hottub/src/providers/hdporn92.rs

use crate::DbPool;
use crate::api::ClientVersion;
use crate::providers::{
    Provider, report_provider_error, report_provider_error_background, requester_or_default,
};
use crate::status::*;
use crate::videos::{ServerOptions, VideoFormat, VideoItem};
use async_trait::async_trait;
use error_chain::error_chain;
use futures::stream::{self, StreamExt};
use regex::Regex;
use scraper::{Html, Selector};

pub const CHANNEL_METADATA: crate::providers::ProviderChannelMetadata =
    crate::providers::ProviderChannelMetadata {
        group_id: "mainstream-tube",
        tags: &["tube", "mixed", "movies"],
    };

const BASE_URL: &str = "https://hdporn92.com";
const CHANNEL_ID: &str = "hdporn92";
const DEFAULT_PER_PAGE: usize = 24;
const ENRICH_CONCURRENCY: usize = 4;

const BROWSER_UA: &str =
    "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) \
     Chrome/124.0.0.0 Safari/537.36";

error_chain! {
    foreign_links {
        Io(std::io::Error);
        HttpRequest(wreq::Error);
    }
    errors {
        Parse(msg: String) {
            description("parse error")
            display("parse error: {}", msg)
        }
    }
}

#[derive(Debug, Clone)]
pub struct Hdporn92Provider;

#[derive(Debug, Clone)]
enum Target {
    Latest,
    Search { query: String },
}

#[derive(Debug, Clone)]
struct CardStub {
    id: String,
    title: String,
    url: String,
    thumb: String,
    quality: Option<String>,
    category_slug: Option<String>,
}

impl Hdporn92Provider {
    pub fn new() -> Self {
        Self
    }

    fn build_channel(&self, _clientversion: ClientVersion) -> Channel {
        Channel {
            id: CHANNEL_ID.to_string(),
            name: "HDPorn92".to_string(),
            description:
                "HD Porn videos from network studios on hdporn92.com. Plays through a \
                 monetized morencius.com (VidHide-family) embed; the HLS URL is signed \
                 and Referer-locked, so the page itself does not expose direct media."
                    .to_string(),
            premium: false,
            favicon: "https://www.google.com/s2/favicons?sz=64&domain=hdporn92.com".to_string(),
            status: "active".to_string(),
            categories: vec![],
            options: vec![],
            nsfw: true,
            cacheDuration: Some(1800),
            ytdlpCommand: Some("yt-dlp --impersonate chrome-120".to_string()),
        }
    }

    fn selector(value: &str) -> Result<Selector> {
        Selector::parse(value)
            .map_err(|error| Error::from(format!("selector `{value}` failed: {error}")))
    }

    fn pick_target(query: Option<&str>) -> Target {
        if let Some(q) = query.map(str::trim).filter(|q| !q.is_empty()) {
            return Target::Search { query: q.to_string() };
        }
        Target::Latest
    }

    fn build_listing_url(target: &Target, page: u16) -> String {
        let page = page.max(1);
        match target {
            Target::Latest => {
                if page <= 1 {
                    format!("{BASE_URL}/")
                } else {
                    format!("{BASE_URL}/page/{page}/")
                }
            }
            Target::Search { query } => {
                let encoded: String =
                    url::form_urlencoded::byte_serialize(query.as_bytes()).collect();
                if page <= 1 {
                    format!("{BASE_URL}/?s={encoded}")
                } else {
                    format!("{BASE_URL}/page/{page}/?s={encoded}")
                }
            }
        }
    }

    fn html_headers(referer: &str) -> Vec<(String, String)> {
        vec![
            ("User-Agent".to_string(), BROWSER_UA.to_string()),
            ("Referer".to_string(), referer.to_string()),
        ]
    }

    async fn fetch_html(options: &ServerOptions, url: &str, referer: &str) -> Result<String> {
        let mut requester = requester_or_default(options, CHANNEL_ID, "fetch_html");
        requester
            .get_with_headers(url, Self::html_headers(referer), None)
            .await
            .map_err(|error| Error::from(format!("request failed for {url}: {error}")))
    }

