基于文件名日期删除30天前文件的技术求助（CSV格式文件名）

Got it, let's work through this problem. Since you want to base the deletion on the date in the filename (not the file's modification time, which is what find -mtime +30 uses), we can build a custom script to parse those dates and set up a recurring task to run it every 30 days.

Step 1: Build a Script to Delete Files Based on Filename Date #

First, we'll create a bash script that:

• Calculates the date 30 days prior to today
• Scans your CSV directory
• Extracts the date from each filename
• Compares it to the threshold and deletes old files

Example Script (for prefix_YYYYMMDD.csv format)

Let's assume your files look like sales_report_20240101.csv (date in YYYYMMDD format at the end, after an underscore). Adjust the date extraction logic to match your actual filename pattern.

#!/bin/bash

# Replace this with the path to your CSV files
TARGET_DIR="/home/youruser/csv_files"

# Calculate the threshold date (30 days ago) in YYYYMMDD format
THRESHOLD=$(date -d "30 days ago" +%Y%m%d)

# Loop through all CSV files in the target directory
for file in "$TARGET_DIR"/*.csv; do
    # Skip if the path isn't a valid file (e.g., no CSV files found)
    [ -f "$file" ] || continue

    # Get just the filename (without the full path)
    filename=$(basename "$file")

    # Extract the date from the filename (adjust this to match your pattern!)
    # For "sales_report_20240101.csv": remove everything before the last underscore, then remove .csv
    file_date=${filename#*_}
    file_date=${file_date%.csv}

    # Validate the extracted date is a valid 8-digit number (optional but safe)
    if ! [[ "$file_date" =~ ^[0-9]{8}$ ]]; then
        echo "Skipping $filename: invalid date format detected"
        continue
    fi

    # Compare dates: if the file's date is older than the threshold, delete it
    if [ "$file_date" -lt "$THRESHOLD" ]; then
        echo "Deleting old file: $file"
        # Uncomment the line below once you've tested and confirmed it works
        # rm "$file"
    fi
done

Adjusting for Other Date Formats

If your filenames use YYYY-MM-DD (like report-2024-01-01.csv), modify these parts:

1. Update the threshold to match the format:

   THRESHOLD=$(date -d "30 days ago" +%Y-%m-%d)
   

2. Extract the date with a regex to catch the hyphenated format:

   file_date=$(echo "$filename" | grep -oE '[0-9]{4}-[0-9]{2}-[0-9]{2}')
   

3. Compare using string comparison (since hyphenated dates are lex order-matching chronological order):

   if [[ "$file_date" < "$THRESHOLD" ]]; then
   

Step 2: Set Up a Recurring 30-Day Task #

Cron doesn't have a direct "every 30 days" option, but we have two reliable ways to set this up:

Option 1: Approximate Every 30 Days (Monthly Twice)

Run the script on the 1st and 21st of each month (roughly 30 days apart):

1. Open your crontab editor:

   crontab -e
   

2. Add this line (replace the script path with your actual file location):

   0 0 1,21 * * /path/to/your/delete_old_csvs.sh >> /var/log/old_csv_deletions.log 2>&1
   

   This runs at midnight (00:00) on the 1st and 21st, and logs output to a file for debugging.

Option 2: Strict 30-Day Intervals

If you need exact 30-day gaps, use a cron job that checks the elapsed time since the last run. Alternatively, use anacron (if your system supports it) which handles intervals better for non-24/7 systems.

Key Notes #

• Test First: Run the script manually first with the rm line commented out to verify it lists the correct files to delete.
• Permissions: Make the script executable with chmod +x /path/to/your/delete_old_csvs.sh.
• Special Characters: If your filenames have spaces or special characters, the script uses double quotes around variables to avoid breaking paths.
• Customize Extraction: Tweak the file_date extraction logic to match your specific filename pattern (e.g., if the date is at the start like 20240101_sales.csv, use file_date=${filename%_*} instead).

内容的提问来源于stack exchange，提问作者user9495954