Merge branch 'fix_lang' into 'master'

fix: visibility

See merge request veretcle/scootaloo!33

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "scootaloo"
-version = "0.6.0"
+version = "0.9.4"
 authors = ["VC <veretcle+framagit@mateu.be>"]
 edition = "2021"
 
@@ -10,16 +10,17 @@ edition = "2021"
 chrono = "^0.4"
 serde = { version = "1.0", features = ["derive"] }
 toml = "^0.5"
-clap = "^2.34"
+clap = "^4"
-futures = "^0.3"
 egg-mode = "^0.16"
 rusqlite = "^0.27"
-tokio = { version = "1", features = ["full"]}
+tokio = { version = "^1", features = ["full"]}
+futures = "^0.3"
 elefren = "^0.22"
 html-escape = "^0.2"
 reqwest = "^0.11"
 log = "^0.4"
 simple_logger = "^2.1"
+mime = "^0.3"
 
 [profile.release]
 strip = true

README.md

@@ -10,23 +10,25 @@ If any of the last steps failed, the Toot gets published with the exact same tex
 RT are excluded, replies are included when considered part of a thread (reply to self), not the actual replies to other Twitter users.
 
 # Usage
+## Configuring
 
 First up, create a configuration file (default path is `/usr/local/etc/scootaloo.toml`). It will look like this:
 
 ```toml
 [scootaloo]
-
+db_path = "/var/lib/scootaloo/scootaloo.sqlite" ## file containing the SQLite Tweet corresponding Toot DB, must be writeable
-db_path="/var/lib/scootaloo/scootaloo.sqlite" ## file containing the SQLite Tweet corresponding Toot DB, must be writeable
+cache_path = "/tmp/scootaloo" ## a dir where the temporary files will be download, must be writeable
-cache_path="/tmp/scootaloo" ## a dir where the temporary files will be download, must be writeable
+rate_limiting = 4 ## optional, default 4, number of accounts handled simultaneously
 
 [twitter]
-username="NintendojoFR" ## User Timeline to copy
-
 ## Consumer/Access key for Twitter (can be generated at https://developer.twitter.com/en/apps)
-consumer_key="MYCONSUMERKEY"
+page_size = 20 ## optional, default 200, max number of tweet retrieved
-consumer_secret="MYCONSUMERSECRET"
+consumer_key = "MYCONSUMERKEY"
-access_key="MYACCESSKEY"
+consumer_secret = "MYCONSUMERSECRET"
-access_secret="MYACCESSSECRET"
+access_key = "MYACCESSKEY"
+access_secret = "MYACCESSSECRET"
+
+[mastodon]
 ```
 
 Then run the command with the `init` subcommand to initiate the DB:
@@ -44,7 +46,8 @@ scootaloo register --host https://m.nintendojo.fr
 This will give you the end of the TOML file. It will look like this:
 
 ```toml
-[mastodon]
+[mastodon.nintendojofr] ## account
+twitter_screen_name="NintendojoFR" ## User Timeline to copy
 base = "https://m.nintendojo.fr"
 client_id = "MYCLIENTID"
 client_secret = "MYCLIENTSECRET"
@@ -52,6 +55,15 @@ redirect = "urn:ietf:wg:oauth:2.0:oob"
 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:
 
 ```sh
@@ -71,6 +83,7 @@ OPTIONS:
 SUBCOMMANDS:
     help        Prints this message or the help of the given subcommand(s)
     init        Command to init Scootaloo DB
+    migrate     Command to migrate Scootaloo DB
     register    Command to register to a Mastodon Instance
 ```
 
@@ -86,5 +99,17 @@ sqlite3 /var/lib/scootaloo/scootaloo.sqlite
 And inserting the data:
 
 ```sql
-INSERT INTO tweet_to_toot VALUES (1383782580412030982, "");
+INSERT INTO tweet_to_toot VALUES ("<twitter_screen_name>", 1383782580412030982, "<twitter_screen_name>");
 ```
+
+The last value is supposed to be the Toot ID. It cannot be null, so you better initialize it with something unique, like the Twitter Screen Name for example.
+
+# Migrating from Scootaloo ⩽ 0.6.1
+
+The DB scheme has change between version 0.6.x and 0.7.x (this is due to the multi-account nature of Scootaloo from 0.7.x onward). You need to migrate your DB. You can do so by issuing the command:
+
+```
+scootaloo migrate
+```
+
+You can optionnally specify a screen name with the `--name` option. By default, it’ll take the first screen name in the config file.