    fn parse_listing_page(html: &str) -> Result<Vec<CardStub>> {
        let document = Html::parse_document(html);
        let card_sel = Self::selector("article.loop-video[data-video-id]")?;
        let link_sel = Self::selector("a[href]")?;
        let thumb_attr_sel = Self::selector("[data-main-thumb]")?;
        let title_sel = Self::selector("header.entry-header span")?;
        let views_sel = Self::selector("span.views")?;
        let class_regex = Regex::new(r"\bpost-(\d+)\b")
            .map_err(|error| Error::from(format!("regex failed: {error}")))?;
        let category_regex = Regex::new(r"\bcategory-([a-z0-9-]+)")
            .map_err(|error| Error::from(format!("regex failed: {error}")))?;

        let mut stubs = Vec::new();
        let mut seen_ids = std::collections::HashSet::new();

        for card in document.select(&card_sel) {
            let class_attr = card.value().attr("class").unwrap_or_default();

            let id = class_attr
                .split_whitespace()
                .find_map(|c| c.strip_prefix("post-"))
                .map(str::to_string)
                .or_else(|| {
                    card.value()
                        .attr("data-post-id")
                        .map(str::to_string)
                })
                .unwrap_or_default();
            if id.is_empty() || !seen_ids.insert(id.clone()) {
                continue;
            }

            let url = card
                .select(&link_sel)
                .next()
                .and_then(|a| a.value().attr("href"))
                .map(str::to_string)
                .unwrap_or_default();
            if url.is_empty() {
                continue;
            }

            let title = card
                .select(&title_sel)
                .next()
                .map(|s| s.text().collect::<Vec<_>>().join(""))
                .map(|t| t.trim().to_string())
                .filter(|t| !t.is_empty())
                .unwrap_or_else(|| id.clone());

            let thumb = card
                .select(&thumb_attr_sel)
                .next()
                .and_then(|el| el.value().attr("data-main-thumb"))
                .map(str::to_string)
                .filter(|t| !t.is_empty())
                .or_else(|| {
                    card.select(&Self::selector("img[src]").ok()?)
                        .next()
                        .and_then(|img| img.value().attr("src"))
                        .map(str::to_string)
                })
                .unwrap_or_default();

            let quality = card
                .select(&views_sel)
                .next()
                .map(|s| s.text().collect::<Vec<_>>().join(""))
                .map(|t| t.trim().to_string())
                .filter(|t| !t.is_empty() && t != "0");

            let category_slug = class_regex
                .captures(class_attr)
                .and_then(|caps| caps.get(1))
                .map(|m| m.as_str().to_string())
                .or_else(|| {
                    category_regex
                        .captures(class_attr)
                        .and_then(|caps| caps.get(1))
                        .map(|m| m.as_str().to_string())
                });

            stubs.push(CardStub {
                id,
                title,
                url,
                thumb,
                quality,
                category_slug,
            });
        }

        Ok(stubs)
    }

    /// Extracts the morencius.com embed URL from a hdporn92 detail page.
    ///
    /// The detail page renders the embed inside an `<IFRAME SRC="...">` (note
    /// the uppercase tag and attribute names — the retrotube WordPress theme
    /// emits mixed-case markup).
    fn extract_embed_url(html: &str) -> Option<String> {
        // Find the morencius embed host fragment and walk back to the scheme
        // (the iframe `SRC` is a quoted URL like `SRC="https://morencius..."`).
        let needle = "morencius.com/embed/";
        let needle_idx = html.find(needle)?;
        // Walk back to the start of the URL (find `https://` or `http://`).
        let head = &html[..=needle_idx];
        let scheme_idx = head
            .rfind("https://")
            .or_else(|| head.rfind("http://"))?;
        let after_scheme = &html[scheme_idx..];
        // End of the URL: the next quote, space, or `>`.
        let end = after_scheme
            .find(|c: char| c == '"' || c == '\'' || c == ' ' || c == '>')
            .unwrap_or(after_scheme.len());
        Some(after_scheme[..end].to_string())
    }

