| 1 | + | package main |
| 2 | + | |
| 3 | + | import ( |
| 4 | + | "bufio" |
| 5 | + | "encoding/json" |
| 6 | + | "errors" |
| 7 | + | "flag" |
| 8 | + | "fmt" |
| 9 | + | "io" |
| 10 | + | "net/http" |
| 11 | + | "os" |
| 12 | + | "time" |
| 13 | + | ) |
| 14 | + | |
| 15 | + | const ( |
| 16 | + | wbSnapshotApiURL = "https://web.archive.org/cdx/search/xd?output=json&url=%s&fl=timestamp,original&collapse=digest&gzip=false&filter=statuscode:200" |
| 17 | + | wbFileURL = "https://web.archive.org/web/%sid_/%s" |
| 18 | + | ) |
| 19 | + | |
| 20 | + | func main() { |
| 21 | + | var url string |
| 22 | + | |
| 23 | + | flag.Parse() |
| 24 | + | |
| 25 | + | if flag.NArg() > 0 { |
| 26 | + | url = flag.Arg(0) |
| 27 | + | } else { |
| 28 | + | sc := bufio.NewScanner(os.Stdin) |
| 29 | + | for sc.Scan() { |
| 30 | + | url = sc.Text() |
| 31 | + | } |
| 32 | + | |
| 33 | + | if err := sc.Err(); err != nil { |
| 34 | + | fmt.Fprintf(os.Stderr, "failed to read input: %s\n", err) |
| 35 | + | } |
| 36 | + | } |
| 37 | + | |
| 38 | + | client := http.Client{ |
| 39 | + | Timeout: time.Second * 5, |
| 40 | + | } |
| 41 | + | |
| 42 | + | snapshots, err := getSnapshots(client, url) |
| 43 | + | if err != nil { |
| 44 | + | fmt.Fprintf(os.Stderr, "failed to getting snapshots %s\n", err) |
| 45 | + | } |
| 46 | + | |
| 47 | + | lastSnapshot := snapshots[len(snapshots)-1] |
| 48 | + | lastSnapshotTs := lastSnapshot[0] |
| 49 | + | lastSnapshotURL := lastSnapshot[1] |
| 50 | + | |
| 51 | + | request, err := http.NewRequest("GET", fmt.Sprintf(wbFileURL, lastSnapshotTs, lastSnapshotURL), nil) |
| 52 | + | if err != nil { |
| 53 | + | fmt.Fprintf(os.Stderr, "failed to generate request waybackmachine api: %s\n", err) |
| 54 | + | } |
| 55 | + | |
| 56 | + | request.Header.Add("Accept-Encoding", "plain") |
| 57 | + | |
| 58 | + | response, err := client.Do(request) |
| 59 | + | if err != nil { |
| 60 | + | fmt.Fprintf(os.Stderr, "failed to send request waybackmachine api: %s\n", err) |
| 61 | + | } |
| 62 | + | |
| 63 | + | defer response.Body.Close() |
| 64 | + | |
| 65 | + | io.Copy(os.Stdout, response.Body) |
| 66 | + | } |
| 67 | + | |
| 68 | + | func getSnapshots(c http.Client, url string) ([][]string, error) { |
| 69 | + | request, err := http.NewRequest("GET", fmt.Sprintf(wbSnapshotApiURL, url), nil) |
| 70 | + | if err != nil { |
| 71 | + | return [][]string{}, fmt.Errorf("getSnapshots: failed to generate request waybackmachine api: %w", err) |
| 72 | + | } |
| 73 | + | |
| 74 | + | response, err := c.Do(request) |
| 75 | + | if err != nil { |
| 76 | + | return [][]string{}, fmt.Errorf("getSnapshots: failed to send request waybackmachine api: %w", err) |
| 77 | + | } |
| 78 | + | defer response.Body.Close() |
| 79 | + | |
| 80 | + | var r [][]string |
| 81 | + | dec := json.NewDecoder(response.Body) |
| 82 | + | |
| 83 | + | err = dec.Decode(&r) |
| 84 | + | if err != nil { |
| 85 | + | return [][]string{}, fmt.Errorf("getSnapshots: error while decoding response %w", err) |
| 86 | + | } |
| 87 | + | |
| 88 | + | if len(r) < 1 { |
| 89 | + | return [][]string{}, errors.New("getSnapshots: no results found for this url") |
| 90 | + | } |
| 91 | + | |
| 92 | + | return r[1:], nil |
| 93 | + | } |
| 94 | + | |