src/config.rs

@@ -1,25 +1,28 @@
+use std::{collections::HashMap, fs::read_to_string};
+
 use serde::Deserialize;
-use std::fs::read_to_string;
 
 /// General configuration Struct
 #[derive(Debug, Deserialize)]
 pub struct Config {
     pub twitter: TwitterConfig,
-    pub mastodon: MastodonConfig,
+    pub mastodon: HashMap<String, MastodonConfig>,
     pub scootaloo: ScootalooConfig,
 }
 
 #[derive(Debug, Deserialize)]
 pub struct TwitterConfig {
-    pub username: String,
     pub consumer_key: String,
     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,
@@ -31,6 +34,7 @@ pub struct MastodonConfig {
 pub struct ScootalooConfig {
     pub db_path: String,
     pub cache_path: String,
+    pub rate_limit: Option<usize>,
 }
 
 /// Parses the TOML file into a Config Struct

src/error.rs

@@ -1,8 +1,12 @@
 use std::{
+    boxed::Box,
+    convert::From,
     error::Error,
     fmt::{Display, Formatter, Result},
 };
 
+use elefren::Error as elefrenError;
+
 #[derive(Debug)]
 pub struct ScootalooError {
     details: String,
@@ -23,3 +27,15 @@ impl Display for ScootalooError {
         write!(f, "{}", self.details)
     }
 }
+
+impl From<Box<dyn Error>> for ScootalooError {
+    fn from(error: Box<dyn Error>) -> Self {
+        ScootalooError::new(&format!("Error in a subset crate: {}", error))
+    }
+}
+
+impl From<elefrenError> for ScootalooError {
+    fn from(error: elefrenError) -> Self {
+        ScootalooError::new(&format!("Error in elefren crate: {}", error))
+    }
+}

src/lib.rs

@@ -13,145 +13,141 @@ mod twitter;
 use twitter::*;
 
 mod util;
+use crate::util::generate_media_ids;
 
 mod state;
-pub use state::init_db;
+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 log::{debug, error, info, warn};
+use futures::StreamExt;
+use log::info;
 use rusqlite::Connection;
-use std::borrow::Cow;
+use std::sync::Arc;
-use tokio::fs::remove_file;
+use tokio::{spawn, sync::Mutex};
+
+const DEFAULT_RATE_LIMIT: usize = 4;
+const DEFAULT_PAGE_SIZE: i32 = 200;
 
 /// This is where the magic happens
 #[tokio::main]
 pub async fn run(config: Config) {
     // open the SQLite connection
-    let conn = Connection::open(&config.scootaloo.db_path).unwrap_or_else(|e| {
+    let conn = Arc::new(Mutex::new(
-        panic!(
+        Connection::open(&config.scootaloo.db_path).unwrap_or_else(|e| {
-            "Something went wrong when opening the DB {}: {}",
-            &config.scootaloo.db_path, e
-        )
-    });
-    // retrieve the last tweet ID for the username
-    let last_tweet_id = read_state(&conn, None)
-        .unwrap_or_else(|e| panic!("Cannot retrieve last_tweet_id: {}", e))
-        .map(|s| s.tweet_id);
-
-    // get OAuth2 token
-    let token = get_oauth2_token(&config.twitter);
-
-    // get Mastodon instance
-    let mastodon = get_mastodon_token(&config.mastodon);
-
-    // get user timeline feed (Vec<tweet>)
-    let mut feed = get_user_timeline(&config.twitter, token, last_tweet_id)
-        .await
-        .unwrap_or_else(|e| {
             panic!(
-                "Something went wrong when trying to retrieve {}’s timeline: {}",
+                "Something went wrong when opening the DB {}: {}",
-                &config.twitter.username, e
+                &config.scootaloo.db_path, e
             )
-        });
+        }),
+    ));
 
-    // empty feed -> exiting
+    let mut stream = futures::stream::iter(config.mastodon.into_values())
-    if feed.is_empty() {
+        .map(|mastodon_config| {
-        info!("Nothing to retrieve since last time, exiting…");
+            // calculate Twitter page size
-        return;
+            let page_size = mastodon_config
-    }
+                .twitter_page_size
+                .unwrap_or_else(|| config.twitter.page_size.unwrap_or(DEFAULT_PAGE_SIZE));
 
-    // order needs to be chronological
+            // create temporary value for each task
-    feed.reverse();
+            let scootaloo_cache_path = config.scootaloo.cache_path.clone();
+            let token = get_oauth2_token(&config.twitter);
+            let task_conn = conn.clone();
 
-    for tweet in &feed {
+            spawn(async move {
-        debug!("Treating Tweet {} inside feed", tweet.id);
+                info!("Starting treating {}", &mastodon_config.twitter_screen_name);
-        // initiate the toot_reply_id var
+                // retrieve the last tweet ID for the username
-        let mut toot_reply_id: Option<String> = None;
+                let lconn = task_conn.lock().await;
-        // determine if the tweet is part of a thread (response to self) or a standard response
+                let last_tweet_id = read_state(&lconn, &mastodon_config.twitter_screen_name, None)?
