diff options
| author | rtz12 <koenig@fagott.pw> | 2016-03-29 15:58:55 (UTC) |
|---|---|---|
| committer | rtz12 <koenig@fagott.pw> | 2016-03-29 15:58:55 (UTC) |
| commit | 1400381e9c193eb7f681d95616eed9070edaab56 (patch) | |
| tree | 2db90f612fcf10c0b4efada9f0e5a2e4f825a1a6 /tools/charcrawler/crawler/acd.go | |
| parent | 8661775d15c703ebfba4d3779d796849939c705b (diff) | |
Crawler hinzugefuegt
Diffstat (limited to 'tools/charcrawler/crawler/acd.go')
| -rw-r--r-- | tools/charcrawler/crawler/acd.go | 81 |
1 files changed, 81 insertions, 0 deletions
diff --git a/tools/charcrawler/crawler/acd.go b/tools/charcrawler/crawler/acd.go new file mode 100644 index 0000000..31efb0d --- /dev/null +++ b/tools/charcrawler/crawler/acd.go | |||
| @@ -0,0 +1,81 @@ | |||
| 1 | package crawler | ||
| 2 | |||
| 3 | import ( | ||
| 4 | "log" | ||
| 5 | "strconv" | ||
| 6 | "strings" | ||
| 7 | |||
| 8 | "github.com/PuerkitoBio/goquery" | ||
| 9 | ) | ||
| 10 | |||
| 11 | type ACDCrawler struct{} | ||
| 12 | |||
| 13 | func (a ACDCrawler) Name() string { | ||
| 14 | return "ACD" | ||
| 15 | } | ||
| 16 | |||
| 17 | func (a ACDCrawler) Crawl(id int) (CharacterData, error) { | ||
| 18 | c := make(CharacterData) | ||
| 19 | doc, err := goquery.NewDocument("http://www.animecharactersdatabase.com/character.php?id=" + strconv.Itoa(id)) | ||
| 20 | if err != nil { | ||
| 21 | log.Println(err) | ||
| 22 | return nil, CrawlError | ||
| 23 | } | ||
| 24 | text := doc.Text() | ||
| 25 | if strings.Contains(text, "bad character : try") { | ||
| 26 | return nil, CharacterNotFound | ||
| 27 | } | ||
| 28 | dataTr := doc.Find("#besttable").Next().ChildrenFiltered("tbody").ChildrenFiltered("tr") | ||
| 29 | leftSide := dataTr.Children().Eq(0) | ||
| 30 | rightSide := dataTr.Children().Eq(1) | ||
| 31 | imageCols := rightSide.ChildrenFiltered("table").First().Find("tbody > tr > td") | ||
| 32 | val, _ := imageCols.Eq(0).Find("img").Attr("src") | ||
| 33 | c["__thumb"] = val | ||
| 34 | val, _ = imageCols.Eq(1).Find("a").Attr("href") | ||
| 35 | c["__img"] = val | ||
| 36 | leftSide.ChildrenFiltered("table").Eq(1).Find("tr").Each(func(i int, s *goquery.Selection) { | ||
| 37 | c[s.Find("th").Text()] = s.Find("td").Text() | ||
| 38 | }) | ||
| 39 | var key string | ||
| 40 | leftSide. | ||
| 41 | ChildrenFiltered("table"). | ||
| 42 | Eq(0). | ||
| 43 | Find("td"). | ||
| 44 | Eq(1). | ||
| 45 | Find("dl"). | ||
| 46 | Children(). | ||
| 47 | Each(func(i int, s *goquery.Selection) { | ||
| 48 | switch goquery.NodeName(s) { | ||
| 49 | case "dt": | ||
| 50 | key = s.Text() | ||
| 51 | case "dd": | ||
| 52 | c[key] = s.Text() | ||
| 53 | } | ||
| 54 | }) | ||
| 55 | tags := make([]string, 0) | ||
| 56 | leftSide.ChildrenFiltered("div").Eq(0).Find("a").Each(func(i int, s *goquery.Selection) { | ||
| 57 | tags = append(tags, s.Text()) | ||
| 58 | }) | ||
| 59 | c["__tags"] = tags | ||
| 60 | vas := make([]string, 0) | ||
| 61 | leftSide.ChildrenFiltered("div").Eq(1).Find("a").Each(func(i int, s *goquery.Selection) { | ||
| 62 | vas = append(vas, s.Text()) | ||
| 63 | }) | ||
| 64 | c["__vas"] = vas | ||
| 65 | leftSide.ChildrenFiltered("dl").Children().Each(func(i int, s *goquery.Selection) { | ||
| 66 | switch goquery.NodeName(s) { | ||
| 67 | case "dt": | ||
| 68 | key = s.Text() | ||
| 69 | case "dd": | ||
| 70 | c[key] = s.Text() | ||
| 71 | } | ||
| 72 | }) | ||
| 73 | apps := make([]int, 0) | ||
| 74 | rightSide.Find(".tile3top").Each(func(i int, s *goquery.Selection) { | ||
| 75 | val, _ = s.Find("a").Attr("href") | ||
| 76 | id, _ := strconv.Atoi(strings.Split(val, "=")[1]) | ||
| 77 | apps = append(apps, id) | ||
| 78 | }) | ||
| 79 | c["__appearances"] = apps | ||
| 80 | return c, nil | ||
| 81 | } | ||
