superdeduper/src/main.rs

use std::collections::HashMap;
use std::error::Error;
use std::io::Write;
use std::time::Duration;

use human_format::Formatter;
use human_format::Scales;
use humantime;
use lazy_static::lazy_static;
use log::info;
use regex::Regex;
use structopt::StructOpt;
use tabwriter::TabWriter;

use superdeduper::clean_path_parent;
use superdeduper::CompactMetadata;
use superdeduper::MovieLibrary;

const MOVIE_DIR: &str = "/home/wathiede/Movies";
const TO_BE_REMOVED_DIR: &str = "/storage/media/to-be-deleted/";

lazy_static! {
    static ref CLEAN_TITLE_CHARS: Regex = Regex::new("[^ 0-9[:alpha:]]").unwrap();
}

fn normalize(path: &str) -> String {
    CLEAN_TITLE_CHARS
        .replace_all(&path, "")
        .to_ascii_lowercase()
}

lazy_static! {
    static ref YEAR_SUFFIX: Regex = Regex::new(r" \d{4}$").unwrap();
}

fn print_dupes(lib: &MovieLibrary) {
    let videos = lib.movies().expect("couldn't get videos from library");
    let mut fmtr = Formatter::new();
    fmtr.with_separator("");
    fmtr.with_scales(Scales::Binary());
    for (keep, deletes) in videos.duplicate_candidates() {
        let p = &keep.files.first().unwrap().0;
        println!("{}", &p[..p.find("/").unwrap()]);
        println!("  Keeping:");
        for (p, md) in &keep.files {
            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!("  Need to remove:");
        for delete in &deletes {
            for (p, md) in &delete.files {
                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!();
    }
}

fn print_all(videos: HashMap<String, CompactMetadata>) {
    let mut names = videos.keys().collect::<Vec<_>>();
    names.sort();

    let mut fmtr = Formatter::new();
    fmtr.with_separator("");
    fmtr.with_scales(Scales::Binary());
    let mut tw = TabWriter::new(vec![]);
    for name in names {
        let md = &videos[name];
        write!(
            &mut tw,
            "{}B/s\t{}\t{}\t{}\t{}\n",
            fmtr.format(md.bit_rate as f64),
            md.largest_dimension().unwrap(),
            fmtr.format(md.size as f64),
            humantime::Duration::from(Duration::from_secs(md.duration as u64)),
            name,
        )
        .unwrap();
        //&p[p.rfind("/").unwrap() + 1..]
    }
    println!("{}", String::from_utf8(tw.into_inner().unwrap()).unwrap());
}

fn print_video_groups(video_groups: &HashMap<String, Vec<(String, CompactMetadata)>>) {
    let mut names = video_groups.keys().collect::<Vec<_>>();
    names.sort();

    let mut fmtr = Formatter::new();
    fmtr.with_separator("");
    fmtr.with_scales(Scales::Binary());
    for name in names {
        let paths = &video_groups[name];
        if paths.len() < 2 {
            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(),
                fmtr.format(md.size as f64),
                humantime::Duration::from(Duration::from_secs(md.duration as u64)),
                &p[p.rfind("/").unwrap() + 1..]
            );
        }
    }
}

fn print_videos(videos: &HashMap<String, CompactMetadata>, filter: Option<&Regex>) {
    let mut names = videos.keys().collect::<Vec<_>>();
    names.sort();
    let mut fmtr = Formatter::new();
    fmtr.with_separator("");
    fmtr.with_scales(Scales::Binary());
    for name in names {
        if let Some(re) = filter {
            if !re.is_match(name) {
                continue;
            }
        }
        let md = &videos[name];
        info!(
            "{:>9} {:>8} {} {}",
            md.largest_dimension().unwrap(),
            fmtr.format(md.size as f64),
            humantime::Duration::from(Duration::from_secs(md.duration as u64)),
            &name[MOVIE_DIR.len() + 1..]
        );
        println!("mv '{}' '{}'", name, TO_BE_REMOVED_DIR);
    }
}

#[derive(StructOpt)]
enum Command {
    #[structopt(name = "samples", about = "Print video files deemed to be samples")]
    Samples,
    #[structopt(name = "groups", about = "Print videos grouped by root name")]
    Groups,
    #[structopt(
        name = "compact-metadata",
        about = "Read full metadata file and write compact file."
    )]
    CompactMetadata,
    #[structopt(name = "print-all", about = "Print useful metadata about all files")]
    PrintAll,
    #[structopt(name = "print-dupes", about = "Print duplicate movies")]
    PrintDupes,
    #[structopt(name = "update-metadata", about = "Write full metadata files")]
    UpdateMetadata,
    #[structopt(
        name = "update-compact-metadata",
        about = "Write full metadata files and update compact file on changes"
    )]
    UpdateAndCompactMetadata,
}

#[derive(StructOpt)]
#[structopt(
    name = "superdeduper",
    about = "Tool for pruning extra videos in collection"
)]
struct SuperDeduper {
    #[structopt(
        short = "v",
        help = "Sets the level of verbosity",
        parse(from_occurrences)
    )]
    verbose: usize,
    #[structopt(long = "module", help = "Additional log target to enable")]
    module: Option<String>,
    #[structopt(subcommand)] // Note that we mark a field as a subcommand
    cmd: Command,
}

fn main() -> Result<(), Box<dyn Error>> {
    let app = SuperDeduper::from_args();
    let mut modules = vec![module_path!().to_string()];
    if let Some(module) = app.module {
        modules.push(module);
    }
    stderrlog::new()
        .verbosity(app.verbose)
        .timestamp(stderrlog::Timestamp::Millisecond)
        .modules(modules)
        .init()
        .unwrap();

    match app.cmd {
        Command::Samples => {
            let lib = MovieLibrary::new(MOVIE_DIR);
            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 videos = lib.videos()?;

            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));
            }

            print_video_groups(&video_groups);
        }
        Command::CompactMetadata => {
            let lib = MovieLibrary::new(MOVIE_DIR);
            lib.compact_metadata()?;
        }
        Command::PrintDupes => {
            let lib = MovieLibrary::new(MOVIE_DIR);
            print_dupes(&lib);
        }
        Command::PrintAll => {
            let lib = MovieLibrary::new(MOVIE_DIR);
            let videos = lib.videos()?;

            print_all(videos);
        }
        Command::UpdateMetadata => {
            let lib = MovieLibrary::new(MOVIE_DIR);
            lib.update_metadata()?;
        }
        Command::UpdateAndCompactMetadata => {
            let lib = MovieLibrary::new(MOVIE_DIR);
            let new_videos = lib.update_metadata()?;
            if !new_videos.is_empty() {
                info!(
                    "{} new videos added, recompacting metadata",
                    new_videos.len()
                );
                lib.compact_metadata()?;
            }
        }
    }
    Ok(())
}