    /// Extracts the embed ID from a `https://morencius.com/embed/{id}` URL.
    fn extract_embed_id(embed_url: &str) -> Option<String> {
        let prefix = "morencius.com/embed/";
        let idx = embed_url.find(prefix)?;
        let after = &embed_url[idx + prefix.len()..];
        let end = after
            .find(|c: char| c == '?' || c == '/' || c == '"' || c == '\'')
            .unwrap_or(after.len());
        if end == 0 {
            None
        } else {
            Some(after[..end].to_string())
        }
    }

    /// Decodes every P.A.C.K.E.R. `eval(function(p,a,c,k,e,d){...}(...))` block
    /// found in the morencius embed page and returns the deobfuscated JS
    /// source for each. The embed page carries more than one such block (an
    /// ad/VAST setup block plus the actual jwplayer setup block), so callers
    /// must scan all of them for the data they need rather than assume the
    /// first one is the relevant one.
    fn decode_packered_blocks(source: &str) -> Vec<String> {
        // Match: eval(function(p,a,c,k,e,d){...}('P',A,C,'K'.split('|')
        // We capture P, A, C, and the dictionary K.
        let Ok(re) = Regex::new(
            r#"eval\(function\(p,a,c,k,e,d\)\{.*?\}\('([\s\S]*?)',(\d+),(\d+),'([\s\S]*?)'\.split\('\|'\)\)\)"#,
        ) else {
            return Vec::new();
        };

        re.captures_iter(source)
            .filter_map(|caps| {
                let p = caps.get(1)?.as_str();
                let a: u32 = caps.get(2)?.as_str().parse().ok()?;
                let c: usize = caps.get(3)?.as_str().parse().ok()?;
                let k: Vec<&str> = caps.get(4)?.as_str().split('|').collect();

                // P.A.C.K.E.R. deobfuscation: iterate c-1..=0; replace
                // \b<token>\b with dictionary[i] (the original JS uses a
                // `\w+` lookalike via the `e` parameter, but in practice
                // every token is a base-A digit string).
                let mut result = p.to_string();
                for i in (0..c).rev() {
                    let value = k.get(i).copied().unwrap_or("");
                    if value.is_empty() {
                        continue;
                    }
                    let token = to_base(i, a);
                    // Build a word-boundary regex for the token. Base-A
                    // tokens are always made of [0-9a-z], so \b is a real
                    // word boundary.
                    let token_re =
                        Regex::new(&format!(r"\b{}\b", regex::escape(&token))).ok()?;
                    result = token_re.replace_all(&result, value).into_owned();
                }
                Some(result)
            })
            .collect()
    }

    /// Decodes the first P.A.C.K.E.R. block found in `source`. Kept for
    /// callers/tests that only care about a single block; production code
    /// that needs to find the block with a specific key (e.g. `hls4`) should
    /// use [`Self::decode_packered_blocks`] and scan all of them instead.
    fn decode_packered(source: &str) -> Option<String> {
        Self::decode_packered_blocks(source).into_iter().next()
    }

    /// Pulls the morencius HLS URL out of the decoded JS. The player itself
    /// picks `links.hls4 || links.hls3 || links.hls2` (not every tier is
    /// always populated, and hls4 is frequently absent), so mirror that
    /// priority order. At least one tier has been observed serving a `.txt`
    /// decoy instead of a real `.m3u8` playlist, so within that priority
    /// order prefer a candidate that actually looks like a stream URL.
    fn extract_hls_url(decoded_js: &str) -> Option<String> {
        let re = Regex::new(r#""(hls[234])"\s*:\s*"([^"]+)""#).ok()?;
        let candidates: Vec<(String, String)> = re
            .captures_iter(decoded_js)
            .filter_map(|caps| {
                let tier = caps.get(1)?.as_str().to_string();
                let raw = caps.get(2)?.as_str().replace("\\/", "/");
                Some((tier, raw))
            })
            .collect();

        let normalize = |path: &str| {
            if path.starts_with("http") {
                path.to_string()
            } else if path.starts_with('/') {
                format!("https://morencius.com{path}")
            } else {
                format!("https://morencius.com/{path}")
            }
        };

        const TIERS: [&str; 3] = ["hls4", "hls3", "hls2"];
        TIERS
            .iter()
            .filter_map(|tier| candidates.iter().find(|(t, _)| t == tier))
            .map(|(_, raw)| normalize(raw))
            .find(|url| url.to_lowercase().ends_with(".m3u8"))
            .or_else(|| {
                TIERS
                    .iter()
                    .filter_map(|tier| candidates.iter().find(|(t, _)| t == tier))
                    .map(|(_, raw)| normalize(raw))
                    .next()
            })
    }

