onetxt

Break through chat-AI attachment limits - merge a directory of text files into one clean context file for easy ingestion with local or hosted LLMs. Designed for simplicity and ease of use.

If you need splitting instead of merging, check this tool.

Overview

onetxt scans the current directory for a glob pattern (default *.txt), concatenates matches, ensures clean trailing newlines, and skips the output file to avoid self‑append loops. It’s a single‑file, zero‑dependency Python CLI using standard library modules like argparse and pathlib for portability.

Features

• Merge all files matching a glob while excluding the output filename if present.
• Stream line by line to handle large files without high memory usage.
• Auto‑add a newline if a file’s last line lacks one, and insert a blank line between documents.
• Control input encoding and decode behavior via --encoding and --errors.

Installation

Place onetxt.py with the source files and run with Python.

Usage

Windows

• Double‑click: on Windows, double‑clicking onetxt.py runs with defaults.

• Basic (PowerShell/CMD): run from the folder containing the files.

python onetxt.py

• Custom output and pattern with globbing.

python onetxt.py --out all_notes.txt --pattern "*.txt"

• Merge Markdown with strict decoding and no pause (for scripts/CI).

python onetxt.py --out dataset.md --pattern "*.md" --encoding utf-8 --errors strict --no-pause

Others (macOS/Linux)

• Basic: run from the folder containing the files.

python3 onetxt.py

• Custom output and pattern.

python3 onetxt.py --out all_notes.txt --pattern "*.txt"

• Merge Markdown with strict decoding and no pause (for scripts/CI).

python3 onetxt.py --out dataset.md --pattern "*.md" --encoding utf-8 --errors strict --no-pause

Formatting example

• Example output structure (merged_file.txt) showing per‑document wrappers suitable for chunking and retrieval.

<context>
  <doc id="1" name="file1">
contents1
  </doc>

  <doc id="2" name="file2">
contents2
  </doc>

  <doc id="3" name="file3">
contents3
  </doc>
</context>

• Each document is wrapped in a element with id/name derived from the base filename, and documents are separated by blank lines for clean boundaries.

Options

• --out: Output filename, default merged_file.txt
• --pattern: Glob for inputs, default *.txt
• --encoding: Input encoding, default utf‑8
• --errors: strict | ignore | replace (default replace)
• --no-pause: Do not prompt before exit (useful for scripted runs)

Notes

onetxt writes a blank line between documents and guarantees each file ends with a newline so downstream chunkers and parsers see clean boundaries on the merged output.