bash script for creating context info for LLMs

make-context.sh

#!/bin/bash

# --- Script Description ---
# Finds all text-based files in the current directory and its subdirectories,
# excluding dotfiles/dot-directories and specified filenames/directories,
# and prints their content in a Markdown-friendly format.

# --- Usage ---
# ./script_name.sh [-e file.txt] [-d node_modules] ...

# --- Initialization ---
# Array for file-based exclusions (uses -name)
FIND_EXCLUDE_FILES=(-not -name 'context.md')

# Array for path-based exclusions (uses -path for directories)
FIND_EXCLUDE_PATHS=()

# --- Option Parsing ---
# Use getopts to parse command-line options.
# -e: Exclude a file by its name.
# -d: Exclude a directory by its name/path.
while getopts ":e:d:" opt; do
  case ${opt} in
    e)
      # Exclude based on filename.
      FIND_EXCLUDE_FILES+=(-not -name "$OPTARG")
      ;;
    d)
      # Exclude based on path. The "*/.../*" pattern ensures we exclude
      # any file within a directory of that name, anywhere in the tree.
      # Example: -d 'node_modules' becomes -not -path '*/node_modules/*'
      FIND_EXCLUDE_PATHS+=(-not -path "*/$OPTARG/*")
      ;;
    \?)
      echo "Invalid Option: -$OPTARG" >&2
      exit 1
      ;;
    :)
      echo "Option -$OPTARG requires an argument." >&2
      exit 1
      ;;
  esac
done

# --- Main Execution ---
# The find command now uses both arrays to build its exclusion list.
find . \
  -type f \
  -not -path '*/.*' \
  "${FIND_EXCLUDE_FILES[@]}" \
  "${FIND_EXCLUDE_PATHS[@]}" \
  -exec sh -c '
    # For each found file, check if it is a text-based or empty file.
    if file "$1" | grep -qE "text|empty|JSON|XML|HTML"; then
      # Print the file contents in a Markdown code block with a header.
      echo "### File: $1"
      echo "´´´"
      cat "$1"
      printf "\n´´´\n"
    fi
' _ {} \;