4 Commits

Author SHA1 Message Date
VC
31aea7e1a6 Merge branch 'fix_french_inclusive_writing' into 'master'
feat: allow user to remove some links, replace some links by others

See merge request veretcle/scootaloo!38
2022-11-23 12:49:46 +00:00
Clément VERET
851f95d516 doc: regexp + alt services 2022-11-23 13:17:00 +01:00
Clément VERET
ffb9522ce2 feat: main logic for regex + url filtering 2022-11-23 13:16:56 +01:00
Clément VERET
b6df8c6230 test: add tests for scootaloo alt services + regexp 2022-11-23 13:09:03 +01:00
7 changed files with 1716 additions and 943 deletions

1856
Cargo.lock generated

File diff suppressed because it is too large Load Diff

View File

@@ -1,22 +1,22 @@
[package] [package]
name = "scootaloo" name = "scootaloo"
version = "1.0.0" version = "0.11.0"
authors = ["VC <veretcle+framagit@mateu.be>"] authors = ["VC <veretcle+framagit@mateu.be>"]
edition = "2021" edition = "2021"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies] [dependencies]
chrono = "^0.4"
regex = "^1" regex = "^1"
serde = { version = "1.0", features = ["derive"] } serde = { version = "1.0", features = ["derive"] }
toml = "^0.5" toml = "^0.5"
clap = "^4" clap = "^4"
egg-mode = "^0.16" egg-mode = "^0.16"
rusqlite = "^0.27" rusqlite = "^0.27"
isolang = "^2" tokio = { version = "^1", features = ["full"]}
tokio = { version = "^1", features = ["rt"]}
futures = "^0.3" futures = "^0.3"
megalodon = "^0.2" elefren = "^0.22"
html-escape = "^0.2" html-escape = "^0.2"
reqwest = "^0.11" reqwest = "^0.11"
log = "^0.4" log = "^0.4"
@@ -25,4 +25,3 @@ mime = "^0.3"
[profile.release] [profile.release]
strip = true strip = true
lto = true

View File

@@ -19,7 +19,7 @@ pub struct TwitterConfig {
pub page_size: Option<i32>, pub page_size: Option<i32>,
} }
#[derive(Debug, Deserialize, Clone)] #[derive(Debug, Deserialize)]
pub struct MastodonConfig { pub struct MastodonConfig {
pub twitter_screen_name: String, pub twitter_screen_name: String,
pub mastodon_screen_name: Option<String>, pub mastodon_screen_name: Option<String>,

View File

@@ -5,7 +5,7 @@ use std::{
fmt::{Display, Formatter, Result}, fmt::{Display, Formatter, Result},
}; };
use megalodon::error::Error as megalodonError; use elefren::Error as elefrenError;
#[derive(Debug)] #[derive(Debug)]
pub struct ScootalooError { pub struct ScootalooError {
@@ -34,8 +34,8 @@ impl From<Box<dyn Error>> for ScootalooError {
} }
} }
impl From<megalodonError> for ScootalooError { impl From<elefrenError> for ScootalooError {
fn from(error: megalodonError) -> Self { fn from(error: elefrenError) -> Self {
ScootalooError::new(&format!("Error in megalodon crate: {}", error)) ScootalooError::new(&format!("Error in elefren crate: {}", error))
} }
} }

