Add local season/episode matching and on-disk rename (skip-on-conflict, confirm-first)

crates/renamr-core/Cargo.toml

@@ -15,4 +15,9 @@ thiserror = "1.0"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 
+# Filename-based season/episode detection (local, no lookahead/lookbehind support
+# unlike Java's regex engine used by the original SeasonEpisodeMatcher, so patterns
+# are adapted to rely on \b word boundaries instead).
+regex = "1"
+
 [dev-dependencies]

crates/renamr-core/src/lib.rs

@@ -6,6 +6,8 @@
 //! parsing, web datasource clients, rename engine are not yet ported).
 
 pub mod format;
+pub mod matching;
+pub mod rename;
 
 use rhai::Dynamic;
 use std::collections::HashMap;

crates/renamr-core/src/matching/mod.rs

@@ -0,0 +1,156 @@
+//! Local, filename-based season/episode detection.
+//!
+//! This ports the pattern-matching half of the original FileBot's
+//! `similarity/SeasonEpisodeMatcher.java`: given a filename, guess a season/episode
+//! number pair and the series-name text that precedes it. Unlike the original (which
+//! runs on Java's regex engine and leans on lookahead/lookbehind), Rust's `regex` crate
+//! has neither, so every pattern here is rewritten to rely on `\b` word boundaries and
+//! character classes instead.
+//!
+//! Deliberately scoped to *local* filename parsing only - there is no metadata lookup
+//! here, no online API calls, and no fuzzy title matching against a show database (that
+//! is the much larger `EpisodeMetrics`-based engine, not yet ported). This is the
+//! "works offline the moment you load real files" increment.
+
+use regex::Regex;
+use serde::Serialize;
+use std::sync::OnceLock;
+
+/// A season/episode number pair guessed from a filename, plus the raw series-name text
+/// that appeared before the match.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
+#[serde(rename_all = "camelCase")]
+pub struct EpisodeGuess {
+    pub series_name: String,
+    pub season: i32,
+    pub episode: i32,
+}
+
+impl EpisodeGuess {
+    /// Formats as FileBot's conventional `{s00e00}` binding value, e.g. `S01E01`.
+    pub fn season_episode_label(&self) -> String {
+        format!("S{:02}E{:02}", self.season, self.episode)
+    }
+}
+
+fn season_episode_pattern() -> &'static Regex {
+    static RE: OnceLock<Regex> = OnceLock::new();
+    RE.get_or_init(|| {
+        Regex::new(r"(?i)\bseason\s*(\d{1,4})\D{0,5}episode\s*(\d{1,3})\b").unwrap()
+    })
+}
+
+fn s00e00_pattern() -> &'static Regex {
+    static RE: OnceLock<Regex> = OnceLock::new();
+    RE.get_or_init(|| Regex::new(r"(?i)\bs(\d{1,2})[^a-zA-Z0-9]{0,3}e(\d{1,3})\b").unwrap())
+}
+
+fn sxe_pattern() -> &'static Regex {
+    static RE: OnceLock<Regex> = OnceLock::new();
+    RE.get_or_init(|| Regex::new(r"\b(\d{1,2})x(\d{2,3})\b").unwrap())
+}
+
+/// Cleans up the raw text preceding a season/episode match into a plausible series
+/// name: `.`/`_` used as word separators become spaces, and leading/trailing
+/// whitespace/dash/dot clutter is trimmed.
+fn clean_series_name(raw: &str) -> String {
+    let replaced: String = raw
+        .chars()
+        .map(|c| if c == '.' || c == '_' { ' ' } else { c })
+        .collect();
+
+    let trimmed = replaced.trim_matches(|c: char| c.is_whitespace() || c == '-' || c == '.');
+
+    trimmed.split_whitespace().collect::<Vec<_>>().join(" ")
+}
+