-        if let Some(r) = &tweet.in_reply_to_screen_name {
+                    .map(|r| r.tweet_id);
-            if r.to_lowercase() != config.twitter.username.to_lowercase() {
+                drop(lconn);
-                // we are responding not threading
-                info!("Tweet is a direct response, skipping");
-                continue;
-            }
-            info!("Tweet is a thread");
-            toot_reply_id = read_state(&conn, tweet.in_reply_to_status_id)
-                .unwrap_or(None)
-                .map(|s| s.toot_id);
-        };
 
-        // build basic status by just yielding text and dereferencing contained urls
+                // get reversed, curated user timeline
-        let mut status_text = build_basic_status(tweet);
+                let feed = get_user_timeline(
+                    &mastodon_config.twitter_screen_name,
+                    &token,
+                    last_tweet_id,
+                    page_size,
+                )
+                .await?;
 
-        let mut status_medias: Vec<String> = vec![];
+                // get Mastodon instance
-        // reupload the attachments if any
+                let mastodon = get_mastodon_token(&mastodon_config);
-        if let Some(m) = &tweet.extended_entities {
+
-            for media in &m.media {
+                for tweet in &feed {
-                let local_tweet_media_path =
+                    info!("Treating Tweet {} inside feed", tweet.id);
-                    match get_tweet_media(media, &config.scootaloo.cache_path).await {
+
-                        Ok(m) => m,
+                    let lconn = task_conn.lock().await;
-                        Err(e) => {
+                    // initiate the toot_reply_id var and retrieve the corresponding toot_id
-                            error!("Cannot get tweet media for {}: {}", &media.url, e);
+                    let toot_reply_id: Option<String> = tweet.in_reply_to_user_id.and_then(|_| {
-                            continue;
+                        read_state(
+                            &lconn,
+                            &mastodon_config.twitter_screen_name,
+                            tweet.in_reply_to_status_id,
+                        )
+                        .unwrap_or(None)
+                        .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);
+
+                    // building associative media list
+                    let (media_url, status_medias) =
+                        generate_media_ids(tweet, &scootaloo_cache_path, &mastodon).await;
+
+                    status_text = status_text.replace(&media_url, "");
+
+                    info!("Building corresponding Mastodon status");
+
+                    let mut status_builder = StatusBuilder::new();
+
+                    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);
+
+                    let status = status_builder.build()?;
+
+                    let published_status = mastodon.new_status(status)?;
+                    // this will return if it cannot publish the status preventing the last_tweet from
+                    // being written into db
+
+                    let ttt_towrite = TweetToToot {
+                        twitter_screen_name: mastodon_config.twitter_screen_name.clone(),
+                        tweet_id: tweet.id,
+                        toot_id: published_status.id,
                     };
 
-                let mastodon_media_ids = match mastodon
+                    // write the current state (tweet ID and toot ID) to avoid copying it another time
-                    .media(Cow::from(local_tweet_media_path.to_owned()))
+                    let lconn = task_conn.lock().await;
-                {
+                    write_state(&lconn, ttt_towrite)?;
-                    Ok(m) => {
+                    drop(lconn);
-                        remove_file(&local_tweet_media_path)
+                }
-                            .await
+                Ok::<(), ScootalooError>(())
-                            .unwrap_or_else(|e|
+            })
-                                warn!("Attachment for {} has been uploaded, but I’m unable to remove the existing file: {}", &local_tweet_media_path, e)
+        })
-                            );
+        .buffer_unordered(config.scootaloo.rate_limit.unwrap_or(DEFAULT_RATE_LIMIT));
-                        m.id
-                    }
-                    Err(e) => {
-                        error!(
-                            "Attachment {} cannot be uploaded to Mastodon Instance: {}",
-                            &local_tweet_media_path, e
-                        );
-                        continue;
-                    }
-                };
 
-                status_medias.push(mastodon_media_ids);
+    // launch and wait for every handle
-
+    while let Some(result) = stream.next().await {
-                // last step, removing the reference to the media from with the toot’s text
+        match result {
-                status_text = status_text.replace(&media.url, "");
+            Ok(Err(e)) => eprintln!("Error within thread: {}", e),
-            }
+            Err(e) => eprintln!("Error with thread: {}", e),
+            _ => (),
         }
-        // finished reuploading attachments, now let’s do the toot baby!
-
-        debug!("Building corresponding Mastodon status");
-
-        let mut status_builder = StatusBuilder::new();
-
-        status_builder.status(&status_text).media_ids(status_medias);
-
