missav

src/providers/missav.rs

@@ -0,0 +1,375 @@
+use std::vec;
+use std::env;
+use error_chain::error_chain;
+use htmlentity::entity::{decode, ICodedDataTrait};
+use futures::future::join_all;
+use wreq::Client;
+use wreq::Proxy;
+use wreq_util::Emulation;
+use crate::db;
+use crate::providers::Provider;
+use crate::util::cache::VideoCache;
+use crate::util::flaresolverr::{FlareSolverrRequest, Flaresolverr};
+use crate::videos::ServerOptions;
+use crate::videos::{VideoItem};
+use crate::DbPool;
+
+
+error_chain! {
+    foreign_links {
+        Io(std::io::Error);
+        HttpRequest(wreq::Error);
+        JsonError(serde_json::Error);
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct MissavProvider {
+    url: String,
+}
+impl MissavProvider {
+    pub fn new() -> Self {
+        MissavProvider {
+            url: "https://missav.ws".to_string(),
+        }
+    }
+    async fn get(&self, cache:VideoCache, pool:DbPool, page: u8) -> Result<Vec<VideoItem>> {
+
+        let url_str = format!("{}/en/new?page={}", self.url, page);
+        
+        let old_items = match cache.get(&url_str) {
+            Some((time, items)) => {
+                if time.elapsed().unwrap_or_default().as_secs() < 60 * 60 {
+                    return Ok(items.clone());
+                }
+                else{
+                    items.clone()
+                }
+            }
+            None => {
+                vec![]
+            }
+        };
+
+        
+        let client = Client::builder().cert_verification(false).emulation(Emulation::Firefox136).build()?;
+
+        let proxy = Proxy::all("http://192.168.0.103:8081").unwrap();
+        
+        let mut response = client.get(url_str.clone())
+        .proxy(proxy.clone())
+        .send().await?;
+        if response.status().is_redirection(){
+            response = client.get(response.headers()["Location"].to_str().unwrap())
+            .proxy(proxy.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(), pool).await;
+            if !video_items.is_empty() {
+                cache.remove(&url_str);
+                cache.insert(url_str.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_str.clone(),
+                    maxTimeout: 60000,
+                })
+                .await;
+            let video_items = match result {
+                Ok(res) => {
+                    // println!("FlareSolverr response: {}", res);
+                    self.get_video_items_from_html(res.solution.response, pool).await
+                }
+                Err(e) => {
+                    println!("Error solving FlareSolverr: {}", e);
+                    return Err("Failed to solve FlareSolverr".into());
+                }
+            };
+            if !video_items.is_empty() {
+                cache.remove(&url_str);
+                cache.insert(url_str.clone(), video_items.clone());
+            } else {
+                return Ok(old_items);
+            }
+            Ok(video_items)
+        }
+    }
+    async fn query(&self, cache: VideoCache, pool:DbPool, page: u8, query: &str) -> Result<Vec<VideoItem>> {
+        let search_string = query.replace(" ", "%20");
+        let url_str = format!(
+            "{}/en/search/{}?page={}",
+            self.url, search_string, page
+        );
+        // Check our Video Cache. If the result is younger than 1 hour, we return it.
+        let old_items = match cache.get(&url_str) {
+            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 client = Client::builder().cert_verification(false).emulation(Emulation::Firefox136).build()?;
+
+        let proxy = Proxy::all("http://192.168.0.103:8081").unwrap();
+        
+        let mut response = client.get(url_str.clone())
+        .proxy(proxy.clone())
+        .send().await?;
+        if response.status().is_redirection(){
+            response = client.get(response.headers()["Location"].to_str().unwrap())
+            .proxy(proxy.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(), pool).await;
+            if !video_items.is_empty() {
+                cache.remove(&url_str);
+                cache.insert(url_str.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_str.clone(),
+                    maxTimeout: 60000,
+                })
+                .await;
+            let video_items = match result {
+                Ok(res) => {
+                    self.get_video_items_from_html(res.solution.response, pool).await
+                }
+                Err(e) => {
+                    println!("Error solving FlareSolverr: {}", e);
+                    return Err("Failed to solve FlareSolverr".into());
+                }
+            };
+            if !video_items.is_empty() {
+                cache.remove(&url_str);
+                cache.insert(url_str.clone(), video_items.clone());
+            } else{
+                return Ok(old_items);
+            }
+            Ok(video_items)
+        }
+    }
+
+    async fn get_video_items_from_html(&self, html: String, pool: DbPool) -> Vec<VideoItem> {
+        if html.is_empty() {
+            println!("HTML is empty");
+            return vec![];
+        }
+        let raw_videos = html
+            .split("@mouseenter=\"setPreview(\'")
+            .collect::<Vec<&str>>()[1..]
