Use upstream human_format

Handle missing subtitle encoding.

Cargo.toml

@@ -9,7 +9,7 @@ edition = "2018"
 [dependencies]
 failure = "0.1"
 glob = "0.3"
-human_format = { git ="https://github.com/wathiede/human-format-rs" }
+human_format = "1.1.0"
 humantime = "1"
 lazy_static = "1.4"
 log = "0.4"

src/lib.rs

@@ -1,37 +1,37 @@
-use std::collections::HashMap;
+use std::{
-use std::env;
+    cmp::Ordering,
-use std::ffi::OsStr;
+    collections::{HashMap, HashSet},
-use std::fmt;
+    env,
-use std::fmt::Display;
+    ffi::OsStr,
-use std::fmt::Formatter;
+    fmt,
-use std::fs::File;
+    fmt::{Display, Formatter},
-use std::io::BufReader;
+    fs::File,
-use std::io::BufWriter;
+    io::{BufReader, BufWriter},
-use std::path::Path;
+    path::{Path, PathBuf},
-use std::path::PathBuf;
+    process::Command,
-use std::process::Command;
+    str::FromStr,
-use std::str::FromStr;
+};
 
-use failure::bail;
+use failure::{bail, Error, ResultExt};
-use failure::Error;
-use failure::ResultExt;
 use glob::glob;
 use lazy_static::lazy_static;
-use log::error;
+use log::{error, info};
-use log::info;
+use rayon::{iter::ParallelBridge, prelude::ParallelIterator};
-use rayon::iter::ParallelBridge;
-use rayon::prelude::ParallelIterator;
 use regex::Regex;
-use serde::de;
+use serde::{de, de::Deserializer, Deserialize, Serialize};
-use serde::de::Deserializer;
-use serde::Deserialize;
-use serde::Serialize;
 use serde_json::Value;
 
 const FULL_METADATA_FILENAME: &str = "metadata.json";
 const COMPACT_METADATA_FILENAME: &str = "metadata.compact.json";
 
-#[derive(Clone, Deserialize, Debug, PartialEq, Serialize)]
+pub fn clean_path_parent<P: AsRef<Path>>(path: P) -> PathBuf {
+    let path = path.as_ref();
+    let path = path.parent().unwrap();
+    let path = path.to_str().unwrap();
+    PathBuf::from(path)
+}
+
+#[derive(Clone, Deserialize, Debug, Eq, PartialEq, Serialize)]
 pub struct Resolution(usize, usize);
 impl From<(usize, usize)> for Resolution {
     fn from(res: (usize, usize)) -> Self {
@@ -46,6 +46,20 @@ impl Display for Resolution {
     }
 }
 
+impl Ord for Resolution {
+    fn cmp(&self, other: &Resolution) -> Ordering {
+        let pixels = self.0.checked_mul(self.1).unwrap_or(usize::max_value());
+        let other_pixels = other.0.checked_mul(other.1).unwrap_or(usize::max_value());
+        pixels.cmp(&other_pixels)
+    }
+}
+
+impl PartialOrd for Resolution {
+    fn partial_cmp(&self, other: &Resolution) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
 fn option_from_str<'de, T, D>(deserializer: D) -> Result<Option<T>, D::Error>
 where
     T: FromStr,
@@ -121,7 +135,7 @@ enum Stream {
         codec_long_name: String,
         coded_height: usize,
         coded_width: usize,
-        display_aspect_ratio: String,
+        display_aspect_ratio: Option<String>,
         #[serde(default, deserialize_with = "from_str")]
         duration: f32,
         height: usize,
@@ -133,13 +147,13 @@ enum Stream {
         codec_name: String,
         codec_long_name: String,
         channels: usize,
-        channel_layout: String,
+        channel_layout: Option<String>,
         tags: Option<Tags>,
     },
     #[serde(rename = "subtitle")]
     Subtitle {
-        codec_name: String,
+        codec_name: Option<String>,
-        codec_long_name: String,
+        codec_long_name: Option<String>,
         tags: Option<Tags>,
     },
     #[serde(rename = "attachment")]
@@ -185,7 +199,8 @@ pub struct AudioFormat {
     short_name: String,
     long_name: String,
     channels: usize,
-    channel_layout: String,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    channel_layout: Option<String>,
     #[serde(skip_serializing_if = "Option::is_none")]
     title: Option<String>,
     #[serde(skip_serializing_if = "Option::is_none")]
@@ -194,8 +209,8 @@ pub struct AudioFormat {
 