View File

@@ -7,7 +7,7 @@ use config::Config;
mod mastodon; mod mastodon;
pub use mastodon::register; pub use mastodon::register;
use mastodon::*; use mastodon::{build_basic_status, get_mastodon_token};
mod twitter; mod twitter;
use twitter::*; use twitter::*;
@@ -19,14 +19,12 @@ mod state;
pub use state::{init_db, migrate_db}; pub use state::{init_db, migrate_db};
use state::{read_state, write_state, TweetToToot}; use state::{read_state, write_state, TweetToToot};
use elefren::{prelude::*, status_builder::StatusBuilder, Language};
use futures::StreamExt; use futures::StreamExt;
use html_escape::decode_html_entities;
use isolang::Language;
use log::info; use log::info;
use megalodon::{megalodon::PostStatusInputOptions, Megalodon};
use regex::Regex; use regex::Regex;
use rusqlite::Connection; use rusqlite::Connection;
use std::sync::Arc; use std::{collections::HashMap, sync::Arc};
use tokio::{spawn, sync::Mutex}; use tokio::{spawn, sync::Mutex};
const DEFAULT_RATE_LIMIT: usize = 4; const DEFAULT_RATE_LIMIT: usize = 4;
@@ -45,7 +43,21 @@ pub async fn run(config: Config) {
}), }),
)); ));
let global_mastodon_config = Arc::new(Mutex::new(config.mastodon.clone())); 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 display_url_re = config let display_url_re = config
.scootaloo .scootaloo
@@ -64,11 +76,11 @@ pub async fn run(config: Config) {
// create temporary value for each task // create temporary value for each task
let scootaloo_cache_path = config.scootaloo.cache_path.clone(); let scootaloo_cache_path = config.scootaloo.cache_path.clone();
let scootaloo_mentions = scootaloo_mentions.clone();
let scootaloo_alt_services = config.scootaloo.alternative_services_for.clone(); let scootaloo_alt_services = config.scootaloo.alternative_services_for.clone();
let display_url_re = display_url_re.clone(); let display_url_re = display_url_re.clone();
let token = get_oauth2_token(&config.twitter); let token = get_oauth2_token(&config.twitter);
let task_conn = conn.clone(); let task_conn = conn.clone();
let global_mastodon_config = global_mastodon_config.clone();
spawn(async move { spawn(async move {
info!("Starting treating {}", &mastodon_config.twitter_screen_name); info!("Starting treating {}", &mastodon_config.twitter_screen_name);
@@ -93,71 +105,26 @@ pub async fn run(config: Config) {
for tweet in &feed { for tweet in &feed {
info!("Treating Tweet {} inside feed", tweet.id); info!("Treating Tweet {} inside feed", tweet.id);
// basic toot text
let mut status_text = tweet.text.clone();
// add mentions and smart mentions
if !&tweet.entities.user_mentions.is_empty() {
info!("Tweet contains mentions, add them!");
let global_mastodon_config = global_mastodon_config.lock().await;
twitter_mentions(
&mut status_text,
&tweet.entities.user_mentions,
&global_mastodon_config,
);
drop(global_mastodon_config);
}
if !&tweet.entities.urls.is_empty() {
info!("Tweet contains links, add them!");
let mut associated_urls =
associate_urls(&tweet.entities.urls, &display_url_re);
if let Some(q) = &tweet.quoted_status {
if let Some(u) = &q.user {
info!(
"Tweet {} contains a quote, we try to find it within the DB",
tweet.id
);
// we know we have a quote and a user, we can lock both the
// connection to DB and global_config
// we will release them manually as soon as theyre useless
let lconn = task_conn.lock().await; let lconn = task_conn.lock().await;
let global_mastodon_config = global_mastodon_config.lock().await; // initiate the toot_reply_id var and retrieve the corresponding toot_id
if let Ok(Some(r)) = read_state(&lconn, &u.screen_name, Some(q.id)) let toot_reply_id: Option<String> = tweet.in_reply_to_user_id.and_then(|_| {
{ read_state(
info!("We have found the associated toot({})", &r.toot_id); &lconn,
// drop conn immediately after the request: we wont need it &mastodon_config.twitter_screen_name,
// any more and the treatment there might be time-consuming tweet.in_reply_to_status_id,
drop(lconn);
if let Some((m, t)) =
find_mastodon_screen_name_by_twitter_screen_name(
&r.twitter_screen_name,
&global_mastodon_config,
) )
{ .unwrap_or(None)
// drop the global conf, we have all we required, no need .map(|s| s.toot_id)
// to block it further });
drop(global_mastodon_config); drop(lconn);
replace_tweet_by_toot(
&mut associated_urls, // build basic status by just yielding text and dereferencing contained urls
&r.twitter_screen_name, let mut status_text = build_basic_status(
q.id, tweet,
&m, &scootaloo_mentions,
&t, &display_url_re,
&r.toot_id, &scootaloo_alt_services,
); );
}
}
}
}
if let Some(a) = &scootaloo_alt_services {
replace_alt_services(&mut associated_urls, a);
}
decode_urls(&mut status_text, &associated_urls);
}
// building associative media list // building associative media list
let (media_url, status_medias) = let (media_url, status_medias) =
@@ -165,55 +132,30 @@ pub async fn run(config: Config) {
status_text = status_text.replace(&media_url, ""); status_text = status_text.replace(&media_url, "");
// now that the text wont be altered anymore, we can safely remove HTML
// entities
status_text = decode_html_entities(&status_text).to_string();
info!("Building corresponding Mastodon status"); info!("Building corresponding Mastodon status");
let mut post_status = PostStatusInputOptions { let mut status_builder = StatusBuilder::new();
media_ids: None,
poll: None,
in_reply_to_id: None,
sensitive: None,
spoiler_text: None,
visibility: None,
scheduled_at: None,
language: None,
quote_id: None,
};
if !status_medias.is_empty() { status_builder.status(&status_text).media_ids(status_medias);
post_status.media_ids = Some(status_medias);
}
// thread if necessary // theard if necessary
if tweet.in_reply_to_user_id.is_some() { if let Some(i) = toot_reply_id {
let lconn = task_conn.lock().await; status_builder.in_reply_to(&i);
if let Ok(Some(r)) = read_state(
&lconn,
&mastodon_config.twitter_screen_name,
tweet.in_reply_to_status_id,
) {
post_status.in_reply_to_id = Some(r.toot_id.to_owned());
}
drop(lconn);
} }
// language if any // language if any
if let Some(l) = &tweet.lang { if let Some(l) = &tweet.lang {
if let Some(r) = Language::from_639_1(l) { if let Some(r) = Language::from_639_1(l) {
post_status.language = Some(r.to_string()); status_builder.language(r);
} }
} }
// can be activated for test purposes // can be activated for test purposes
// post_status.visibility = Some(megalodon::entities::StatusVisibility::Direct); // status_builder.visibility(elefren::status_builder::Visibility::Private);
let published_status = mastodon let status = status_builder.build()?;
.post_status(status_text, Some(&post_status))
.await? let published_status = mastodon.new_status(status)?;
.json();
// this will return if it cannot publish the status preventing the last_tweet from // this will return if it cannot publish the status preventing the last_tweet from
// being written into db // being written into db

