cloutier--bird.makeup/src/BirdsiteLive.Twitter/TwitterTweetsService.cs

326 lines
16 KiB
C#
Raw Normal View History

2021-01-25 23:40:30 -05:00
using System;
using System.Collections.Generic;
2020-07-18 23:35:19 -04:00
using System.Linq;
2022-05-13 11:29:28 -04:00
using System.IO;
2022-05-05 20:15:07 -04:00
using System.Net.Http;
using System.Text.Json;
2022-11-25 15:05:05 -05:00
using System.Text.Json.Nodes;
2022-05-05 20:15:07 -04:00
using System.Threading.Tasks;
2020-03-22 01:29:51 -04:00
using BirdsiteLive.Common.Settings;
using BirdsiteLive.Statistics.Domain;
2020-03-22 01:29:51 -04:00
using BirdsiteLive.Twitter.Models;
2021-01-30 00:22:29 -05:00
using BirdsiteLive.Twitter.Tools;
2021-01-25 23:40:30 -05:00
using Microsoft.Extensions.Logging;
2022-05-10 17:32:07 -04:00
using System.Text.RegularExpressions;
2020-03-21 18:58:23 -04:00
namespace BirdsiteLive.Twitter
{
public interface ITwitterTweetsService
2020-03-21 18:58:23 -04:00
{
2020-07-22 20:19:40 -04:00
ExtractedTweet GetTweet(long statusId);
ExtractedTweet[] GetTimeline(string username, int nberTweets, long fromTweetId = -1);
2020-03-21 18:58:23 -04:00
}
public class TwitterTweetsService : ITwitterTweetsService
2020-03-21 18:58:23 -04:00
{
2021-01-30 00:22:29 -05:00
private readonly ITwitterAuthenticationInitializer _twitterAuthenticationInitializer;
private readonly ITwitterStatisticsHandler _statisticsHandler;
private readonly ITwitterUserService _twitterUserService;
2021-01-25 23:40:30 -05:00
private readonly ILogger<TwitterTweetsService> _logger;
2022-05-05 20:15:07 -04:00
private HttpClient _httpClient = new HttpClient();
2020-03-21 18:58:23 -04:00
#region Ctor
2022-05-08 19:19:09 -04:00
public TwitterTweetsService(ITwitterAuthenticationInitializer twitterAuthenticationInitializer, ITwitterStatisticsHandler statisticsHandler, ITwitterUserService twitterUserService, ILogger<TwitterTweetsService> logger)
2020-03-21 18:58:23 -04:00
{
2021-01-30 00:22:29 -05:00
_twitterAuthenticationInitializer = twitterAuthenticationInitializer;
_statisticsHandler = statisticsHandler;
_twitterUserService = twitterUserService;
2021-01-25 23:40:30 -05:00
_logger = logger;
2020-03-21 18:58:23 -04:00
}
#endregion
2021-01-25 23:40:30 -05:00
2022-05-05 20:15:07 -04:00
2020-07-22 20:19:40 -04:00
public ExtractedTweet GetTweet(long statusId)
2022-05-05 20:15:07 -04:00
{
return GetTweetAsync(statusId).Result;
}
public async Task<ExtractedTweet> GetTweetAsync(long statusId)
2020-07-01 22:45:43 -04:00
{
2021-01-25 23:40:30 -05:00
try
{
2022-11-25 13:48:49 -05:00
var client = await _twitterAuthenticationInitializer.MakeHttpClient();
2022-05-05 20:15:07 -04:00
JsonDocument tweet;
2022-05-13 18:51:23 -04:00
var reqURL = "https://api.twitter.com/2/tweets/" + statusId
+ "?expansions=author_id,referenced_tweets.id,attachments.media_keys,entities.mentions.username,referenced_tweets.id.author_id&tweet.fields=id,created_at,text,author_id,in_reply_to_user_id,referenced_tweets,attachments,withheld,geo,entities,public_metrics,possibly_sensitive,source,lang,context_annotations,conversation_id,reply_settings&user.fields=id,created_at,name,username,protected,verified,withheld,profile_image_url,location,url,description,entities,pinned_tweet_id,public_metrics&media.fields=media_key,duration_ms,height,preview_image_url,type,url,width,public_metrics,alt_text,variants";
using (var request = new HttpRequestMessage(new HttpMethod("GET"), reqURL))
2022-11-25 13:48:49 -05:00
{
var httpResponse = await client.SendAsync(request);
2022-05-05 20:15:07 -04:00
httpResponse.EnsureSuccessStatusCode();
var c = await httpResponse.Content.ReadAsStringAsync();
tweet = JsonDocument.Parse(c);