 #[derive(Clone, Deserialize, Debug, PartialEq, Serialize)]
 pub struct SubtitleFormat {
-    short_name: String,
+    short_name: Option<String>,
-    long_name: String,
+    long_name: Option<String>,
     #[serde(skip_serializing_if = "Option::is_none")]
     title: Option<String>,
     #[serde(skip_serializing_if = "Option::is_none")]
@@ -256,7 +271,7 @@ pub struct MetadataFile {
 
 #[derive(PartialEq, Debug)]
 pub struct MovieLibrary {
-    root: String,
+    pub root: String,
 }
 
 fn json_metadata_for_path<P: AsRef<Path> + AsRef<OsStr>>(path: P) -> Result<String, Error> {
@@ -273,7 +288,7 @@ fn json_metadata_for_path<P: AsRef<Path> + AsRef<OsStr>>(path: P) -> Result<Stri
     ])
     .arg(Path::new("./").join(path));
     info!(target: "json", "cmd {:?}", cmd);
-    let output = cmd.output()?;
+    let output = cmd.output().context(format!("failed to run {:?}", cmd))?;
     if output.status.success() {
         return Ok(String::from_utf8(output.stdout)?);
     }
@@ -291,8 +306,60 @@ lazy_static! {
 }
 
 #[derive(Default, Debug, PartialEq)]
-struct Movie {
+pub struct Movie {
-    files: Vec<(String, CompactMetadata)>,
+    pub files: Vec<(String, CompactMetadata)>,
+}
+
+impl Movie {
+    fn max_pixel_count(&self) -> Option<usize> {
+        if self.files.is_empty() {
+            None
+        } else {
+            Some(self.files.iter().fold(usize::min_value(), |acc, (_, cmd)| {
+                let min = cmd.video.iter().fold(usize::min_value(), |acc, v| {
+                    std::cmp::max(acc, v.width * v.height)
+                });
+                std::cmp::max(acc, min)
+            }))
+        }
+    }
+    fn min_pixel_count(&self) -> Option<usize> {
+        if self.files.is_empty() {
+            None
+        } else {
+            Some(self.files.iter().fold(usize::max_value(), |acc, (_, cmd)| {
+                let min = cmd.video.iter().fold(usize::max_value(), |acc, v| {
+                    std::cmp::min(acc, v.width * v.height)
+                });
+                std::cmp::min(acc, min)
+            }))
+        }
+    }
+    fn min_bit_rate(&self) -> Option<usize> {
+        if self.files.is_empty() {
+            None
+        } else {
+            Some(self.files.iter().fold(usize::max_value(), |acc, (_, cmd)| {
+                std::cmp::min(acc, cmd.bit_rate)
+            }))
+        }
+    }
+}
+
+impl Display for Movie {
+    fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
+        let p = &self.files.first().unwrap().0;
+        write!(f, "{}", &p[..p.find("/").unwrap()])?;
+        for (path, cmd) in &self.files {
+            write!(
+                f,
+                " {} {}",
+                &path[path.rfind("/").unwrap()..],
+                cmd.largest_dimension().unwrap()
+            )?;
+        }
+        Ok(())
+    }
 }
 
 #[derive(Debug, PartialEq)]
@@ -300,6 +367,74 @@ pub struct Movies {
     movies: Vec<Movie>,
 }
 
