Merge branch 'filter_tweet' into 'master'

refactor: eliminate response tweet earlier

See merge request veretcle/scootaloo!30

Cargo.lock

@@ -2103,7 +2103,7 @@ dependencies = [
 
 [[package]]
 name = "scootaloo"
-version = "0.9.0"
+version = "0.9.2"
 dependencies = [
  "chrono",
  "clap",

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "scootaloo"
-version = "0.9.0"
+version = "0.9.2"
 authors = ["VC <veretcle+framagit@mateu.be>"]
 edition = "2021"
 

README.md

@@ -22,6 +22,7 @@ rate_limiting = 4 ## optional, default 4, number of accounts handled simultaneou
 
 [twitter]
 ## Consumer/Access key for Twitter (can be generated at https://developer.twitter.com/en/apps)
+page_size = 20 ## optional, default 200, max number of tweet retrieved
 consumer_key = "MYCONSUMERKEY"
 consumer_secret = "MYCONSUMERSECRET"
 access_key = "MYACCESSKEY"
@@ -56,6 +57,11 @@ token = "MYTOKEN"
 
 You can add other account if you like, after the `[mastodon]` moniker. Scootaloo would theorically support an unlimited number of accounts.
 
+You can also add a custom twitter page size in this section that would override the global (under the `twitter` moniker) and default one (200), like so:
+```
+twitter_page_size = 40
+```
+
 ## Running
 
 You can then run the application via `cron` for example. Here is the generic usage:

src/config.rs

@@ -16,11 +16,13 @@ pub struct TwitterConfig {
     pub consumer_secret: String,
     pub access_key: String,
     pub access_secret: String,
+    pub page_size: Option<i32>,
 }
 
 #[derive(Debug, Deserialize)]
 pub struct MastodonConfig {
     pub twitter_screen_name: String,
+    pub twitter_page_size: Option<i32>,
     pub base: String,
     pub client_id: String,
     pub client_secret: String,

src/lib.rs

@@ -28,6 +28,7 @@ use tokio::{spawn, sync::Mutex};
 use futures::StreamExt;
 
 const DEFAULT_RATE_LIMIT: usize = 4;
+const DEFAULT_PAGE_SIZE: i32 = 200;
 
 /// This is where the magic happens
 #[tokio::main]
@@ -44,6 +45,11 @@ pub async fn run(config: Config) {
 
     let mut stream = futures::stream::iter(config.mastodon.into_values())
         .map(|mastodon_config| {
+            // calculate Twitter page size
+            let page_size = mastodon_config
+                .twitter_page_size
+                .unwrap_or_else(|| config.twitter.page_size.unwrap_or(DEFAULT_PAGE_SIZE));
+
             // create temporary value for each task
             let scootaloo_cache_path = config.scootaloo.cache_path.clone();
             let token = get_oauth2_token(&config.twitter);
@@ -58,46 +64,32 @@ pub async fn run(config: Config) {
                     .map(|r| r.tweet_id);
                 drop(lconn);
 
-                // get user timeline feed (Vec<tweet>)
-                let mut feed =
-                    get_user_timeline(&mastodon_config.twitter_screen_name, &token, last_tweet_id)
-                        .await?;
-
-                // empty feed -> exiting
-                if feed.is_empty() {
-                    info!("Nothing to retrieve since last time, exiting…");
-                    return Ok(());
-                }
+                // get reversed, curated user timeline
+                let feed = get_user_timeline(
+                    &mastodon_config.twitter_screen_name,
+                    &token,
+                    last_tweet_id,
+                    page_size,
+                )
+                .await?;
 
                 // get Mastodon instance
                 let mastodon = get_mastodon_token(&mastodon_config);
 
-                // order needs to be chronological
-                feed.reverse();
-
                 for tweet in &feed {
                     info!("Treating Tweet {} inside feed", tweet.id);
-                    // initiate the toot_reply_id var
-                    let mut toot_reply_id: Option<String> = None;
-                    // determine if the tweet is part of a thread (response to self) or a standard response
-                    if let Some(r) = &tweet.in_reply_to_screen_name {
-                        if r.to_lowercase() != mastodon_config.twitter_screen_name.to_lowercase() {
-                            // we are responding not threading
-                            info!("Tweet is a direct response, skipping");
-                            continue;
-                        }
-                        info!("Tweet is a thread");
-                        // get the corresponding toot id
-                        let lconn = task_conn.lock().await;
-                        toot_reply_id = read_state(
+                    let lconn = task_conn.lock().await;
+                    // initiate the toot_reply_id var and retrieve the corresponding toot_id
+                    let toot_reply_id: Option<String> = tweet.in_reply_to_user_id.and_then(|_| {
+                        read_state(
                             &lconn,
                             &mastodon_config.twitter_screen_name,
                             tweet.in_reply_to_status_id,
                         )
                         .unwrap_or(None)
-                        .map(|s| s.toot_id);
-                        drop(lconn);
-                    };
+                        .map(|s| s.toot_id)
+                    });
+                    drop(lconn);
 
                     // build basic status by just yielding text and dereferencing contained urls
                     let mut status_text = build_basic_status(tweet);

src/twitter.rs

@@ -27,19 +27,31 @@ pub fn get_oauth2_token(config: &TwitterConfig) -> Token {
     }
 }
 
-/// Gets Twitter user timeline
+/// Gets Twitter user timeline, eliminate responses to others and reverse it
 pub async fn get_user_timeline(
     screen_name: &str,
     token: &Token,
     lid: Option<u64>,
+    page_size: i32,
 ) -> Result<Vec<Tweet>, Box<dyn Error>> {
     // fix the page size to 200 as it is the maximum Twitter authorizes
     let (_, feed) = user_timeline(UserID::from(screen_name.to_owned()), true, false, token)
-        .with_page_size(200)
+        .with_page_size(page_size)
         .older(lid)
         .await?;
 