}
2021-01-30 00:22:29 -05:00
2021-01-25 23:40:30 -05:00
_statisticsHandler.CalledTweetApi();
if (tweet == null) return null; //TODO: test this
2022-05-13 11:29:28 -04:00
2022-05-17 18:17:05 -04:00
JsonElement mediaExpension = default;
try
{
tweet.RootElement.GetProperty("includes").TryGetProperty("media", out mediaExpension);
}
catch (Exception)
{ }
2022-05-13 11:29:28 -04:00
2022-05-13 18:54:40 -04:00
//return tweet.RootElement.GetProperty("data").EnumerateArray().Select<JsonElement, ExtractedTweet>(x => Extract(x, mediaExpension)).ToArray().First();
return Extract( tweet.RootElement.GetProperty("data"), mediaExpension);
2021-01-25 23:40:30 -05:00
}
catch (Exception e)
{
_logger.LogError(e, "Error retrieving tweet {TweetId}", statusId);
return null;
}
2020-07-22 20:19:40 -04:00
}
public ExtractedTweet[] GetTimeline(string username, int nberTweets, long fromTweetId = -1)
2020-07-18 23:35:19 -04:00
{
2022-05-05 20:15:07 -04:00
return GetTimelineAsync(username, nberTweets, fromTweetId).Result;
}
public async Task<ExtractedTweet[]> GetTimelineAsync(string username, int nberTweets, long fromTweetId = -1)
{
2022-05-17 17:51:21 -04:00
if (nberTweets < 5)
nberTweets = 5;
if (nberTweets > 100)
nberTweets = 100;
2021-01-30 00:22:29 -05:00
2022-11-25 13:48:49 -05:00
var client = await _twitterAuthenticationInitializer.MakeHttpClient();
2021-01-29 23:10:02 -05:00
var user = _twitterUserService.GetUser(username);
if (user == null || user.Protected) return new ExtractedTweet[0];
2022-11-25 13:48:49 -05:00
2022-11-26 14:08:13 -05:00
var reqURL = "https://twitter.com/i/api/graphql/s0hG9oAmWEYVBqOLJP-TBQ/UserTweetsAndReplies?variables=%7B%22userId%22%3A%22"
2022-05-07 18:35:35 -04:00
+ user.Id +
2022-11-25 13:48:49 -05:00
"%22%2C%22count%22%3A40%2C%22includePromotedContent%22%3Atrue%2C%22withQuickPromoteEligibilityTweetFields%22%3Atrue%2C%22withSuperFollowsUserFields%22%3Atrue%2C%22withDownvotePerspective%22%3Afalse%2C%22withReactionsMetadata%22%3Afalse%2C%22withReactionsPerspective%22%3Afalse%2C%22withSuperFollowsTweetFields%22%3Atrue%2C%22withVoice%22%3Atrue%2C%22withV2Timeline%22%3Atrue%7D&features=%7B%22responsive_web_twitter_blue_verified_badge_is_enabled%22%3Atrue%2C%22verified_phone_label_enabled%22%3Afalse%2C%22responsive_web_graphql_timeline_navigation_enabled%22%3Atrue%2C%22unified_cards_ad_metadata_container_dynamic_card_content_query_enabled%22%3Atrue%2C%22tweetypie_unmention_optimization_enabled%22%3Atrue%2C%22responsive_web_uc_gql_enabled%22%3Atrue%2C%22vibe_api_enabled%22%3Atrue%2C%22responsive_web_edit_tweet_api_enabled%22%3Atrue%2C%22graphql_is_translatable_rweb_tweet_is_translatable_enabled%22%3Afalse%2C%22standardized_nudges_misinfo%22%3Atrue%2C%22tweet_with_visibility_results_prefer_gql_limited_actions_policy_enabled%22%3Afalse%2C%22interactive_text_enabled%22%3Atrue%2C%22responsive_web_text_conversations_enabled%22%3Afalse%2C%22responsive_web_enhance_cards_enabled%22%3Atrue%7D";
JsonDocument results;
List<ExtractedTweet> extractedTweets = new List<ExtractedTweet>();
2022-05-05 20:15:07 -04:00
try
{
2022-05-07 18:35:35 -04:00
using (var request = new HttpRequestMessage(new HttpMethod("GET"), reqURL))
2022-11-25 13:48:49 -05:00
{
2022-11-25 14:42:26 -05:00
2022-11-25 13:48:49 -05:00
var httpResponse = await client.SendAsync(request);
2022-05-05 20:15:07 -04:00
httpResponse.EnsureSuccessStatusCode();
var c = await httpResponse.Content.ReadAsStringAsync();
2022-11-25 13:48:49 -05:00
results = JsonDocument.Parse(c);
2022-05-05 20:15:07 -04:00
}
_statisticsHandler.CalledTweetApi();
2022-11-25 13:48:49 -05:00
if (results == null) return null; //TODO: test this
