be-gay-do-crimes
/
NewPipeExtractor
mirror of https://github.com/TeamNewPipe/NewPipeExtractor
1
0
Fork 0
Code Issues Releases Wiki Activity

Use more often playerResponse in yt stream extractor 

This enhances performance and should make the extractor more reliable since it get info from a stable json structure that shouldn't be subject to many changes. Fallback html methods have been kept. In case of error the thrown exception contains the data about the playerResponse failure, that should be clearer than a NPE caused by not-found html tags.
This commit is contained in:
Stypox 2020-01-06 20:45:57 +01:00
parent d83787a5ca
commit 1ed89aad3e
No known key found for this signature in database
GPG Key ID: 4BDF1B40A49FDD23
1 changed files with 100 additions and 79 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

167
extractor/src/main/java/org/schabi/newpipe/extractor/services/youtube/extractors/YoutubeStreamExtractor.java
View File

@ -106,20 +106,22 @@ public class YoutubeStreamExtractor extends StreamExtractor {
@Override
public String getName() throws ParsingException {
assertPageFetched();
String name = getStringFromMetaData("title");
if(name == null) {
// Fallback to HTML method
try {
return playerResponse.getObject("videoDetails").getString("title");
} catch (Exception e) {
// fallback HTML method
String name = null;
try {
name = doc.select("meta[name=title]").attr(CONTENT);
} catch (Exception e) {
throw new ParsingException("Could not get the title", e);
}
}
if(name == null || name.isEmpty()) {
throw new ParsingException("Could not get the title");
} catch (Exception ignored) {}
if (name == null) {
throw new ParsingException("Could not get name", e);
}
return name;
}
}
@Override
public String getTextualUploadDate() throws ParsingException {
@ -128,10 +130,18 @@ public class YoutubeStreamExtractor extends StreamExtractor {
}
try {
return doc.select("meta[itemprop=datePublished]").attr(CONTENT);
} catch (Exception e) {//todo: add fallback method
return playerResponse.getObject("microformat").getObject("playerMicroformatRenderer").getString("publishDate");
} catch (Exception e) {
String uploadDate = null;
try {
uploadDate = doc.select("meta[itemprop=datePublished]").attr(CONTENT);
} catch (Exception ignored) {}
if (uploadDate == null) {
throw new ParsingException("Could not get upload date", e);
}
return uploadDate;
}
}
@Override
@ -149,24 +159,23 @@ public class YoutubeStreamExtractor extends StreamExtractor {
@Override
public String getThumbnailUrl() throws ParsingException {
assertPageFetched();
// Try to get high resolution thumbnail first, if it fails, use low res from the player instead
try {
return doc.select("link[itemprop=\"thumbnailUrl\"]").first().attr("abs:href");
} catch (Exception ignored) {
// Try other method...
}
JsonArray thumbnails = playerResponse.getObject("videoDetails").getObject("thumbnail").getArray("thumbnails");
// the last thumbnail is the one with the highest resolution
return thumbnails.getObject(thumbnails.size()-1).getString("url");
try {
if (playerArgs != null && playerArgs.isString("thumbnail_url")) return playerArgs.getString("thumbnail_url");
} catch (Exception ignored) {
// Try other method...
}
try {
return videoInfoPage.get("thumbnail_url");
} catch (Exception e) {
String url = null;
try {
url = doc.select("link[itemprop=\"thumbnailUrl\"]").first().attr("abs:href");
} catch (Exception ignored) {}
if (url == null) {
throw new ParsingException("Could not get thumbnail url", e);
}
return url;
}
}
@Nonnull
@ -174,11 +183,17 @@ public class YoutubeStreamExtractor extends StreamExtractor {
public String getDescription() throws ParsingException {
assertPageFetched();
try {
// first try to get html-formatted description
return parseHtmlAndGetFullLinks(doc.select("p[id=\"eow-description\"]").first().html());
} catch (Exception e) {
try {
// fallback to raw non-html description
return playerResponse.getObject("videoDetails").getString("shortDescription");
} catch (Exception ignored) {
throw new ParsingException("Could not get the description", e);
}
}
}
// onclick="yt.www.watch.player.seekTo(0*3600+00*60+00);return false;"
// :00 is NOT recognized as a timestamp in description or comments.
@ -269,25 +284,22 @@ public class YoutubeStreamExtractor extends StreamExtractor {
public long getLength() throws ParsingException {
assertPageFetched();
// try getting duration from playerargs
try {
String durationMs = playerResponse
.getObject("streamingData")
.getArray("formats")
.getObject(0)
.getString("approxDurationMs");
return Long.parseLong(durationMs)/1000;
} catch (Exception e) {
}
//try getting value from age gated video
try {
String duration = playerResponse
.getObject("videoDetails")
.getString("lengthSeconds");
return Long.parseLong(duration);
} catch (Exception e) {
throw new ParsingException("Every methode to get the duration has failed: ", e);
try {
String durationMs = playerResponse
.getObject("streamingData")
.getArray("formats")
.getObject(0)
.getString("approxDurationMs");
