main.py

import re
import os
from collections import Counter

# Common prefixes that indicate an automated/system email rather than a person
SYSTEM_PREFIXES = (
    "no-reply", "noreply", "donotreply", "do-not-reply",
    "notifications", "notification", "alerts", "alert",
    "automated", "system", "mailer", "postmaster",
    "bounce", "bounces", "newsletter",
)


def extract_emails(input_file):
    # Check if file exists
    if not os.path.exists(input_file):
        print(f"[!] File not found: {input_file}")
        return []

    with open(input_file, "r", encoding="utf-8") as f:
        content = f.read()

    # Regex pattern to match email addresses
    pattern = r'[a-zA-Z0-9._%+\-]+@[a-zA-Z0-9.\-]+\.[a-zA-Z]{2,}'
    emails = re.findall(pattern, content)

    # Remove duplicates while preserving order
    seen = set()
    unique_emails = []
    for email in emails:
        if email.lower() not in seen:
            seen.add(email.lower())
            unique_emails.append(email)

    return unique_emails


def categorize_email(email):
    """Return 'system' if the email looks automated, else 'personal'."""
    local_part = email.split("@")[0].lower()
    for prefix in SYSTEM_PREFIXES:
        if local_part.startswith(prefix) or local_part == prefix:
            return "system"
    return "personal"


def get_domain(email):
    return email.split("@")[1].lower()


def build_stats(emails):
    categories = Counter()
    domains = Counter()
    domain_categories = {}

    for email in emails:
        category = categorize_email(email)
        domain = get_domain(email)

        categories[category] += 1
        domains[domain] += 1
        domain_categories.setdefault(domain, []).append(category)

    return categories, domains, domain_categories


def save_emails(emails, output_file="extracted_emails.txt"):
    categories, domains, domain_categories = build_stats(emails)

    with open(output_file, "w", encoding="utf-8") as f:
        f.write("Email Extraction Report\n")
        f.write("=" * 40 + "\n")
        f.write(f"Total emails found: {len(emails)}\n")
        f.write("=" * 40 + "\n\n")

        # ── Category breakdown ──────────────────────────────────────────────
        f.write("Category Breakdown:\n")
        f.write(f"  Personal/Work emails: {categories['personal']}\n")
        f.write(f"  System/No-reply emails: {categories['system']}\n")
        f.write("\n")

        # ── Domain breakdown ─────────────────────────────────────────────────
        f.write("Domain Breakdown:\n")
        for domain, count in domains.most_common():
            f.write(f"  {domain}: {count}\n")
        f.write("\n")

        # ── Personal/Work emails ─────────────────────────────────────────────
        f.write("-" * 40 + "\n")
        f.write("Personal/Work Emails\n")
        f.write("-" * 40 + "\n")
        for email in emails:
            if categorize_email(email) == "personal":
                f.write(email + "\n")
        f.write("\n")

        # ── System/No-reply emails ──────────────────────────────────────────
        f.write("-" * 40 + "\n")
        f.write("System/No-reply Emails\n")
        f.write("-" * 40 + "\n")
        for email in emails:
            if categorize_email(email) == "system":
                f.write(email + "\n")

    return output_file


def main():
    print("=" * 40)
    print("  Email Extractor — Task Automation")
    print("=" * 40)

    input_file = input("\nEnter the .txt filename (e.g. sample.txt): ").strip()

    print(f"\n[*] Scanning {input_file} for email addresses...")
    emails = extract_emails(input_file)

    if not emails:
        print("[!] No email addresses found.")
        return

    print(f"[+] Found {len(emails)} unique email(s):")
    for email in emails:
        print(f"    {email}")

    categories, domains, _ = build_stats(emails)

    print("\n[*] Category Breakdown:")
    print(f"    Personal/Work: {categories['personal']}")
    print(f"    System/No-reply: {categories['system']}")

    print("\n[*] Top Domains:")
    for domain, count in domains.most_common(5):
        print(f"    {domain}: {count}")

    output_file = save_emails(emails)
    print(f"\n[✓] Saved to: {output_file}")


if __name__ == "__main__":
    main()