2020-07-18 23:35:19 -04:00
}
2022-05-05 20:15:07 -04:00
catch (Exception e)
2020-07-18 23:35:19 -04:00
{
2022-05-05 20:15:07 -04:00
_logger.LogError(e, "Error retrieving timeline ", username);
return null;
2020-07-18 23:35:19 -04:00
}
2022-11-25 13:48:49 -05:00
var timeline = results.RootElement.GetProperty("data").GetProperty("user").GetProperty("result")
.GetProperty("timeline_v2").GetProperty("timeline").GetProperty("instructions").EnumerateArray();
foreach (JsonElement timelineElement in timeline)
2022-05-17 18:17:05 -04:00
{
2022-11-25 13:48:49 -05:00
if (timelineElement.GetProperty("type").GetString() != "TimelineAddEntries")
continue;
2022-05-17 18:17:05 -04:00
2022-11-25 13:48:49 -05:00
foreach (JsonElement tweet in timelineElement.GetProperty("entries").EnumerateArray())
{
2022-11-25 14:48:10 -05:00
try
2022-11-25 13:48:49 -05:00
{
2022-11-25 14:48:10 -05:00
2022-11-26 14:08:13 -05:00
JsonElement retweet;
TwitterUser OriginalAuthor;
bool isRetweet = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.TryGetProperty("retweeted_status_result", out retweet);
string MessageContent;
2022-11-26 14:16:15 -05:00
if (!isRetweet)
2022-11-26 14:08:13 -05:00
{
MessageContent = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.GetProperty("full_text").GetString();
OriginalAuthor = null;
}
else
{
MessageContent = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.GetProperty("retweeted_status_result").GetProperty("result")
.GetProperty("legacy").GetProperty("full_text").GetString();
string OriginalAuthorUsername = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.GetProperty("retweeted_status_result").GetProperty("result")
.GetProperty("core").GetProperty("user_results").GetProperty("result")
.GetProperty("legacy").GetProperty("screen_name").GetString();
OriginalAuthor = _twitterUserService.GetUser(OriginalAuthorUsername);
}
2022-11-25 15:29:47 -05:00
string creationTime = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
2022-11-25 15:51:58 -05:00
.GetProperty("created_at").GetString().Replace(" +0000", "");
2022-11-25 14:48:10 -05:00
var extractedTweet = new ExtractedTweet
{
Id = Int64.Parse(tweet.GetProperty("sortIndex").GetString()),
InReplyToStatusId = null,
InReplyToAccount = null,
2022-11-26 14:08:13 -05:00
MessageContent = MessageContent,
2022-11-25 15:51:58 -05:00
CreatedAt = DateTime.ParseExact(creationTime, "ddd MMM dd HH:mm:ss yyyy", System.Globalization.CultureInfo.InvariantCulture),
2022-11-25 14:48:10 -05:00
IsReply = false,
IsThread = false,
2022-11-26 14:08:13 -05:00
IsRetweet = isRetweet,
2022-11-25 14:48:10 -05:00
Media = null,
RetweetUrl = "https://t.co/123",
2022-11-26 14:08:13 -05:00
OriginalAuthor = OriginalAuthor,
2022-11-25 14:48:10 -05:00
};
2022-11-25 15:05:05 -05:00
extractedTweets.Add(extractedTweet);
2022-11-25 14:48:10 -05:00
}
catch (Exception e)
{
2022-11-25 15:05:05 -05:00
_logger.LogError("Tried getting timeline from user " + username + ", but got error: \n" + e.Message + e.StackTrace + e.Source
+ JsonObject.Create(tweet).ToJsonString(new JsonSerializerOptions { WriteIndented = true }));
2022-11-25 14:48:10 -05:00
}
2022-11-25 13:48:49 -05:00
}
}
2022-05-13 11:29:28 -04:00
2022-11-25 13:48:49 -05:00
return extractedTweets.ToArray();
2022-05-08 19:19:09 -04:00
}
2022-05-13 11:29:28 -04:00
private ExtractedTweet Extract(JsonElement tweet, JsonElement media)
2022-05-08 19:19:09 -04:00
{
2022-05-17 18:00:02 -04:00
var id = Int64.Parse(tweet.GetProperty("id").GetString());
2022-05-08 19:19:09 -04:00
bool IsRetweet = false;
bool IsReply = false;
long? replyId = null;
JsonElement replyAccount;
string? replyAccountString = null;
JsonElement referenced_tweets;