+impl Movies {
+    /// Find all movies with multiple copies.  The returned vec contains a tuple of (Movie to keep,
+    /// One or more Movies to remove).  The highest resolution movie is kept.
+    /// Movies with differing years are considered distinct movies.
+    /// If there is a yearless movie and one or more movies with a year exist, then the yearless
+    /// movie will be removed
+    pub fn duplicate_candidates(&self) -> Vec<(&Movie, Vec<&Movie>)> {
+        lazy_static! {
+            static ref MULTIPLE_SPACES: Regex = Regex::new(r"\s+").unwrap();
+        }
+        let date_re = Regex::new(r"\(\d{4}\)$").unwrap();
+        let mut movie_counter = HashMap::new();
+        let mut movies_without_date_counter = HashMap::new();
+        for m in &self.movies {
+            let (path, _cmd) = m.files.first().unwrap();
+            let parent = MULTIPLE_SPACES
+                .replace_all(
+                    &clean_path_parent(path)
+                        .to_string_lossy()
+                        .to_ascii_lowercase()
+                        .replace("-", " ")
+                        .replace("'", " "),
+                    " ",
+                )
+                .to_string();
+            if date_re.is_match(&parent) {
+                movie_counter.entry(parent).or_insert(Vec::new()).push(m);
+            } else {
+                movies_without_date_counter
+                    .entry(parent)
+                    .or_insert(Vec::new())
+                    .push(m);
+            }
+        }
+        let mut dupes: Vec<(&Movie, Vec<&Movie>)> = Vec::new();
+        for (parent, movies) in movie_counter.iter() {
+            let dateless_parent = parent[..parent.len() - 7].to_string();
+            if let Some(movie) = movies_without_date_counter.remove(&dateless_parent) {
+                let tuple = (movies[0], movie);
+                dupes.push(tuple);
+            }
+        }
+
+        for (_parent, mut movies) in movie_counter.into_iter() {
+            if movies.len() > 1 {
+                // Sort, lowest resolution  movie first
+                movies.sort_by(|a, b| a.max_pixel_count().cmp(&b.max_pixel_count()));
+                // Flip order, we care about the largest.
+                movies.reverse();
+                // Take the largest image, return the rest for removal.
+                let tuple = (movies.remove(0), movies);
+                dupes.push(tuple);
+            }
+        }
+        // Sort to make this function deterministic.
+        dupes.sort_by(|(a_keep, _), (b_keep, _)| {
+            a_keep
+                .files
+                .first()
+                .unwrap()
+                .0
+                .partial_cmp(&b_keep.files.first().unwrap().0)
+                .unwrap()
+        });
+        dupes
+    }
+}
+
 fn movies_from_paths_compact_metadata(mut p_cmd: HashMap<String, CompactMetadata>) -> Movies {
     let multidisc = collapse_multidisc(&p_cmd.keys().map(|s| s.to_string()).collect());
     let movies = multidisc
@@ -336,7 +471,6 @@ impl MovieLibrary {
         let f = File::open(&path).context(format!("open {}", path.display()))?;
         let r = BufReader::new(f);
 
-        // Read the JSON contents of the file as an instance of `User`.
         let mdf: MetadataFile = serde_json::from_reader(r)
             .context(format!("serde_json::from_reader {}", path.display()))?;
 
@@ -391,7 +525,7 @@ impl MovieLibrary {
                                 short_name: codec_name.to_string(),
                                 long_name: codec_long_name.to_string(),
                                 channels: *channels,
-                                channel_layout: channel_layout.to_string(),
+                                channel_layout: channel_layout.clone(),
                                 title: tags.as_ref().and_then(|t| t.title()),
                                 language: tags.as_ref().and_then(|t| t.language()),
                             })
@@ -412,8 +546,8 @@ impl MovieLibrary {
                         } = s
                         {
                             Some(SubtitleFormat {
-                                short_name: codec_name.to_string(),
+                                short_name: codec_name.clone(),
-                                long_name: codec_long_name.to_string(),
+                                long_name: codec_long_name.clone(),
                                 title: tags.as_ref().and_then(|t| t.title()),
                                 language: tags.as_ref().and_then(|t| t.language()),
                             })
@@ -446,7 +580,7 @@ impl MovieLibrary {
     pub fn update_metadata(&self) -> Result<Vec<String>, Error> {
         let path = Path::new(&self.root).join(FULL_METADATA_FILENAME);
 
-        let old_metadata: HashMap<String, Value> = match File::open(&path) {
+        let mut old_metadata: HashMap<String, Value> = match File::open(&path) {
             Ok(f) => {
                 let r = BufReader::new(f);
                 serde_json::from_reader(r)?
@@ -458,6 +592,29 @@ impl MovieLibrary {
         };
 
         info!("Read metadata, {} videos found", old_metadata.len());
+        // Filter out stale metadata (where the file no longer exists).
+        let old_metadata: HashMap<String, Value> = self
+            .iter_video_files()
+            .filter(|r| r.is_ok())
+            .filter_map(|r| {
+                let path = r
+                    .as_ref()
+                    .unwrap()
+                    .strip_prefix(&self.root)
+                    .unwrap()
+                    .to_str()
+                    .unwrap()
+                    .to_owned();
+                match old_metadata.remove(&path) {
+                    Some(v) => Some((path, v)),
+                    None => None,
+                }
+            })
+            .collect();
+        info!(
+            "After removing stale metadata, {} videos found",
+            old_metadata.len()
+        );
 
