feat: add archive scraper

feat: add more information to errors

src/main.rs

@@ -71,14 +71,6 @@ enum Commands {
     Scrape {
         // The source to scrape from.
         source: scrapers::Source,
-
-        /// The file extension to search for, defaults to mp3.
-        #[clap(long, short, default_value = "mp3")]
-        extension: String,
-
-        /// Whether to include the full HTTP URL or just the distinguishing part.
-        #[clap(long, short)]
-        include_full: bool,
     },
 }
 
@@ -101,12 +93,9 @@ async fn main() -> eyre::Result<()> {
         match command {
             // TODO: Actually distinguish between sources.
             #[cfg(feature = "scrape")]
-            Commands::Scrape {
-                source,
-                extension,
-                include_full,
-            } => match source {
-                Source::Lofigirl => scrapers::lofigirl::scrape(extension, include_full).await?,
+            Commands::Scrape { source } => match source {
+                Source::Archive => scrapers::archive::scrape().await?,
+                Source::Lofigirl => scrapers::lofigirl::scrape().await?,
                 Source::Chillhop => scrapers::chillhop::scrape().await?,
             },
         }

src/player.rs

@@ -41,7 +41,8 @@ pub use error::Error;
 pub mod mpris;
 
 /// The time to wait in between errors.
-const TIMEOUT: Duration = Duration::from_secs(3);
+/// TODO: Make this configurable.
+const TIMEOUT: Duration = Duration::from_secs(5);
 
 /// Main struct responsible for queuing up & playing tracks.
 // TODO: Consider refactoring [Player] from being stored in an [Arc], into containing many smaller [Arc]s.

src/player/downloader.rs

@@ -49,7 +49,7 @@ impl Downloader {
             Ok(track) => self.player.tracks.write().await.push_back(track),
             Err(error) if !error.is_timeout() => {
                 if debug {
-                    panic!("{}", error)
+                    panic!("{error}")
                 }
 
                 sleep(TIMEOUT).await;

src/player/queue.rs

@@ -10,7 +10,7 @@ use crate::{
 impl Player {
     /// Fetches the next track from the queue, or a random track if the queue is empty.
     /// This will also set the current track to the fetched track's info.
-    async fn fetch(&self) -> Result<tracks::DecodedTrack, tracks::TrackError> {
+    async fn fetch(&self) -> Result<tracks::DecodedTrack, tracks::Error> {
         // TODO: Consider replacing this with `unwrap_or_else` when async closures are stablized.
         let track = self.tracks.write().await.pop_front();
         let track = if let Some(track) = track {
@@ -66,7 +66,7 @@ impl Player {
             Err(error) => {
                 if !error.is_timeout() {
                     if debug {
-                        panic!("{error:?}")
+                        panic!("{error}")
                     }
 
                     sleep(TIMEOUT).await;

src/scrapers.rs

@@ -8,12 +8,14 @@ use tokio::{
     io::AsyncWriteExt,
 };
 
+pub mod archive;
 pub mod chillhop;
 pub mod lofigirl;
 
 #[derive(Clone, Copy, PartialEq, Eq, Debug, ValueEnum)]
 pub enum Source {
     Lofigirl,
+    Archive,
     Chillhop,
 }
 
@@ -21,6 +23,7 @@ impl Source {
     pub fn cache_dir(&self) -> &'static str {
         match self {
             Source::Lofigirl => "lofigirl",
+            Source::Archive => "archive",
             Source::Chillhop => "chillhop",
         }
     }
@@ -28,6 +31,7 @@ impl Source {
     pub fn url(&self) -> &'static str {
         match self {
             Source::Chillhop => "https://chillhop.com",
+            Source::Archive => "https://ia601004.us.archive.org/31/items/lofigirl",
             Source::Lofigirl => "https://lofigirl.com/wp-content/uploads",
         }
     }