if(tweet.TryGetProperty("in_reply_to_user_id", out replyAccount))
{
replyAccountString = replyAccount.GetString();
}
if(tweet.TryGetProperty("referenced_tweets", out referenced_tweets))
{
var first = referenced_tweets.EnumerateArray().ToList()[0];
if (first.GetProperty("type").GetString() == "retweeted")
{
IsRetweet = true;
2022-05-10 18:03:32 -04:00
var regex = new Regex("RT @([A-Za-z0-9_]+):");
2022-05-10 17:32:07 -04:00
var match = regex.Match(tweet.GetProperty("text").GetString());
var originalAuthor = _twitterUserService.GetUser(match.Groups[1].Value);
2022-05-08 19:19:09 -04:00
var statusId = Int64.Parse(first.GetProperty("id").GetString());
var extracted = GetTweet(statusId);
2022-05-17 18:00:02 -04:00
extracted.RetweetId = id;
2022-05-08 19:19:09 -04:00
extracted.IsRetweet = true;
2022-05-09 20:31:18 -04:00
extracted.OriginalAuthor = originalAuthor;
2022-05-08 19:19:09 -04:00
return extracted;
}
if (first.GetProperty("type").GetString() == "replied_to")
{
IsReply = true;
replyId = Int64.Parse(first.GetProperty("id").GetString());
}
if (first.GetProperty("type").GetString() == "quoted")
{
IsReply = true;
replyId = Int64.Parse(first.GetProperty("id").GetString());
}
}
2022-05-13 11:29:28 -04:00
var extractedMedia = Array.Empty<ExtractedMedia>();
JsonElement attachments;
2022-05-14 11:19:35 -04:00
try
2022-05-13 11:29:28 -04:00
{
2022-05-14 11:19:35 -04:00
if (tweet.TryGetProperty("attachments", out attachments))
2022-05-13 11:29:28 -04:00
{
2022-05-14 11:19:35 -04:00
foreach (JsonElement m in attachments.GetProperty("media_keys").EnumerateArray())
2022-05-13 18:58:31 -04:00
{
2022-05-14 11:19:35 -04:00
var mediaInfo = media.EnumerateArray().Where(x => x.GetProperty("media_key").GetString() == m.GetString()).First();
var mediaType = mediaInfo.GetProperty("type").GetString();
if (mediaType != "photo")
2022-05-13 11:29:28 -04:00
{
2022-05-14 11:19:35 -04:00
continue;
2022-05-13 11:29:28 -04:00
}
2022-05-14 11:19:35 -04:00
var url = mediaInfo.GetProperty("url").GetString();
extractedMedia.Append(
new ExtractedMedia
{
Url = url,
MediaType = GetMediaType(mediaType, url),
}
);
2022-05-13 11:29:28 -04:00
2022-05-14 11:19:35 -04:00
}
2022-05-13 11:29:28 -04:00
}
2022-05-14 11:19:35 -04:00
}
catch (Exception e)
{
2022-05-17 18:00:02 -04:00
_logger.LogError("Tried getting media from tweet " + id + ", but got error: \n" + e.Message + e.StackTrace + e.Source);
2022-05-14 11:19:35 -04:00
2022-05-13 11:29:28 -04:00
}
2022-05-08 19:19:09 -04:00
var extractedTweet = new ExtractedTweet
{
2022-05-17 18:00:02 -04:00
Id = id,
2022-05-08 19:19:09 -04:00
InReplyToStatusId = replyId,
InReplyToAccount = replyAccountString,
MessageContent = tweet.GetProperty("text").GetString(),
2022-05-17 17:51:21 -04:00
CreatedAt = tweet.GetProperty("created_at").GetDateTime(),
2022-05-08 19:19:09 -04:00
IsReply = IsReply,
IsThread = false,
IsRetweet = IsRetweet,
2022-05-13 11:29:28 -04:00
Media = extractedMedia,
2022-05-09 20:31:18 -04:00
RetweetUrl = "https://t.co/123",
OriginalAuthor = null,
2022-05-08 19:19:09 -04:00
};
return extractedTweet;
2020-07-18 23:35:19 -04:00
}
2022-05-13 11:29:28 -04:00
private string GetMediaType(string mediaType, string mediaUrl)
{
switch (mediaType)
{
case "photo":
var pExt = Path.GetExtension(mediaUrl);
switch (pExt)
{
case ".jpg":
case ".jpeg":
return "image/jpeg";
case ".png":
return "image/png";
}
return null;
case "animated_gif":
var vExt = Path.GetExtension(mediaUrl);
switch (vExt)
{
case ".gif":
return "image/gif";
case ".mp4":
return "video/mp4";
}
return "image/gif";
case "video":
return "video/mp4";
}
return null;
}
2020-03-21 18:58:23 -04:00
}
2022-05-07 18:54:06 +00:00
}