View File

@@ -1,155 +1,123 @@
use crate::config::MastodonConfig; use crate::config::MastodonConfig;
use egg_mode::entities::{MentionEntity, UrlEntity}; use egg_mode::{
use megalodon::{generator, mastodon::Mastodon, megalodon::AppInputOptions}; entities::{MentionEntity, UrlEntity},
tweet::Tweet,
};
use elefren::{apps::App, prelude::*, scopes::Read, scopes::Scopes, scopes::Write};
use html_escape::decode_html_entities;
use regex::Regex; use regex::Regex;
use std::{collections::HashMap, io::stdin}; use std::{borrow::Cow, collections::HashMap, io::stdin};
/// Decodes the Twitter mention to something that will make sense once Twitter has joined the /// Decodes the Twitter mention to something that will make sense once Twitter has joined the
/// Fediverse. Users in the global user list of Scootaloo are rewritten, as they are Mastodon users /// Fediverse
/// as well fn twitter_mentions(ums: &[MentionEntity]) -> HashMap<String, String> {
pub fn twitter_mentions( ums.iter()
toot: &mut String,
ums: &[MentionEntity],
masto: &HashMap<String, MastodonConfig>,
) {
let tm: HashMap<String, String> = ums
.iter()
.map(|s| { .map(|s| {
( (
format!("@{}", s.screen_name), format!("@{}", s.screen_name),
format!("@{}@twitter.com", s.screen_name), format!("@{}@twitter.com", s.screen_name),
) )
}) })
.chain( .collect()
masto
.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::<HashMap<String, String>>(),
)
.collect();
for (k, v) in tm {
*toot = toot.replace(&k, &v);
}
} }
/// Decodes urls in toot /// Decodes urls from UrlEntities
pub fn decode_urls(toot: &mut String, urls: &HashMap<String, String>) { fn decode_urls(
for (k, v) in urls { urls: &[UrlEntity],
*toot = toot.replace(k, v); re: &Option<Regex>,
} alt_urls: &Option<HashMap<String, String>>,
} ) -> HashMap<String, String> {
/// Reassociates source url with destination url for rewritting
/// this takes a Teet UrlEntity and an optional Regex
pub fn associate_urls(urls: &[UrlEntity], re: &Option<Regex>) -> HashMap<String, String> {
urls.iter() urls.iter()
.filter(|s| s.expanded_url.is_some()) .filter(|s| s.expanded_url.is_some())
.map(|s| { .map(|s| {
(s.url.to_owned(), { (s.url.to_owned(), {
let mut def = s.expanded_url.as_deref().unwrap().to_owned(); let mut def = s.expanded_url.as_deref().unwrap().to_owned();
if let Some(r) = re { if let Some(a) = &alt_urls {
for (url_source, url_destination) in a {
def = def.replace(
&format!("https://{}", url_source),
&format!("https://{}", url_destination),
);
}
}
if let Some(r) = &re {
if r.is_match(s.expanded_url.as_deref().unwrap()) { if r.is_match(s.expanded_url.as_deref().unwrap()) {
def = s.display_url.to_owned(); def = s.display_url.clone();
} }
} }
def def
}) })
}) })
.collect::<HashMap<String, String>>() .collect()
}
/// Replaces the commonly used services by mirrors, if asked to
pub fn replace_alt_services(urls: &mut HashMap<String, String>, alts: &HashMap<String, String>) {
for val in urls.values_mut() {
for (k, v) in alts {
*val = val.replace(&format!("/{}/", k), &format!("/{}/", v));
}
}
}
/// Finds a Mastodon screen_name/base_url from a MastodonConfig
pub fn find_mastodon_screen_name_by_twitter_screen_name(
twitter_screen_name: &str,
masto: &HashMap<String, MastodonConfig>,
) -> Option<(String, String)> {
masto.iter().find_map(|(_, v)| {
if twitter_screen_name == v.twitter_screen_name && v.mastodon_screen_name.is_some() {
Some((
v.mastodon_screen_name.as_ref().unwrap().to_owned(),
v.base.to_owned(),
))
} else {
None
}
})
}
/// Replaces the original quoted tweet by the corresponding toot
pub fn replace_tweet_by_toot(
