[Youtube] Add cookies to youtube mix request

This way youtube wont return duplicates when getting more items of the mix (but youtube can also track us)

Author: XiangRongLin

extractor/src/main/java/org/schabi/newpipe/extractor/services/youtube/YoutubeParsingHelper.java

@@ -5,8 +5,10 @@
 import com.grack.nanojson.JsonParser;
 import com.grack.nanojson.JsonParserException;
 import com.grack.nanojson.JsonWriter;
+
 import org.jsoup.Jsoup;
 import org.jsoup.nodes.Document;
+import org.schabi.newpipe.extractor.Page;
 import org.schabi.newpipe.extractor.downloader.Response;
 import org.schabi.newpipe.extractor.exceptions.ContentNotAvailableException;
 import org.schabi.newpipe.extractor.exceptions.ExtractionException;
@@ -21,6 +23,7 @@
 import java.net.MalformedURLException;
 import java.net.URL;
 import java.net.URLDecoder;
+import java.nio.charset.StandardCharsets;
 import java.time.LocalDate;
 import java.time.OffsetDateTime;
 import java.time.ZoneOffset;
@@ -35,6 +38,7 @@
 import static org.schabi.newpipe.extractor.utils.Utils.HTTP;
 import static org.schabi.newpipe.extractor.utils.Utils.HTTPS;
 import static org.schabi.newpipe.extractor.utils.Utils.isNullOrEmpty;
+import static org.schabi.newpipe.extractor.utils.Utils.join;
 
 /*
  * Created by Christian Schabesberger on 02.03.16.
@@ -61,6 +65,12 @@ public class YoutubeParsingHelper {
     private YoutubeParsingHelper() {
     }
 
+    /**
+     * The official youtube app supports intents in this format, where after the ':' is the videoId.
+     * Accordingly there are other apps sharing streams in this format.
+     */
+    public final static String BASE_YOUTUBE_INTENT_URL = "vnd.youtube";
+
     private static final String HARDCODED_CLIENT_VERSION = "2.20200214.04.00";
     private static String clientVersion;
 
@@ -193,22 +203,22 @@ public static OffsetDateTime parseDateFrom(String textualUploadDate) throws Pars
     }
 
     /**
-     * Checks if the given playlist id is a youtube mix (auto-generated playlist)
-     * Ids from a youtube mix start with "RD"
+     * Checks if the given playlist id is a YouTube Mix (auto-generated playlist)
+     * Ids from a YouTube Mix start with "RD"
      * @param playlistId
-     * @return Whether given id belongs to a youtube mix
+     * @return Whether given id belongs to a YouTube Mix
      */
-    public static boolean isYoutubeMixId(String playlistId) {
+    public static boolean isYoutubeMixId(final String playlistId) {
         return playlistId.startsWith("RD") && !isYoutubeMusicMixId(playlistId);
     }
 
     /**
-     * Checks if the given playlist id is a youtube music mix (auto-generated playlist)
-     * Ids from a youtube music mix start with "RD"
+     * Checks if the given playlist id is a YouTube Music Mix (auto-generated playlist)
+     * Ids from a YouTube Music Mix start with "RD"
      * @param playlistId
-     * @return Whether given id belongs to a youtube music mix
+     * @return Whether given id belongs to a YouTube Music Mix
      */
-    public static boolean isYoutubeMusicMixId(String playlistId) {
+    public static boolean isYoutubeMusicMixId(final String playlistId) {
         return playlistId.startsWith("RDAMVM");
     }
 
@@ -352,7 +362,7 @@ public static boolean areHardcodedYoutubeMusicKeysValid() throws IOException, Re
                 .end()
                 .value("query", "test")
                 .value("params", "Eg-KAQwIARAAGAAgACgAMABqChAEEAUQAxAKEAk%3D")
-            .end().done().getBytes("UTF-8");
+            .end().done().getBytes(StandardCharsets.UTF_8);
         // @formatter:on
 
