
 fun versionCommand [] =
-      let val version = "0.15.0"
+      let val version = "0.16.0"
       in Output (version ^ "\n")
       end
   | versionCommand [_] = versionCommand []
   | versionCommand (cmd :: rest) =
       Error (invalidUsage ^ (formatArgs (cmd :: rest)) ^ "\n" ^ usage ^ cmd)
 
+type options = {ignoreCase: bool, passThrough: bool}
+
+fun idOptions (opts: options) : Id.options = {ignoreCase = #ignoreCase opts}
+
 fun getCommand (opts: options) [_, filename] =
       processFile (#passThrough opts) (fn _ => true) (fn x => x) filename
   | getCommand opts [_, filename, section] =
@@ -139,7 +164,7 @@ fun getCommand (opts: options) [_, filename] =
         val q = Ini.SelectProperty {section = section, key = key}
         val parsed =
           ((Ini.select (idOptions opts) q) o (Ini.parse (#passThrough opts))
-           o readLines) filename
+           o checkWrongEncoding o readLines) filename
         val allItems = List.concat
           (List.map (fn {name = _, contents = xs} => xs) parsed)
         val values =
@@ -275,11 +300,16 @@ fun processArgs (opts: options) [] = helpCommand []
   | processArgs opts (cmd :: _) =
       Error (unknownCommand ^ (formatArgs [cmd]) ^ "\n" ^ availableCommands)
 
+fun handleException (message: string) =
+  exitWithError "" ("Error: " ^ message)
+
 val args = CommandLine.arguments ()
 
 val result =
   processArgs {ignoreCase = false, passThrough = false} args
-  handle Ini.Tokenization (message) => exitWithError "" ("Error: " ^ message)
+  handle
+    Encoding message => handleException message
+  | Ini.Tokenization message => handleException message
 val _ =
   case result of
     Output s => printFlush TextIO.stdOut s