#!/bin/bash
# download_wikipedia.sh
# This script downloads a daily backup of the latest Wikipedia archive
# from kiwix.org, then (on schedule) promotes backups to weekly, monthly,
# quarterly, and yearly folders. Finally, it deletes old backups according
# to the specified retention rules.
#
# Retention rules:
# - Daily backups older than 7 days are deleted.
# - Weekly backups older than 31 days are deleted.
# - Monthly backups older than 93 days are deleted.
# - Quarterly backups older than 365 days are deleted.

set -e

#################################
# Directories for Backups
#################################
DAILY_DIR="$HOME/wikipedia_archives/daily"
WEEKLY_DIR="$HOME/wikipedia_archives/weekly"
MONTHLY_DIR="$HOME/wikipedia_archives/monthly"
QUARTERLY_DIR="$HOME/wikipedia_archives/quarterly"
YEARLY_DIR="$HOME/wikipedia_archives/yearly"

# Ensure all backup directories exist.
for dir in "$DAILY_DIR" "$WEEKLY_DIR" "$MONTHLY_DIR" "$QUARTERLY_DIR" "$YEARLY_DIR"; do
    mkdir -p "$dir"
done

#################################
# Define the URL of the Wikipedia archive
#################################
URL=https://download.kiwix.org/zim/wikipedia_en_all.zim

# Get today's date (YYYY-MM-DD) and define the output filename.
TODAY=$(date +%F)
OUTPUT_FILE="wikipedia_en_all_${TODAY}.zim"

#################################
# Step 1: Daily Download
#################################
cd "$DAILY_DIR" || { echo "Failed to change directory to $DAILY_DIR"; exit 1; }
wget -O "$OUTPUT_FILE" "$URL"
echo "$(date '+%F') - Daily download: Saved as ${OUTPUT_FILE}"

#################################
# Step 2: Weekly Backup (On Sundays)
#################################
# date +%u returns 7 on Sunday.
if [ "$(date +%u)" -eq 7 ]; then
    OLDEST_DAILY=$(ls -1tr "$DAILY_DIR"/wikipedia_en_all_*.zim 2>/dev/null | head -n 1)
    if [ -n "$OLDEST_DAILY" ]; then
        cp "$OLDEST_DAILY" "$WEEKLY_DIR"
        echo "$(date '+%F') - Weekly backup: Copied $(basename "$OLDEST_DAILY") to weekly directory."
    else
        echo "$(date '+%F') - Weekly backup: No daily backup file found."
    fi
fi

#################################
# Helper Function: Is Last Day of Month?
#################################
is_last_day_of_month() {
    # If tomorrow's day is 01 then today is the last day of the month.
    [ "$(date -d tomorrow +%d)" = "01" ]
}

#################################
# Step 3: Monthly Backup (On Last Day of Month)
#################################
if is_last_day_of_month; then
    OLDEST_WEEKLY=$(ls -1tr "$WEEKLY_DIR"/wikipedia_en_all_*.zim 2>/dev/null | head -n 1)
    if [ -n "$OLDEST_WEEKLY" ]; then
        cp "$OLDEST_WEEKLY" "$MONTHLY_DIR"
        echo "$(date '+%F') - Monthly backup: Copied $(basename "$OLDEST_WEEKLY") to monthly directory."
    else
        echo "$(date '+%F') - Monthly backup: No weekly backup file found."
    fi
fi

#################################
# Step 4: Quarterly Backup (On Last Day of Month for Quarter-Ending Months)
#################################
if is_last_day_of_month; then
    MONTH=$(date +%m)
    if [[ "$MONTH" == "03" || "$MONTH" == "06" || "$MONTH" == "09" || "$MONTH" == "12" ]]; then
        OLDEST_MONTHLY=$(ls -1tr "$MONTHLY_DIR"/wikipedia_en_all_*.zim 2>/dev/null | head -n 1)
        if [ -n "$OLDEST_MONTHLY" ]; then
            cp "$OLDEST_MONTHLY" "$QUARTERLY_DIR"
            echo "$(date '+%F') - Quarterly backup: Copied $(basename "$OLDEST_MONTHLY") to quarterly directory."
        else
            echo "$(date '+%F') - Quarterly backup: No monthly backup file found."
        fi
    fi
fi

#################################
# Step 5: Yearly Backup (On December 31)
#################################
if [ "$(date +%m-%d)" = "12-31" ]; then
    OLDEST_QUARTERLY=$(ls -1tr "$QUARTERLY_DIR"/wikipedia_en_all_*.zim 2>/dev/null | head -n 1)
    if [ -n "$OLDEST_QUARTERLY" ]; then
        cp "$OLDEST_QUARTERLY" "$YEARLY_DIR"
        echo "$(date '+%F') - Yearly backup: Copied $(basename "$OLDEST_QUARTERLY") to yearly directory."
    else
        echo "$(date '+%F') - Yearly backup: No quarterly backup file found."
    fi
fi

#################################
# Step 6: Deletion Operations (Run from Daily to Quarterly)
#################################
# Delete daily files older than 7 days.
find "$DAILY_DIR" -type f -name "wikipedia_en_all_*.zim" -mtime +7 -delete
# Delete weekly files older than 31 days.
find "$WEEKLY_DIR" -type f -name "wikipedia_en_all_*.zim" -mtime +31 -delete
# Delete monthly files older than 93 days.
find "$MONTHLY_DIR" -type f -name "wikipedia_en_all_*.zim" -mtime +93 -delete
# Delete quarterly files older than 365 days.
find "$QUARTERLY_DIR" -type f -name "wikipedia_en_all_*.zim" -mtime +365 -delete

echo "$(date '+%F') - All backup operations completed."