155 lines
3.4 KiB
Go
Raw Normal View History

2018-03-08 18:54:08 +00:00
package main
import (
2018-03-08 19:07:03 +00:00
"fmt"
2018-03-08 20:00:51 +00:00
"os"
2018-03-08 18:54:08 +00:00
"time"
"github.com/animenotifier/arn/osutils"
2018-03-08 19:07:03 +00:00
"github.com/animenotifier/arn"
"github.com/blitzprog/color"
2018-03-08 19:07:03 +00:00
2018-03-08 18:54:08 +00:00
"github.com/aerogo/crawler"
)
2018-03-08 20:00:51 +00:00
const (
// The maximum age of files we accept until we force a refresh.
maxAge = 7 * 24 * time.Hour
2018-03-26 00:12:06 +00:00
delayBetweenRequests = 1100 * time.Millisecond
2018-10-29 23:59:35 +00:00
userAgent = "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/71.0.3578.20 Safari/537.36"
animeDirectory = "anime"
2018-10-30 00:35:30 +00:00
characterDirectory = "character"
2018-03-08 20:00:51 +00:00
)
2018-03-08 18:54:08 +00:00
2018-04-12 16:57:28 +00:00
var headers = map[string]string{
"User-Agent": userAgent,
"Accept-Encoding": "gzip",
}
2018-03-08 18:54:08 +00:00
func main() {
2018-04-12 16:57:28 +00:00
defer color.Green("Finished.")
2018-10-31 15:04:47 +00:00
// Create directories in case they're missing
os.Mkdir(animeDirectory, 0777)
os.Mkdir(characterDirectory, 0777)
2018-04-12 16:57:28 +00:00
// Called with arguments?
if InvokeShellArgs() {
return
}
2018-03-08 20:00:51 +00:00
// Filter anime with MAL ID
2018-10-30 00:35:30 +00:00
var animes []*arn.Anime
2018-03-08 19:07:03 +00:00
2018-10-30 00:35:30 +00:00
if objectType == "all" || objectType == "anime" {
animes = arn.FilterAnime(func(anime *arn.Anime) bool {
malID := anime.GetMapping("myanimelist/anime")
if malID == "" {
return false
}
return !newOnly || !osutils.Exists(animeFilePath(malID))
2018-10-30 00:35:30 +00:00
})
2018-03-08 19:07:03 +00:00
2018-10-30 00:35:30 +00:00
color.Yellow("Found %d anime", len(animes))
2018-03-08 19:07:03 +00:00
2018-10-30 00:35:30 +00:00
// Sort so that we download the most important ones first
arn.SortAnimeByQuality(animes)
2018-03-08 20:00:51 +00:00
}
2018-10-30 00:35:30 +00:00
// Filter characters with MAL ID
var characters []*arn.Character
if objectType == "all" || objectType == "character" {
characters = arn.FilterCharacters(func(character *arn.Character) bool {
malID := character.GetMapping("myanimelist/character")
if malID == "" {
return false
}
return !newOnly || !osutils.Exists(characterFilePath(malID))
2018-10-30 00:35:30 +00:00
})
color.Yellow("Found %d characters", len(characters))
// Sort so that we download the most important ones first
arn.SortCharactersByLikes(characters)
}
2018-03-08 20:00:51 +00:00
2018-03-09 01:34:50 +00:00
// We don't need the database anymore
arn.Node.Close()
2018-03-08 20:00:51 +00:00
// Create crawler
malCrawler := crawler.New(
2018-04-12 16:57:28 +00:00
headers,
2018-03-08 23:22:36 +00:00
delayBetweenRequests,
2018-10-30 00:35:30 +00:00
len(animes)+len(characters),
2018-03-08 20:00:51 +00:00
)
// Queue up URLs
count := 0
for _, anime := range animes {
2018-10-30 00:35:30 +00:00
queueAnime(anime, malCrawler)
count++
}
for _, character := range characters {
queueCharacter(character, malCrawler)
2018-03-08 19:07:03 +00:00
count++
}
2018-03-08 18:54:08 +00:00
2018-03-28 22:26:19 +00:00
// Log number of links
2018-03-08 19:07:03 +00:00
color.Yellow("Queued up %d links", count)
2018-03-28 22:26:19 +00:00
// Wait for completion
2018-03-08 18:54:08 +00:00
malCrawler.Wait()
2018-04-12 16:57:28 +00:00
}
func animeFilePath(malID string) string {
return fmt.Sprintf("%s/%s.html.gz", animeDirectory, malID)
}
func characterFilePath(malID string) string {
return fmt.Sprintf("%s/%s.html.gz", characterDirectory, malID)
}
2018-10-30 00:35:30 +00:00
func queueAnime(anime *arn.Anime, malCrawler *crawler.Crawler) {
2018-04-12 16:57:28 +00:00
malID := anime.GetMapping("myanimelist/anime")
url := "https://myanimelist.net/anime/" + malID
filePath := animeFilePath(malID)
2018-04-12 16:57:28 +00:00
fileInfo, err := os.Stat(filePath)
if err == nil && time.Since(fileInfo.ModTime()) <= maxAge {
// fmt.Println(color.YellowString(url), "skip")
return
}
2018-03-28 22:26:19 +00:00
2018-04-12 16:57:28 +00:00
malCrawler.Queue(&crawler.Task{
URL: url,
Destination: filePath,
2018-10-29 23:59:35 +00:00
Raw: true,
2018-04-12 16:57:28 +00:00
})
2018-03-08 18:54:08 +00:00
}
2018-10-30 00:35:30 +00:00
func queueCharacter(character *arn.Character, malCrawler *crawler.Crawler) {
malID := character.GetMapping("myanimelist/character")
url := "https://myanimelist.net/character/" + malID
filePath := characterFilePath(malID)
2018-10-30 00:35:30 +00:00
fileInfo, err := os.Stat(filePath)
if err == nil && time.Since(fileInfo.ModTime()) <= maxAge {
// fmt.Println(color.YellowString(url), "skip")
return
}
malCrawler.Queue(&crawler.Task{
URL: url,
Destination: filePath,
Raw: true,
})
}