glenux-archive/cloutier--bird.makeup
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

better RT extraction

Browse source
This commit is contained in:
Nicolas Constant 2021-02-02 20:05:59 -05:00
parent 32b53e09e2
commit 7ddda8d18c
No known key found for this signature in database
GPG key ID: 1E9F677FB01A5688
1 changed files with 27 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

35
src/BirdsiteLive.Twitter/Extractors/TweetExtractor.cs
View file

@ -23,7 +23,7 @@ namespace BirdsiteLive.Twitter.Extractors
InReplyToStatusId = tweet.InReplyToStatusId, InReplyToStatusId = tweet.InReplyToStatusId,
InReplyToAccount = tweet.InReplyToScreenName, InReplyToAccount = tweet.InReplyToScreenName,
MessageContent = ExtractMessage(tweet), MessageContent = ExtractMessage(tweet),
Media = ExtractMedia(tweet.Media), Media = ExtractMedia(tweet),
CreatedAt = tweet.CreatedAt.ToUniversalTime(), CreatedAt = tweet.CreatedAt.ToUniversalTime(),
IsReply = tweet.InReplyToUserId != null, IsReply = tweet.InReplyToUserId != null,
IsThread = tweet.InReplyToUserId != null && tweet.InReplyToUserId == tweet.CreatedBy.Id, IsThread = tweet.InReplyToUserId != null && tweet.InReplyToUserId == tweet.CreatedBy.Id,
@ -36,10 +36,17 @@ namespace BirdsiteLive.Twitter.Extractors
private string ExtractRetweetUrl(ITweet tweet) private string ExtractRetweetUrl(ITweet tweet)
{ {
if (tweet.IsRetweet && tweet.FullText.Contains("https://t.co/")) if (tweet.IsRetweet)
{ {
var retweetId = tweet.FullText.Split(new[] { "https://t.co/" }, StringSplitOptions.RemoveEmptyEntries).Last(); if (tweet.RetweetedTweet != null)
return $"https://t.co/{retweetId}"; {
return tweet.RetweetedTweet.Url;
}
if (tweet.FullText.Contains("https://t.co/"))
{
var retweetId = tweet.FullText.Split(new[] { "https://t.co/" }, StringSplitOptions.RemoveEmptyEntries).Last();
return $"https://t.co/{retweetId}";
}
} }
return null; return null;
@ -47,8 +54,15 @@ namespace BirdsiteLive.Twitter.Extractors
public string ExtractMessage(ITweet tweet) public string ExtractMessage(ITweet tweet)
{ {
var tweetUrls = tweet.Media.Select(x => x.URL).Distinct();
var message = tweet.FullText; var message = tweet.FullText;
var tweetUrls = tweet.Media.Select(x => x.URL).Distinct();
if (tweet.IsRetweet && tweet.QuotedStatusId == null && message.StartsWith("RT") && tweet.RetweetedTweet != null)
{
message = tweet.RetweetedTweet.FullText;
tweetUrls = tweet.RetweetedTweet.Media.Select(x => x.URL).Distinct();
}
foreach (var tweetUrl in tweetUrls) foreach (var tweetUrl in tweetUrls)
{ {
if(tweet.IsRetweet) if(tweet.IsRetweet)
@ -60,8 +74,10 @@ namespace BirdsiteLive.Twitter.Extractors
if (tweet.QuotedTweet != null) message = $"[Quote {{RT}}]{Environment.NewLine}{message}"; if (tweet.QuotedTweet != null) message = $"[Quote {{RT}}]{Environment.NewLine}{message}";
if (tweet.IsRetweet) if (tweet.IsRetweet)
{ {
if (tweet.RetweetedTweet != null) if (tweet.RetweetedTweet != null && !message.StartsWith("RT"))
message = $"[{{RT}} @{tweet.RetweetedTweet.CreatedBy.ScreenName}]{Environment.NewLine}{message}"; message = $"[{{RT}} @{tweet.RetweetedTweet.CreatedBy.ScreenName}]{Environment.NewLine}{message}";
else if (tweet.RetweetedTweet != null && message.StartsWith($"RT @{tweet.RetweetedTweet.CreatedBy.ScreenName}:"))
message = message.Replace($"RT @{tweet.RetweetedTweet.CreatedBy.ScreenName}:", $"[{{RT}} @{tweet.RetweetedTweet.CreatedBy.ScreenName}]{Environment.NewLine}");
else else
message = message.Replace("RT", "[{{RT}}]"); message = message.Replace("RT", "[{{RT}}]");
} }
@ -73,10 +89,13 @@ namespace BirdsiteLive.Twitter.Extractors
return message; return message;
} }
public ExtractedMedia[] ExtractMedia(List<IMediaEntity> media) public ExtractedMedia[] ExtractMedia(ITweet tweet)
{ {
var result = new List<ExtractedMedia>(); var media = tweet.Media;
if (tweet.IsRetweet && tweet.RetweetedTweet != null)
media = tweet.RetweetedTweet.Media;
var result = new List<ExtractedMedia>();
foreach (var m in media) foreach (var m in media)
{ {
var mediaUrl = GetMediaUrl(m); var mediaUrl = GetMediaUrl(m);