         Map<String, List<String>> headers = new HashMap<>();
@@ -436,10 +446,14 @@ public static String getUrlFromNavigationEndpoint(JsonObject navigationEndpoint)
         } else if (navigationEndpoint.has("watchEndpoint")) {
             StringBuilder url = new StringBuilder();
             url.append("https://www.youtube.com/watch?v=").append(navigationEndpoint.getObject("watchEndpoint").getString("videoId"));
-            if (navigationEndpoint.getObject("watchEndpoint").has("playlistId"))
-                url.append("&list=").append(navigationEndpoint.getObject("watchEndpoint").getString("playlistId"));
-            if (navigationEndpoint.getObject("watchEndpoint").has("startTimeSeconds"))
-                url.append("&t=").append(navigationEndpoint.getObject("watchEndpoint").getInt("startTimeSeconds"));
+            if (navigationEndpoint.getObject("watchEndpoint").has("playlistId")) {
+                url.append("&amp;list=").append(navigationEndpoint.getObject("watchEndpoint")
+                        .getString("playlistId"));
+            }
+            if (navigationEndpoint.getObject("watchEndpoint").has("startTimeSeconds")) {
+                url.append("&amp;t=").append(navigationEndpoint.getObject("watchEndpoint")
+                        .getInt("startTimeSeconds"));
+            }
             return url.toString();
         } else if (navigationEndpoint.has("watchPlaylistEndpoint")) {
             return "https://www.youtube.com/playlist?list=" +
@@ -467,7 +481,6 @@ public static String getTextFromObject(JsonObject textObject, boolean html) thro
             if (html && ((JsonObject) textPart).has("navigationEndpoint")) {
                 String url = getUrlFromNavigationEndpoint(((JsonObject) textPart).getObject("navigationEndpoint"));
                 if (!isNullOrEmpty(url)) {
-                    url = url.replaceAll("&", "&amp;");
                     textBuilder.append("<a href=\"").append(url).append("\">").append(text).append("</a>");
                     continue;
                 }
@@ -506,8 +519,8 @@ public static String fixThumbnailUrl(String thumbnailUrl) {
     public static String getValidJsonResponseBody(final Response response)
             throws ParsingException, MalformedURLException {
         if (response.responseCode() == 404) {
-            throw new ContentNotAvailableException("Not found" +
-                    " (\"" + response.responseCode() + " " + response.responseMessage() + "\")");
+            throw new ContentNotAvailableException("Not found"
+                    + " (\"" + response.responseCode() + " " + response.responseMessage() + "\")");
         }
 
         final String responseBody = response.responseBody();
@@ -527,22 +540,64 @@ public static String getValidJsonResponseBody(final Response response)
         final String responseContentType = response.getHeader("Content-Type");
         if (responseContentType != null
                 && responseContentType.toLowerCase().contains("text/html")) {
-            throw new ParsingException("Got HTML document, expected JSON response" +
-                    " (latest url was: \"" + response.latestUrl() + "\")");
+            throw new ParsingException("Got HTML document, expected JSON response"
+                    + " (latest url was: \"" + response.latestUrl() + "\")");
         }
 
         return responseBody;
     }
 
+    public static Response getResponse(final String url, final Localization localization)
+            throws IOException, ExtractionException {
+        final Map<String, List<String>> headers = new HashMap<>();
+        headers.put("X-YouTube-Client-Name", Collections.singletonList("1"));
+        headers.put("X-YouTube-Client-Version", Collections.singletonList(getClientVersion()));
+
+        final Response response = getDownloader().get(url, headers, localization);
+        getValidJsonResponseBody(response);
+
+        return response;
+    }
+
+    public static String extractCookieValue(final String cookieName, final Response response) {
+        final List<String> cookies = response.responseHeaders().get("Set-Cookie");
+        int startIndex;
+        String result = "";
+        for (final String cookie : cookies) {
+            startIndex = cookie.indexOf(cookieName);
+            if (startIndex != -1) {
+                result = cookie.substring(startIndex + cookieName.length() + "=".length(),
+                        cookie.indexOf(";", startIndex));
+            }
+        }
+        return result;
+    }
+
     public static JsonArray getJsonResponse(final String url, final Localization localization)
             throws IOException, ExtractionException {
         Map<String, List<String>> headers = new HashMap<>();
         headers.put("X-YouTube-Client-Name", Collections.singletonList("1"));
         headers.put("X-YouTube-Client-Version", Collections.singletonList(getClientVersion()));
         final Response response = getDownloader().get(url, headers, localization);
 