         let mut metadata: HashMap<_, _> = self
             .iter_video_files()
@@ -484,11 +641,17 @@ impl MovieLibrary {
                         Some((path.to_string_lossy().into_owned(), json))
                     }
                     Err(e) => {
-                        error!("{}", e);
+                        error!("Failed to open {}: {}", path.to_string_lossy(), e);
                         None
                     }
                 }
             })
+            .inspect(|(path, json)| {
+                if let Err(err) = serde_json::from_str::<Metadata>(&json) {
+                    error!("Can't parse metadata for {}: {}", path, err);
+                    error!("{}", json);
+                }
+            })
             .map(|(path, json)| (path, serde_json::from_str::<Value>(&json).unwrap()))
             .collect();
         let new_videos = metadata.keys().cloned().collect();
@@ -527,7 +690,7 @@ impl MovieLibrary {
         Ok(movies_from_paths_compact_metadata(p_cmd))
     }
 
-    pub fn videos(&self) -> Result<(HashMap<String, CompactMetadata>), Error> {
+    pub fn videos(&self) -> Result<HashMap<String, CompactMetadata>, Error> {
         let path = Path::new(&self.root).join(COMPACT_METADATA_FILENAME);
         let f = File::open(&path).context(format!("open {}", path.display()))?;
         let r = BufReader::new(f);
@@ -535,6 +698,28 @@ impl MovieLibrary {
         Ok(serde_json::from_reader(r)
             .context(format!("serde_json::from_reader {}", path.display()))?)
     }
+
+    pub fn empty_dirs(&self) -> Result<Vec<String>, Error> {
+        let mut all_dirs = HashSet::new();
+        let root_len = self.root.len() + 1; // +1 for trailing slash
+        for de in Path::new(&self.root).read_dir()? {
+            let de = de?;
+            if de.metadata()?.is_dir() {
+                let path = de.path().to_string_lossy().to_string();
+                all_dirs.insert(path[root_len..].to_string());
+            }
+        }
+        for path in self.videos()?.keys() {
+            let dir = match path.find("/") {
+                Some(idx) => path[..idx].to_string(),
+                None => path.to_string(),
+            };
+            all_dirs.remove(&dir);
+        }
+        let mut empty_dirs: Vec<_> = all_dirs.into_iter().collect();
+        empty_dirs.sort();
+        Ok(empty_dirs)
+    }
 }
 
 #[cfg(test)]

src/main.rs

@@ -1,12 +1,6 @@
-use std::collections::HashMap;
+use std::{collections::HashMap, error::Error, io::Write, path::Path, time::Duration};
-use std::error::Error;
-use std::io::Write;
-use std::path::Path;
-use std::path::PathBuf;
-use std::time::Duration;
 
-use human_format::Formatter;
+use human_format::{Formatter, Scales};
-use human_format::Scales;
 use humantime;
 use lazy_static::lazy_static;
 use log::info;
@@ -14,25 +8,10 @@ use regex::Regex;
 use structopt::StructOpt;
 use tabwriter::TabWriter;
 
-use superdeduper::is_multidisc;
+use superdeduper::{clean_path_parent, CompactMetadata, MovieLibrary};
-use superdeduper::CompactMetadata;
-use superdeduper::MovieLibrary;
 
 const MOVIE_DIR: &str = "/home/wathiede/Movies";
-const TO_BE_REMOVED_DIR: &str = "/storage/media/to-be-deleted/";
+const TO_BE_REMOVED_DIR: &str = "/home/wathiede/to-be-deleted/";
-
-fn clean_path_parent<P: AsRef<Path>>(path: P) -> PathBuf {
-    let path = path.as_ref();
-    let path = path.parent().unwrap();
-    let path = path.to_str().unwrap();
-    /*
-    // Strip year
-    if path.ends_with(')') {
-        path = &path[..path.len() - 7];
-    }
-    */
-    PathBuf::from(path)
-}
 
 lazy_static! {
     static ref CLEAN_TITLE_CHARS: Regex = Regex::new("[^ 0-9[:alpha:]]").unwrap();
@@ -47,47 +26,18 @@ fn normalize(path: &str) -> String {
 lazy_static! {
     static ref YEAR_SUFFIX: Regex = Regex::new(r" \d{4}$").unwrap();
 }
-fn print_dupes(videos: HashMap<String, CompactMetadata>) {
-    let mut video_groups: HashMap<String, Vec<(String, CompactMetadata)>> = HashMap::new();
-    for (name, md) in videos.into_iter() {
-        let clean_name = normalize(clean_path_parent(&name).to_str().unwrap());
-        let paths = video_groups.entry(clean_name).or_insert(Vec::new());
-        paths.push((name.to_string(), md));
-    }
 
