Merge branch 'fix_docs' into 'master'

docs: add mastodon_screen_name

See merge request veretcle/scootaloo!36

Cargo.lock

@@ -2103,7 +2103,7 @@ dependencies = [
 
 [[package]]
 name = "scootaloo"
-version = "0.9.0"
+version = "0.10.0"
 dependencies = [
  "chrono",
  "clap",

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "scootaloo"
-version = "0.9.0"
+version = "0.10.0"
 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"
@@ -46,7 +47,8 @@ This will give you the end of the TOML file. It will look like this:
 
 ```toml
 [mastodon.nintendojofr] ## account
-twitter_screen_name="NintendojoFR" ## User Timeline to copy
+twitter_screen_name = "NintendojoFR" ## User Timeline to copy
+mastodon_screen_name = "nintendojofr" ## optional, Mastodon account name used for smart mentions
 base = "https://m.nintendojo.fr"
 client_id = "MYCLIENTID"
 client_secret = "MYCLIENTSECRET"
@@ -56,6 +58,11 @@ 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 +16,14 @@ 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 mastodon_screen_name: Option<String>,
+    pub twitter_page_size: Option<i32>,
     pub base: String,
     pub client_id: String,
     pub client_secret: String,

src/lib.rs

@@ -19,15 +19,15 @@ mod state;
 pub use state::{init_db, migrate_db};
 use state::{read_state, write_state, TweetToToot};
 
-use elefren::{prelude::*, status_builder::StatusBuilder};
+use elefren::{prelude::*, status_builder::StatusBuilder, Language};
+use futures::StreamExt;
 use log::info;
 use rusqlite::Connection;
-use std::sync::Arc;
+use std::{collections::HashMap, sync::Arc};
 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]
@@ -42,65 +42,73 @@ pub async fn run(config: Config) {
         }),
     ));
 
+    let scootaloo_mentions: HashMap<String, String> = config
+        .mastodon
+        .values()
+        .filter(|s| s.mastodon_screen_name.is_some())
+        .map(|s| {
+            (
+                format!("@{}", s.twitter_screen_name),
+                format!(
+                    "{}@{}",
+                    s.mastodon_screen_name.as_ref().unwrap(),
+                    s.base.split('/').last().unwrap()
+                ),
+            )
+        })
+        .collect();
+
     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 scootaloo_mentions = scootaloo_mentions.clone();
             let token = get_oauth2_token(&config.twitter);
             let task_conn = conn.clone();
 
             spawn(async move {
                 info!("Starting treating {}", &mastodon_config.twitter_screen_name);
-
                 // retrieve the last tweet ID for the username
                 let lconn = task_conn.lock().await;
                 let last_tweet_id = read_state(&lconn, &mastodon_config.twitter_screen_name, None)?
                     .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);
+                    let mut status_text = build_basic_status(tweet, &scootaloo_mentions);
 
                     // building associative media list
                     let (media_url, status_medias) =
@@ -114,10 +122,18 @@ pub async fn run(config: Config) {
 
                     status_builder.status(&status_text).media_ids(status_medias);
 
+                    // theard if necessary
                     if let Some(i) = toot_reply_id {
                         status_builder.in_reply_to(&i);
                     }
 
+                    // language if any
+                    if let Some(l) = &tweet.lang {
+                        if let Some(r) = Language::from_639_1(l) {
+                            status_builder.language(r);
+                        }
+                    }
+
                     // can be activated for test purposes
                     // status_builder.visibility(elefren::status_builder::Visibility::Private);
 

src/main.rs

@@ -15,10 +15,7 @@ fn main() {
                 .short('c')
                 .long("config")
                 .value_name("CONFIG_FILE")
-                .help(&format!(
-                    "TOML config file for scootaloo (default {})",
-                    DEFAULT_CONFIG_PATH
-                ))
+                .help("TOML config file for scootaloo")
                 .num_args(1)
                 .default_value(DEFAULT_CONFIG_PATH)
                 .display_order(1),
@@ -28,7 +25,7 @@ fn main() {
                 .short('l')
                 .long("loglevel")
                 .value_name("LOGLEVEL")
-                .help("Log level. Valid values are: Off, Warn, Error, Info, Debug")
+                .help("Log level")
                 .num_args(1)
                 .value_parser(["Off", "Warn", "Error", "Info", "Debug"])
                 .display_order(2),

src/mastodon.rs

@@ -4,7 +4,7 @@ use egg_mode::{
     entities::{MentionEntity, UrlEntity},
     tweet::Tweet,
 };
-use elefren::{apps::App, prelude::*, scopes::Scopes};
+use elefren::{apps::App, prelude::*, scopes::Read, scopes::Scopes, scopes::Write};
 use html_escape::decode_html_entities;
