dynamic network loading for omgxxx

This commit is contained in:
Simon
2025-10-04 15:41:24 +00:00
parent 28a4c57616
commit ef625527a2
3 changed files with 224 additions and 67 deletions

View File

@@ -51,6 +51,7 @@ async fn main() -> std::io::Result<()> {
let cache: util::cache::VideoCache = crate::util::cache::VideoCache::new();
providers::init_providers_now();
web::HttpServer::new(move || {
web::App::new()

View File

@@ -47,6 +47,12 @@ pub static ALL_PROVIDERS: Lazy<HashMap<&'static str, DynProvider>> = Lazy::new(|
m
});
pub fn init_providers_now() {
// Idempotent & thread-safe: runs the Lazy init exactly once.
Lazy::force(&ALL_PROVIDERS);
}
#[async_trait]
pub trait Provider: Send + Sync {
async fn get_videos(

View File

@@ -1,15 +1,17 @@
use crate::api::ClientVersion;
use crate::status::*;
use crate::util::parse_abbreviated_number;
use crate::DbPool;
use crate::api::ClientVersion;
use crate::providers::Provider;
use crate::util::cache::VideoCache;
use crate::util::parse_abbreviated_number;
use crate::util::time::parse_time_to_seconds;
use crate::videos::{ServerOptions, VideoItem};
use crate::{status::*, util};
use async_trait::async_trait;
use error_chain::error_chain;
use htmlentity::entity::{ICodedDataTrait, decode};
use std::sync::{Arc, RwLock};
use std::thread;
use std::vec;
use async_trait::async_trait;
error_chain! {
foreign_links {
@@ -21,22 +23,127 @@ error_chain! {
#[derive(Debug, Clone)]
pub struct OmgxxxProvider {
url: String,
sites: Vec<FilterOption>,
networks: Vec<FilterOption>,
sites: Arc<RwLock<Vec<FilterOption>>>,
networks: Arc<RwLock<Vec<FilterOption>>>,
}
impl OmgxxxProvider {
pub fn new() -> Self {
OmgxxxProvider {
println!("new");
let provider = OmgxxxProvider {
url: "https://www.omg.xxx".to_string(),
sites: vec![],
networks: vec![],
sites: Arc::new(RwLock::new(vec![])),
networks: Arc::new(RwLock::new(vec![])),
};
// Kick off the background load but return immediately
provider.spawn_initial_load();
provider
}
fn spawn_initial_load(&self) {
println!("spawn_initial_load");
let url = self.url.clone();
// let sites = Arc::clone(&self.sites);
let networks = Arc::clone(&self.networks);
thread::spawn(move || {
// Create a tiny runtime just for these async tasks
let rt = tokio::runtime::Builder::new_current_thread()
.enable_all()
.build()
.expect("build tokio runtime");
rt.block_on(async move {
// If you have a streaming sites loader, call it here too
// if let Err(e) = Self::load_sites_into(&url, sites).await {
// eprintln!("load_sites_into failed: {e}");
// }
if let Err(e) = Self::load_networks(&url, networks).await {
eprintln!("load_networks failed: {e}");
}
});
});
}
/// === separate, self-contained logic ===
/// Replace the bodies with your real fetching/parsing code.
// async fn load_sites(base_url: &str) -> Result<Vec<FilterOption>> {
// // Example stub:
// // let html = reqwest::blocking::get(format!("{}/sites/", base_url))?.text()?;
// // Ok(parse_sites_from_html(&html))
// Ok(vec![
// FilterOption {
// id: "site-a".into(),
// title: "Site A".into(),
// },
// FilterOption {
// id: "site-b".into(),
// title: "Site B".into(),
// },
// ])
// }
async fn load_networks(base_url: &str, networks: Arc<RwLock<Vec<FilterOption>>>) -> Result<()> {
println!("load_networks");
let mut requester = util::requester::Requester::new();
let text = requester.get(&base_url).await.unwrap();
let networks_div = text.split("class=\"sites__list\"").collect::<Vec<&str>>()[1]
.split("</div>")
.collect::<Vec<&str>>()[0];
let mut networks_vec: Vec<FilterOption> = vec![];
for network_element in networks_div.split("sites__item").collect::<Vec<&str>>()[1..].to_vec() {
if network_element.contains("sites__all"){continue;}
let network_url = network_element.split("href=\"").collect::<Vec<&str>>()[1]
.split("\"")
.collect::<Vec<&str>>()[0];
let network_id = network_url.split("/").collect::<Vec<&str>>()[4].to_string();
let network_name = network_element.split(">").collect::<Vec<&str>>()[1]
.split("<")
.collect::<Vec<&str>>()[0]
.to_string();
networks_vec.push(FilterOption {