-        if let Some(i) = toot_reply_id {
-            status_builder.in_reply_to(&i);
-        }
-
-        let status = status_builder
-            .build()
-            .unwrap_or_else(|_| panic!("Cannot build status with text {}", &status_text));
-
-        // publish status
-        // again unwrap is safe here as we are in the main thread
-        let published_status = mastodon.new_status(status).unwrap();
-        // this will panic if it cannot publish the status, which is a good thing, it allows the
-        // last_tweet gathered not to be written
-
-        let ttt_towrite = TweetToToot {
-            tweet_id: tweet.id,
-            toot_id: published_status.id,
-        };
-
-        // write the current state (tweet ID and toot ID) to avoid copying it another time
-        write_state(&conn, ttt_towrite)
-            .unwrap_or_else(|e| panic!("Can’t write the last tweet retrieved: {}", e));
     }
 }

src/main.rs

@@ -1,5 +1,5 @@
-use clap::{App, Arg, SubCommand};
+use clap::{Arg, Command};
-use log::{error, LevelFilter};
+use log::LevelFilter;
 use scootaloo::*;
 use simple_logger::SimpleLogger;
 use std::str::FromStr;
@@ -7,91 +7,133 @@ use std::str::FromStr;
 const DEFAULT_CONFIG_PATH: &str = "/usr/local/etc/scootaloo.toml";
 
 fn main() {
-    let matches = App::new(env!("CARGO_PKG_NAME"))
+    let matches = Command::new(env!("CARGO_PKG_NAME"))
         .version(env!("CARGO_PKG_VERSION"))
         .about("A Twitter to Mastodon bot")
         .arg(
-            Arg::with_name("config")
+            Arg::new("config")
-                .short("c")
+                .short('c')
                 .long("config")
                 .value_name("CONFIG_FILE")
-                .help(&format!(
+                .help("TOML config file for scootaloo")
-                    "TOML config file for scootaloo (default {})",
+                .num_args(1)
-                    DEFAULT_CONFIG_PATH
+                .default_value(DEFAULT_CONFIG_PATH)
-                ))
-                .takes_value(true)
                 .display_order(1),
         )
         .arg(
-            Arg::with_name("log_level")
+            Arg::new("log_level")
-                .short("l")
+                .short('l')
                 .long("loglevel")
                 .value_name("LOGLEVEL")
-                .help("Log level. Valid values are: Off, Warn, Error, Info, Debug")
+                .help("Log level")
-                .takes_value(true)
+                .num_args(1)
+                .value_parser(["Off", "Warn", "Error", "Info", "Debug"])
                 .display_order(2),
         )
         .subcommand(
-            SubCommand::with_name("register")
+            Command::new("register")
                 .version(env!("CARGO_PKG_VERSION"))
                 .about("Command to register to a Mastodon Instance")
                 .arg(
-                    Arg::with_name("host")
+                    Arg::new("host")
-                        .short("H")
+                        .short('H')
                         .long("host")
                         .value_name("HOST")
                         .help("Base URL of the Mastodon instance to register to (no default)")
-                        .takes_value(true)
+                        .num_args(1)
                         .required(true)
-                        .display_order(1),
+                        .display_order(1)
+                    )
+                .arg(
+                    Arg::new("name")
+                        .short('n')
+                        .long("name")
+                        .help("Twitter Screen Name (like https://twitter.com/screen_name, no default)")
+                        .num_args(1)
+                        .required(true)
+                        .display_order(2)
                 ),
         )
         .subcommand(
-            SubCommand::with_name("init")
+            Command::new("init")
                 .version(env!("CARGO_PKG_VERSION"))
                 .about("Command to init Scootaloo DB")
                 .arg(
-                    Arg::with_name("config")
+                    Arg::new("config")
-                        .short("c")
+                        .short('c')
                         .long("config")
                         .value_name("CONFIG_FILE")
                         .help(&format!(
                             "TOML config file for scootaloo (default {})",
                             DEFAULT_CONFIG_PATH
                         ))
-                        .takes_value(true)
+                        .default_value(DEFAULT_CONFIG_PATH)
+                        .num_args(1)
                         .display_order(1),
                 ),
         )
+        .subcommand(
+            Command::new("migrate")
+            .version(env!("CARGO_PKG_VERSION"))
+            .about("Command to migrate Scootaloo DB")
+            .arg(
+                Arg::new("config")
+                    .short('c')
+                    .long("config")
+                    .value_name("CONFIG_FILE")
+                    .help(&format!("TOML config file for scootaloo (default {})", DEFAULT_CONFIG_PATH))
+                    .default_value(DEFAULT_CONFIG_PATH)
+                    .num_args(1)
+                    .display_order(1),
+            )
+            .arg(
+                    Arg::new("name")
+                        .short('n')
+                        .long("name")
+                        .help("Twitter Screen Name (like https://twitter.com/screen_name, no default)")
