Summary

Issue #47 (closed) fixed the UnicodeEncodeError crash caused by box-drawing characters in searcher.py on Windows. The same class of crash is still present in several other modules that print ✓ (U+2713) or other non-ASCII glyphs, and there is no sys.stdout.reconfigure(encoding="utf-8", ...) anywhere in the package.

Reproduction

Windows 11, Python 3.13, default console encoding cp1251, mempalace 3.0.14.

# Crash 1: mempalace init
mempalace init --yes D:/path/to/project

Crashes while printing detected-entities confirmation block (Cyrillic content from project docs):

File "C:\Program Files\Python313\Lib\encodings\cp1251.py", line 19, in encode
    return codecs.charmap_encode(input,self.errors,encoding_table)[0]
UnicodeEncodeError: 'charmap' codec can't encode characters in position 30-34: character maps to <undefined>

# Crash 2: mempalace mine (project mode)
mempalace mine D:/path/to/project --wing vibeOps

  print(f"  ✓ [{i:4}/{len(files)}] {filepath.name[:50]:50} +{drawers}")
  File "cp1251.py", line 19, in encode
    return codecs.charmap_encode(input,self.errors,encoding_table)[0]
UnicodeEncodeError: 'charmap' codec can't encode character '\u2713' in position 2: character maps to <undefined>

# Crash 3: mempalace mine --mode convos
mempalace mine C:/Users/.../.claude/projects/... --mode convos --extract general

Same \u2713 crash from convo_miner.py:359.

Affected print sites (U+2713 checkmark)

• mempalace/miner.py:596 — print(f" ✓ [{i:4}/{len(files)}] ...")
• mempalace/convo_miner.py:359 — print(f" ✓ [{i:4}/{len(files)}] ...")
• mempalace/split_mega_files.py:227 — print(f" ✓ {name} ...")

Plus anywhere in entity_detector.py that prints user project text (Cyrillic / non-ASCII file contents) during init's confirmation step.

grep reconfigure\(encoding .../mempalace/ returns zero hits — the #47 fix is not applied anywhere in the installed package.

Workaround

set PYTHONIOENCODING=utf-8
set PYTHONUTF8=1

Both commands then succeed. But this is a bad UX — the tool ships crashing on default Windows consoles for any user whose project directory contains non-ASCII text or any mining run that prints a progress checkmark (i.e. all of them).

Proposed fix

Call this once at the top of cli.py::main() (so every subcommand benefits):

import sys
if sys.platform == "win32" and hasattr(sys.stdout, "reconfigure"):
    sys.stdout.reconfigure(encoding="utf-8", errors="replace")
    sys.stderr.reconfigure(encoding="utf-8", errors="replace")

This is exactly the suggested fix from #47 but applied at the CLI entry point instead of inside one module.

Environment

• Windows 11, Python 3.13.x
• mempalace 3.0.14
• Console: default Windows Terminal (cp1251 on Russian locale, cp1252 on en-US)