feat(anime_detect): perform tag trimming on series titles

Author: jonathanlmc

Cargo.lock

@@ -8,3 +8,4 @@ license.workspace = true
 tap = "1.0.1"
 thiserror = "2.0.18"
 tracing = "0.1.44"
+winnow = { version = "0.7.14", features = ["simd"] }

anime_detect/Cargo.toml

@@ -0,0 +1,49 @@
+use winnow::{
+    Parser, Result,
+    ascii::Caseless,
+    combinator::{alt, delimited, eof, opt, repeat, repeat_till},
+    token::{any, take_until},
+};
+
+const RESOLUTION_TAGS: [Caseless<&str>; 5] = [
+    Caseless("480p"),
+    Caseless("720p"),
+    Caseless("1080p"),
+    Caseless("2160p"),
+    Caseless("4320p"),
+];
+
+pub fn trim_name(mut name: &str) -> Result<String> {
+    let many_tags = repeat(0.., tag).map(|()| ()).void();
+
+    let title_until_tag_or_end =
+        repeat_till(1.., any, alt((tag, eof))).map(|(chars, _)| -> Vec<char> { chars });
+
+    (many_tags, title_until_tag_or_end)
+        .parse_next(&mut name)
+        .map(|(_, name)| name.into_iter().collect())
+}
+
+fn tag<'a>(input: &mut &'a str) -> Result<&'a str> {
+    (
+        opt(whitespace),
+        alt((brackets, parens, alt(RESOLUTION_TAGS))),
+        opt(whitespace),
+    )
+        .map(|(_, tag, _)| tag)
+        .parse_next(input)
+}
+
+fn parens<'a>(input: &mut &'a str) -> Result<&'a str> {
+    delimited('(', take_until(0.., ')'), ')').parse_next(input)
+}
+
+fn brackets<'a>(input: &mut &'a str) -> Result<&'a str> {
+    delimited('[', take_until(0.., ']'), ']').parse_next(input)
+}
+
+fn whitespace(input: &mut &str) -> Result<()> {
+    repeat(1.., alt((' ', '_', '.')))
+        .map(|()| ())
+        .parse_next(input)
+}

anime_detect/src/series/combinator.rs

@@ -1,9 +1,11 @@
+mod combinator;
+
 use std::{
     collections::HashMap,
     path::{Path, PathBuf},
 };
 
-use tap::TapOptional;
+use tap::TapFallible;
 
 use crate::{Episode, EpisodeSet};
 
@@ -34,16 +36,16 @@ impl Series {
             .ok_or(ParseSeriesError::InvalidPath)
             .map(|name| name.to_string_lossy())
             .map(|name| {
-                trim_name(&name)
-                    .tap_none(|| {
+                combinator::trim_name(&name)
+                    .tap_err(|err| {
                         tracing::warn!(
-                            "no series name could be determined for `{}`, \
-                            defaulting to full directory name",
+                            %err,
+                            "failed to parse series name at `{}`; \
+                            using full directory name",
                             dir.display()
                         );
                     })
-                    .unwrap_or(&name)
-                    .to_string()
+                    .unwrap_or_else(|_| name.into_owned())
             })?;
 
         let mut season_episodes = HashMap::new();
@@ -57,11 +59,6 @@ impl Series {
     }
 }
 
-fn trim_name(name: &str) -> Option<&str> {
-    // TODO
-    Some(name)
-}
-
 fn collect_episodes_in_dir(
     dir: &Path,
     season_episodes: &mut HashMap<u32, EpisodeSet>,

anime_detect/src/series.rs anime_detect/src/series/mod.rsanime_detect/src/series.rs renamed to anime_detect/src/series/mod.rs

@@ -1,7 +1,5 @@
-use tracing_subscriber::util::SubscriberInitExt;
-
 fn main() {
-    tracing_subscriber::registry().init();
+    tracing_subscriber::fmt().init();
 
     let dir = std::env::args().nth(1).expect("no path provided");