+/// Tries each season/episode pattern in priority order (most explicit/specific first)
+/// and returns the first match, with the series name guessed from the text before it.
+///
+/// Returns `None` if no pattern matches - this is expected for movies, already-clean
+/// filenames, or anything that just doesn't carry a season/episode marker in its name.
+pub fn guess_episode(filename: &str) -> Option<EpisodeGuess> {
+    // `_` counts as a word character to regex's `\b`, so "show_name_s02e03" would
+    // otherwise fail to find a boundary right before the `s`. Match against a
+    // space-normalized copy instead (same length, so match byte-offsets still line up
+    // with the original string for slicing out the series-name prefix).
+    let normalized: String = filename
+        .chars()
+        .map(|c| if c == '_' { ' ' } else { c })
+        .collect();
+
+    let patterns = [season_episode_pattern(), s00e00_pattern(), sxe_pattern()];
+
+    for pattern in patterns {
+        if let Some(caps) = pattern.captures(&normalized) {
+            let whole = caps.get(0).unwrap();
+            let season: i32 = caps.get(1)?.as_str().parse().ok()?;
+            let episode: i32 = caps.get(2)?.as_str().parse().ok()?;
+
+            let prefix = &filename[..whole.start()];
+            let series_name = clean_series_name(prefix);
+
+            return Some(EpisodeGuess {
+                series_name,
+                season,
+                episode,
+            });
+        }
+    }
+
+    None
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn matches_dotted_s00e00_style() {
+        let guess = guess_episode("Breaking.Bad.S05E14.720p.HDTV.x264-KILLERS").unwrap();
+        assert_eq!(guess.series_name, "Breaking Bad");
+        assert_eq!(guess.season, 5);
+        assert_eq!(guess.episode, 14);
+        assert_eq!(guess.season_episode_label(), "S05E14");
+    }
+
+    #[test]
+    fn matches_dashed_sxe_style() {
+        let guess = guess_episode("Reacher - 1x01 - Welcome to Margrave").unwrap();
+        assert_eq!(guess.series_name, "Reacher");
+        assert_eq!(guess.season, 1);
+        assert_eq!(guess.episode, 1);
+        assert_eq!(guess.season_episode_label(), "S01E01");
+    }
+
+    #[test]
+    fn matches_verbose_season_episode_words() {
+        let guess = guess_episode("The Office Season 4 Episode 9 - Dinner Party").unwrap();
+        assert_eq!(guess.series_name, "The Office");
+        assert_eq!(guess.season, 4);
+        assert_eq!(guess.episode, 9);
+    }
+
+    #[test]
+    fn matches_underscored_prefix() {
+        let guess = guess_episode("some_show_name_s02e03_episode_title").unwrap();
+        assert_eq!(guess.series_name, "some show name");
+        assert_eq!(guess.season, 2);
+        assert_eq!(guess.episode, 3);
+    }
+
+    #[test]
+    fn returns_none_for_movie_filenames() {
+        assert!(guess_episode("Inception.2010.1080p.BluRay.x264").is_none());
+        assert!(guess_episode("The Matrix (1999)").is_none());
+    }
+
+    #[test]
+    fn prefers_more_specific_pattern_over_sxe() {
+        // s00e00 pattern should win over the looser 1x01-style pattern when both could
+        // technically apply to fragments of the same string.
+        let guess = guess_episode("Show.Name.S01E02.1080p").unwrap();
+        assert_eq!(guess.season, 1);
+        assert_eq!(guess.episode, 2);
+    }
+}

