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

337 lines
17 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;
2022-12-29 09:58:08 -05:00
using BirdsiteLive.DAL.Contracts;
using BirdsiteLive.DAL.Models;
2020-03-21 18:58:23 -04:00
namespace BirdsiteLive.Twitter
{
public interface ITwitterTweetsService
2020-03-21 18:58:23 -04:00
{
Task<ExtractedTweet> GetTweetAsync(long statusId);
2022-12-26 11:13:00 -05:00
Task<ExtractedTweet[]> GetTimelineAsync(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;
2022-12-28 15:17:48 -05:00
private readonly ICachedTwitterUserService _twitterUserService;
2022-12-29 09:58:08 -05:00
private readonly ITwitterUserDal _twitterUserDal;
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
2022-12-29 09:58:08 -05:00
public TwitterTweetsService(ITwitterAuthenticationInitializer twitterAuthenticationInitializer, ITwitterStatisticsHandler statisticsHandler, ICachedTwitterUserService twitterUserService, ITwitterUserDal twitterUserDal, 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;
2022-12-29 09:58:08 -05:00
_twitterUserDal = twitterUserDal;
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
public async Task<ExtractedTweet> GetTweetAsync(long statusId)
2020-07-01 22:45:43 -04:00
{
2022-11-26 16:42:15 -05:00
var client = await _twitterAuthenticationInitializer.MakeHttpClient();
string reqURL = "https://twitter.com/i/api/graphql/BoHLKeBvibdYDiJON1oqTg/TweetDetail?variables=%7B%22focalTweetId%22%3A%22"
+ statusId + "%22%2C%22referrer%22%3A%22profile%22%2C%22rux_context%22%3A%22HHwWgICypZb4saYsAAAA%22%2C%22with_rux_injections%22%3Atrue%2C%22includePromotedContent%22%3Atrue%2C%22withCommunity%22%3Atrue%2C%22withQuickPromoteEligibilityTweetFields%22%3Atrue%2C%22withBirdwatchNotes%22%3Afalse%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";
2021-01-25 23:40:30 -05:00
try
{
2022-05-05 20:15:07 -04:00
JsonDocument tweet;
2022-05-13 18:51:23 -04:00
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
2022-05-13 11:29:28 -04:00
2022-11-26 16:42:15 -05:00
var timeline = tweet.RootElement.GetProperty("data").GetProperty("threaded_conversation_with_injections_v2")
2022-11-27 11:33:52 -05:00
.GetProperty("instructions").EnumerateArray().First().GetProperty("entries").EnumerateArray();
2022-05-13 11:29:28 -04:00
2023-01-14 11:16:21 -05:00
return await Extract( timeline.Where(x => x.GetProperty("entryId").GetString() == "tweet-" + statusId).ToArray().First() );
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
}
2022-05-05 20:15:07 -04:00
public async Task<ExtractedTweet[]> GetTimelineAsync(string username, int nberTweets, long fromTweetId = -1)
{
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
2022-12-29 09:58:08 -05:00
long userId;
SyncTwitterUser user = await _twitterUserDal.GetTwitterUserAsync(username);
if (user.TwitterUserId == default)
{
var user2 = await _twitterUserService.GetUserAsync(username);
userId = user2.Id;
await _twitterUserDal.UpdateTwitterUserIdAsync(username, user2.Id);
}
else
{
userId = user.TwitterUserId;
}
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"
2022-12-29 09:58:08 -05:00
+ userId +
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())
{
2022-11-26 14:34:09 -05:00
if (tweet.GetProperty("content").GetProperty("entryType").GetString() != "TimelineTimelineItem")
continue;
2022-12-29 09:58:08 -05:00
try
{
JsonElement userDoc = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("core").GetProperty("user_results");
TwitterUser tweetUser = _twitterUserService.Extract(userDoc);
_twitterUserService.AddUser(tweetUser);
}
catch (Exception _)
{}
2022-11-25 14:48:10 -05:00
try
2022-11-26 16:42:15 -05:00
{
2022-12-28 10:23:46 -05:00
var extractedTweet = await Extract(tweet);
2022-12-28 15:17:48 -05:00
2022-12-28 14:36:16 -05:00
if (extractedTweet.Id == fromTweetId)
break;
2022-11-26 16:42:15 -05:00
2022-12-29 13:02:38 -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
}
2022-12-28 15:17:48 -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
}
2022-12-28 10:23:46 -05:00
private async Task<ExtractedTweet> Extract(JsonElement tweet)
2022-05-08 19:19:09 -04:00
{
2022-11-26 16:42:15 -05:00
JsonElement retweet;
TwitterUser OriginalAuthor;
2022-11-26 16:58:35 -05:00
JsonElement inReplyToPostIdElement;
JsonElement inReplyToUserElement;
string inReplyToUser = null;
long? inReplyToPostId = null;