    /// Pulls the integer duration (in seconds, may be decimal) from the decoded
    /// jwplayer setup block. The JS has `duration:"<sec>.<frac>"`.
    fn extract_duration(decoded_js: &str) -> Option<u32> {
        let re = Regex::new(r#"duration\s*:\s*"?(\d+(?:\.\d+)?)"?"#).ok()?;
        let caps = re.captures(decoded_js)?;
        let raw = caps.get(1)?.as_str().parse::<f64>().ok()?;
        Some(raw.round() as u32)
    }

    async fn enrich_card(stub: CardStub, options: &ServerOptions) -> Result<VideoItem> {
        // Step 1: fetch the detail page to find the morencius embed.
        let detail_html = Self::fetch_html(options, &stub.url, BASE_URL).await?;
        let embed_url = Self::extract_embed_url(&detail_html)
            .ok_or_else(|| Error::from(format!("no morencius embed in {}", stub.url)))?;
        // Sanity check the embed id — the URL is what we use as Referer, but
        // we want to fail early if the URL is malformed rather than silently
        // building a broken `formats[0]`.
        let _embed_id = Self::extract_embed_id(&embed_url)
            .ok_or_else(|| Error::from(format!("could not parse embed id from {embed_url}")))?;

        // Step 2: fetch the morencius embed page (referer = the hdporn92 detail).
        let player_html = Self::fetch_html(options, &embed_url, &stub.url).await?;
        let decoded_blocks = Self::decode_packered_blocks(&player_html);
        if decoded_blocks.is_empty() {
            return Err(Error::from(format!("no eval block in {embed_url}")));
        }
        // A page may carry multiple P.A.C.K.E.R. blocks (e.g. an ad/VAST
        // setup block); only one decodes to JS containing the player's
        // hls2/3/4 stream keys, so scan all of them rather than assuming
        // the first.
        let decoded = decoded_blocks
            .iter()
            .find(|block| Self::extract_hls_url(block).is_some())
            .ok_or_else(|| Error::from(format!("no hls url in decoded JS of {embed_url}")))?;
        let stream_path = Self::extract_hls_url(decoded)
            .ok_or_else(|| Error::from(format!("no hls url in decoded JS of {embed_url}")))?;
        let duration = Self::extract_duration(decoded).unwrap_or(0);

        // Build the format. morencius serves the m3u8 only when Referer is the
        // embed URL — without it, the response is 403.
        let mut format =
            VideoFormat::m3u8(stream_path, "auto".to_string(), "m3u8".to_string());
        format.add_http_header("Referer".to_string(), embed_url.clone());
        format.add_http_header("User-Agent".to_string(), BROWSER_UA.to_string());

        let mut item = VideoItem::new(
            stub.id,
            stub.title,
            stub.url,
            CHANNEL_ID.to_string(),
            stub.thumb,
            duration,
        );
        item.formats = Some(vec![format]);
        if let Some(slug) = stub.category_slug {
            item.tags = Some(vec![slug]);
        }

        Ok(item)
    }

    async fn fetch_page(
        target: Target,
        page: u16,
        per_page: usize,
        options: &ServerOptions,
    ) -> Result<Vec<VideoItem>> {
        let url = Self::build_listing_url(&target, page);
        let html = Self::fetch_html(options, &url, BASE_URL).await?;
        let stubs = Self::parse_listing_page(&html)?;
        if stubs.is_empty() {
            return Ok(vec![]);
        }

        let limited: Vec<_> = stubs.into_iter().take(per_page).collect();
        let options = options.clone();

        let items = stream::iter(limited.into_iter().map(|stub| {
            let options = options.clone();
            async move {
                match Self::enrich_card(stub, &options).await {
                    Ok(item) => Some(item),
                    Err(error) => {
                        report_provider_error_background(
                            CHANNEL_ID,
                            "fetch_page.enrich_card",
                            &error.to_string(),
                        );
                        None
                    }
                }
            }
        }))
        .buffer_unordered(ENRICH_CONCURRENCY)
        .filter_map(async move |v| v)
        .collect::<Vec<_>>()
        .await;

