package main

import (
	"bufio"
	"bytes"
	"regexp"

	"github.com/asaskevich/govalidator"

	"gitlab.smoothnet.org/nhyatt/bind-response-policy-zone-creator/internal/log"
)

func parseAdBlock(data []byte) []string {
	var domains []string

	// convert data to reader for line-by-line reading
	r := bytes.NewReader(data)

	// process combined files line-by-line
	scanner := bufio.NewScanner(r)
	for scanner.Scan() {
		line := scanner.Text()

		// skip lines with the AdBlock header
		if regexp.MustCompile(`^\[Adblock Plus\]`).MatchString(line) {
			continue
		}

		// skip lines where the first non-whitespace character is '#' or '//'
		if regexp.MustCompile(`^(\s+)?(#|\/\/|\!)`).MatchString(line) {
			continue
		}

		// skip lines with no characters and/or whitespace only
		if regexp.MustCompile(`^(\s+)?$`).MatchString(line) {
			continue
		}

		// remove line header
		d := regexp.MustCompile(`^\|\|`).ReplaceAllString(line, "")

		// remove line footer
		d = regexp.MustCompile(`\^$`).ReplaceAllString(d, "")

		if govalidator.IsDNSName(d) {
			domains = append(domains, d)
		} else {
			log.Debug("host invalid", "host", d)
		}
	}

	return domains
}