+            .to_vec();
+        let mut urls: Vec<String> = 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.to_string().trim());
+            // }
+           
+            let url_str = video_segment.split("<a href=\"").collect::<Vec<&str>>()[1]
+                .split("\"")
+                .collect::<Vec<&str>>()[0]
+                .to_string();
+            urls.push(url_str.clone());
+            
+        }
+        let futures = urls.into_iter().map(|el| self.get_video_item(el.clone(), pool.clone()));
+        let results: Vec<Result<VideoItem>> = join_all(futures).await;
+        let video_items: Vec<VideoItem> = results
+            .into_iter()
+            .filter_map(Result::ok)
+            .collect();   
+
+        return video_items;
+    }
+
+    async fn get_video_item(&self, url_str: String, pool: DbPool) -> Result<VideoItem> {
+        let mut conn = pool.get().expect("couldn't get db connection from pool");
+        let db_result = db::get_video(&mut conn,url_str.clone());
+        match db_result {
+            Ok(Some(entry)) => {
+                let video_item: VideoItem = serde_json::from_str(entry.as_str()).unwrap();
+                return Ok(video_item)
+            }
+            Ok(None) => {
+            }
+            Err(e) => {
+                println!("Error fetching video from database: {}", e);
+            }
+        }
+        drop(conn);
+        
+
+        let client = Client::builder().cert_verification(false).emulation(Emulation::Firefox136).build()?;
+
+        let proxy = Proxy::all("http://192.168.0.103:8081").unwrap();
+        
+        let mut response = client.get(url_str.clone())
+        .proxy(proxy.clone())
+        .send().await?;
+        if response.status().is_redirection(){
+            response = client.get(response.headers()["Location"].to_str().unwrap())
+            .proxy(proxy.clone())
+            .send().await?;
+        }
+        let vid;
+        if response.status().is_success() {
+            vid = response.text().await?;
+        } else {
+            return Err("Failed to fetch video details".into());
+        }
+        let mut title = vid.split("<meta property=\"og:title\" content=\"").collect::<Vec<&str>>()[1]
+            .split("\"")
+            .collect::<Vec<&str>>()[0]
+            .to_string();
+        title = decode(title.as_bytes()).to_string().unwrap_or(title);
+
+        let thumb = vid.split("<meta property=\"og:image\" content=\"").collect::<Vec<&str>>()[1]
+            .split("\"")
+            .collect::<Vec<&str>>()[0]
+            .to_string();
+
+        let raw_duration = vid.split("<meta property=\"og:video:duration\" content=\"").collect::<Vec<&str>>()[1]
+            .split("\"")
+            .collect::<Vec<&str>>()[0]
+            .to_string();
+        let duration = raw_duration.parse::<u32>().unwrap_or(0);
+
+        let id = url_str.split("/").collect::<Vec<&str>>()[4]
+            .to_string();
+        let mut tags = vec![];
+        if vid.contains("<span>Actress:</span>"){
+            for actress_snippet in vid.split("<span>Actress:</span>").collect::<Vec<&str>>()[1]
+            .split("</div>").collect::<Vec<&str>>()[0].split("class=\"text-nord13 font-medium\">"){
+                let tag = actress_snippet.split("<").collect::<Vec<&str>>()[0].trim()
+                    .to_string();
+                if !tag.is_empty(){
+                    tags.push(format!("@genre:{}", tag));
+                }
+            }
+        }
+        if vid.contains("<span>Actor:</span>"){
+            for actor_snippet in vid.split("<span>Actor:</span>").collect::<Vec<&str>>()[1]