bool isReply = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.TryGetProperty("in_reply_to_status_id_str", out inReplyToPostIdElement);
tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.TryGetProperty("in_reply_to_screen_name", out inReplyToUserElement);
if (isReply)
{
inReplyToPostId = Int64.Parse(inReplyToPostIdElement.GetString());
inReplyToUser = inReplyToUserElement.GetString();
}
2022-11-26 16:42:15 -05:00
bool isRetweet = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.TryGetProperty("retweeted_status_result", out retweet);
string MessageContent;
if (!isRetweet)
2022-05-13 11:29:28 -04:00
{
2022-11-26 16:42:15 -05:00
MessageContent = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.GetProperty("full_text").GetString();
OriginalAuthor = null;
2022-05-14 11:19:35 -04:00
}
2022-11-26 16:42:15 -05:00
else
2022-05-14 11:19:35 -04:00
{
2022-11-26 16:42:15 -05:00
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();
2022-12-28 10:23:46 -05:00
OriginalAuthor = await _twitterUserService.GetUserAsync(OriginalAuthorUsername);
2022-05-13 11:29:28 -04:00
}
2022-11-26 16:42:15 -05:00
string creationTime = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.GetProperty("created_at").GetString().Replace(" +0000", "");
2022-11-27 15:41:55 -05:00
JsonElement extendedEntities;
bool hasMedia = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.TryGetProperty("extended_entities", out extendedEntities);
2022-12-16 10:23:48 -05:00
JsonElement.ArrayEnumerator urls = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.GetProperty("entities").GetProperty("urls").EnumerateArray();
foreach (JsonElement url in urls)
{
string tco = url.GetProperty("url").GetString();
string goodUrl = url.GetProperty("expanded_url").GetString();
MessageContent = MessageContent.Replace(tco, goodUrl);
}
2022-11-27 15:41:55 -05:00
List<ExtractedMedia> Media = new List<ExtractedMedia>();
if (hasMedia)
{
foreach (JsonElement media in extendedEntities.GetProperty("media").EnumerateArray())
{
2022-12-19 19:53:18 -05:00
var type = media.GetProperty("type").GetString();
string url = "";
2023-01-13 10:11:50 -05:00
if (type == "video" || type == "animated_gif")
2022-12-19 19:53:18 -05:00
{
2023-01-13 10:11:50 -05:00
var bitrate = -1;
2022-12-19 19:53:18 -05:00
foreach (JsonElement v in media.GetProperty("video_info").GetProperty("variants").EnumerateArray())
{
if (v.GetProperty("content_type").GetString() != "video/mp4")
continue;
int vBitrate = v.GetProperty("bitrate").GetInt32();
if (vBitrate > bitrate)
{
bitrate = vBitrate;
url = v.GetProperty("url").GetString();
}
}
}
else
{
url = media.GetProperty("media_url_https").GetString();
}
2022-11-27 15:41:55 -05:00
var m = new ExtractedMedia
{
2022-12-19 19:53:18 -05:00
MediaType = GetMediaType(type, media.GetProperty("media_url_https").GetString()),
Url = url,
2022-11-27 15:41:55 -05:00
};
Media.Add(m);
MessageContent = MessageContent.Replace(media.GetProperty("url").GetString(), "");
2022-11-27 15:41:55 -05:00
}
}
2023-01-13 10:36:38 -05:00
bool isQuoteTweet = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.GetProperty("is_quote_status").GetBoolean();
if (isQuoteTweet)
{
string quoteTweetLink = tweet.GetProperty("content").GetProperty("itemContent")
.GetProperty("tweet_results").GetProperty("result").GetProperty("legacy")
.GetProperty("quoted_status_permalink").GetProperty("expanded").GetString();
MessageContent = MessageContent + "\n" + quoteTweetLink;
}
2022-05-08 19:19:09 -04:00
var extractedTweet = new ExtractedTweet
{
2022-11-26 16:42:15 -05:00
Id = Int64.Parse(tweet.GetProperty("sortIndex").GetString()),
2022-11-26 16:58:35 -05:00
InReplyToStatusId = inReplyToPostId,
InReplyToAccount = inReplyToUser,
MessageContent = MessageContent.Trim(),
2022-11-26 16:42:15 -05:00
CreatedAt = DateTime.ParseExact(creationTime, "ddd MMM dd HH:mm:ss yyyy", System.Globalization.CultureInfo.InvariantCulture),
2022-11-26 16:58:35 -05:00
IsReply = isReply,
2022-05-08 19:19:09 -04:00
IsThread = false,
2022-11-26 16:42:15 -05:00
IsRetweet = isRetweet,
2022-11-27 15:41:55 -05:00
Media = Media.Count() == 0 ? null : Media.ToArray(),
2022-05-09 20:31:18 -04:00
RetweetUrl = "https://t.co/123",
2022-11-26 16:42:15 -05:00
OriginalAuthor = OriginalAuthor,
2022-05-08 19:19:09 -04:00
};
2022-11-26 16:42:15 -05:00
2022-05-08 19:19:09 -04:00
return extractedTweet;
2022-11-26 16:42:15 -05:00
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
}