src/scrapers/archive.rs

@@ -0,0 +1,74 @@
+//! Has all of the functions for the `scrape` command.
+//!
+//! This command is completely optional, and as such isn't subject to the same
+//! quality standards as the rest of the codebase.
+
+use futures::{stream::FuturesOrdered, StreamExt};
+use lazy_static::lazy_static;
+use reqwest::Client;
+use scraper::{Html, Selector};
+
+use crate::scrapers::{get, Source};
+
+lazy_static! {
+    static ref SELECTOR: Selector = Selector::parse("html > body > pre > a").unwrap();
+}
+
+async fn parse(client: &Client, path: &str) -> eyre::Result<Vec<String>> {
+    let document = get(client, path, super::Source::Lofigirl).await?;
+    let html = Html::parse_document(&document);
+
+    Ok(html
+        .select(&SELECTOR)
+        .skip(1)
+        .map(|x| String::from(x.attr("href").unwrap()))
+        .collect())
+}
+
+/// This function basically just scans the entire file server, and returns a list of paths to mp3 files.
+///
+/// It's a bit hacky, and basically works by checking all of the years, then months, and then all of the files.
+/// This is done as a way to avoid recursion, since async rust really hates recursive functions.
+async fn scan() -> eyre::Result<Vec<String>> {
+    let client = Client::new();
+
+    let mut releases = parse(&client, "/").await?;
+    releases.truncate(releases.len() - 4);
+
+    // A little bit of async to run all of the months concurrently.
+    let mut futures = FuturesOrdered::new();
+
+    for release in releases {
+        let client = client.clone();
+        futures.push_back(async move {
+            let items = parse(&client, &release).await.unwrap();
+            items
+                .into_iter()
+                .filter_map(|x| {
+                    if x.ends_with(".mp3") {
+                        Some(format!("{release}{x}"))
+                    } else {
+                        None
+                    }
+                })
+                .collect::<Vec<String>>()
+        });
+    }
+
+    let mut files = Vec::new();
+    while let Some(mut result) = futures.next().await {
+        files.append(&mut result);
+    }
+
+    eyre::Result::Ok(files)
+}
+
+pub async fn scrape() -> eyre::Result<()> {
+    println!("{}/", Source::Lofigirl.url());
+    let files = scan().await?;
+    for file in files {
+        println!("{file}");
+    }
+
+    Ok(())
+}

src/scrapers/lofigirl.rs

@@ -8,7 +8,7 @@ use lazy_static::lazy_static;
 use reqwest::Client;
 use scraper::{Html, Selector};
 
-use crate::scrapers::{get, Source};
+use crate::scrapers::get;
 
 lazy_static! {
     static ref SELECTOR: Selector = Selector::parse("html > body > pre > a").unwrap();
@@ -29,10 +29,8 @@ async fn parse(client: &Client, path: &str) -> eyre::Result<Vec<String>> {
 ///
 /// It's a bit hacky, and basically works by checking all of the years, then months, and then all of the files.
 /// This is done as a way to avoid recursion, since async rust really hates recursive functions.
-async fn scan(extension: &str, include_full: bool) -> eyre::Result<Vec<String>> {
+async fn scan() -> eyre::Result<Vec<String>> {
     let client = Client::new();
-    let extension = &format!(".{}", extension);
-
     let items = parse(&client, "/").await?;
 
     let mut years: Vec<u32> = items
@@ -60,12 +58,8 @@ async fn scan(extension: &str, include_full: bool) -> eyre::Result<Vec<String>>
                 items
                     .into_iter()
                     .filter_map(|x| {
-                        if x.ends_with(extension) {
-                            if include_full {
-                                Some(format!("{}/{path}{x}", Source::Lofigirl.url()))
-                            } else {
-                                Some(format!("{path}{x}"))
-                            }
+                        if x.ends_with(".mp3") {
+                            Some(format!("{path}{x}"))
                         } else {
                             None
                         }
@@ -83,8 +77,8 @@ async fn scan(extension: &str, include_full: bool) -> eyre::Result<Vec<String>>
     eyre::Result::Ok(files)
 }
 
