feat(yt_dlp): Support a DeArrow post processor

2 files changed, 128 insertions, 0 deletions

diff --git a/crates/yt_dlp/src/post_processors/dearrow.rs b/crates/yt_dlp/src/post_processors/dearrow.rs
new file mode 100644
index 0000000..110beeb
--- /dev/null
+++ b/crates/yt_dlp/src/post_processors/dearrow.rs
@@ -0,0 +1,108 @@
+use log::{info, warn};
+use serde::{Deserialize, Serialize};
+
+use crate::{InfoJson, json_get};
+
+use super::PostProcessor;
+
+#[derive(Debug, Clone, Copy)]
+pub struct DeArrowPP;
+
+impl PostProcessor for DeArrowPP {
+    fn extractors(&self) -> &'static [&'static str] {
+        &["Youtube"]
+    }
+
+    fn process(&self, mut info: InfoJson) -> Result<InfoJson, super::Error> {
+        let mut output: DeArrowApi = reqwest::blocking::get(format!(
+            "https://sponsor.ajay.app/api/branding?videoID={}",
+            json_get!(info, "id", as_str)
+        ))?
+        .json()?;
+
+        output.titles.reverse();
+
+        let title_len = output.titles.len();
+        loop {
+            let Some(title) = output.titles.pop() else {
+                break;
+            };
+
+            if (title.locked || title.votes < 1) && title_len > 1 {
+                info!(
+                    "Skipping title {:#?}, as it is not good enough",
+                    title.value
+                );
+                // Skip titles that are not “good” enough.
+                continue;
+            }
+
+            if let Some(old_title) = info.insert(
+                "title".to_owned(),
+                serde_json::Value::String(title.value.clone()),
+            ) {
+                warn!("Updating title from {:#?} to {:#?}", old_title, title.value);
+                info.insert("original_title".to_owned(), old_title);
+            } else {
+                warn!("Setting title to {:#?}", title.value);
+            }
+
+            break;
+        }
+
+        Ok(info)
+    }
+}
+
+#[derive(Serialize, Deserialize)]
+/// See: <https://wiki.sponsor.ajay.app/w/API_Docs/DeArrow>
+struct DeArrowApi {
+    titles: Vec<Title>,
+    thumbnails: Vec<Thumbnail>,
+
+    #[serde(alias = "randomTime")]
+    random_time: Option<f64>,
+
+    #[serde(alias = "videoDuration")]
+    video_duration: Option<f64>,
+
+    #[serde(alias = "casualVotes")]
+    casual_votes: Vec<String>,
+}
+
+#[derive(Serialize, Deserialize)]
+struct Title {
+    /// Note: Titles will sometimes contain > before a word.
+    /// This tells the auto-formatter to not format a word.
+    /// If you have no auto-formatter, you can ignore this and replace it with an empty string
+    #[serde(alias = "title")]
+    value: String,
+
+    original: bool,
+    votes: u64,
+    locked: bool,
+
+    #[serde(alias = "UUID")]
+    uuid: String,
+
+    /// only present if requested
+    #[serde(alias = "userID")]
+    user_id: Option<String>,
+}
+
+#[derive(Serialize, Deserialize)]
+struct Thumbnail {
+    // null if original is true
+    timestamp: Option<f64>,
+
+    original: bool,
+    votes: u64,
+    locked: bool,
+
+    #[serde(alias = "UUID")]
+    uuid: String,
+
+    /// only present if requested
+    #[serde(alias = "userID")]
+    user_id: Option<String>,
+}
diff --git a/crates/yt_dlp/src/post_processors/mod.rs b/crates/yt_dlp/src/post_processors/mod.rs
new file mode 100644
index 0000000..6067c7a
--- /dev/null
+++ b/crates/yt_dlp/src/post_processors/mod.rs
@@ -0,0 +1,20 @@
+use crate::InfoJson;
+
+pub mod dearrow;
+
+pub trait PostProcessor: std::fmt::Debug + Send {
+    /// Process a [`InfoJson`] object and return the updated one.
+    ///
+    /// # Errors
+    /// If the processing steps failed.
+    fn process(&self, info: InfoJson) -> Result<InfoJson, Error>;
+
+    /// The supported extractors for this post processor
+    fn extractors(&self) -> &'static [&'static str];
+}
+
+#[derive(thiserror::Error, Debug)]
+pub enum Error {
+    #[error("Failed to access a api: {0}")]
+    Get(#[from] reqwest::Error),
+}