        Ok(items)
    }
}

/// Convert a non-negative integer to a string in the given base (radix).
/// The result uses lowercase 0-9a-z. Used for the P.A.C.K.E.R. base-A tokens.
fn to_base(mut value: usize, base: u32) -> String {
    debug_assert!(base >= 2 && base <= 36);
    if value == 0 {
        return "0".to_string();
    }
    let mut digits = Vec::new();
    let base = base as usize;
    while value > 0 {
        let d = (value % base) as u8;
        digits.push(if d < 10 {
            (b'0' + d) as char
        } else {
            (b'a' + (d - 10)) as char
        });
        value /= base;
    }
    digits.into_iter().rev().collect()
}

#[async_trait]
impl Provider for Hdporn92Provider {
    async fn get_videos(
        &self,
        _cache: crate::util::cache::VideoCache,
        _pool: DbPool,
        _sort: String,
        query: Option<String>,
        page: String,
        per_page: String,
        options: ServerOptions,
    ) -> Vec<VideoItem> {
        let page = page.parse::<u16>().unwrap_or(1).max(1);
        let per_page = per_page
            .parse::<usize>()
            .unwrap_or(DEFAULT_PER_PAGE)
            .clamp(1, 48);

        let normalized_query = query
            .as_deref()
            .map(str::trim)
            .filter(|q| !q.is_empty())
            .map(ToOwned::to_owned);

        let target = Self::pick_target(normalized_query.as_deref());

        match Self::fetch_page(target, page, per_page, &options).await {
            Ok(items) => items,
            Err(error) => {
                report_provider_error(CHANNEL_ID, "get_videos", &error.to_string()).await;
                vec![]
            }
        }
    }

    fn get_channel(&self, clientversion: ClientVersion) -> Option<Channel> {
        Some(self.build_channel(clientversion))
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn builds_listing_urls() {
        assert_eq!(
            Hdporn92Provider::build_listing_url(&Target::Latest, 1),
            "https://hdporn92.com/"
        );
        assert_eq!(
            Hdporn92Provider::build_listing_url(&Target::Latest, 2),
            "https://hdporn92.com/page/2/"
        );
        assert_eq!(
            Hdporn92Provider::build_listing_url(
                &Target::Search {
                    query: "isla wild".to_string()
                },
                1
            ),
            "https://hdporn92.com/?s=isla+wild"
        );
        assert_eq!(
            Hdporn92Provider::build_listing_url(
                &Target::Search {
                    query: "milf".to_string()
                },
                2
            ),
            "https://hdporn92.com/page/2/?s=milf"
        );
    }

    #[test]
    fn picks_search_for_query() {
        match Hdporn92Provider::pick_target(Some("test")) {
            Target::Search { query } => assert_eq!(query, "test"),
            other => panic!("expected Search, got {:?}", other),
        }
    }

    #[test]
    fn picks_latest_for_no_query() {
        assert!(matches!(
            Hdporn92Provider::pick_target(None),
            Target::Latest
        ));
        assert!(matches!(
            Hdporn92Provider::pick_target(Some("")),
            Target::Latest
        ));
        assert!(matches!(
            Hdporn92Provider::pick_target(Some("   ")),
            Target::Latest
        ));
    }

    #[test]
    fn extracts_embed_id_from_url() {
        assert_eq!(
            Hdporn92Provider::extract_embed_id("https://morencius.com/embed/vc6vwjo6ngbn"),
            Some("vc6vwjo6ngbn".to_string())
        );
        assert_eq!(
            Hdporn92Provider::extract_embed_id("https://morencius.com/embed/abc123?foo=bar"),
            Some("abc123".to_string())
        );
        assert_eq!(
            Hdporn92Provider::extract_embed_id("https://other.com/embed/abc"),
            None
        );
    }