-    let mut names = video_groups.keys().collect::<Vec<_>>();
+fn print_dupes(lib: &MovieLibrary) {
-    names.sort();
+    let videos = lib.movies().expect("couldn't get videos from library");
-
-    for name in &names {
-        if YEAR_SUFFIX.is_match(&name) {
-            let yearless = &name[..&name.len() - 5];
-            info!("is '{}' in map", yearless);
-            if let Some(yearless_vg) = video_groups.get(yearless) {
-                println!("Possible dupe between movie with year and no year:");
-                println!("  {:?}", video_groups.get(name.as_str()).unwrap());
-                println!("  {:?}", yearless_vg);
-            }
-        }
-    }
     let mut fmtr = Formatter::new();
     fmtr.with_separator("");
     fmtr.with_scales(Scales::Binary());
-    for name in names {
+    let mut delete_paths = Vec::new();
-        let possible_dupes = &video_groups[name];
+    for (keep, deletes) in videos.duplicate_candidates() {
-        if possible_dupes.len() < 2 {
+        let p = &keep.files.first().unwrap().0;
-            continue;
+        println!("{}", &p[..p.find("/").unwrap()]);
-        }
+        println!("  Keeping:");
-        let paths: Vec<String> = possible_dupes
+        for (p, md) in &keep.files {
-            .iter()
-            .map(|(name, _)| name.to_string())
-            .collect();
-        if is_multidisc(&paths) {
-            continue;
-        }
-        let mut file: Vec<_> = video_groups[name].iter().collect();
-        file.sort_by(|(n1, _), (n2, _)| n1.partial_cmp(n2).unwrap());
-        println!("{}:", name);
-        for (p, md) in file {
             println!(
                 "    {:>9} {:>9} {} {}",
                 md.largest_dimension().unwrap(),
@@ -96,6 +46,28 @@ fn print_dupes(videos: HashMap<String, CompactMetadata>) {
                 &p[p.rfind("/").unwrap() + 1..]
             );
         }
+        println!("  Need to remove:");
+        for delete in &deletes {
+            for (p, md) in &delete.files {
+                delete_paths.push(p);
+                println!(
+                    "    {:>9} {:>9} {} {}",
+                    md.largest_dimension().unwrap(),
+                    fmtr.format(md.size as f64),
+                    humantime::Duration::from(Duration::from_secs(md.duration as u64)),
+                    &p[p.rfind("/").unwrap() + 1..]
+                );
+            }
+        }
+        println!();
+    }
+    delete_paths.sort();
+    let root = Path::new(&lib.root);
+    for path in &delete_paths {
+        println!(r#"rm "{}""#, root.join(path).to_string_lossy(),);
+    }
+    if delete_paths.len() > 0 {
+        println!("superdeduper update-compact-metadata && superdeduper empty-dirs")
     }
 }
 
@@ -171,7 +143,7 @@ fn print_videos(videos: &HashMap<String, CompactMetadata>, filter: Option<&Regex
             humantime::Duration::from(Duration::from_secs(md.duration as u64)),
             &name[MOVIE_DIR.len() + 1..]
         );
-        println!("mv '{}' '{}'", name, TO_BE_REMOVED_DIR);
+        println!("rm '{}'", name);
     }
 }
 
@@ -197,6 +169,8 @@ enum Command {
         about = "Write full metadata files and update compact file on changes"
     )]
     UpdateAndCompactMetadata,
+    #[structopt(about = "Print directories with no movie files in them")]
+    EmptyDirs,
 }
 
