Merge pull request #66 from Who23/youtube-sources

Add ability to specifiy youtube URLs manually

Looks like wonderful stuff, I appreciate the contribution and time investment to my small homebrew project, I hope you've been getting good stuff out of it.

README.md

@@ -54,6 +54,8 @@ Arguments:
     -s, --song <song>           Specify song name to download
     -A, --album <album>         Specify the album name to download
     -p, --playlist <playlist>   Specify the playlist name to download
+    -u, --url <url>             Specify the youtube url to download from (for single songs only)
+    -g, --give-url              Specify the youtube url sources while downloading (for albums or playlists only only)
 
 Examples:
     $ irs --song "Bohemian Rhapsody" --artist "Queen"

src/bottle/cli.cr

@@ -20,6 +20,8 @@ class CLI
     [["-s", "--song"], "song", "string"],
     [["-A", "--album"], "album", "string"],
     [["-p", "--playlist"], "playlist", "string"],
+    [["-u", "--url"], "url", "string"],
+    [["-g", "--give-url"], "give-url", "bool"],
   ]
 
   @args : Hash(String, String)
@@ -48,6 +50,8 @@ class CLI
         #{Style.blue "-s, --song <song>"}           Specify song name to download
         #{Style.blue "-A, --album <album>"}         Specify the album name to download
         #{Style.blue "-p, --playlist <playlist>"}   Specify the playlist name to download
+        #{Style.blue "-u, --url <url>"}             Specify the youtube url to download from (for single songs only)
+        #{Style.blue "-g, --give-url"}              Specify the youtube url sources while downloading (for albums or playlists only)
 
     #{Style.bold "Examples:"}
         $ #{Style.green %(irs --song "Bohemian Rhapsody" --artist "Queen")}
@@ -82,17 +86,17 @@ class CLI
     elsif @args["song"]? && @args["artist"]?
       s = Song.new(@args["song"], @args["artist"])
       s.provide_client_keys(Config.client_key, Config.client_secret)
-      s.grab_it
+      s.grab_it(@args["url"]?)
       s.organize_it(Config.music_directory)
       exit
     elsif @args["album"]? && @args["artist"]?
       a = Album.new(@args["album"], @args["artist"])
       a.provide_client_keys(Config.client_key, Config.client_secret)
-      a.grab_it
+      a.grab_it(!!@args["give-url"])
     elsif @args["playlist"]? && @args["artist"]?
       p = Playlist.new(@args["playlist"], @args["artist"])
       p.provide_client_keys(Config.client_key, Config.client_secret)
-      p.grab_it
+      p.grab_it(!!@args["give-url"])
     else
       puts Style.red("Those arguments don't do anything when used that way.")
       puts "Type `irs -h` to see usage."

src/glue/list.cr

@@ -17,6 +17,9 @@ abstract class SpotifyList
     "searching" => [
       Style.bold("Searching for %l by %a ... \r"),
       Style.green("+ ") + Style.bold("%l by %a                                 \n")
+    ],
+    "url" => [
+      Style.bold("When prompted for a URL, provide a youtube URL or press enter to scrape for one\n")
     ]
   }
 
@@ -24,11 +27,15 @@ abstract class SpotifyList
   end
 
   # Finds the list, and downloads all of the songs using the `Song` class
-  def grab_it
+  def grab_it(ask_url : Bool = false)
     if !@spotify_searcher.authorized?
       raise("Need to call provide_client_keys on Album or Playlist class.")
     end
 
+    if ask_url
+      outputter("url", 0)
+    end
+
     outputter("searching", 0)
     list = find_it()
     outputter("searching", 1)
@@ -47,7 +54,7 @@ abstract class SpotifyList
       song.provide_metadata(data)
 
       puts Style.bold("[#{data["track_number"]}/#{contents.size}]")
-      song.grab_it
+      song.grab_it ask_url: ask_url
 
       organize(song)
 

src/glue/song.cr

@@ -24,7 +24,10 @@ class Song
     ],
     "url" => [
       "  Searching for URL ...\r",
-      Style.green("  + ") + Style.dim("URL found                       \n")
+      Style.green("  + ") + Style.dim("URL found                       \n"),
+      "  Validating URL ...\r",
+      Style.green("  + ") + Style.dim("URL validated                   \n"),
+      "URL?: "
     ],
     "download" => [
       "  Downloading video:\n",
@@ -47,11 +50,12 @@ class Song
   end
 
   # Find, downloads, and tags the mp3 song that this class represents.
+  # Optionally takes a youtube URL to download from
   # 
   # ```
   # Song.new("Bohemian Rhapsody", "Queen").grab_it
   # ```
-  def grab_it
+  def grab_it(url : (String | Nil) = nil, ask_url : Bool = false)
     outputter("intro", 0)
 
     if !@spotify_searcher.authorized? && !@metadata
@@ -81,14 +85,31 @@ class Song
     data = @metadata.as(JSON::Any)
     @filename = data["track_number"].to_s + " - #{data["name"].to_s}.mp3"
 
-    outputter("url", 0)
-    url = Youtube.find_url(@song_name, @artist_name, search_terms: "lyrics")
-    if !url
-      raise("There was no url found on youtube for " +
-            %("#{@song_name}" by "#{@artist_name}. ) +
-            "Check your input and try again.")
+    if ask_url
+      outputter("url", 4)
+      url = gets
+      if !url.nil? && url.strip == ""
+        url = nil
+      end
+    end
+
+    if !url
+      outputter("url", 0)
+      url = Youtube.find_url(@song_name, @artist_name, search_terms: "lyrics")
+      if !url
+        raise("There was no url found on youtube for " +
+              %("#{@song_name}" by "#{@artist_name}. ) +
+              "Check your input and try again.")
+      end
+      outputter("url", 1)
+    else
+      outputter("url", 2)
+      if !Youtube.is_valid_url(url)
+        raise("The url '#{url}' is an invalid youtube URL " +
+              "Check the URL and try again")
+      end
+      outputter("url", 3)
     end
-    outputter("url", 1)
 
     outputter("download", 0)
     Ripper.download_mp3(url.as(String), @filename)

src/search/youtube.cr

@@ -1,6 +1,7 @@
 require "http"
 require "xml"
 require "json"
+require "uri"
 
 
 module Youtube
@@ -23,6 +24,45 @@ module Youtube
 
   alias NODES_CLASS = Array(Hash(String, String))
 
+  # Checks if the given URL is a valid youtube URL
+  #
+  # ```
+  # Youtube.is_valid_url("https://www.youtube.com/watch?v=NOTANACTUALVIDEOID")
+  # => false
+  # ```
+  def is_valid_url(url : String) : Bool
+    uri = URI.parse url
+
+    # is it a video on youtube, with a query
+    query = uri.query
+    if uri.host != "www.youtube.com" || uri.path != "/watch" || !query
+      return false
+    end
+
+
+    queries = query.split('&')
+
+    # find the video ID
+    i = 0
+    while i < queries.size
+      if queries[i].starts_with?("v=")
+        vID = queries[i][2..-1]
+        break
+      end
+      i += 1
+    end
+
+    if !vID
+      return false
+    end
+
+
+    # this is an internal endpoint to validate the video ID
+    response = HTTP::Client.get "https://www.youtube.com/get_video_info?video_id=#{vID}"
+
+    return response.body.includes?("status=ok")
+  end
+
   # Finds a youtube url based off of the given information.
   # The query to youtube is constructed like this:
   #   "<song_name> <artist_name> <search terms>"