+                        .num_args(1)
+                        .display_order(2)
+            )
+        )
         .get_matches();
 
     match matches.subcommand() {
-        ("register", Some(sub_m)) => {
+        Some(("register", sub_m)) => {
-            register(sub_m.value_of("host").unwrap());
+            register(
+                sub_m.get_one::<String>("host").unwrap(),
+                sub_m.get_one::<String>("name").unwrap(),
+            );
             return;
         }
-        ("init", Some(sub_m)) => {
+        Some(("init", sub_m)) => {
-            let config = parse_toml(sub_m.value_of("config").unwrap_or(DEFAULT_CONFIG_PATH));
+            let config = parse_toml(sub_m.get_one::<String>("config").unwrap());
             init_db(&config.scootaloo.db_path).unwrap();
             return;
         }
+        Some(("migrate", sub_m)) => {
+            let config = parse_toml(sub_m.get_one::<String>("config").unwrap());
+            let config_twitter_screen_name =
+                &config.mastodon.values().next().unwrap().twitter_screen_name;
+            migrate_db(
+                &config.scootaloo.db_path,
+                sub_m
+                    .get_one::<String>("name")
+                    .unwrap_or(config_twitter_screen_name),
+            )
+            .unwrap();
+            return;
+        }
         _ => (),
     }
 
-    if matches.is_present("log_level") {
+    if let Some(level) = matches.get_one::<String>("log_level") {
-        match LevelFilter::from_str(matches.value_of("log_level").unwrap()) {
+        SimpleLogger::new()
-            Ok(level) => SimpleLogger::new().with_level(level).init().unwrap(),
+            .with_level(LevelFilter::from_str(level).unwrap())
-            Err(e) => {
+            .init()
-                SimpleLogger::new()
+            .unwrap();
-                    .with_level(LevelFilter::Error)
-                    .init()
-                    .unwrap();
-                error!("Unknown log level filter: {}", e);
-            }
-        };
     }
 
-    let config = parse_toml(matches.value_of("config").unwrap_or(DEFAULT_CONFIG_PATH));
+    let config = parse_toml(matches.get_one::<String>("config").unwrap());
 
     run(config);
 }

src/mastodon.rs

@@ -65,7 +65,7 @@ pub fn build_basic_status(tweet: &Tweet) -> String {
 /// Generic register function
 /// As this function is supposed to be run only once, it will panic for every error it encounters
 /// Most of this function is a direct copy/paste of the official `elefren` crate
-pub fn register(host: &str) {
+pub fn register(host: &str, screen_name: &str) {
     let mut builder = App::builder();
     builder
         .client_name(Cow::from(env!("CARGO_PKG_NAME").to_string()))
@@ -100,7 +100,12 @@ pub fn register(host: &str) {
     let toml = toml::to_string(&*mastodon).unwrap();
 
     println!(
-        "Please insert the following block at the end of your configuration file:\n[mastodon]\n{}",
+        "Please insert the following block at the end of your configuration file:
+[mastodon.{}]
+twitter_screen_name = \"{}\"
+{}",
+        screen_name.to_lowercase(),
+        screen_name,
         toml
     );
 }

src/state.rs

@@ -1,10 +1,13 @@
-use log::debug;
-use rusqlite::{params, Connection, OptionalExtension};
 use std::error::Error;
 
+use log::debug;
+
+use rusqlite::{params, Connection, OptionalExtension};
+
 /// Struct for each query line
 #[derive(Debug)]
 pub struct TweetToToot {
+    pub twitter_screen_name: String,
     pub tweet_id: u64,
     pub toot_id: String,
 }
@@ -13,12 +16,13 @@ pub struct TweetToToot {
 /// if a tweet_id is passed, read this particular tweet from DB
 pub fn read_state(
     conn: &Connection,
+    n: &str,
     s: Option<u64>,
 ) -> Result<Option<TweetToToot>, Box<dyn Error>> {
     debug!("Reading tweet_id {:?}", s);
     let query: String = match s {
-        Some(i) => format!("SELECT * FROM tweet_to_toot WHERE tweet_id = {}", i),
+        Some(i) => format!("SELECT * FROM tweet_to_toot WHERE tweet_id = {} and twitter_screen_name = \"{}\"", i, n),
-        None => "SELECT * FROM tweet_to_toot ORDER BY tweet_id DESC LIMIT 1".to_string(),
+        None => format!("SELECT * FROM tweet_to_toot WHERE twitter_screen_name = \"{}\" ORDER BY tweet_id DESC LIMIT 1", n),
     };
 
     let mut stmt = conn.prepare(&query)?;
@@ -26,8 +30,9 @@ pub fn read_state(
     let t = stmt
         .query_row([], |row| {
             Ok(TweetToToot {
-                tweet_id: row.get(0)?,
+                twitter_screen_name: row.get("twitter_screen_name")?,
-                toot_id: row.get(1)?,
+                tweet_id: row.get("tweet_id")?,
+                toot_id: row.get("toot_id")?,
             })
         })
         .optional()?;
@@ -39,8 +44,8 @@ pub fn read_state(
 pub fn write_state(conn: &Connection, t: TweetToToot) -> Result<(), Box<dyn Error>> {
     debug!("Write struct {:?}", t);
     conn.execute(
-        "INSERT INTO tweet_to_toot (tweet_id, toot_id) VALUES (?1, ?2)",
+        "INSERT INTO tweet_to_toot (twitter_screen_name, tweet_id, toot_id) VALUES (?1, ?2, ?3)",
-        params![t.tweet_id, t.toot_id],
+        params![t.twitter_screen_name, t.tweet_id, t.toot_id],
     )?;
 
     Ok(())
@@ -53,8 +58,9 @@ pub fn init_db(d: &str) -> Result<(), Box<dyn Error>> {
 
     conn.execute(
         "CREATE TABLE IF NOT EXISTS tweet_to_toot (
-            tweet_id        INTEGER PRIMARY KEY,
+            twitter_screen_name TEXT NOT NULL,
-            toot_id         TEXT UNIQUE