-        final String responseBody = getValidJsonResponseBody(response);
+        return toJsonArray(getValidJsonResponseBody(response));
+    }
+
+    public static JsonArray getJsonResponse(final Page page, final Localization localization)
+            throws IOException, ExtractionException {
+        final Map<String, List<String>> headers = new HashMap<>();
+        if (!isNullOrEmpty(page.getCookies())) {
+            headers.put("Cookie", Collections.singletonList(join(";", "=", page.getCookies())));
+        }
+        headers.put("X-YouTube-Client-Name", Collections.singletonList("1"));
+        headers.put("X-YouTube-Client-Version", Collections.singletonList(getClientVersion()));
+
+        final Response response = getDownloader().get(page.getUrl(), headers, localization);
+
+        return toJsonArray(getValidJsonResponseBody(response));
+    }
 
+    public static JsonArray toJsonArray(final String responseBody) throws ParsingException {
         try {
             return JsonParser.array().from(responseBody);
         } catch (JsonParserException e) {

extractor/src/main/java/org/schabi/newpipe/extractor/services/youtube/YoutubeService.java

@@ -110,7 +110,7 @@ public ChannelExtractor getChannelExtractor(ListLinkHandler linkHandler) {
     }
 
     @Override
-    public PlaylistExtractor getPlaylistExtractor(ListLinkHandler linkHandler) {
+    public PlaylistExtractor getPlaylistExtractor(final ListLinkHandler linkHandler) {
         if (YoutubeParsingHelper.isYoutubeMixId(linkHandler.getId())) {
             return new YoutubeMixPlaylistExtractor(this, linkHandler);
         } else {

extractor/src/main/java/org/schabi/newpipe/extractor/services/youtube/extractors/YoutubeMixPlaylistExtractor.java

@@ -4,8 +4,10 @@
 import com.grack.nanojson.JsonObject;
 
 import org.schabi.newpipe.extractor.ListExtractor;
+import org.schabi.newpipe.extractor.Page;
 import org.schabi.newpipe.extractor.StreamingService;
 import org.schabi.newpipe.extractor.downloader.Downloader;
+import org.schabi.newpipe.extractor.downloader.Response;
 import org.schabi.newpipe.extractor.exceptions.ExtractionException;
 import org.schabi.newpipe.extractor.exceptions.ParsingException;
 import org.schabi.newpipe.extractor.linkhandler.ListLinkHandler;
@@ -15,34 +17,50 @@
 import org.schabi.newpipe.extractor.stream.StreamInfoItemsCollector;
 
 import java.io.IOException;
+import java.util.Collections;
+import java.util.List;
 
 import javax.annotation.Nonnull;
 import javax.annotation.Nullable;
 
+import static org.schabi.newpipe.extractor.services.youtube.YoutubeParsingHelper.extractCookieValue;
 import static org.schabi.newpipe.extractor.services.youtube.YoutubeParsingHelper.getJsonResponse;
+import static org.schabi.newpipe.extractor.services.youtube.YoutubeParsingHelper.getResponse;
 import static org.schabi.newpipe.extractor.services.youtube.YoutubeParsingHelper.getUrlFromNavigationEndpoint;
+import static org.schabi.newpipe.extractor.services.youtube.YoutubeParsingHelper.toJsonArray;
 
 /**
- * A YoutubePlaylistExtractor for a mix (auto-generated playlist). It handles urls in the format of
- * "youtube.com/watch?v=videoId&list=playlistId"
+ * A {@link YoutubePlaylistExtractor} for a mix (auto-generated playlist).
+ * It handles URLs in the format of
+ * {@code youtube.com/watch?v=videoId&list=playlistId}
  */
 public class YoutubeMixPlaylistExtractor extends PlaylistExtractor {
 