 #[derive(StructOpt)]
@@ -205,6 +179,13 @@ enum Command {
     about = "Tool for pruning extra videos in collection"
 )]
 struct SuperDeduper {
+    #[structopt(
+        short = "r",
+        long = "root",
+        help = "Root directory to store files.",
+        default_value = MOVIE_DIR,
+    )]
+    root: String,
     #[structopt(
         short = "v",
         help = "Sets the level of verbosity",
@@ -232,14 +213,14 @@ fn main() -> Result<(), Box<dyn Error>> {
 
     match app.cmd {
         Command::Samples => {
-            let lib = MovieLibrary::new(MOVIE_DIR);
+            let lib = MovieLibrary::new(app.root);
             let videos = lib.videos()?;
 
             let samples_re = Regex::new(r"(?i).*sample.*").unwrap();
             print_videos(&videos, Some(&samples_re));
         }
         Command::Groups => {
-            let lib = MovieLibrary::new(MOVIE_DIR);
+            let lib = MovieLibrary::new(app.root);
             let videos = lib.videos()?;
 
             let mut video_groups: HashMap<String, Vec<(String, CompactMetadata)>> = HashMap::new();
@@ -252,35 +233,38 @@ fn main() -> Result<(), Box<dyn Error>> {
             print_video_groups(&video_groups);
         }
         Command::CompactMetadata => {
-            let lib = MovieLibrary::new(MOVIE_DIR);
+            let lib = MovieLibrary::new(app.root);
             lib.compact_metadata()?;
         }
         Command::PrintDupes => {
-            let lib = MovieLibrary::new(MOVIE_DIR);
+            let lib = MovieLibrary::new(app.root);
-            let videos = lib.videos()?;
+            print_dupes(&lib);
-
-            print_dupes(videos);
         }
         Command::PrintAll => {
-            let lib = MovieLibrary::new(MOVIE_DIR);
+            let lib = MovieLibrary::new(app.root);
             let videos = lib.videos()?;
 
             print_all(videos);
         }
         Command::UpdateMetadata => {
-            let lib = MovieLibrary::new(MOVIE_DIR);
+            let lib = MovieLibrary::new(app.root);
             lib.update_metadata()?;
         }
         Command::UpdateAndCompactMetadata => {
-            let lib = MovieLibrary::new(MOVIE_DIR);
+            let lib = MovieLibrary::new(app.root);
-            let new_videos = lib.update_metadata()?;
+            lib.update_metadata()?;
-            if !new_videos.is_empty() {
-                info!(
-                    "{} new videos added, recompacting metadata",
-                    new_videos.len()
-                );
             lib.compact_metadata()?;
         }
+        Command::EmptyDirs => {
+            let lib = MovieLibrary::new(app.root);
+            let dirs = lib.empty_dirs()?;
+            let root = Path::new(&lib.root);
+            if !dirs.is_empty() {
+                println!("Empty directories:");
+                for d in dirs {
+                    println!(r#"rm -rf "{}""#, root.join(d).to_string_lossy());
+                }
+            }
         }
     }
     Ok(())

src/movielibrary_test.rs

@@ -37,7 +37,7 @@ fn test_simple_library() {
                             short_name: "mp3".to_string(),
                             long_name: "MP3 (MPEG audio layer 3)".to_string(),
                             channels: 2,
-                            channel_layout: "stereo".to_string(),
+                            channel_layout: Some("stereo".to_string()),
                             title: None,
                             language: None,
                         }],
@@ -49,7 +49,7 @@ fn test_simple_library() {
     );
 }
 
-fn build_tuple<R>(path: &str, res: R) -> (String, CompactMetadata)
+fn build_tuple<R>(path: &str, res: R, bit_rate: usize) -> (String, CompactMetadata)
 where
     R: Into<Resolution>,
 {
@@ -58,7 +58,7 @@ where
         path.to_string(),
         CompactMetadata {
             filename: format!("./{}", path),
-            bit_rate: 1,
+            bit_rate,
             duration: 1.0,
             format_name: "test_format".to_string(),
             size: 1,
@@ -72,43 +72,71 @@ where
     )
 }
 
