package findlinks

import (
	"fmt"
	"io"
	"strings"

	"golang.org/x/net/html"
	"golang.org/x/net/html/atom"
)

// A Link encapsulates the data harvested from a link.
type Link struct {
	Href string
	Text string
}

// FindLinks consumes the given [io.Reader], scraping it of anchor
// tags. Each anchor tag is "unmarshalled" into a [Link]. The
// resulting slice of Links is returned, along with an error.
func FindLinks(r io.Reader) ([]Link, error) {
	doc, err := html.Parse(r)
	if err != nil {
		return nil, fmt.Errorf("can't parse html reader: %w", err)
	}

	links := iterHTML(doc, nil)

	return links, nil
}

func iterHTML(n *html.Node, buffer []Link) []Link {
	if n.Type == html.ElementNode && n.DataAtom == atom.A {
		var link Link

		// Href
		link.Href = extractHref(n)

		// Text
		chunks := extractText(n, nil)
		link.Text = strings.Join(chunks, " ")

		buffer = append(buffer, link)
	} else {
		for c := n.FirstChild; c != nil; c = c.NextSibling {
			buffer = iterHTML(c, buffer)
		}
	}

	return buffer
}

// extractHref returns the first href attribute of anchor.
func extractHref(anchor *html.Node) string {
	var href string

	for _, a := range anchor.Attr {
		if a.Key == atom.Href.String() {
			href = a.Val
			break
		}
	}

	return href
}

// extractText recursively scans anchor to return the various nested
// pieces of text content.
func extractText(anchor *html.Node, buffer []string) []string {
	for c := anchor.FirstChild; c != nil; c = c.NextSibling {
		switch c.Type {
		case html.TextNode:
			buffer = append(buffer, c.Data)
		case html.ElementNode:
			buffer = extractText(c, buffer)
		}
	}

	return buffer
}