urls: &mut HashMap<String, String>,
twitter_screen_name: &str,
tweet_id: u64,
mastodon_screen_name: &str,
base_url: &str,
toot_id: &str,
) {
for val in urls.values_mut() {
if val.to_lowercase().starts_with(&format!(
"https://twitter.com/{}/status/{}",
twitter_screen_name.to_lowercase(),
tweet_id
)) {
*val = format!("{}/@{}/{}", base_url, mastodon_screen_name, toot_id);
}
}
} }
/// Gets Mastodon Data /// Gets Mastodon Data
pub fn get_mastodon_token(masto: &MastodonConfig) -> Mastodon { pub fn get_mastodon_token(masto: &MastodonConfig) -> Mastodon {
Mastodon::new(masto.base.to_string(), Some(masto.token.to_string()), None) let data = Data {
base: Cow::from(masto.base.to_owned()),
client_id: Cow::from(masto.client_id.to_owned()),
client_secret: Cow::from(masto.client_secret.to_owned()),
redirect: Cow::from(masto.redirect.to_owned()),
token: Cow::from(masto.token.to_owned()),
};
Mastodon::from(data)
}
/// Builds toot text from tweet
pub fn build_basic_status(
tweet: &Tweet,
mentions: &HashMap<String, String>,
url_regex_filter: &Option<Regex>,
url_alt_services: &Option<HashMap<String, String>>,
) -> String {
let mut toot = tweet.text.to_owned();
for decoded_url in decode_urls(&tweet.entities.urls, url_regex_filter, url_alt_services) {
toot = toot.replace(&decoded_url.0, &decoded_url.1);
}
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);
}
decode_html_entities(&toot).to_string()
} }
/// Generic register function /// Generic register function
/// As this function is supposed to be run only once, it will panic for every error it encounters /// 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 /// Most of this function is a direct copy/paste of the official `elefren` crate
#[tokio::main] pub fn register(host: &str, screen_name: &str) {
pub async fn register(host: &str, screen_name: &str) { let mut builder = App::builder();
let mastodon = generator(megalodon::SNS::Mastodon, host.to_string(), None, None); 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(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(),
));
let options = AppInputOptions { let app = builder.build().expect("Cannot build the app");
redirect_uris: None,
scopes: Some(
[
"read:accounts".to_string(),
"write:accounts".to_string(),
"write:media".to_string(),
"write:statuses".to_string(),
]
.to_vec(),
),
website: Some("https://framagit.org/veretcle/scootaloo".to_string()),
};
let app_data = mastodon let registration = Registration::new(host)
.register_app(env!("CARGO_PKG_NAME").to_string(), &options) .register(app)
.await .expect("Cannot build registration object");
.expect("Cannot build registration object!"); let url = registration
.authorize_url()
let url = app_data.url.expect("Cannot generate registration URI!"); .expect("Cannot generate registration URI!");
println!("Click this link to authorize on Mastodon: {}", url); println!("Click this link to authorize on Mastodon: {}", url);
println!("Paste the returned authorization code: "); println!("Paste the returned authorization code: ");
@@ -159,47 +127,27 @@ pub async fn register(host: &str, screen_name: &str) {
.read_line(&mut input) .read_line(&mut input)
.expect("Unable to read back registration code!"); .expect("Unable to read back registration code!");
let token_data = mastodon let code = input.trim();
.fetch_access_token( let mastodon = registration
app_data.client_id.to_owned(), .complete(code)
app_data.client_secret.to_owned(),
input.trim().to_string(),
megalodon::default::NO_REDIRECT.to_string(),
)
.await
.expect("Unable to create access token!"); .expect("Unable to create access token!");