-fn build_movie<R>(paths: Vec<(&str, R)>) -> Movie
+fn build_movie<R>(paths: Vec<(&str, R, usize)>) -> Movie
 where
     R: Into<Resolution>,
 {
     Movie {
         files: paths
             .into_iter()
-            .map(|(path, res)| build_tuple(path, res))
+            .map(|(path, res, bit_rate)| build_tuple(path, res, bit_rate))
             .collect(),
     }
 }
 
 fn build_complex_metadata() -> HashMap<String, CompactMetadata> {
     vec![
+        build_tuple("Remake (2019)/new.mkv", (1920, 1080), 10),
+        build_tuple("Remake (1960)/old.mkv", (1920, 1080), 1),
         build_tuple(
             "One Movie With Year (2019)/abcdef123456789.mkv",
             (1920, 1080),
+            1,
         ),
         build_tuple(
             "One Movie With Two Parts (2019)/abcdef123456789 part 1.mkv",
             (1280, 720),
+            1,
         ),
         build_tuple(
             "One Movie With Two Parts (2019)/abcdef123456789 part 2.mkv",
             (1280, 720),
+            1,
         ),
         build_tuple(
             "Two Movies With Multi Parts (2019)/abcdef123456789 part 1.mkv",
             (1280, 720),
+            1000,
         ),
         build_tuple(
             "Two Movies With Multi Parts (2019)/abcdef123456789 part 2.mkv",
             (1280, 720),
+            1000,
         ),
         build_tuple(
             "Two Movies With Multi Parts (2019)/somethingelse.mkv",
             (1920, 1080),
+            5000,
+        ),
+        build_tuple(
+            "two movies different caps (2019)/larger.mkv",
+            (1920, 1080),
+            100,
+        ),
+        build_tuple(
+            "Two Movies Different Caps (2019)/smaller.mkv",
+            (640, 480),
+            1,
+        ),
+        build_tuple(
+            "Movie with and without year (2019)/with year.mkv",
+            (1920, 1080),
+            10,
+        ),
+        build_tuple(
+            "Movie with and without year/without year.mkv",
+            (1280, 720),
+            1,
         ),
     ]
     .into_iter()
@@ -118,33 +146,61 @@ fn build_complex_metadata() -> HashMap<String, CompactMetadata> {
 fn build_complex_movies() -> Movies {
     let mut m = Movies {
         movies: vec![
+            build_movie(vec![("Remake (2019)/new.mkv", (1920, 1080), 10)]),
+            build_movie(vec![("Remake (1960)/old.mkv", (1920, 1080), 1)]),
             build_movie(vec![(
                 "One Movie With Year (2019)/abcdef123456789.mkv",
                 (1920, 1080),
+                1,
             )]),
             build_movie(vec![
                 (
                     "One Movie With Two Parts (2019)/abcdef123456789 part 1.mkv",
                     (1280, 720),
+                    1,
                 ),
                 (
                     "One Movie With Two Parts (2019)/abcdef123456789 part 2.mkv",
                     (1280, 720),
+                    1,
                 ),
             ]),
             build_movie(vec![
                 (
                     "Two Movies With Multi Parts (2019)/abcdef123456789 part 1.mkv",
                     (1280, 720),
+                    1000,
                 ),
                 (
                     "Two Movies With Multi Parts (2019)/abcdef123456789 part 2.mkv",
                     (1280, 720),
+                    1000,
                 ),
             ]),
             build_movie(vec![(
                 "Two Movies With Multi Parts (2019)/somethingelse.mkv",
                 (1920, 1080),
+                5000,
+            )]),
+            build_movie(vec![(
+                "two movies different caps (2019)/larger.mkv",
+                (1920, 1080),
+                100,
+            )]),
+            build_movie(vec![(
+                "Two Movies Different Caps (2019)/smaller.mkv",
+                (640, 480),
+                1,
+            )]),
+            build_movie(vec![(
+                "Movie with and without year (2019)/with year.mkv",
+                (1920, 1080),
+                10,
+            )]),
+            build_movie(vec![(
+                "Movie with and without year/without year.mkv",
+                (1280, 720),
+                1,
             )]),
         ],
     };
@@ -173,6 +229,150 @@ fn test_roundtrip_library() -> Result<(), Box<dyn Error>> {
     let got = ml.movies().expect("failed to build movies");
     assert_eq!(got.movies.len(), want.movies.len());
