cmd/channeldump.go

   1 package cmd
   2
   3 import (
   4         "github.com/spf13/cobra"
   5         "net/url"
   6         "fmt"
   7         "os"
   8         "strings"
   9         "time"
  10         "bufio"
  11         "log"
  12         "github.com/terorie/yt-mango/api"
  13 )
  14
  15 var channelDumpCmd = cobra.Command{
  16         Use: "dumpurls <channel ID> <file>",
  17         Short: "Get all public video URLs from channel",
  18         Long: "Write all videos URLs of a channel to a file",
  19         Args: cobra.ExactArgs(2),
  20         Run: func(cmd *cobra.Command, args []string) {
  21                 channelID := args[0]
  22                 fileName := args[1]
  23
  24                 if !matchChannelID.MatchString(channelID) {
  25                         // Check if youtube.com domain
  26                         _url, err := url.Parse(channelID)
  27                         if err != nil || (_url.Host != "www.youtube.com" && _url.Host != "youtube.com") {
  28                                 fmt.Fprintln(os.Stderr, "Not a channel ID:", channelID)
  29                                 os.Exit(1)
  30                         }
  31
  32                         // Check if old /user/ URL
  33                         if strings.HasPrefix(_url.Path, "/user/") {
  34                                 // TODO Implement extraction of channel ID
  35                                 fmt.Fprintln(os.Stderr, "New /channel/ link is required!\n" +
  36                                         "The old /user/ links do not work.")
  37                                 os.Exit(1)
  38                         }
  39
  40                         // Remove /channel/ path
  41                         channelID = strings.TrimPrefix(_url.Path, "/channel/")
  42                         if len(channelID) == len(_url.Path) {
  43                                 // No such prefix to be removed
  44                                 fmt.Fprintln(os.Stderr, "Not a channel ID:", channelID)
  45                                 os.Exit(1)
  46                         }
  47
  48                         // Remove rest of path from channel ID
  49                         slashIndex := strings.IndexRune(channelID, '/')
  50                         if slashIndex != -1 {
  51                                 channelID = channelID[:slashIndex]
  52                         }
  53                 }
  54
  55                 log.Printf("Starting work on channel ID \"%s\".", channelID)
  56                 startTime := time.Now()
  57
  58                 var flags int
  59                 if force {
  60                         flags = os.O_WRONLY | os.O_CREATE | os.O_TRUNC
  61                 } else {
  62                         flags = os.O_WRONLY | os.O_CREATE | os.O_EXCL
  63                 }
  64
  65                 file, err := os.OpenFile(fileName, flags, 0640)
  66                 defer file.Close()
  67                 writer := bufio.NewWriter(file)
  68                 defer writer.Flush()
  69
  70                 if err != nil {
  71                         fmt.Fprintln(os.Stderr, err)
  72                         os.Exit(1)
  73                 }
  74
  75                 totalURLs := 0
  76                 for i := offset; true; i++ {
  77                         channelURLs, err := api.DefaultAPI.GetChannelVideoURLs(channelID, uint(i))
  78                         if err != nil {
  79                                 log.Printf("Aborting on error %v.", err)
  80                                 break
  81                         }
  82                         if len(channelURLs) == 0 {
  83                                 log.Printf("Page %d returned no videos.", i)
  84                                 break
  85                         }
  86                         totalURLs += len(channelURLs)
  87                         log.Printf("Received page %d: %d videos.", i, len(channelURLs))
  88
  89                         for _, _url:= range channelURLs {
  90                                 _, err := writer.WriteString(_url + "\n")
  91                                 if err != nil { panic(err) }
  92                         }
  93                 }
  94
  95                 duration := time.Since(startTime)
  96                 log.Printf("Got %d URLs in %s.", totalURLs, duration.String())
  97         },
  98 }