+            tweet_id            INTEGER PRIMARY KEY,
+            toot_id             TEXT UNIQUE
             )",
         [],
     )?;
@@ -62,6 +68,31 @@ pub fn init_db(d: &str) -> Result<(), Box<dyn Error>> {
     Ok(())
 }
 
+/// Migrate DB from 0.6.x to 0.7.x
+pub fn migrate_db(d: &str, s: &str) -> Result<(), Box<dyn Error>> {
+    debug!("Migrating DB for Scootaloo");
+
+    let conn = Connection::open(d)?;
+
+    let res = conn.execute(
+        &format!(
+            "ALTER TABLE tweet_to_toot
+        ADD COLUMN twitter_screen_name TEXT NOT NULL
+        DEFAULT \"{}\"",
+            s
+        ),
+        [],
+    );
+
+    match res {
+        Err(e) => match e.to_string().as_str() {
+            "duplicate column name: twitter_screen_name" => Ok(()),
+            _ => Err(e.into()),
+        },
+        _ => Ok(()),
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -93,9 +124,9 @@ mod tests {
         let conn = Connection::open(d).unwrap();
 
         conn.execute(
-            "INSERT INTO tweet_to_toot
+            "INSERT INTO tweet_to_toot (twitter_screen_name, tweet_id, toot_id)
             VALUES
-            (100, 'A');",
+            ('tamerelol', 100, 'A');",
             [],
         )
         .unwrap();
@@ -114,6 +145,7 @@ mod tests {
         let conn = Connection::open(d).unwrap();
 
         let t_in = TweetToToot {
+            twitter_screen_name: "tamerelol".to_string(),
             tweet_id: 123456789,
             toot_id: "987654321".to_string(),
         };
@@ -125,14 +157,16 @@ mod tests {
         let t_out = stmt
             .query_row([], |row| {
                 Ok(TweetToToot {
-                    tweet_id: row.get(0).unwrap(),
+                    twitter_screen_name: row.get("twitter_screen_name").unwrap(),
-                    toot_id: row.get(1).unwrap(),
+                    tweet_id: row.get("tweet_id").unwrap(),
+                    toot_id: row.get("toot_id").unwrap(),
                 })
             })
             .unwrap();
 
+        assert_eq!(&t_out.twitter_screen_name, "tamerelol");
         assert_eq!(t_out.tweet_id, 123456789);
-        assert_eq!(t_out.toot_id, "987654321".to_string());
+        assert_eq!(&t_out.toot_id, "987654321");
 
         remove_file(d).unwrap();
     }
@@ -146,15 +180,15 @@ mod tests {
         let conn = Connection::open(d).unwrap();
 
         conn.execute(
-            "INSERT INTO tweet_to_toot (tweet_id, toot_id)
+            "INSERT INTO tweet_to_toot (twitter_screen_name, tweet_id, toot_id)
             VALUES
-            (101, 'A'),
+            ('tamerelol', 101, 'A'),
-            (102, 'B');",
+            ('tamerelol', 102, 'B');",
             [],
         )
         .unwrap();
 
-        let t_out = read_state(&conn, None).unwrap().unwrap();
+        let t_out = read_state(&conn, "tamerelol", None).unwrap().unwrap();
 
         remove_file(d).unwrap();
 
@@ -170,7 +204,7 @@ mod tests {
 
         let conn = Connection::open(d).unwrap();
 
-        let t_out = read_state(&conn, None).unwrap();
+        let t_out = read_state(&conn, "tamerelol", None).unwrap();
 
         remove_file(d).unwrap();
 
@@ -186,14 +220,14 @@ mod tests {
         let conn = Connection::open(d).unwrap();
 
         conn.execute(
-            "INSERT INTO tweet_to_toot (tweet_id, toot_id)
+            "INSERT INTO tweet_to_toot (twitter_screen_name, tweet_id, toot_id)
             VALUES
-            (100, 'A');",
+            ('tamerelol', 100, 'A');",
             [],
         )
         .unwrap();
 
-        let t_out = read_state(&conn, Some(101)).unwrap();
+        let t_out = read_state(&conn, "tamerelol", Some(101)).unwrap();
 
         remove_file(d).unwrap();
 
@@ -209,18 +243,62 @@ mod tests {
         let conn = Connection::open(d).unwrap();
 
         conn.execute(
-            "INSERT INTO tweet_to_toot (tweet_id, toot_id)
+            "INSERT INTO tweet_to_toot (twitter_screen_name, tweet_id, toot_id)
             VALUES
-            (100, 'A');",
+            ('tamerelol', 100, 'A');",
             [],
         )
         .unwrap();
 
-        let t_out = read_state(&conn, Some(100)).unwrap().unwrap();
+        let t_out = read_state(&conn, "tamerelol", Some(100)).unwrap().unwrap();
 
         remove_file(d).unwrap();
 
         assert_eq!(t_out.tweet_id, 100);
         assert_eq!(t_out.toot_id, "A");
     }
+
+    #[test]
+    fn test_migrate_db_add_column() {
+        let d = "/tmp/test_migrate_db_add_column.sqlite";
+
+        let conn = Connection::open(d).unwrap();
+
+        conn.execute(
+            "CREATE TABLE IF NOT EXISTS tweet_to_toot (
+            tweet_id            INTEGER PRIMARY KEY,
+            toot_id             TEXT UNIQUE
+            )",
+            [],
+        )
+        .unwrap();
+
+        migrate_db(d, "tamerelol").unwrap();
+
+        let mut stmt = conn.prepare("PRAGMA table_info(tweet_to_toot);").unwrap();
+
+        let mut t = stmt.query([]).unwrap();
+
+        while let Some(row) = t.next().unwrap() {
+            if row.get::<usize, u8>(0).unwrap() == 2 {
+                assert_eq!(
+                    row.get::<usize, String>(1).unwrap(),
+                    "twitter_screen_name".to_string()
+                );
+            }
+        }
+
+        remove_file(d).unwrap();
+    }
+
+    #[test]
+    fn test_migrate_db_no_add_column() {
+        let d = "/tmp/test_migrate_db_no_add_column.sqlite";
+
+        init_db(d).unwrap();
+
+        migrate_db(d, "tamerelol").unwrap();
+
+        remove_file(d).unwrap();