+    /**
+     * YouTube identifies mixes based on this cookie. With this information it can generate
+     * continuations without duplicates.
+     */
+    private static final String COOKIE_NAME = "VISITOR_INFO1_LIVE";
+
     private JsonObject initialData;
     private JsonObject playlistData;
+    private String cookieValue;
 
-    public YoutubeMixPlaylistExtractor(StreamingService service, ListLinkHandler linkHandler) {
+    public YoutubeMixPlaylistExtractor(final StreamingService service,
+                                       final ListLinkHandler linkHandler) {
         super(service, linkHandler);
     }
 
     @Override
-    public void onFetchPage(@Nonnull Downloader downloader)
-        throws IOException, ExtractionException {
+    public void onFetchPage(@Nonnull final Downloader downloader)
+            throws IOException, ExtractionException {
         final String url = getUrl() + "&pbj=1";
-        final JsonArray ajaxJson = getJsonResponse(url, getExtractorLocalization());
+        final Response response = getResponse(url, getExtractorLocalization());
+        final JsonArray ajaxJson = toJsonArray(response.responseBody());
         initialData = ajaxJson.getObject(3).getObject("response");
         playlistData = initialData.getObject("contents").getObject("twoColumnWatchNextResults")
-            .getObject("playlist").getObject("playlist");
+                .getObject("playlist").getObject("playlist");
+        cookieValue = extractCookieValue(COOKIE_NAME, response);
     }
 
     @Nonnull
@@ -58,16 +76,15 @@ public String getName() throws ParsingException {
     @Override
     public String getThumbnailUrl() throws ParsingException {
         try {
-            final String playlistId = playlistData.getString("playlistId");
+            return getThumbnailUrlFromPlaylistId(playlistData.getString("playlistId"));
+        } catch (final Exception e) {
             try {
-                return getThumbnailUrlFromPlaylistId(playlistId);
-            } catch (ParsingException e) {
                 //fallback to thumbnail of current video. Always the case for channel mix
                 return getThumbnailUrlFromVideoId(
                     initialData.getObject("currentVideoEndpoint").getObject("watchEndpoint")
                         .getString("videoId"));
+            } catch (final Exception ignored) {
             }
-        } catch (Exception e) {
             throw new ParsingException("Could not get playlist thumbnail", e);
         }
     }
@@ -104,63 +121,66 @@ public long getStreamCount() {
     @Nonnull
     @Override
     public InfoItemsPage<StreamInfoItem> getInitialPage() throws ExtractionException {
-        StreamInfoItemsCollector collector = new StreamInfoItemsCollector(getServiceId());
+        final StreamInfoItemsCollector collector = new StreamInfoItemsCollector(getServiceId());
         collectStreamsFrom(collector, playlistData.getArray("contents"));
-        return new InfoItemsPage<>(collector, getNextPageUrl());
+        return new InfoItemsPage<>(collector,
+                new Page(getNextPageUrl(), Collections.singletonMap(COOKIE_NAME, cookieValue)));
     }
 
-    @Override
-    public String getNextPageUrl() throws ExtractionException {
+    private String getNextPageUrl() throws ExtractionException {
         return getNextPageUrlFrom(playlistData);
     }
 
-    private String getNextPageUrlFrom(JsonObject playlistData) throws ExtractionException {
-        final JsonObject lastStream = ((JsonObject) playlistData.getArray("contents")
-            .get(playlistData.getArray("contents").size() - 1));
+    private String getNextPageUrlFrom(final JsonObject playlistJson) throws ExtractionException {
+        final JsonObject lastStream = ((JsonObject) playlistJson.getArray("contents")
+                .get(playlistJson.getArray("contents").size() - 1));
         if (lastStream == null || lastStream.getObject("playlistPanelVideoRenderer") == null) {
             throw new ExtractionException("Could not extract next page url");
         }
-        //Index of video in mix is missing, but adding it doesn't appear to have any effect.
-        //And since the index needs to be tracked by us, it is left out
+
         return getUrlFromNavigationEndpoint(
-            lastStream.getObject("playlistPanelVideoRenderer").getObject("navigationEndpoint"))
-            + "&pbj=1";
+                lastStream.getObject("playlistPanelVideoRenderer").getObject("navigationEndpoint"))
+                + "&pbj=1";
     }
 