let mastodon = generator( let toml = toml::to_string(&*mastodon).unwrap();
megalodon::SNS::Mastodon,
host.to_string(),
Some(token_data.access_token.to_owned()),
None,
);
let current_account = mastodon let current_account = mastodon
.verify_account_credentials() .verify_credentials()
.await .expect("Unable to access account information!");
.expect("Unable to access account information!")
.json();
println!( println!(
r#"Please insert the following block at the end of your configuration file: "Please insert the following block at the end of your configuration file:
[mastodon.{}] [mastodon.{}]
twitter_screen_name = "{}" twitter_screen_name = \"{}\"
mastodon_screen_name = "{}" mastodon_screen_name = \"{}\"
base = "{}" {}",
client_id = "{}"
client_secret = "{}"
redirect = "{}"
token = "{}""#,
screen_name.to_lowercase(), screen_name.to_lowercase(),
screen_name, screen_name,
current_account.username, current_account.username,
host, toml
app_data.client_id,
app_data.client_secret,
app_data.redirect_uri,
token_data.access_token,
); );
} }
@@ -207,209 +155,74 @@ token = "{}""#,
mod tests { mod tests {
use super::*; use super::*;
#[test] use chrono::prelude::*;
fn test_replace_tweet_by_toot() { use egg_mode::tweet::TweetEntities;
let mut associated_urls = HashMap::from([
(
"https://t.co/perdudeouf".to_string(),
"https://www.perdu.com".to_string(),
),
(
"https://t.co/realquoteshere".to_string(),
"https://twitter.com/nintendojofr/status/1590047921633755136".to_string(),
),
(
"https://t.co/almostthere".to_string(),
"https://twitter.com/NintendojoFR/status/nope".to_string(),
),
(
"http://t.co/yetanotherone".to_string(),
"https://twitter.com/NINTENDOJOFR/status/1590047921633755136".to_string(),
),
]);
let expected_urls = HashMap::from([
(
"https://t.co/perdudeouf".to_string(),
"https://www.perdu.com".to_string(),
),
(
"https://t.co/realquoteshere".to_string(),
"https://m.nintendojo.fr/@nintendojofr/109309605486908797".to_string(),
),
(
"https://t.co/almostthere".to_string(),
"https://twitter.com/NintendojoFR/status/nope".to_string(),
),
(
"http://t.co/yetanotherone".to_string(),
"https://m.nintendojo.fr/@nintendojofr/109309605486908797".to_string(),
),
]);
replace_tweet_by_toot(
&mut associated_urls,
"NintendojoFR",
1590047921633755136,
"nintendojofr",
"https://m.nintendojo.fr",
"109309605486908797",
);
assert_eq!(associated_urls, expected_urls);
}
#[test]
fn test_find_mastodon_screen_name_by_twitter_screen_name() {
let masto_config = HashMap::from([
(
"test".to_string(),
MastodonConfig {
twitter_screen_name: "tonpere".to_string(),
mastodon_screen_name: Some("lalali".to_string()),
twitter_page_size: None,
base: "https://mstdn.net".to_string(),
client_id: "".to_string(),
client_secret: "".to_string(),
redirect: "".to_string(),
token: "".to_string(),
},
),
(
"test2".to_string(),
MastodonConfig {
twitter_screen_name: "tamerelol".to_string(),
mastodon_screen_name: None,
twitter_page_size: None,
base: "https://mastoot.fr".to_string(),
client_id: "".to_string(),
client_secret: "".to_string(),
redirect: "".to_string(),
token: "".to_string(),
},
),
(
"test3".to_string(),
MastodonConfig {
twitter_screen_name: "NintendojoFR".to_string(),
mastodon_screen_name: Some("nintendojofr".to_string()),
twitter_page_size: None,
base: "https://m.nintendojo.fr".to_string(),
client_id: "".to_string(),
client_secret: "".to_string(),
redirect: "".to_string(),
token: "".to_string(),
},
),
]);
// case sensitiveness, to avoid any mistake
assert_eq!(
None,
find_mastodon_screen_name_by_twitter_screen_name("nintendojofr", &masto_config)
);
assert_eq!(
Some((