    #[test]
    fn extracts_embed_url_from_detail_html() {
        let html = r#"<IFRAME SRC="https://morencius.com/embed/vc6vwjo6ngbn" FRAMEBORDER=0>"#;
        assert_eq!(
            Hdporn92Provider::extract_embed_url(html).as_deref(),
            Some("https://morencius.com/embed/vc6vwjo6ngbn")
        );
    }

    #[test]
    fn decodes_hls4_url_from_packered_js() {
        // A minimal P.A.C.K.E.R. block that decodes to JSON containing an
        // hls4 key. We construct it so the base-36 token `0` decodes to
        // "hls4" and `1` decodes to the m3u8 path. The `2` is the empty
        // string (no key beyond index 1) and the loop skips it.
        //
        // a = 36, c = 2, k = ["hls4", "/stream/abc/secret/123/456/master.m3u8"]
        // p = '{"0":"1"}'  (after substitution: {"hls4":"/stream/.../master.m3u8"})
        //
        // Reverse iteration:
        //   i=1: token="1", k[1]="..." → '{"0":"/stream/.../master.m3u8"}'
        //   i=0: token="0", k[0]="hls4" → '{"hls4":"/stream/.../master.m3u8"}'
        let source = r#"eval(function(p,a,c,k,e,d){e=function(c){return c};if(!''.replace(/^/,String)){while(c--){d[c.toString(a)]=k[c]||c.toString(a)};k=[function(e){return d[e]}];e=function(){return'\w+'};c=1};while(c--){if(k[c]){p=p.replace(new RegExp('\b'+e(c)+'\b','g'),k[c])}}return p}('{"0":"1"}',36,2,'hls4|/stream/abc/secret/123/456/master.m3u8'.split('|')));"#;
        let decoded = Hdporn92Provider::decode_packered(source)
            .expect("deobfuscator should return Some");
        // Sanity: the decoded text contains the hls4 key.
        assert!(decoded.contains("\"hls4\""), "decoded: {decoded}");
        let url = Hdporn92Provider::extract_hls_url(&decoded)
            .expect("hls4 should be present after decode");
        // The path was /stream/abc/secret/123/456/master.m3u8.
        assert_eq!(url, "https://morencius.com/stream/abc/secret/123/456/master.m3u8");
    }

    #[test]
    fn extracts_duration_from_decoded_js() {
        let decoded = r#"jwplayer().setup({duration:"1709.40",preload:'auto'})"#;
        assert_eq!(
            Hdporn92Provider::extract_duration(decoded),
            Some(1709)
        );
    }

    #[test]
    fn base_conversion() {
        assert_eq!(to_base(0, 36), "0");
        assert_eq!(to_base(35, 36), "z");
        assert_eq!(to_base(36, 36), "10");
        // 49 in base 36 is "1d" (the dictionary key for "hls3")
        assert_eq!(to_base(49, 36), "1d");
        // 51 in base 36 is "1f" (the dictionary key for "hls4")
        assert_eq!(to_base(51, 36), "1f");
    }

    #[test]
    fn decodes_real_morencius_html() {
        // Exercise the deobfuscator against a real captured morencius embed
        // page. The fixture is the iframe served to the iOS webview when
        // opening `hdporn92.com/mypervyfamily-isla-wild-.../`.
        let html = std::fs::read_to_string(
            "src/providers/testdata/hdporn92_morencius.html",
        )
        .expect("test fixture should exist");
        // The page carries several P.A.C.K.E.R. blocks (ad setup, player
        // setup, ...); only the player block decodes to JS containing hls4.
        let blocks = Hdporn92Provider::decode_packered_blocks(&html);
        assert!(!blocks.is_empty(), "real morencius HTML should be deobfuscable");
        let decoded = blocks
            .iter()
            .find(|block| Hdporn92Provider::extract_hls_url(block).is_some())
            .expect("one decoded block should contain an hls2/3/4 key");
        let url = Hdporn92Provider::extract_hls_url(decoded)
            .expect("an hls url should be present after decode");
        assert!(url.starts_with("https://"), "got {url}");
        // The duration is also present.
        let duration = Hdporn92Provider::extract_duration(decoded);
        assert!(duration.unwrap_or(0) > 0, "duration should be > 0");
    }
}