+    }
 }

src/twitter.rs

@@ -27,52 +27,167 @@ 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(
-    config: &TwitterConfig,
+    screen_name: &str,
-    token: Token,
+    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(
+    let (_, feed) = user_timeline(UserID::from(screen_name.to_owned()), true, false, token)
-        UserID::from(config.username.to_owned()),
+        .with_page_size(page_size)
-        true,
+        .older(lid)
-        false,
+        .await?;
-        &token,
-    )
-    .with_page_size(200)
-    .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
 pub async fn get_tweet_media(m: &MediaEntity, t: &str) -> Result<String, Box<dyn Error>> {
     match m.media_type {
-        MediaType::Photo => {
+        MediaType::Photo => cache_media(&m.media_url_https, t).await,
-            return cache_media(&m.media_url_https, t).await;
-        }
         _ => match &m.video_info {
-            Some(v) => {
+            Some(v) => match &v.variants.iter().find(|&x| x.content_type == "video/mp4") {
-                for variant in &v.variants {
+                Some(u) => cache_media(&u.url, t).await,
-                    if variant.content_type == "video/mp4" {
+                None => Err(ScootalooError::new(&format!(
-                        return cache_media(&variant.url, t).await;
-                    }
-                }
-                return Err(ScootalooError::new(&format!(
                     "Media Type for {} is video but no mp4 file URL is available",
                     &m.url
                 ))
-                .into());
+                .into()),
-            }
+            },
-            None => {
+            None => Err(ScootalooError::new(&format!(
-                return Err(ScootalooError::new(&format!(
+                "Media Type for {} is video but does not contain any video_info",
-                    "Media Type for {} is video but does not contain any video_info",
+                &m.url
-                    &m.url
+            ))
-                ))
+            .into()),
-                .into());
-            }
         },
-    };
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use egg_mode::entities::{
+        MediaSize, MediaSizes,
+        MediaType::{Gif, Photo},
+        ResizeMode::Crop,
+        ResizeMode::Fit,
+        VideoInfo, VideoVariant,
+    };
+    use std::fs::remove_dir_all;
+
+    const TMP_DIR: &'static str = "/tmp/scootaloo_get_tweet_media_test";
+
+    #[tokio::test]
+    async fn test_get_tweet_media() {
+        let m_photo = MediaEntity {
+            display_url: "pic.twitter.com/sHrwmP69Yv".to_string(),
+            expanded_url: "https://twitter.com/NintendojoFR/status/1555473821121056771/photo/1"
+                .to_string(),
+            id: 1555473771280080896,
+            range: (91, 114),
+            media_url: "http://pbs.twimg.com/media/FZYnJ1qWIAAReHt.jpg".to_string(),
+            media_url_https: "https://pbs.twimg.com/media/FZYnJ1qWIAAReHt.jpg"
+                .to_string(),
+            sizes: MediaSizes {
+                thumb: MediaSize {
+                    w: 150,
+                    h: 150,
+                    resize: Crop
+                },
+                small: MediaSize {
+                    w: 680,
+                    h: 510,
+                    resize: Fit
+                },
+                medium: MediaSize {
+                    w: 1200,
+                    h: 900,
+                    resize: Fit
+                },
+                large: MediaSize {
+                    w: 1280,
+                    h: 960,
+                    resize: Fit
+                }
+            },
+            source_status_id: None,
+            media_type: Photo,
+            url: "https://t.co/sHrwmP69Yv".to_string(),
+            video_info: None,
+            ext_alt_text: Some("Le menu «\u{a0}Classes » du jeu vidéo Xenoblade Chronicles 3 (Switch). L’affinité du personnage pour la classe est notée par quatre lettres : C, A, C, A (caca)."
+                .to_string())
+        };
+        let m_video = MediaEntity {
+            display_url: "pic.twitter.com/xDln0RrkjU".to_string(),
+            expanded_url: "https://twitter.com/NintendojoFR/status/1551822196833673218/photo/1"
+                .to_string(),
+            id: 1551822189711790081,
+            range: (275, 298),
+            media_url: "http://pbs.twimg.com/tweet_video_thumb/FYkuD0RXEAE-iDx.jpg".to_string(),
+            media_url_https: "https://pbs.twimg.com/tweet_video_thumb/FYkuD0RXEAE-iDx.jpg"
+                .to_string(),
+            sizes: MediaSizes {
+                thumb: MediaSize {
+                    w: 150,
+                    h: 150,
+                    resize: Crop,
+                },
+                small: MediaSize {
+                    w: 320,
+                    h: 240,
+                    resize: Fit,
+                },
+                medium: MediaSize {
+                    w: 320,
+                    h: 240,
+                    resize: Fit,
+                },
+                large: MediaSize {
+                    w: 320,
+                    h: 240,
+                    resize: Fit,
+                },
+            },
+            source_status_id: None,
+            media_type: Gif,
+            url: "https://t.co/xDln0RrkjU".to_string(),
+            video_info: Some(VideoInfo {
+                aspect_ratio: (4, 3),
+                duration_millis: None,
+                variants: vec![VideoVariant {