"nintendojofr".to_string(),
"https://m.nintendojo.fr".to_string()
)),
find_mastodon_screen_name_by_twitter_screen_name("NintendojoFR", &masto_config)
);
// should return None if twitter_screen_name is undefined
assert_eq!(
None,
find_mastodon_screen_name_by_twitter_screen_name("tamerelol", &masto_config)
);
assert_eq!(
Some(("lalali".to_string(), "https://mstdn.net".to_string())),
find_mastodon_screen_name_by_twitter_screen_name("tonpere", &masto_config)
);
}
#[test] #[test]
fn test_twitter_mentions() { fn test_twitter_mentions() {
let mention_entities = vec![ let mention_entity = MentionEntity {
MentionEntity {
id: 12345, id: 12345,
range: (1, 3), range: (1, 3),
name: "Ta Mere l0l".to_string(), name: "Ta Mere l0l".to_string(),
screen_name: "tamerelol".to_string(), screen_name: "tamerelol".to_string(),
}, };
MentionEntity {
id: 6789,
range: (1, 3),
name: "TONPERE".to_string(),
screen_name: "tonpere".to_string(),
},
];
let mut toot = ":kikoo: @tamerelol @tonpere !".to_string(); let twitter_ums = vec![mention_entity];
let masto_config = HashMap::from([( let mut expected_mentions = HashMap::new();
"test".to_string(), expected_mentions.insert(
(MastodonConfig { "@tamerelol".to_string(),
twitter_screen_name: "tonpere".to_string(), "@tamerelol@twitter.com".to_string(),
mastodon_screen_name: Some("lalali".to_string()), );
twitter_page_size: None,
base: "https://mstdn.net".to_string(),
client_id: "".to_string(),
client_secret: "".to_string(),
redirect: "".to_string(),
token: "".to_string(),
}),
)]);
twitter_mentions(&mut toot, &mention_entities, &masto_config); let decoded_mentions = twitter_mentions(&twitter_ums);
assert_eq!(&toot, ":kikoo: @tamerelol@twitter.com @lalali@mstdn.net !"); assert_eq!(expected_mentions, decoded_mentions);
} }
#[test] #[test]
fn test_decode_urls() { fn test_decode_urls() {
let urls = HashMap::from([ let url_entity1 = UrlEntity {
(
"https://t.co/thisisatest".to_string(),
"https://www.nintendojo.fr/dojobar".to_string(),
),
(
"https://t.co/nopenotinclusive".to_string(),
"invité.es".to_string(),
),
]);
let mut toot =
"Rendez-vous sur https://t.co/thisisatest avec nos https://t.co/nopenotinclusive !"
.to_string();
decode_urls(&mut toot, &urls);
assert_eq!(
&toot,
"Rendez-vous sur https://www.nintendojo.fr/dojobar avec nos invité.es !"
);
}
#[test]
fn test_associate_urls() {
let urls = vec![
UrlEntity {
display_url: "tamerelol".to_string(), display_url: "tamerelol".to_string(),
expanded_url: Some("https://www.nintendojo.fr/dojobar".to_string()), expanded_url: Some("https://www.nintendojo.fr/dojobar".to_string()),
range: (1, 3), range: (1, 3),
url: "https://t.me/tamerelol".to_string(), url: "https://t.me/tamerelol".to_string(),
}, };
UrlEntity {
display_url: "sadcat".to_string(), let url_entity2 = UrlEntity {
display_url: "tamerelol".to_string(),
expanded_url: None, expanded_url: None,
range: (1, 3), range: (1, 3),
url: "https://t.me/sadcat".to_string(), url: "https://t.me/tamerelol".to_string(),
}, };
UrlEntity {
let wrong_url_entity = UrlEntity {
display_url: "invité.es".to_string(), display_url: "invité.es".to_string(),
expanded_url: Some("http://xn--invit-fsa.es".to_string()), expanded_url: Some("http://xn--invit-fsa.es".to_string()),
range: (85, 108), range: (85, 108),
url: "https://t.co/WAUgnpHLmo".to_string(), url: "https://t.co/WAUgnpHLmo".to_string(),
}, };
let rewritten_url_entity = UrlEntity {
display_url: "youtu.be/w5TrSaoYmZ8".to_string(),
expanded_url: Some("https://youtu.be/w5TrSaoYmZ8".to_string()),
range: (0, 23),
url: "https://t.co/fUVYXuF7tg".to_string(),
};
let re = Regex::new("(.+)\\.es$").ok();