     assert_eq!(got, want);
-    //assert_eq!(got, want, "Got {:#?}\nWant {:#?}", got, want);
+    Ok(())
+}
+
+fn validate_duplicates(got: Vec<(&Movie, Vec<&Movie>)>, want: Vec<(Movie, Vec<Movie>)>) {
+    let want: Vec<(&Movie, Vec<&Movie>)> = want
+        .iter()
+        .map(|(keep, delete)| (keep, delete.iter().map(|m| m).collect()))
+        .collect();
+    assert_eq!(got, want);
+}
+
+#[test]
+fn test_duplicate_candidates() -> Result<(), Box<dyn Error>> {
+    let movies = build_complex_movies();
+    let got = movies.duplicate_candidates();
+    let want = vec![
+        (
+            build_movie(vec![(
+                "Movie with and without year (2019)/with year.mkv",
+                (1920, 1080),
+                10,
+            )]),
+            vec![build_movie(vec![(
+                "Movie with and without year/without year.mkv",
+                (1280, 720),
+                1,
+            )])],
+        ),
+        (
+            build_movie(vec![(
+                "Two Movies With Multi Parts (2019)/somethingelse.mkv",
+                (1920, 1080),
+                5000,
+            )]),
+            vec![build_movie(vec![
+                (
+                    "Two Movies With Multi Parts (2019)/abcdef123456789 part 1.mkv",
+                    (1280, 720),
+                    1000,
+                ),
+                (
+                    "Two Movies With Multi Parts (2019)/abcdef123456789 part 2.mkv",
+                    (1280, 720),
+                    1000,
+                ),
+            ])],
+        ),
+        (
+            build_movie(vec![(
+                "two movies different caps (2019)/larger.mkv",
+                (1920, 1080),
+                100,
+            )]),
+            vec![build_movie(vec![(
+                "Two Movies Different Caps (2019)/smaller.mkv",
+                (640, 480),
+                1,
+            )])],
+        ),
+    ];
+    validate_duplicates(got, want);
+    Ok(())
+}
+
+#[test]
+fn test_fullmetal() -> Result<(), Box<dyn Error>> {
+    let mut movies = Movies {
+        movies: vec![
+            build_movie(vec![(
+                "Full Metal Jacket (1987)/Full Metal Jacket.mp4",
+                (1280, 720),
+                2581935,
+            )]),
+            build_movie(vec![(
+                "Full Metal Jacket (1987)/1776f8e2fb614a6fb77a66cde601bb45.mkv",
+                (1920, 1080),
+                5719802,
+            )]),
+        ],
+    };
+    movies.movies.sort_by(|a, b| {
+        a.files
+            .first()
+            .unwrap()
+            .0
+            .partial_cmp(&b.files.first().unwrap().0)
+            .unwrap()
+    });
+    let got = movies.duplicate_candidates();
+    let want = vec![(
+        build_movie(vec![(
+            "Full Metal Jacket (1987)/1776f8e2fb614a6fb77a66cde601bb45.mkv",
+            (1920, 1080),
+            5719802,
+        )]),
+        vec![build_movie(vec![(
+            "Full Metal Jacket (1987)/Full Metal Jacket.mp4",
+            (1280, 720),
+            2581935,
+        )])],
+    )];
+    validate_duplicates(got, want);
+    Ok(())
+}
+
+#[test]
+fn test_keep_higher_res_lower_bit_rate() -> Result<(), Box<dyn Error>> {
+    let mut movies = Movies {
+        movies: vec![
+            build_movie(vec![(
+                "X Men The Last Stand (2006)/X.Men.The.Last.Stand.2006.1080p.BluRay.x264.DTS-ES.PRoDJi.mkv",
+                (1920, 800),
+                11349705,
+            )]),
+            build_movie(vec![(
+                "X Men The Last Stand (2006)/948f08a4ba784626ac13de77b77559dd.mkv",
+                (1920, 1080),
+                6574160,
+            )]),
+        ],
+    };
+    movies.movies.sort_by(|a, b| {
+        a.files
+            .first()
+            .unwrap()
+            .0
+            .partial_cmp(&b.files.first().unwrap().0)
+            .unwrap()
+    });
+    let got = movies.duplicate_candidates();
+    let want = vec![(
+        build_movie(vec![(
+                "X Men The Last Stand (2006)/948f08a4ba784626ac13de77b77559dd.mkv",
+                (1920, 1080),
+                6574160,
+        )]),
+        vec![
+        build_movie(vec![(
+                "X Men The Last Stand (2006)/X.Men.The.Last.Stand.2006.1080p.BluRay.x264.DTS-ES.PRoDJi.mkv",
+                (1920, 800),
+                11349705,
+        )])
+        ],
+    )];
+    validate_duplicates(got, want);
     Ok(())
 }