-pub async fn scrape(extension: String, include_full: bool) -> eyre::Result<()> {
-    let files = scan(&extension, include_full).await?;
+pub async fn scrape() -> eyre::Result<()> {
+    let files = scan().await?;
     for file in files {
         println!("{file}");
     }

src/tracks.rs

@@ -20,41 +20,15 @@ use std::{io::Cursor, path::Path, time::Duration};
 use bytes::Bytes;
 use inflector::Inflector as _;
 use rodio::{Decoder, Source as _};
-use thiserror::Error;
-use tokio::io;
 use unicode_segmentation::UnicodeSegmentation;
 use url::form_urlencoded;
 
+pub mod error;
 pub mod list;
 
-/// The error type for the track system, which is used to handle errors that occur
-/// while downloading, decoding, or playing tracks.
-#[derive(Debug, Error)]
-pub enum TrackError {
-    #[error("timeout")]
-    Timeout,
+pub use error::Error;
 
-    #[error("unable to decode")]
-    Decode(#[from] rodio::decoder::DecoderError),
-
-    #[error("invalid name")]
-    InvalidName,
-
-    #[error("invalid file path")]
-    InvalidPath,
-
-    #[error("unable to read file")]
-    File(#[from] io::Error),
-
-    #[error("unable to fetch data")]
-    Request(#[from] reqwest::Error),
-}
-
-impl TrackError {
-    pub const fn is_timeout(&self) -> bool {
-        matches!(self, Self::Timeout)
-    }
-}
+use crate::tracks::error::Context;
 
 /// Just a shorthand for a decoded [Bytes].
 pub type DecodedData = Decoder<Cursor<Bytes>>;
@@ -92,7 +66,7 @@ impl QueuedTrack {
     /// This will actually decode and format the track,
     /// returning a [`DecodedTrack`] which can be played
     /// and also has a duration & formatted name.
-    pub fn decode(self) -> eyre::Result<DecodedTrack, TrackError> {
+    pub fn decode(self) -> eyre::Result<DecodedTrack, Error> {
         DecodedTrack::new(self)
     }
 }
@@ -134,13 +108,13 @@ impl Info {
     /// Formats a name with [Inflector].
     /// This will also strip the first few numbers that are
     /// usually present on most lofi tracks.
-    fn format_name(name: &str) -> eyre::Result<String, TrackError> {
+    fn format_name(name: &str) -> eyre::Result<String, Error> {
         let path = Path::new(name);
 
         let stem = path
             .file_stem()
             .and_then(|x| x.to_str())
-            .ok_or(TrackError::InvalidName)?;
+            .ok_or((name, error::Kind::InvalidName))?;
         let formatted = Self::decode_url(stem)
             .to_lowercase()
             .to_title_case()
@@ -181,7 +155,7 @@ impl Info {
         name: TrackName,
         full_path: String,
         decoded: &DecodedData,
-    ) -> eyre::Result<Self, TrackError> {
+    ) -> eyre::Result<Self, Error> {
         let (display_name, custom_name) = match name {
             TrackName::Raw(raw) => (Self::format_name(&raw)?, false),
             TrackName::Formatted(custom) => (custom, true),
@@ -210,11 +184,12 @@ pub struct DecodedTrack {
 impl DecodedTrack {
     /// Creates a new track.
     /// This is equivalent to [`QueuedTrack::decode`].
-    pub fn new(track: QueuedTrack) -> eyre::Result<Self, TrackError> {
+    pub fn new(track: QueuedTrack) -> eyre::Result<Self, Error> {
         let data = Decoder::builder()
             .with_byte_len(track.data.len().try_into().unwrap())
             .with_data(Cursor::new(track.data))
-            .build()?;
+            .build()
+            .track(track.full_path.clone())?;
 
         let info = Info::new(track.name, track.full_path, &data)?;
 