let alt: HashMap<String, String> = HashMap::from([
("youtube.com".to_string(), "invidio.us".to_string()),
("youtu.be".to_string(), "invidio.us".to_string()),
("www.youtube.com".to_string(), "invidio.us".to_string()),
]);
let twitter_urls = vec![
url_entity1,
url_entity2,
wrong_url_entity,
rewritten_url_entity,
]; ];
let expected_urls = HashMap::from([ let expected_urls = HashMap::from([
@@ -421,72 +234,83 @@ mod tests {
"https://t.co/WAUgnpHLmo".to_string(), "https://t.co/WAUgnpHLmo".to_string(),
"invité.es".to_string(), "invité.es".to_string(),
), ),
(
"https://t.co/fUVYXuF7tg".to_string(),
"https://invidio.us/w5TrSaoYmZ8".to_string(),
),
]); ]);
let re = Regex::new("(.+)\\.es$").ok(); let decoded_urls = decode_urls(&twitter_urls, &re, &Some(alt));
let associated_urls = associate_urls(&urls, &re); assert_eq!(expected_urls, decoded_urls);
assert_eq!(associated_urls, expected_urls);
} }
#[test] #[test]
fn test_replace_alt_services() { fn test_build_basic_status() {
let mut associated_urls = HashMap::from([ let t = Tweet {
( coordinates: None,
"https://t.co/youplaboom".to_string(), created_at: Utc::now(),
"https://www.youtube.com/watch?v=dQw4w9WgXcQ".to_string(), current_user_retweet: None,
), display_text_range: None,
( entities: TweetEntities {
"https://t.co/thisisfine".to_string(), hashtags: vec![],
"https://twitter.com/Nintendo/status/1594590628771688448".to_string(), symbols: vec![],
), urls: vec![
( UrlEntity {
"https://t.co/nopenope".to_string(), display_url: "youtube.com/watch?v=w5TrSa…".to_string(),
"https://www.nintendojo.fr/dojobar".to_string(), expanded_url: Some("https://www.youtube.com/watch?v=w5TrSaoYmZ8".to_string()),
), range: (93, 116),
( url: "https://t.co/zXw0FfX2Nt".to_string(),
"https://t.co/broken".to_string(), }
"http://youtu.be".to_string(), ],
), user_mentions: vec![
( MentionEntity {
"https://t.co/alsobroken".to_string(), id: 491500016,
"https://youtube.com".to_string(), 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,
},
extended_entities: None,
favorite_count: 0,
favorited: None,
filter_level: None,
id: 1491541246984306693,
in_reply_to_user_id: None,
in_reply_to_screen_name: None,
in_reply_to_status_id: None,
lang: None,
place: None,
possibly_sensitive: None,
quoted_status: None,
quoted_status_id: None,
retweet_count: 0,
retweeted: None,
retweeted_status: None,
source: None,
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,
withheld_in_countries: None,
withheld_scope: None,
};
let alt_services = HashMap::from([ let s: HashMap<String, String> = HashMap::from([(
("twitter.com".to_string(), "nitter.net".to_string()), "@WillyWonka".to_string(),
("youtu.be".to_string(), "invidio.us".to_string()), "@WillyWonka@chocolatefactory.org".to_string(),
("www.youtube.com".to_string(), "invidio.us".to_string()), )]);
("youtube.com".to_string(), "invidio.us".to_string()),
]);
let expected_urls = HashMap::from([ let t_out = build_basic_status(&t, &s, &None, &None);
(
"https://t.co/youplaboom".to_string(),
"https://invidio.us/watch?v=dQw4w9WgXcQ".to_string(),
),
(
"https://t.co/thisisfine".to_string(),
"https://nitter.net/Nintendo/status/1594590628771688448".to_string(),
),
(
"https://t.co/nopenope".to_string(),
"https://www.nintendojo.fr/dojobar".to_string(),
),
(
"https://t.co/broken".to_string(),
"http://youtu.be".to_string(),
),
(
"https://t.co/alsobroken".to_string(),
"https://youtube.com".to_string(),
),
]);
replace_alt_services(&mut associated_urls, &alt_services); 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");
assert_eq!(associated_urls, expected_urls);
} }
} }

