package crawler

import (
	"context"
	"io"
	"net/http"
	"sync"

	"urlcrawler/internal/htmlx"
	"urlcrawler/internal/urlutil"
)

type task struct {
	url   string
	depth int
}

// Crawl visits pages up to maxDepth and returns visited set, per-URL errors, and per-page outgoing links.
// The visitedCallback and errorCallback functions are called when a page is successfully visited or encounters an error.
// visitedCallback receives the URL, its depth, and the current number of pending tasks in the queue.
// errorCallback receives the URL, the error, and the current number of pending tasks in the queue.
func Crawl(ctx context.Context, startURL string, maxDepth int, concurrency int, sameHostOnly bool, client *http.Client, userAgent string, visitedCallback func(string, int, int), errorCallback func(string, error, int)) (map[string]struct{}, map[string]error, map[string]map[string]struct{}) {
	visited := make(map[string]struct{})
	errs := make(map[string]error)
	outlinks := make(map[string]map[string]struct{})
	var mu sync.Mutex

	origin := urlutil.Origin(startURL)

	tasks := make(chan task, concurrency*2)
	wgWorkers := sync.WaitGroup{}
	wgTasks := sync.WaitGroup{}

	enqueue := func(t task) {
		wgTasks.Add(1)
		tasks <- t
	}

	worker := func() {
		defer wgWorkers.Done()
		for tk := range tasks {
			if ctx.Err() != nil {
				wgTasks.Done()
				return
			}
			mu.Lock()
			if _, seen := visited[tk.url]; seen {
				mu.Unlock()
				wgTasks.Done()
				continue
			}
			visited[tk.url] = struct{}{}
			mu.Unlock()

			if visitedCallback != nil {
				visitedCallback(tk.url, tk.depth, len(tasks))
			}

			req, _ := http.NewRequestWithContext(ctx, http.MethodGet, tk.url, nil)
			req.Header.Set("User-Agent", userAgent)
			resp, err := client.Do(req)
			if err != nil {
				mu.Lock()
				errs[tk.url] = err
				mu.Unlock()

				if errorCallback != nil {
					errorCallback(tk.url, err, len(tasks))
				}
				wgTasks.Done()
				continue
			}
			func() {
				defer resp.Body.Close()
				ct := resp.Header.Get("Content-Type")
				if resp.StatusCode != http.StatusOK || ct == "" || (ct != "text/html" && !hasPrefix(ct, "text/html")) {
					return
				}
				body, _ := io.ReadAll(resp.Body)
				hrefs := htmlx.ExtractAnchors(stringsReader(string(body)))
				var toEnqueue []string
				for _, href := range hrefs {
					abs, ok := urlutil.Normalize(tk.url, href)
					if !ok {
						continue
					}
					mu.Lock()
					m, ok2 := outlinks[tk.url]
					if !ok2 {
						m = make(map[string]struct{})
						outlinks[tk.url] = m
					}
					m[abs] = struct{}{}
					mu.Unlock()

					if tk.depth < maxDepth {
						if !sameHostOnly || urlutil.SameHost(origin, abs) {
							toEnqueue = append(toEnqueue, abs)
						}
					}
				}
				for _, u := range toEnqueue {
					enqueue(task{url: u, depth: tk.depth + 1})
				}
			}()
			wgTasks.Done()
		}
	}

	for i := 0; i < concurrency; i++ {
		wgWorkers.Add(1)
		go worker()
	}

	// Close the tasks channel when all enqueued tasks are processed.
	go func() {
		wgTasks.Wait()
		close(tasks)
	}()

	enqueue(task{url: startURL, depth: 0})
	wgWorkers.Wait()

	return visited, errs, outlinks
}

func hasPrefix(s string, prefix string) bool {
	return len(s) >= len(prefix) && s[:len(prefix)] == prefix
}

// stringsReader avoids importing strings at package top for a single use.
func stringsReader(s string) io.Reader {
	return &stringReader{str: s}
}

type stringReader struct{ str string }

func (r *stringReader) Read(p []byte) (int, error) {
	if len(r.str) == 0 {
		return 0, io.EOF
	}
	n := copy(p, r.str)
	r.str = r.str[n:]
	return n, nil
}