Why Your Current Approach Is Slow #

Your intuition is spot-on. Running find in a loop for every filename means:

• Each find process starts fresh and scans both ~/dir1 and ~/dir2 from the root.
• If you have N files, you're doing N full directory traversals—total waste of time and system resources.
• On top of that, ls | while read name is unsafe for filenames containing spaces, newlines, or special characters (like * or ?), which can break your script unexpectedly.

Optimized Solutions #

All the fixes below reduce the number of directory scans to one, which is the biggest win for speed. Here are the best options, ordered by safety and ease of use:

1. Use a Bash Associative Array for Safe, Fast Filtering #

This method handles all filename types (spaces, special chars) and only runs find once:

# Step 1: Create a lookup set of your target filenames
declare -A target_files
for name in *; do
  target_files["$name"]=1
done

# Step 2: Find all files in dir1/dir2, then filter against the set
find ~/dir1 ~/dir2 -type f -exec bash -c '
  for file do
    # Extract the filename from the full path
    basename="${file##*/}"
    # Check if it exists in our target set
    if [[ -n "${target_files[$basename]}" ]]; then
      echo "$file"
    fi
  done' bash {} +

The -exec ... {} + passes as many files as possible to each bash instance, minimizing process overhead.

2. Single find Command with Multiple -name Patterns #

If you don't need to handle super tricky filenames, you can build a single find command that checks all your target names at once:

# Generate a list of -name patterns, then run find once
find ~/dir1 ~/dir2 \( $(printf "-name '%s' -o " *) | sed 's/ -o $//' \)

• The printf creates a string like -name 'file1' -o -name 'file2' -o ...
• sed removes the trailing -o to avoid a syntax error
• The parentheses () group all the -name conditions so find treats them as a single OR list

3. Pattern File for Extra Safety #

If your filenames have quotes or backslashes, use a temporary file to store patterns:

# Write each filename as a -name line to a temp file
printf "-name '%s'\n" * > /tmp/find_patterns.txt
# Adjust to group OR conditions properly
sed -i '$ s/$/ -o /' /tmp/find_patterns.txt
echo "-false" >> /tmp/find_patterns.txt

# Run find with the pattern file
find ~/dir1 ~/dir2 -f /tmp/find_patterns.txt

# Clean up the temp file
rm /tmp/find_patterns.txt

This avoids command-line parsing issues with special characters.

Key Takeaway #

The core problem with your original script is redundant directory traversal. By running find once and checking all filenames in a single pass, you'll see a massive speed improvement—especially if you have a lot of target files or large directories.

内容的提问来源于stack exchange，提问作者ZhaoGang