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

@@ -1,32 +1,24 @@
-use std::cmp::Ordering;
+use std::{
-use std::collections::HashMap;
+    cmp::Ordering,
-use std::env;
+    collections::{HashMap, HashSet},
-use std::ffi::OsStr;
+    env,
-use std::fmt;
+    ffi::OsStr,
-use std::fmt::Display;
+    fmt,
-use std::fmt::Formatter;
+    fmt::{Display, Formatter},
-use std::fs::File;
+    fs::File,
-use std::io::BufReader;
+    io::{BufReader, BufWriter},
-use std::io::BufWriter;
+    path::{Path, PathBuf},
-use std::path::Path;
+    process::Command,
-use std::path::PathBuf;
+    str::FromStr,
-use std::process::Command;
+};
-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";
@@ -36,12 +28,6 @@ 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();
-    /*
-    // Strip year
-    if path.ends_with(')') {
-        path = &path[..path.len() - 7];
-    }
-    */
     PathBuf::from(path)
 }
 
@@ -166,8 +152,8 @@ enum Stream {
     },
     #[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")]
@@ -223,8 +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")]
@@ -285,7 +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> {
@@ -302,7 +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)?);
     }
@@ -325,6 +311,30 @@ pub struct Movie {
 }
 
 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
@@ -334,17 +344,6 @@ impl Movie {
             }))
         }
     }
-
-    fn min_resolution(&self) -> Option<Resolution> {
-        if self.files.is_empty() {
-            None
-        } else {
-            Some(self.files.iter().fold(
-                Resolution(usize::max_value(), usize::max_value()),
-                |acc, (_, cmd)| std::cmp::min(acc, cmd.largest_dimension().unwrap()),
-            ))
-        }
-    }
 }
 
 impl Display for Movie {
@@ -370,23 +369,51 @@ pub struct Movies {
 
 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, TODO(wathiede): with
+    /// One or more Movies to remove).  The highest resolution movie is kept.
-    /// higher bit rate breaking ties.
+    /// 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 = clean_path_parent(path)
+            let parent = MULTIPLE_SPACES
-                .to_string_lossy()
+                .replace_all(
-                .to_ascii_lowercase()
+                    &clean_path_parent(path)
+                        .to_string_lossy()
+                        .to_ascii_lowercase()
+                        .replace("-", " ")
+                        .replace("'", " "),
+                    " ",
+                )
                 .to_string();
-            movie_counter.entry(parent).or_insert(Vec::new()).push(m);
+            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::new();
+        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 bit_rate movie first
+                // Sort, lowest resolution  movie first
-                movies.sort_by(|a, b| a.min_bit_rate().cmp(&b.min_bit_rate()));
+                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.
@@ -394,14 +421,16 @@ impl Movies {
                 dupes.push(tuple);
             }
         }
-        for d in &dupes {
+        // Sort to make this function deterministic.
-            let (biggest, deletes) = d;
+        dupes.sort_by(|(a_keep, _), (b_keep, _)| {
-            eprintln!("biggest: {}", biggest);
+            a_keep
-            for (i, delete) in deletes.iter().enumerate() {
+                .files
-                eprintln!("{}. delete: {}", i + 1, delete);
+                .first()
-            }
+                .unwrap()
-        }
+                .0
-
+                .partial_cmp(&b_keep.files.first().unwrap().0)
+                .unwrap()
+        });
         dupes
     }
 }
@@ -517,8 +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()),
                             })
@@ -612,11 +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();
@@ -655,7 +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);
@@ -663,6 +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

@@ -1,10 +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::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;
@@ -12,12 +8,10 @@ use regex::Regex;
 use structopt::StructOpt;
 use tabwriter::TabWriter;
 
