Files
hottub/src/providers/pornhub.rs
2025-07-16 18:50:33 +00:00

259 lines
8.9 KiB
Rust

use crate::DbPool;
use crate::providers::Provider;
use crate::util::cache::VideoCache;
use crate::util::flaresolverr::{FlareSolverrRequest, Flaresolverr};
use crate::util::time::parse_time_to_seconds;
use crate::videos::{VideoItem};
use error_chain::error_chain;
use htmlentity::entity::{ICodedDataTrait, decode};
use std::env;
use std::vec;
use wreq::{Client, Proxy};
use wreq_util::Emulation;
error_chain! {
foreign_links {
Io(std::io::Error);
HttpRequest(wreq::Error);
}
}
#[derive(Debug, Clone)]
pub struct PornhubProvider {
url: String,
}
impl PornhubProvider {
pub fn new() -> Self {
PornhubProvider {
url: "https://www.pornhub.com".to_string(),
}
}
async fn get(
&self,
cache: VideoCache,
page: u8,
) -> Result<Vec<VideoItem>> {
let mut url = format!("{}/video?page={}", self.url, page);
if page == 1 {
url = format!("{}/video", self.url);
}
let old_items = match cache.get(&url) {
Some((time, items)) => {
if time.elapsed().unwrap_or_default().as_secs() < 60 * 60 {
println!("Cache hit for URL: {}", url);
return Ok(items.clone());
} else {
items.clone()
}
}
None => {
vec![]
}
};
let client = Client::builder().emulation(Emulation::Firefox136).build()?;
let response = client.get(url.clone()).send().await?;
if response.status().is_success() {
let text = response.text().await?;
let video_items: Vec<VideoItem> = self.get_video_items_from_html(text.clone());
if !video_items.is_empty() {
cache.remove(&url);
cache.insert(url.clone(), video_items.clone());
} else {
return Ok(old_items);
}
Ok(video_items)
} else {
let flare_url = env::var("FLARE_URL").expect("FLARE_URL not set");
let flare = Flaresolverr::new(flare_url);
let result = flare
.solve(FlareSolverrRequest {
cmd: "request.get".to_string(),
url: url.clone(),
maxTimeout: 60000,
})
.await;
let video_items = match result {
Ok(res) => {
// println!("FlareSolverr response: {}", res);
self.get_video_items_from_html(res.solution.response)
}
Err(e) => {
println!("Error solving FlareSolverr: {}", e);
return Err("Failed to solve FlareSolverr".into());
}
};
if !video_items.is_empty() {
cache.remove(&url);
cache.insert(url.clone(), video_items.clone());
} else {
return Ok(old_items);
}
Ok(video_items)
}
}
async fn query(
&self,
cache: VideoCache,
page: u8,
query: &str,
) -> Result<Vec<VideoItem>> {
let search_string = query.to_lowercase().trim().replace(" ", "+");
let mut url = format!("{}/video/search?search={}&page={}", self.url, search_string, page);
if page == 1 {
url = format!("{}/video/search?search={}", self.url, search_string);
}
// Check our Video Cache. If the result is younger than 1 hour, we return it.
let old_items = match cache.get(&url) {
Some((time, items)) => {
if time.elapsed().unwrap_or_default().as_secs() < 60 * 60 {
return Ok(items.clone());
} else {
let _ = cache.check().await;
return Ok(items.clone());
}
}
None => {
vec![]
}
};
let proxy = Proxy::all("http://192.168.0.103:8081").unwrap();
let client = Client::builder().cert_verification(false).emulation(Emulation::Firefox136).build()?;
let response = client.get(url.clone()).proxy(proxy).send().await?;
if response.status().is_success() {
let text = response.text().await?;
let video_items: Vec<VideoItem> = self.get_video_items_from_html(text.clone());
if !video_items.is_empty() {
cache.remove(&url);
cache.insert(url.clone(), video_items.clone());
} else {
return Ok(old_items);
}
Ok(video_items)
} else {
let flare_url = env::var("FLARE_URL").expect("FLARE_URL not set");
let flare = Flaresolverr::new(flare_url);
let result = flare
.solve(FlareSolverrRequest {
cmd: "request.get".to_string(),
url: url.clone(),
maxTimeout: 60000,
})
.await;
let video_items = match result {
Ok(res) => self.get_video_items_from_html(res.solution.response),
Err(e) => {
println!("Error solving FlareSolverr: {}", e);
return Err("Failed to solve FlareSolverr".into());
}
};
if !video_items.is_empty() {
cache.remove(&url);
cache.insert(url.clone(), video_items.clone());
} else {
return Ok(old_items);
}
Ok(video_items)
}
}
fn get_video_items_from_html(&self, html: String) -> Vec<VideoItem> {
if html.is_empty() {
println!("HTML is empty");
return vec![];
}
let mut items: Vec<VideoItem> = Vec::new();
println!("{}", html.contains("<ul id=\"video"));
let video_listing_content = html.split("<ul id=\"video").collect::<Vec<&str>>()[1];
let raw_videos = video_listing_content
.split("class=\"pcVideoListItem ")
.collect::<Vec<&str>>()[1..]
.to_vec();
for video_segment in &raw_videos {
// let vid = video_segment.split("\n").collect::<Vec<&str>>();
// for (index, line) in vid.iter().enumerate() {
// println!("Line {}: {}", index, line);
// }
let url = format!("{}{}", self.url, video_segment.split("<a href=\"").collect::<Vec<&str>>()[1]
.split("\"")
.collect::<Vec<&str>>()[0]);
if url == "https://www.pornhub.comjavascript:void(0)".to_string() {
continue;
}
let mut title = video_segment.split("\" title=\"").collect::<Vec<&str>>()[1]
.split("\"")
.collect::<Vec<&str>>()[0]
.to_string();
// html decode
title = decode(title.as_bytes()).to_string().unwrap_or(title);
let id = video_segment.split("data-video-id=\"").collect::<Vec<&str>>()[1]
.split("\"")
.collect::<Vec<&str>>()[0]
.to_string();
let raw_duration = video_segment.split("duration").collect::<Vec<&str>>()[1].split(">").collect::<Vec<&str>>()[1]
.split("<")
.collect::<Vec<&str>>()[0]
.to_string();
let duration = parse_time_to_seconds(&raw_duration).unwrap_or(0) as u32;
let thumb = video_segment.split("src=\"").collect::<Vec<&str>>()[1]
.split("\"")
.collect::<Vec<&str>>()[0]
.to_string();
let video_item = VideoItem::new(
id,
title,
url.to_string(),
"pornhub".to_string(),
thumb,
duration,
);
items.push(video_item);
}
return items;
}
}
impl Provider for PornhubProvider {
async fn get_videos(
&self,
cache: VideoCache,
pool: DbPool,
_channel: String,
sort: String,
query: Option<String>,
page: String,
per_page: String,
featured: String,
) -> Vec<VideoItem> {
let _ = per_page;
let _ = sort;
let _ = featured; // Ignored in this implementation
let _ = pool; // Ignored in this implementation
let videos: std::result::Result<Vec<VideoItem>, Error> = match query {
Some(q) => {
self.query(cache, page.parse::<u8>().unwrap_or(1), &q)
.await
}
None => {
self.get(cache, page.parse::<u8>().unwrap_or(1))
.await
}
};
match videos {
Ok(v) => v,
Err(e) => {
println!("Error fetching videos: {}", e);
vec![]
}
}
}
}