Email directory scanner/ logger

#!/bin/bash

MAILDIR="$HOME/Maildir"
JSON_OUTPUT="$HOME/URL4RM.json"
CSV_OUTPUT="$HOME/URL4RM.csv"
TMP_FILE="$(mktemp)"

PATTERNS=("escapechan" "tor2web" "i2p" "onion" "darkweb" "dread")

# Clear previous output
> "$TMP_FILE"

# Check for required tools
if ! command -v urlscan &>/dev/null; then
    echo "urlscan not found. Please install it."
    exit 1
fi

# Function to escape JSON strings
escape_json() {
    echo "$1" | sed 's/\\/\\\\/g; s/"/\\"/g'
}

# Process emails and extract URLs
find "$MAILDIR" -type f | while read -r email_file; do
    timestamp=$(stat -c %y "$email_file" | cut -d'.' -f1)
    urlscan -n < "$email_file" 2>/dev/null | while read -r url; do
        clean_url=$(echo "$url" | sed 's/["><=]*$//')

        if [[ "$clean_url" =~ [[:alnum:]\._%+-]+@[[:alnum:]\.-]+\.[[:alpha:]]{2,} ]]; then
            continue
        fi

        for pattern in "${PATTERNS[@]}"; do
            if [[ "$clean_url" == *"$pattern"* ]]; then
                echo -e "$timestamp\t$email_file\t$clean_url\t$pattern" >> "$TMP_FILE"
                break
            fi
        done
    done
done

# Deduplicate by URL
awk -F'\t' '!seen[$3]++' "$TMP_FILE" > "${TMP_FILE}_dedup"

# Write CSV
echo "timestamp,email_file,url,pattern" > "$CSV_OUTPUT"
awk -F'\t' '{print "\"" $1 "\",\"" $2 "\",\"" $3 "\",\"" $4 "\""}' "${TMP_FILE}_dedup" >> "$CSV_OUTPUT"

# Write JSON
echo "[" > "$JSON_OUTPUT"
awk -F'\t' '
    BEGIN { first = 1 }
    {
        if (!first) { print "," } else { first = 0 }
        printf "  {\n"
        printf "    \"timestamp\": \"%s\",\n", $1
        printf "    \"email_file\": \"%s\",\n", $2
        printf "    \"url\": \"%s\",\n", $3
        printf "    \"pattern\": \"%s\"\n", $4
        printf "  }"
    }
END { print "\n]" }
' "${TMP_FILE}_dedup" >> "$JSON_OUTPUT"

rm "$TMP_FILE" "${TMP_FILE}_dedup"

echo "JSON output: $JSON_OUTPUT"
echo "CSV output: $CSV_OUTPUT"