+                    bitrate: Some(0),
+                    content_type: "video/mp4".parse::<mime::Mime>().unwrap(),
+                    url: "https://video.twimg.com/tweet_video/FYkuD0RXEAE-iDx.mp4".to_string(),
+                }],
+            }),
+            ext_alt_text: Some("Scared Nintendo GIF".to_string()),
+        };
+
+        let tweet_media_photo = get_tweet_media(&m_photo, TMP_DIR).await.unwrap();
+        let tweet_media_video = get_tweet_media(&m_video, TMP_DIR).await.unwrap();
+
+        assert_eq!(
+            tweet_media_photo,
+            format!("{}/FZYnJ1qWIAAReHt.jpg", TMP_DIR)
+        );
+        assert_eq!(
+            tweet_media_video,
+            format!("{}/FYkuD0RXEAE-iDx.mp4", TMP_DIR)
+        );
+
+        remove_dir_all(TMP_DIR).unwrap();
+    }
 }

src/util.rs

@@ -1,11 +1,80 @@
-use crate::ScootalooError;
+use crate::{twitter::get_tweet_media, ScootalooError};
+
+use std::{borrow::Cow, error::Error};
+
+use egg_mode::tweet::Tweet;
+
+use elefren::prelude::*;
+
+use log::{error, info, warn};
+
 use reqwest::Url;
-use std::error::Error;
+
 use tokio::{
-    fs::{create_dir_all, File},
+    fs::{create_dir_all, remove_file, File},
     io::copy,
 };
 
+use futures::{stream, stream::StreamExt};
+
+/// Generate associative table between media ids and tweet extended entities
+pub async fn generate_media_ids(
+    tweet: &Tweet,
+    cache_path: &str,
+    mastodon: &Mastodon,
+) -> (String, Vec<String>) {
+    let mut media_url = "".to_string();
+    let mut media_ids: Vec<String> = vec![];
+
+    if let Some(m) = &tweet.extended_entities {
+        info!("{} medias in tweet", m.media.len());
+
+        let medias = m.media.clone();
+
+        let mut stream = stream::iter(medias)
+            .map(|media| {
+                // attribute media url
+                media_url = media.url.clone();
+
+                // clone everything we need
+                let cache_path = String::from(cache_path);
+                let mastodon = mastodon.clone();
+
+                tokio::task::spawn(async move {
+                    info!("Start treating {}", media.media_url_https);
+                    // get the tweet embedded media
+                    let local_tweet_media_path = get_tweet_media(&media, &cache_path).await?;
+
+                    // upload media to Mastodon
+                    let mastodon_media =
+                        mastodon.media(Cow::from(local_tweet_media_path.to_owned()))?;
+                    // at this point, we can safely erase the original file
+                    // it doesn’t matter if we can’t remove, cache_media fn is idempotent
+                    remove_file(&local_tweet_media_path).await.ok();
+
+                    Ok::<String, ScootalooError>(mastodon_media.id)
+                })
+            })
+            .buffered(4); // there are max four medias per tweet and they need to be treated in
+                          // order
+
+        while let Some(result) = stream.next().await {
+            match result {
+                Ok(Ok(v)) => media_ids.push(v),
+                Ok(Err(e)) => warn!("Cannot treat media: {}", e),
+                Err(e) => error!("Something went wrong when joining the main thread: {}", e),
+            }
+        }
+    } else {
+        info!("No media in tweet");
+    }
+
+    // in case some media_ids slot remained empty due to errors, remove them
+    media_ids.retain(|x| !x.is_empty());
+
+    (media_url, media_ids)
+}
+
 /// Gets and caches Twitter Media inside the determined temp dir
 pub async fn cache_media(u: &str, t: &str) -> Result<String, Box<dyn Error>> {
     // create dir

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");
@@ -9,21 +61,45 @@ 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.username, "tamerelol");
     assert_eq!(parse_good_toml.twitter.consumer_key, "rand consumer key");
     assert_eq!(parse_good_toml.twitter.consumer_secret, "secret");
     assert_eq!(parse_good_toml.twitter.access_key, "rand access key");
     assert_eq!(parse_good_toml.twitter.access_secret, "super secret");
 
-    assert_eq!(parse_good_toml.mastodon.base, "https://m.nintendojo.fr");
-    assert_eq!(parse_good_toml.mastodon.client_id, "rand client id");
-    assert_eq!(parse_good_toml.mastodon.client_secret, "secret");
     assert_eq!(
-        parse_good_toml.mastodon.redirect,
+        &parse_good_toml
+            .mastodon
+            .get("tamerelol")
+            .unwrap()
+            .twitter_screen_name,
+        "tamerelol"
+    );
+    assert_eq!(
+        &parse_good_toml.mastodon.get("tamerelol").unwrap().base,
+        "https://m.nintendojo.fr"
+    );
+    assert_eq!(
+        &parse_good_toml.mastodon.get("tamerelol").unwrap().client_id,
+        "rand client id"
+    );
+    assert_eq!(
+        &parse_good_toml
+            .mastodon
+            .get("tamerelol")
+            .unwrap()
+            .client_secret,
+        "secret"
+    );
+    assert_eq!(
+        &parse_good_toml.mastodon.get("tamerelol").unwrap().redirect,
         "urn:ietf:wg:oauth:2.0:oob"
     );
-    assert_eq!(parse_good_toml.mastodon.token, "super secret");
+    assert_eq!(
+        &parse_good_toml.mastodon.get("tamerelol").unwrap().token,
+        "super secret"
+    );
 }
 
 #[test]

tests/good_test.toml

@@ -4,14 +4,14 @@ db_path="/var/random/scootaloo.sqlite"
 cache_path="/tmp/scootaloo"
 
 [twitter]
-username="tamerelol"
-
 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"

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"