crates/renamr-core/src/rename/mod.rs

@@ -0,0 +1,249 @@
+//! The on-disk rename operation: given a planned `old path -> new filename` mapping,
+//! actually renames the files.
+//!
+//! Deliberately scoped for v1: same-directory renames only (no moving files into a
+//! different folder tree/organizing by naming scheme yet - that's a bigger feature with
+//! its own directory-creation and cross-filesystem-move concerns). Conflict policy is
+//! fixed at **skip**: if the computed target filename already exists and isn't the same
+//! underlying file, that item is left alone and reported as skipped rather than
+//! overwritten. There is no undo yet, so the UI is expected to show the caller a
+//! confirmation of the full plan before calling [`execute_renames`].
+
+use serde::{Deserialize, Serialize};
+use std::fs;
+use std::path::Path;
+
+/// One planned rename: `source` is the file's current full path, `new_name` is the
+/// filename (including extension) it should have afterward, in the same directory.
+#[derive(Debug, Clone, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct RenamePlanItem {
+    pub source: String,
+    pub new_name: String,
+}
+
+/// What happened to one planned rename.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
+#[serde(tag = "status", rename_all = "camelCase")]
+pub enum RenameOutcome {
+    /// Renamed successfully; `new_path` is the file's new full path.
+    Renamed { new_path: String },
+    /// The computed name was already the file's current name - nothing to do.
+    Unchanged,
+    /// A different file already exists at the target path; left untouched.
+    SkippedConflict { existing_path: String },
+    /// The rename was rejected or the OS reported an error.
+    Failed { error: String },
+}
+
+#[derive(Debug, Clone, Serialize)]
+pub struct RenameResult {
+    pub source: String,
+    pub outcome: RenameOutcome,
+}
+
+/// Rejects filenames that could escape the source file's own directory (path
+/// separators, `.`/`..`) or are empty. The rename engine only ever renames a file
+/// within its existing directory, so a `new_name` that isn't a plain filename is a bug
+/// upstream (or a hostile input) rather than something to interpret as a move.
+fn is_plain_filename(name: &str) -> bool {
+    !name.is_empty()
+        && name != "."
+        && name != ".."
+        && !name.contains('/')
+        && !name.contains('\\')
+}
+
+/// True if `a` and `b` refer to the same underlying file (used to allow case-only
+/// renames on case-insensitive filesystems, where the target "already exists" check
+/// would otherwise find the source file itself).
+fn same_file(a: &Path, b: &Path) -> bool {
+    match (fs::canonicalize(a), fs::canonicalize(b)) {
+        (Ok(pa), Ok(pb)) => pa == pb,
+        _ => false,
+    }
+}
+
+fn rename_one(source: &Path, new_name: &str) -> RenameOutcome {
+    if !is_plain_filename(new_name) {
+        return RenameOutcome::Failed {
+            error: format!("\"{new_name}\" is not a valid filename"),
+        };
+    }
+
+    let current_name = match source.file_name().and_then(|n| n.to_str()) {
+        Some(n) => n,
+        None => {
+            return RenameOutcome::Failed {
+                error: "source has no filename".to_string(),
+            }
+        }
+    };
+
+    if current_name == new_name {
+        return RenameOutcome::Unchanged;
+    }
+
+    let parent = match source.parent() {
+        Some(p) => p,
+        None => {
+            return RenameOutcome::Failed {
+                error: "source has no parent directory".to_string(),
+            }
+        }
+    };
+    let target = parent.join(new_name);
+
+    if target.exists() && !same_file(&target, source) {
+        return RenameOutcome::SkippedConflict {
+            existing_path: target.to_string_lossy().to_string(),
+        };
+    }
+
+    match fs::rename(source, &target) {
+        Ok(()) => RenameOutcome::Renamed {
+            new_path: target.to_string_lossy().to_string(),
+        },
+        Err(e) => RenameOutcome::Failed {
+            error: e.to_string(),
+        },
+    }
+}
+
+/// Executes a batch of planned renames in order. Items are processed sequentially (not
+/// in parallel) so that two items whose computed names collide with each other resolve
+/// the same way an on-disk collision would: the first one through wins, the second is
+/// reported as a skipped conflict rather than silently overwriting it.
+pub fn execute_renames(items: &[RenamePlanItem]) -> Vec<RenameResult> {
+    items
+        .iter()