id: network_id.clone(),
title: network_name.clone(),
});
Self::push_unique(
&networks,
FilterOption {
id: network_id,
title: network_name,
},
);
}
return Ok(());
}
// Push one item with minimal lock time and dedup by id
fn push_unique(target: &Arc<RwLock<Vec<FilterOption>>>, item: FilterOption) {
if let Ok(mut vec) = target.write() {
if !vec.iter().any(|x| x.id == item.id) {
vec.push(item);
// Optional: keep it sorted for nicer UX
// vec.sort_by(|a,b| a.title.cmp(&b.title));
}
}
}
fn build_channel(&self, clientversion: ClientVersion) -> Channel {
let _ = clientversion;
let channel: crate::status::Channel = Channel{
let sites: Vec<FilterOption> = self
.sites
.read()
.map(|g| g.clone()) // or: .map(|g| g.to_vec())
.unwrap_or_default(); // or: .unwrap_or_else(|_| Vec::new())
let networks: Vec<FilterOption> = self
.networks
.read()
.map(|g| g.clone()) // or: .map(|g| g.to_vec())
.unwrap_or_default(); // or: .unwrap_or_else(|_| Vec::new())
Channel {
id: "omgxxx".to_string(),
name: "OMG XXX".to_string(),
description: "Free Porn Site".to_string(),
@@ -48,21 +155,21 @@ impl OmgxxxProvider {
ChannelOption {
id: "sort".to_string(),
title: "Sort".to_string(),
description: "Sort the Videos".to_string(), //"Sort the videos by Date or Name.".to_string(),
description: "Sort the Videos".to_string(),
systemImage: "list.number".to_string(),
colorName: "blue".to_string(),
options: vec![
FilterOption {
id: "latest-updates".to_string(),
title: "Latest".to_string(),
id: "latest-updates".into(),
title: "Latest".into(),
},
FilterOption {
id: "most-popular".to_string(),
title: "Most Viewed".to_string(),
id: "most-popular".into(),
title: "Most Viewed".into(),
},
FilterOption {
id: "top-rated".to_string(),
title: "Top Rated".to_string(),
id: "top-rated".into(),
title: "Top Rated".into(),
},
],
multiSelect: false,
@@ -70,28 +177,25 @@ impl OmgxxxProvider {
ChannelOption {
id: "sites".to_string(),
title: "Sites".to_string(),
description: "Sort the Videos".to_string(), //"Sort the videos by Date or Name.".to_string(),
description: "Sort the Videos".to_string(),
systemImage: "list.bullet.indent".to_string(),
colorName: "green".to_string(),
options: self.sites.clone(),
options: sites,
multiSelect: false,
},
ChannelOption {
id: "networks".to_string(),
title: "Networks".to_string(),
description: "Sort the Videos".to_string(), //"Sort the videos by Date or Name.".to_string(),
description: "Sort the Videos".to_string(),
systemImage: "list.dash".to_string(),
colorName: "purple".to_string(),
options: self.networks.clone(),
options: networks,
multiSelect: false,
}
},
],
nsfw: true,
cacheDuration: None,
};
return channel;
}
}
async fn get(
@@ -144,7 +248,17 @@ impl OmgxxxProvider {
if query.starts_with("@models:") {
search_type = "models";
}
let video_url = format!("{}/{}/{}/{}/", self.url, search_type, query.to_lowercase().trim().replace(" ","-").replace("@models:",""), page);
let video_url = format!(
"{}/{}/{}/{}/",
self.url,
search_type,
query
.to_lowercase()
.trim()
.replace(" ", "-")
.replace("@models:", ""),
page
);
// Check our Video Cache. If the result is younger than 1 hour, we return it.
let old_items = match cache.get(&video_url) {
Some((time, items)) => {
@@ -180,9 +294,12 @@ impl OmgxxxProvider {
}
let mut items: Vec<VideoItem> = Vec::new();
let raw_videos = html.split("videos_list_pagination").collect::<Vec<&str>>()[0]
.split(" class=\"pagination\" ").collect::<Vec<&str>>()[0]
.split("class=\"list-videos\"").collect::<Vec<&str>>()[1]
.split("class=\"item\"").collect::<Vec<&str>>()[1..]
.split(" class=\"pagination\" ")
.collect::<Vec<&str>>()[0]
.split("class=\"list-videos\"")
.collect::<Vec<&str>>()[1]
.split("class=\"item\"")
.collect::<Vec<&str>>()[1..]
.to_vec();
for video_segment in &raw_videos {
// let vid = video_segment.split("\n").collect::<Vec<&str>>();
@@ -190,48 +307,81 @@ impl OmgxxxProvider {
// println!("Line {}: {}", index, line);
// }
let video_url: String = video_segment.split("<a href=\"").collect::<Vec<&str>>()[1]
.split("\"")
.collect::<Vec<&str>>()[0].to_string();
.split("\"")