     @Override
-    public InfoItemsPage<StreamInfoItem> getPage(final String pageUrl)
+    public InfoItemsPage<StreamInfoItem> getPage(final Page page)
             throws ExtractionException, IOException {
-        if (pageUrl == null || pageUrl.isEmpty()) {
+        if (page == null || page.getUrl().isEmpty()) {
             throw new ExtractionException(
                 new IllegalArgumentException("Page url is empty or null"));
         }
 
-        StreamInfoItemsCollector collector = new StreamInfoItemsCollector(getServiceId());
-        final JsonArray ajaxJson = getJsonResponse(pageUrl, getExtractorLocalization());
-        JsonObject playlistData =
-            ajaxJson.getObject(3).getObject("response").getObject("contents")
-                .getObject("twoColumnWatchNextResults").getObject("playlist")
-                .getObject("playlist");
-        final JsonArray streams = playlistData.getArray("contents");
-        //Because continuation requests are created with the last video of previous request as start
-        streams.remove(0);
-        collectStreamsFrom(collector, streams);
-        return new InfoItemsPage<>(collector, getNextPageUrlFrom(playlistData));
+        final JsonArray ajaxJson = getJsonResponse(page, getExtractorLocalization());
+        final JsonObject playlistJson =
+                ajaxJson.getObject(3).getObject("response").getObject("contents")
+                        .getObject("twoColumnWatchNextResults").getObject("playlist")
+                        .getObject("playlist");
+        final JsonArray allStreams = playlistJson.getArray("contents");
+        // Sublist because youtube returns up to 24 previous streams in the mix
+        // +1 because the stream of "currentIndex" was already extracted in previous request
+        final List<Object> newStreams =
+                allStreams.subList(playlistJson.getInt("currentIndex") + 1, allStreams.size());
+
+        final StreamInfoItemsCollector collector = new StreamInfoItemsCollector(getServiceId());
+        collectStreamsFrom(collector, newStreams);
+        return new InfoItemsPage<>(collector,
+                new Page(getNextPageUrlFrom(playlistJson), page.getCookies()));
     }
 
     private void collectStreamsFrom(
-            @Nonnull StreamInfoItemsCollector collector,
-            @Nullable JsonArray streams) {
+            @Nonnull final StreamInfoItemsCollector collector,
+            @Nullable final List<Object> streams) {
 
         if (streams == null) {
             return;
         }
 
         final TimeAgoParser timeAgoParser = getTimeAgoParser();
 
-        for (Object stream : streams) {
+        for (final Object stream : streams) {
             if (stream instanceof JsonObject) {
-                JsonObject streamInfo = ((JsonObject) stream)
+                final JsonObject streamInfo = ((JsonObject) stream)
                     .getObject("playlistPanelVideoRenderer");
                 if (streamInfo != null) {
                     collector.commit(new YoutubeStreamInfoItemExtractor(streamInfo, timeAgoParser));
@@ -169,7 +189,7 @@ private void collectStreamsFrom(
         }
     }
 
-    private String getThumbnailUrlFromPlaylistId(String playlistId) throws ParsingException {
+    private String getThumbnailUrlFromPlaylistId(final String playlistId) throws ParsingException {
         final String videoId;
         if (playlistId.startsWith("RDMM")) {
             videoId = playlistId.substring(4);
@@ -184,7 +204,7 @@ private String getThumbnailUrlFromPlaylistId(String playlistId) throws ParsingEx
         return getThumbnailUrlFromVideoId(videoId);
     }
 
-    private String getThumbnailUrlFromVideoId(String videoId) {
+    private String getThumbnailUrlFromVideoId(final String videoId) {
         return "https://i.ytimg.com/vi/" + videoId + "/hqdefault.jpg";
     }