+        .map(|item| RenameResult {
+            source: item.source.clone(),
+            outcome: rename_one(Path::new(&item.source), &item.new_name),
+        })
+        .collect()
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::sync::atomic::{AtomicU32, Ordering};
+
+    static COUNTER: AtomicU32 = AtomicU32::new(0);
+
+    /// Creates a fresh temp directory with one file in it, for tests that need to touch
+    /// the real filesystem. Returns (dir, file_path).
+    fn temp_file_setup(file_name: &str, contents: &str) -> (std::path::PathBuf, std::path::PathBuf) {
+        let n = COUNTER.fetch_add(1, Ordering::SeqCst);
+        let dir = std::env::temp_dir().join(format!("renamr_rename_test_{}_{}", std::process::id(), n));
+        fs::create_dir_all(&dir).unwrap();
+        let file = dir.join(file_name);
+        fs::write(&file, contents).unwrap();
+        (dir, file)
+    }
+
+    #[test]
+    fn renames_a_file_successfully() {
+        let (dir, file) = temp_file_setup("old.txt", "hello");
+        let items = vec![RenamePlanItem {
+            source: file.to_string_lossy().to_string(),
+            new_name: "new.txt".to_string(),
+        }];
+
+        let results = execute_renames(&items);
+        assert_eq!(results.len(), 1);
+        match &results[0].outcome {
+            RenameOutcome::Renamed { new_path } => {
+                assert!(new_path.ends_with("new.txt"));
+                assert!(Path::new(new_path).exists());
+                assert!(!file.exists());
+            }
+            other => panic!("expected Renamed, got {other:?}"),
+        }
+
+        fs::remove_dir_all(&dir).ok();
+    }
+
+    #[test]
+    fn reports_unchanged_when_name_is_already_correct() {
+        let (dir, file) = temp_file_setup("same.txt", "hello");
+        let items = vec![RenamePlanItem {
+            source: file.to_string_lossy().to_string(),
+            new_name: "same.txt".to_string(),
+        }];
+
+        let results = execute_renames(&items);
+        assert_eq!(results[0].outcome, RenameOutcome::Unchanged);
+        assert!(file.exists());
+
+        fs::remove_dir_all(&dir).ok();
+    }
+
+    #[test]
+    fn skips_when_target_already_exists() {
+        let (dir, file) = temp_file_setup("a.txt", "aaa");
+        fs::write(dir.join("b.txt"), "bbb").unwrap();
+
+        let items = vec![RenamePlanItem {
+            source: file.to_string_lossy().to_string(),
+            new_name: "b.txt".to_string(),
+        }];
+
+        let results = execute_renames(&items);
+        match &results[0].outcome {
+            RenameOutcome::SkippedConflict { existing_path } => {
+                assert!(existing_path.ends_with("b.txt"));
+            }
+            other => panic!("expected SkippedConflict, got {other:?}"),
+        }
+        // Neither file should have been touched.
+        assert!(file.exists());
+        assert_eq!(fs::read_to_string(dir.join("b.txt")).unwrap(), "bbb");
+
+        fs::remove_dir_all(&dir).ok();
+    }
+
+    #[test]
+    fn second_item_in_batch_skips_when_it_collides_with_the_first() {
+        let (dir, file_a) = temp_file_setup("a.txt", "aaa");
+        let file_b = dir.join("c.txt");
+        fs::write(&file_b, "ccc").unwrap();
+
+        let items = vec![
+            RenamePlanItem {
+                source: file_a.to_string_lossy().to_string(),
+                new_name: "target.txt".to_string(),
+            },
+            RenamePlanItem {
+                source: file_b.to_string_lossy().to_string(),
+                new_name: "target.txt".to_string(),
+            },
+        ];
+
+        let results = execute_renames(&items);
+        assert!(matches!(results[0].outcome, RenameOutcome::Renamed { .. }));
+        assert!(matches!(
+            results[1].outcome,
+            RenameOutcome::SkippedConflict { .. }
+        ));
+        // The second file's original content should be untouched.
+        assert_eq!(fs::read_to_string(&file_b).unwrap(), "ccc");
+
+        fs::remove_dir_all(&dir).ok();
+    }
+
+    #[test]
+    fn rejects_path_traversal_in_new_name() {
+        let (dir, file) = temp_file_setup("a.txt", "aaa");
+        let items = vec![RenamePlanItem {
+            source: file.to_string_lossy().to_string(),
+            new_name: "../escaped.txt".to_string(),
+        }];
+
+        let results = execute_renames(&items);
+        assert!(matches!(results[0].outcome, RenameOutcome::Failed { .. }));
+        assert!(file.exists());
+
+        fs::remove_dir_all(&dir).ok();
+    }
+}