.collect::<Vec<&str>>()[0]
.to_string();
let mut title = video_segment.split(" title=\"").collect::<Vec<&str>>()[1]
.split("\"").collect::<Vec<&str>>()[0]
.split("\"")
.collect::<Vec<&str>>()[0]
.to_string();
// html decode
title = decode(title.as_bytes()).to_string().unwrap_or(title);
let id = video_url.split("/").collect::<Vec<&str>>()[4].to_string();
let thumb = match video_segment.split("img loading").collect::<Vec<&str>>()[1].contains("data-src=\"") {
true => video_segment.split("img loading").collect::<Vec<&str>>()[1].split("data-src=\"").collect::<Vec<&str>>()[1]
.split("\"").collect::<Vec<&str>>()[0]
.to_string(),
false => video_segment.split("img loading").collect::<Vec<&str>>()[1].split("data-original=\"").collect::<Vec<&str>>()[1]
.split("\"").collect::<Vec<&str>>()[0]
.to_string(),
let thumb = match video_segment.split("img loading").collect::<Vec<&str>>()[1]
.contains("data-src=\"")
{
true => video_segment.split("img loading").collect::<Vec<&str>>()[1]
.split("data-src=\"")
.collect::<Vec<&str>>()[1]
.split("\"")
.collect::<Vec<&str>>()[0]
.to_string(),
false => video_segment.split("img loading").collect::<Vec<&str>>()[1]
.split("data-original=\"")
.collect::<Vec<&str>>()[1]
.split("\"")
.collect::<Vec<&str>>()[0]
.to_string(),
};
let raw_duration = video_segment.split("<span class=\"duration\">").collect::<Vec<&str>>()[1]
.split("<").collect::<Vec<&str>>()[0]
.split(" ").collect::<Vec<&str>>().last().unwrap_or(&"")
.to_string();
let duration = parse_time_to_seconds(raw_duration.as_str()).unwrap_or(0) as u32;
let views = parse_abbreviated_number(video_segment.split("<div class=\"views\">").collect::<Vec<&str>>()[1]
let raw_duration = video_segment
.split("<span class=\"duration\">")
.collect::<Vec<&str>>()[1]
.split("<")
.collect::<Vec<&str>>()[0]
.to_string().as_str()).unwrap_or(0) as u32;
.split(" ")
.collect::<Vec<&str>>()
.last()
.unwrap_or(&"")
.to_string();
let duration = parse_time_to_seconds(raw_duration.as_str()).unwrap_or(0) as u32;
let views = parse_abbreviated_number(
video_segment
.split("<div class=\"views\">")
.collect::<Vec<&str>>()[1]
.split("<")
.collect::<Vec<&str>>()[0]
.to_string()
.as_str(),
)
.unwrap_or(0) as u32;
let preview = video_segment.split("data-preview=\"").collect::<Vec<&str>>()[1]
.split("\"").collect::<Vec<&str>>()[0]
let preview = video_segment
.split("data-preview=\"")
.collect::<Vec<&str>>()[1]
.split("\"")
.collect::<Vec<&str>>()[0]
.to_string();
let tags = match video_segment.contains("class=\"models\">"){
true => video_segment.split("class=\"models\">").collect::<Vec<&str>>()[1]
.split("</div>").collect::<Vec<&str>>()[0]
.split("href=\"").collect::<Vec<&str>>()[1..]
.into_iter().map(
|s| format!("@models:{}", s.split("/").collect::<Vec<&str>>()[4]
.to_string())
).collect::<Vec<String>>().to_vec(),
false => vec![]
}
;
let tags = match video_segment.contains("class=\"models\">") {
true => video_segment
.split("class=\"models\">")
.collect::<Vec<&str>>()[1]
.split("</div>")
.collect::<Vec<&str>>()[0]
.split("href=\"")
.collect::<Vec<&str>>()[1..]
.into_iter()
.map(|s| {
format!(
"@models:{}",
s.split("/").collect::<Vec<&str>>()[4].to_string()
)
})
.collect::<Vec<String>>()
.to_vec(),
false => vec![],
};
let video_item = VideoItem::new(
id,
@@ -243,14 +393,11 @@ impl OmgxxxProvider {
)
.views(views)
.preview(preview)
.tags(tags)
;
.tags(tags);
items.push(video_item);
}
return items;
}
}
#[async_trait]
@@ -269,7 +416,7 @@ impl Provider for OmgxxxProvider {
let _ = pool;
let videos: std::result::Result<Vec<VideoItem>, Error> = match query {
Some(q) => {
self.query(cache, page.parse::<u8>().unwrap_or(1), &q,options)
self.query(cache, page.parse::<u8>().unwrap_or(1), &q, options)
.await
}
None => {
@@ -286,7 +433,10 @@ impl Provider for OmgxxxProvider {
}
}
fn get_channel(&self, clientversion: ClientVersion) -> crate::status::Channel {
println!("Getting channel for omgxxx with client version: {:?}", clientversion);
println!(
"Getting channel for omgxxx with client version: {:?}",
clientversion
);
self.build_channel(clientversion)
}
}