From 87fc24e34aad273b0052115cfcfa9efe18a13af8 Mon Sep 17 00:00:00 2001
From: mbays <mbays@sdf.org>
Date: Sun, 28 May 2023 00:00:00 +0000
Subject: [PATCH] strip non-esc control chars

---
 ANSIColour.hs | 5 ++++-
 diohsc.hs     | 2 +-
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/ANSIColour.hs b/ANSIColour.hs
index 8a4feab..12e7198 100644
--- a/ANSIColour.hs
+++ b/ANSIColour.hs
@@ -28,7 +28,7 @@ module ANSIColour
     , visibleLength
     , splitAtVisible
     , escapePromptCSI
-    , sanitiseNonCSI
+    , sanitiseForDisplay
     , Colour(..)
     ) where
 
@@ -156,6 +156,9 @@ sanitiseNonCSI s =
             isCSI = T.take 1 post' == "["
         in pre <> (if isCSI then "\ESC" else "\\ESC") <> sanitiseNonCSI post'
 
+sanitiseForDisplay :: T.Text -> T.Text
+sanitiseForDisplay = sanitiseNonCSI . T.filter (\c -> c == '\ESC' || wcwidth c >= 0)
+
 -- |append \STX to each CSI sequence, as required in Haskeline prompts.
 -- See https://github.com/judah/haskeline/wiki/ControlSequencesInPrompt
 escapePromptCSI :: String -> String
diff --git a/diohsc.hs b/diohsc.hs
index a3eeab2..489f50a 100644
--- a/diohsc.hs
+++ b/diohsc.hs
@@ -1277,7 +1277,7 @@ handleCommandLine
                     Just cmd -> (T.lines . T.decodeUtf8With T.lenientDecode <$>) .
                         doRestrictedFilter (filterShell cmd []) . BL.concat . (appendNewline . T.encodeUtf8 <$>)
                         where appendNewline = (`BL.snoc` 10)
-            (Right <$>) . applyFilter . (sanitiseNonCSI <$>) $ case textType of
+            (Right <$>) . applyFilter . (sanitiseForDisplay <$>) $ case textType of
                     "gemini" ->
                         let opts = GemRenderOpts ansi' preOpt pageWidth linkDescFirst
                         in printGemDoc opts (showUriRefFull ansi' ais uri) $ parseGemini bodyText
-- 
2.11.4.GIT