src/tracks/error.rs

@@ -0,0 +1,61 @@
+#[derive(Debug, thiserror::Error)]
+pub enum Kind {
+    #[error("timeout")]
+    Timeout,
+
+    #[error("unable to decode: {0}")]
+    Decode(#[from] rodio::decoder::DecoderError),
+
+    #[error("invalid name")]
+    InvalidName,
+
+    #[error("invalid file path")]
+    InvalidPath,
+
+    #[error("unable to read file: {0}")]
+    File(#[from] std::io::Error),
+
+    #[error("unable to fetch data: {0}")]
+    Request(#[from] reqwest::Error),
+}
+
+#[derive(Debug, thiserror::Error)]
+#[error("{kind}\ntrack: {track}")]
+pub struct Error {
+    pub track: String,
+
+    #[source]
+    pub kind: Kind,
+}
+
+impl Error {
+    pub const fn is_timeout(&self) -> bool {
+        matches!(self.kind, Kind::Timeout)
+    }
+}
+
+impl<T, E> From<(T, E)> for Error
+where
+    T: Into<String>,
+    Kind: From<E>,
+{
+    fn from((track, err): (T, E)) -> Self {
+        Error {
+            track: track.into(),
+            kind: Kind::from(err),
+        }
+    }
+}
+
+pub trait Context<T> {
+    fn track(self, name: impl Into<String>) -> Result<T, Error>;
+}
+
+impl<T, E> Context<T> for Result<T, E>
+where
+    (String, E): Into<Error>,
+{
+    fn track(self, name: impl Into<String>) -> Result<T, Error> {
+        self.map_err(|e| (name.into(), e).into())
+    }
+}

src/tracks/list.rs

@@ -7,7 +7,10 @@ use rand::Rng as _;
 use reqwest::Client;
 use tokio::fs;
 
-use crate::{data_dir, tracks::TrackError};
+use crate::{
+    data_dir,
+    tracks::{self, error::Context},
+};
 
 use super::QueuedTrack;
 
@@ -52,7 +55,11 @@ impl List {
     }
 
     /// Downloads a raw track, but doesn't decode it.
-    async fn download(&self, track: &str, client: &Client) -> Result<(Bytes, String), TrackError> {
+    async fn download(
+        &self,
+        track: &str,
+        client: &Client,
+    ) -> Result<(Bytes, String), tracks::Error> {
         // If the track has a protocol, then we should ignore the base for it.
         let full_path = if track.contains("://") {
             track.to_owned()
@@ -62,28 +69,31 @@ impl List {
 
         let data: Bytes = if let Some(x) = full_path.strip_prefix("file://") {
             let path = if x.starts_with('~') {
-                let home_path = dirs::home_dir().ok_or(TrackError::InvalidPath)?;
-                let home = home_path.to_str().ok_or(TrackError::InvalidPath)?;
+                let home_path =
+                    dirs::home_dir().ok_or((track, tracks::error::Kind::InvalidPath))?;
+                let home = home_path
+                    .to_str()
+                    .ok_or((track, tracks::error::Kind::InvalidPath))?;
 
                 x.replace('~', home)
             } else {
                 x.to_owned()
             };
 
-            let result = tokio::fs::read(path).await?;
+            let result = tokio::fs::read(path.clone()).await.track(track)?;
             result.into()
         } else {
             let response = match client.get(full_path.clone()).send().await {
                 Ok(x) => Ok(x),
                 Err(x) => {
                     if x.is_timeout() {
-                        Err(TrackError::Timeout)
+                        Err((track, tracks::error::Kind::Timeout))
                     } else {
-                        Err(TrackError::Request(x))
+                        Err((track, tracks::error::Kind::Request(x)))
                     }
                 }
             }?;
-            response.bytes().await?
+            response.bytes().await.track(track)?
         };
 
         Ok((data, full_path))
@@ -93,7 +103,7 @@ impl List {
     ///
     /// The Result's error is a bool, which is true if a timeout error occured,
     /// and false otherwise. This tells lowfi if it shouldn't wait to try again.
-    pub async fn random(&self, client: &Client) -> Result<QueuedTrack, TrackError> {
+    pub async fn random(&self, client: &Client) -> Result<QueuedTrack, tracks::Error> {
         let (path, custom_name) = self.random_path();
         let (data, full_path) = self.download(&path, client).await?;