-use superdeduper::clean_path_parent;
+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/";
 
 lazy_static! {
     static ref CLEAN_TITLE_CHARS: Regex = Regex::new("[^ 0-9[:alpha:]]").unwrap();
@@ -68,8 +62,12 @@ fn print_dupes(lib: &MovieLibrary) {
         println!();
     }
     delete_paths.sort();
+    let root = Path::new(&lib.root);
     for path in &delete_paths {
-        println!(r#"mv "{}" /storage/media/to-be-deleted/"#, path);
+        println!(r#"rm "{}""#, root.join(path).to_string_lossy(),);
+    }
+    if delete_paths.len() > 0 {
+        println!("superdeduper update-compact-metadata && superdeduper empty-dirs")
     }
 }
 
@@ -145,7 +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);
     }
 }
 
@@ -171,6 +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)]
@@ -179,6 +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",
@@ -206,14 +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();
@@ -226,28 +233,39 @@ 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);
             print_dupes(&lib);
         }
         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);
             lib.update_metadata()?;
             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

@@ -86,6 +86,8 @@ where
 
 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),
@@ -117,15 +119,25 @@ fn build_complex_metadata() -> HashMap<String, CompactMetadata> {
             5000,
         ),
         build_tuple(
-            "two movies different caps (2019)/abcdef123456789.mkv",
+            "two movies different caps (2019)/larger.mkv",
             (1920, 1080),
             100,
         ),
         build_tuple(
-            "Two Movies Different Caps (2019)/abcdef123456789.mkv",
+            "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()
     .collect()
@@ -134,6 +146,8 @@ 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),
@@ -169,15 +183,25 @@ fn build_complex_movies() -> Movies {
                 5000,
             )]),
             build_movie(vec![(
-                "two movies different caps (2019)/abcdef123456789.mkv",
+                "two movies different caps (2019)/larger.mkv",
                 (1920, 1080),
                 100,
             )]),
             build_movie(vec![(
-                "Two Movies Different Caps (2019)/abcdef123456789.mkv",
+                "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,
+            )]),
         ],
     };
     m.movies.sort_by(|a, b| {
@@ -209,11 +233,11 @@ fn test_roundtrip_library() -> Result<(), Box<dyn Error>> {
 }
 
 fn validate_duplicates(got: Vec<(&Movie, Vec<&Movie>)>, want: Vec<(Movie, Vec<Movie>)>) {
-    assert_eq!(got.len(), want.len());
+    let want: Vec<(&Movie, Vec<&Movie>)> = want
-    for (g, w) in got.iter().zip(&want) {
+        .iter()
-        assert_eq!(g.0, &w.0);
+        .map(|(keep, delete)| (keep, delete.iter().map(|m| m).collect()))
-        assert_eq!(g.1, w.1.iter().map(|v| v).collect::<Vec<_>>());
+        .collect();
-    }
+    assert_eq!(got, want);
 }
 
 #[test]
@@ -221,6 +245,18 @@ 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",
@@ -242,12 +278,12 @@ fn test_duplicate_candidates() -> Result<(), Box<dyn Error>> {
         ),
         (
             build_movie(vec![(
-                "two movies different caps (2019)/abcdef123456789.mkv",
+                "two movies different caps (2019)/larger.mkv",
                 (1920, 1080),
                 100,
             )]),
             vec![build_movie(vec![(
-                "Two Movies Different Caps (2019)/abcdef123456789.mkv",
+                "Two Movies Different Caps (2019)/smaller.mkv",
                 (640, 480),
                 1,
             )])],
@@ -299,7 +335,7 @@ fn test_fullmetal() -> Result<(), Box<dyn Error>> {
 }
 
 #[test]
-fn test_keep_lower_res_higher_bit_rate() -> Result<(), Box<dyn Error>> {
+fn test_keep_higher_res_lower_bit_rate() -> Result<(), Box<dyn Error>> {
     let mut movies = Movies {
         movies: vec![
             build_movie(vec![(
@@ -325,15 +361,17 @@ fn test_keep_lower_res_higher_bit_rate() -> Result<(), Box<dyn Error>> {
     let got = movies.duplicate_candidates();
     let want = 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,
-        )]),
-        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(())