+            .split("</div>").collect::<Vec<&str>>()[0].split("class=\"text-nord13 font-medium\">"){
+                let tag = actor_snippet.split("<").collect::<Vec<&str>>()[0].trim()
+                    .to_string();
+                if !tag.is_empty(){
+                    tags.push(format!("@genre:{}", tag));
+                }
+            }
+        }
+        
+        if vid.contains("<span>Maker:</span>"){
+            for maker_snippet in vid.split("<span>Maker:</span>").collect::<Vec<&str>>()[1]
+            .split("</div>").collect::<Vec<&str>>()[0]
+            .split("class=\"text-nord13 font-medium\">"){
+                let tag = maker_snippet.split("<").collect::<Vec<&str>>()[0].trim()
+                    .to_string();
+                if !tag.is_empty(){
+                    tags.push(format!("@genre:{}", tag));
+                }
+            }
+        }
+
+        if vid.contains("<span>Genre:</span>"){
+            for tag_snippet in vid.split("<span>Genre:</span>").collect::<Vec<&str>>()[1]
+            .split("</div>").collect::<Vec<&str>>()[0].split("class=\"text-nord13 font-medium\">"){
+                let tag = tag_snippet.split("<").collect::<Vec<&str>>()[0].trim()
+                    .to_string();
+                if !tag.is_empty(){
+                    tags.push(format!("@genre:{}", tag));
+                }
+            }
+        }
+        
+        let preview = format!("https://fourhoi.com/{}/preview.mp4",id.clone());
+
+        let mut video_url_parts = vid.split("m3u8").collect::<Vec<&str>>()[1]
+            .split("https").collect::<Vec<&str>>()[0]
+            .split("|").collect::<Vec<&str>>();
+        video_url_parts.reverse();  
+        let video_url = format!("https://{}.{}/{}-{}-{}-{}-{}/playlist.m3u8", 
+                                        video_url_parts[1],
+                                        video_url_parts[2],
+                                        video_url_parts[3],
+                                        video_url_parts[4],
+                                        video_url_parts[5],
+                                        video_url_parts[6],
+                                        video_url_parts[7]
+                                    );
+        let mut video_item = VideoItem::new(
+            id,
+            title,
+            video_url.clone(),
+            "missav".to_string(),
+            thumb,
+            duration,
+        )
+        .tags(tags)
+        .preview(preview)
+        ;
+
+        let mut conn = pool.get().expect("couldn't get db connection from pool");
+        let insert_result = db::insert_video(&mut conn, &url_str, &serde_json::to_string(&video_item)?);
+        match insert_result{
+            Ok(_) => (),
+            Err(e) => {println!("{:?}", e); }
+        }
+        drop(conn);
+
+        return Ok(video_item);
+    }
+}
+
+impl Provider for MissavProvider {
+    async fn get_videos(
+        &self,
+        cache: VideoCache,
+        pool: DbPool,
+        sort: String,
+        query: Option<String>,
+        page: String,
+        per_page: String,
+        options: ServerOptions,
+    ) -> Vec<VideoItem> {
+        let _ = options;
+        let _ = per_page;
+        let _ = sort;
+        let videos: std::result::Result<Vec<VideoItem>, Error> = match query {
+            Some(q) => self.query(cache, pool, page.parse::<u8>().unwrap_or(1), &q).await,
+            None => self.get(cache, pool, page.parse::<u8>().unwrap_or(1)).await,
+        };
+        match videos {
+            Ok(v) => v,
+            Err(e) => {
+                println!("Error fetching videos: {}", e);
+                vec![]
+            }
+        }
+    }
+}