-    Ok(feed.to_vec())
+    let mut feed: Vec<Tweet> = feed
+        .iter()
+        .cloned()
+        .filter(|t| match &t.in_reply_to_screen_name {
+            Some(r) => r.to_lowercase() == screen_name.to_lowercase(),
+            None => true,
+        })
+        .collect();
+
+    feed.reverse();
+
+    Ok(feed)
 }
 
 /// Retrieves a single media from a tweet and store it in a temporary file

tests/config.rs

@@ -1,5 +1,57 @@
 use scootaloo::parse_toml;
 
+#[test]
+fn test_page_size() {
+    const DEFAULT_PAGE_SIZE: i32 = 200;
+    let toml = parse_toml("tests/page_size.toml");
+
+    assert_eq!(toml.twitter.page_size, Some(100));
+
+    assert_eq!(toml.mastodon.get("0").unwrap().twitter_page_size, None);
+
+    assert_eq!(toml.mastodon.get("1").unwrap().twitter_page_size, Some(42));
+
+    // this is the exact line that is used inside fn run() to determine the twitter page size
+    // passed to fn get_user_timeline()
+    let page_size_for_0 = toml
+        .mastodon
+        .get("0")
+        .unwrap()
+        .twitter_page_size
+        .unwrap_or_else(|| toml.twitter.page_size.unwrap_or(DEFAULT_PAGE_SIZE));
+    let page_size_for_1 = toml
+        .mastodon
+        .get("1")
+        .unwrap()
+        .twitter_page_size
+        .unwrap_or_else(|| toml.twitter.page_size.unwrap_or(DEFAULT_PAGE_SIZE));
+
+    assert_eq!(page_size_for_0, 100);
+    assert_eq!(page_size_for_1, 42);
+
+    let toml = parse_toml("tests/no_page_size.toml");
+
+    assert_eq!(toml.twitter.page_size, None);
+    assert_eq!(toml.mastodon.get("0").unwrap().twitter_page_size, None);
+
+    // and same here
+    let page_size_for_0 = toml
+        .mastodon
+        .get("0")
+        .unwrap()
+        .twitter_page_size
+        .unwrap_or_else(|| toml.twitter.page_size.unwrap_or(DEFAULT_PAGE_SIZE));
+
+    assert_eq!(page_size_for_0, 200);
+}
+
+#[test]
+fn test_parse_good_toml_rate_limit() {
+    let parse_good_toml = parse_toml("tests/good_test_rate_limit.toml");
+
+    assert_eq!(parse_good_toml.scootaloo.rate_limit, Some(69 as usize));
+}
+
 #[test]
 fn test_parse_good_toml() {
     let parse_good_toml = parse_toml("tests/good_test.toml");
@@ -9,6 +61,7 @@ fn test_parse_good_toml() {
         "/var/random/scootaloo.sqlite"
     );
     assert_eq!(parse_good_toml.scootaloo.cache_path, "/tmp/scootaloo");
+    assert_eq!(parse_good_toml.scootaloo.rate_limit, None);
 
     assert_eq!(parse_good_toml.twitter.consumer_key, "rand consumer key");
     assert_eq!(parse_good_toml.twitter.consumer_secret, "secret");

tests/good_test_rate_limit.toml

@@ -0,0 +1,20 @@
+[scootaloo]
+
+db_path="/var/random/scootaloo.sqlite"
+cache_path="/tmp/scootaloo"
+rate_limit=69
+
+[twitter]
+consumer_key="rand consumer key"
+consumer_secret="secret"
+access_key="rand access key"
+access_secret="super secret"
+
+[mastodon]
+[mastodon.tamerelol]
+twitter_screen_name="tamerelol"
+base = "https://m.nintendojo.fr"
+client_id = "rand client id"
+client_secret = "secret"
+redirect = "urn:ietf:wg:oauth:2.0:oob"
+token = "super secret"

tests/no_page_size.toml

@@ -0,0 +1,19 @@
+[scootaloo]
+
+db_path="/var/random/scootaloo.sqlite"
+cache_path="/tmp/scootaloo"
+
+[twitter]
+consumer_key="rand consumer key"
+consumer_secret="secret"
+access_key="rand access key"
+access_secret="super secret"
+
+[mastodon]
+[mastodon.0]
+twitter_screen_name="tamerelol"
+base = "https://m.nintendojo.fr"
+client_id = "rand client id"
+client_secret = "secret"
+redirect = "urn:ietf:wg:oauth:2.0:oob"
+token = "super secret"

tests/page_size.toml

@@ -0,0 +1,29 @@
+[scootaloo]
+
+db_path="/var/random/scootaloo.sqlite"
+cache_path="/tmp/scootaloo"
+
+[twitter]
+consumer_key="rand consumer key"
+consumer_secret="secret"
+access_key="rand access key"
+access_secret="super secret"
+page_size=100
+
+[mastodon]
+[mastodon.0]
+twitter_screen_name="tamerelol"
+base = "https://m.nintendojo.fr"
+client_id = "rand client id"
+client_secret = "secret"
+redirect = "urn:ietf:wg:oauth:2.0:oob"
+token = "super secret"
+
+[mastodon.1]
+twitter_screen_name="tonperemdr"
+twitter_page_size=42
+base = "https://m.nintendojo.fr"
+client_id = "rand client id"
+client_secret = "secret"
+redirect = "urn:ietf:wg:oauth:2.0:oob"
+token = "super secret"