return Math.round(Long.parseLong(durationMs)/1000.0f);
} catch (Exception ignored) {
throw new ParsingException("Could not get duration", e);
}
}
}
@ -307,11 +319,15 @@ public class YoutubeStreamExtractor extends StreamExtractor {
try {
if (getStreamType().equals(StreamType.LIVE_STREAM)) {
return getLiveStreamWatchingCount();
} else {
return Long.parseLong(playerResponse.getObject("videoDetails").getString("viewCount"));
}
} catch (Exception e) {
try {
return Long.parseLong(doc.select("meta[itemprop=interactionCount]").attr(CONTENT));
} catch (Exception e) {//todo: find fallback method
throw new ParsingException("Could not get number of views", e);
} catch (Exception ignored) {
throw new ParsingException("Could not get view count", e);
}
}
}
@ -373,7 +389,10 @@ public class YoutubeStreamExtractor extends StreamExtractor {
try {
likesString = button.select("span.yt-uix-button-content").first().text();
} catch (NullPointerException e) {
//if this kicks in our button has no content and therefore likes/dislikes are disabled
//if this kicks in our button has no content and therefore ratings must be disabled
if (playerResponse.getObject("videoDetails").getBoolean("allowRatings")) {
throw new ParsingException("Ratings are enabled even though the like button is missing", e);
}
return -1;
}
return Integer.parseInt(Utils.removeNonDigitCharacters(likesString));
@ -393,7 +412,10 @@ public class YoutubeStreamExtractor extends StreamExtractor {
try {
dislikesString = button.select("span.yt-uix-button-content").first().text();
} catch (NullPointerException e) {
//if this kicks in our button has no content and therefore likes/dislikes are disabled
//if this kicks in our button has no content and therefore ratings must be disabled
if (playerResponse.getObject("videoDetails").getBoolean("allowRatings")) {
throw new ParsingException("Ratings are enabled even though the dislike button is missing", e);
}
return -1;
}
return Integer.parseInt(Utils.removeNonDigitCharacters(dislikesString));
@ -409,60 +431,59 @@ public class YoutubeStreamExtractor extends StreamExtractor {
public String getUploaderUrl() throws ParsingException {
assertPageFetched();
try {
return doc.select("div[class=\"yt-user-info\"]").first().children()
.select("a").first().attr("abs:href");
return "https://www.youtube.com/channel/" +
playerResponse.getObject("videoDetails").getString("channelId");
} catch (Exception e) {
String uploaderUrl = null;
try {
uploaderUrl = doc.select("div[class=\"yt-user-info\"]").first().children()
.select("a").first().attr("abs:href");
} catch (Exception ignored) {}
if (uploaderUrl == null) {
throw new ParsingException("Could not get channel link", e);
}
return uploaderUrl;
}
@Nullable
private String getStringFromMetaData(String field) {
assertPageFetched();
String value = null;
if(playerArgs != null) {
// This can not fail
value = playerArgs.getString(field);
}
if(value == null) {
// This can not fail too
value = videoInfoPage.get(field);
}
return value;
}
@Nonnull
@Override
public String getUploaderName() throws ParsingException {
assertPageFetched();
String name = getStringFromMetaData("author");
try {
return playerResponse.getObject("videoDetails").getString("author");
} catch (Exception e) {
String name = null;
try {
name = doc.select("div.yt-user-info").first().text();
} catch (Exception ignored) {}
if (name == null) {
try {
// Fallback to HTML method
name = doc.select("div.yt-user-info").first().text();
} catch (Exception e) {
throw new ParsingException("Could not get uploader name", e);
}
}
if(name == null || name.isEmpty()) {
throw new ParsingException("Could not get uploader name");
}
return name;
}
}
@Nonnull
@Override
public String getUploaderAvatarUrl() throws ParsingException {
assertPageFetched();
String uploaderAvatarUrl = null;
try {
return doc.select("a[class*=\"yt-user-photo\"]").first()
uploaderAvatarUrl = doc.select("a[class*=\"yt-user-photo\"]").first()
.select("img").first()
.attr("abs:data-thumb");
} catch (Exception e) {//todo: add fallback method
throw new ParsingException("Could not get uploader thumbnail URL.", e);
throw new ParsingException("Could not get uploader avatar url", e);
}
if (uploaderAvatarUrl == null) {
throw new ParsingException("Could not get uploader avatar url");
}
return uploaderAvatarUrl;
}
@Nonnull
@ -590,12 +611,12 @@ public class YoutubeStreamExtractor extends StreamExtractor {
public StreamType getStreamType() throws ParsingException {
assertPageFetched();
try {
if (playerArgs != null && (playerArgs.has("ps") && playerArgs.get("ps").toString().equals("live") ||
(!playerResponse.getObject("streamingData").has(FORMATS)))) {
if (!playerResponse.getObject("streamingData").has(FORMATS) ||
(playerArgs != null && playerArgs.has("ps") && playerArgs.get("ps").toString().equals("live"))) {
return StreamType.LIVE_STREAM;
}
} catch (Exception e) {
throw new ParsingException("Could not get hls manifest url", e);
throw new ParsingException("Could not get stream type", e);
}
return StreamType.VIDEO_STREAM;
}