View File

@@ -1,16 +1,22 @@
use crate::{twitter::get_tweet_media, ScootalooError}; use crate::{twitter::get_tweet_media, ScootalooError};
use std::{borrow::Cow, error::Error};
use egg_mode::tweet::Tweet; use egg_mode::tweet::Tweet;
use futures::{stream, stream::StreamExt};
use elefren::prelude::*;
use log::{error, info, warn}; use log::{error, info, warn};
use megalodon::{mastodon::Mastodon, megalodon::Megalodon};
use reqwest::Url; use reqwest::Url;
use std::error::Error;
use tokio::{ use tokio::{
fs::{create_dir_all, remove_file, File}, fs::{create_dir_all, remove_file, File},
io::copy, io::copy,
}; };
use futures::{stream, stream::StreamExt};
/// Generate associative table between media ids and tweet extended entities /// Generate associative table between media ids and tweet extended entities
pub async fn generate_media_ids( pub async fn generate_media_ids(
tweet: &Tweet, tweet: &Tweet,
@@ -40,10 +46,8 @@ pub async fn generate_media_ids(
let local_tweet_media_path = get_tweet_media(&media, &cache_path).await?; let local_tweet_media_path = get_tweet_media(&media, &cache_path).await?;
// upload media to Mastodon // upload media to Mastodon
let mastodon_media = mastodon let mastodon_media =
.upload_media(local_tweet_media_path.to_owned(), None) mastodon.media(Cow::from(local_tweet_media_path.to_owned()))?;
.await?
.json();
// at this point, we can safely erase the original file // at this point, we can safely erase the original file
// it doesnt matter if we cant remove, cache_media fn is idempotent // it doesnt matter if we cant remove, cache_media fn is idempotent
remove_file(&local_tweet_media_path).await.ok(); remove_file(&local_tweet_media_path).await.ok();