 use std::{borrow::Cow, collections::HashMap, io::stdin};
 
@@ -48,14 +48,18 @@ pub fn get_mastodon_token(masto: &MastodonConfig) -> Mastodon {
 }
 
 /// Builds toot text from tweet
-pub fn build_basic_status(tweet: &Tweet) -> String {
+pub fn build_basic_status(tweet: &Tweet, mentions: &HashMap<String, String>) -> String {
     let mut toot = tweet.text.to_owned();
 
     for decoded_url in decode_urls(&tweet.entities.urls) {
         toot = toot.replace(&decoded_url.0, &decoded_url.1);
     }
 
-    for decoded_mention in twitter_mentions(&tweet.entities.user_mentions) {
+    for decoded_mention in twitter_mentions(&tweet.entities.user_mentions)
+        .into_iter()
+        .chain(mentions.to_owned())
+        .collect::<HashMap<String, String>>()
+    {
         toot = toot.replace(&decoded_mention.0, &decoded_mention.1);
     }
 
@@ -70,7 +74,12 @@ pub fn register(host: &str, screen_name: &str) {
     builder
         .client_name(Cow::from(env!("CARGO_PKG_NAME").to_string()))
         .redirect_uris(Cow::from("urn:ietf:wg:oauth:2.0:oob".to_string()))
-        .scopes(Scopes::write_all())
+        .scopes(
+            Scopes::write(Write::Accounts)
+                .and(Scopes::write(Write::Media))
+                .and(Scopes::write(Write::Statuses))
+                .and(Scopes::read(Read::Accounts)),
+        )
         .website(Cow::from(
             "https://framagit.org/veretcle/scootaloo".to_string(),
         ));
@@ -99,13 +108,19 @@ pub fn register(host: &str, screen_name: &str) {
 
     let toml = toml::to_string(&*mastodon).unwrap();
 
+    let current_account = mastodon
+        .verify_credentials()
+        .expect("Unable to access account information!");
+
     println!(
         "Please insert the following block at the end of your configuration file:
 [mastodon.{}]
 twitter_screen_name = \"{}\"
+mastodon_screen_name = \"{}\"
 {}",
         screen_name.to_lowercase(),
         screen_name,
+        current_account.username,
         toml
     );
 }
@@ -192,7 +207,13 @@ mod tests {
                             range: (80, 95),
                             name: "Nintendo France".to_string(),
                             screen_name: "NintendoFrance".to_string(),
-                        }
+                        },
+                        MentionEntity {
+                            id: 999999999,
+                            range: (80, 95),
+                            name: "Willy Wonka".to_string(),
+                            screen_name: "WillyWonka".to_string(),
+                        },
                     ],
                 media: None,
             },
@@ -213,7 +234,7 @@ mod tests {
             retweeted: None,
             retweeted_status: None,
             source: None,
-            text: "Mother 1 &amp; 2 sur le NES/SNES online !\nDispo maintenant. cc @NintendoFrance https://t.co/zXw0FfX2Nt".to_string(),
+            text: "Mother 1 &amp; 2 sur le NES/SNES online !\nDispo maintenant. cc @NintendoFrance @WillyWonka https://t.co/zXw0FfX2Nt".to_string(),
             truncated: false,
             user: None,
             withheld_copyright: false,
@@ -221,8 +242,13 @@ mod tests {
             withheld_scope: None,
         };
 
-        let t_out = build_basic_status(&t);
+        let s: HashMap<String, String> = HashMap::from([(
+            "@WillyWonka".to_string(),
+            "@WillyWonka@chocolatefactory.org".to_string(),
+        )]);
 
-        assert_eq!(&t_out, "Mother 1 & 2 sur le NES/SNES online !\nDispo maintenant. cc @NintendoFrance@twitter.com https://www.youtube.com/watch?v=w5TrSaoYmZ8");
+        let t_out = build_basic_status(&t, &s);
+
+        assert_eq!(&t_out, "Mother 1 & 2 sur le NES/SNES online !\nDispo maintenant. cc @NintendoFrance@twitter.com @WillyWonka@chocolatefactory.org https://www.youtube.com/watch?v=w5TrSaoYmZ8");
     }
 }

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 +1,80 @@
 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_mastodon_screen_name() {
+    let parse_good_toml = parse_toml("tests/good_test_mastodon_screen_name.toml");
+
+    assert_eq!(
+        parse_good_toml
+            .mastodon
+            .get("0")
+            .unwrap()
+            .mastodon_screen_name,
+        Some("tarace".to_string())
+    );
+
+    assert_eq!(
+        parse_good_toml
+            .mastodon
+            .get("1")
+            .unwrap()
+            .mastodon_screen_name,
+        None
+    );
+}
+
 #[test]
 fn test_parse_good_toml() {
     let parse_good_toml = parse_toml("tests/good_test.toml");
@@ -9,6 +84,7 @@ 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_mastodon_screen_name.toml

@@ -0,0 +1,28 @@
+[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"
+mastodon_screen_name="tarace"
+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="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/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"