From 5858baf199f0ca49a642f18b0ef27b50386b180f Mon Sep 17 00:00:00 2001
From: Dmitry Konstantinov <netudima@gmail.com>
Date: Mon, 22 Jun 2026 12:16:56 +0100
Subject: [PATCH] run_cqlsh.py: fix parsing of a multi-byte UTF-8 sequence,
 os.read()/pipe reads can split the sequence across two reads

patch by Dmitry Konstantinov; reviewed by Stefan Miklosovic for CASSANDRA-21466
---
 pylib/cqlshlib/test/run_cqlsh.py | 33 ++++++++++++++++++++++++--------
 1 file changed, 25 insertions(+), 8 deletions(-)

diff --git a/pylib/cqlshlib/test/run_cqlsh.py b/pylib/cqlshlib/test/run_cqlsh.py
index 288d3ed951..507e4d5337 100644
--- a/pylib/cqlshlib/test/run_cqlsh.py
+++ b/pylib/cqlshlib/test/run_cqlsh.py
@@ -19,6 +19,7 @@
 import os
 import sys
 import re
+import codecs
 import contextlib
 import subprocess
 import signal
@@ -120,6 +121,12 @@ class ProcRunner:
             env = {}
         self.env = env
         self.readbuf = ''
+        # os.read()/pipe reads can split a multi-byte UTF-8 sequence across two
+        # reads (the boundary depends on terminal/readline echo flushing, which
+        # changed with the Ubuntu 22.04 / readline 8.1 build image). Decode
+        # incrementally so a partial trailing sequence is buffered until the
+        # rest of its bytes arrive, instead of failing with UnicodeDecodeError.
+        self._decoder = codecs.getincrementaldecoder("utf-8")()
 
         self.start_proc()
 
@@ -166,16 +173,26 @@ class ProcRunner:
         self.proc.stdin.write(data)
 
     def read_tty(self, blksize, timeout=None):
-        buf = os.read(self.childpty, blksize)
-        if isinstance(buf, bytes):
-            buf = buf.decode("utf-8")
-        return buf
+        while True:
+            buf = os.read(self.childpty, blksize)
+            if not isinstance(buf, bytes):
+                return buf
+            decoded = self._decoder.decode(buf)
+            # A non-empty read that decodes to '' means we only got the start of
+            # a multi-byte UTF-8 sequence; keep reading until it completes rather
+            # than returning '', which callers (read_until) treat as EOF.
+            # An empty read (b'') is genuine EOF and must propagate as ''.
+            if decoded != '' or buf == b'':
+                return decoded
 
     def read_pipe(self, blksize, timeout=None):
-        buf = self.proc.stdout.read(blksize)
-        if isinstance(buf, bytes):
-            buf = buf.decode("utf-8")
-        return buf
+        while True:
+            buf = self.proc.stdout.read(blksize)
+            if not isinstance(buf, bytes):
+                return buf
+            decoded = self._decoder.decode(buf)
+            if decoded != '' or buf == b'':
+                return decoded
 
     def read_until(self, until, blksize=4096, timeout=None,
                    flags=0, ptty_timeout=None, replace=[]):