srtune/src/main.rs

#[macro_use]
extern crate log;
#[macro_use]
extern crate lazy_static;

use regex::Regex;
use std::fs::{File, OpenOptions};
use std::io::{BufReader, Write};
use std::io;
use std::str::FromStr;
use std::convert::TryFrom;
use std::io::BufRead;
use std::ops::{Add, Mul, MulAssign, AddAssign, SubAssign, Sub};
use std::fmt::{self, Display};
use serde::export::fmt::Debug;

const LOG_LEVELS: [&str; 5] = ["error", "warn", "info", "debug", "trace"];
const SPAMMY_LIBS: [&str; 5] = ["tokio_reactor", "hyper", "reqwest", "mio", "want"];

fn main() {
    let argv =
        clap::App::new("srtune")
            .version(env!("CARGO_PKG_VERSION"))
            .arg(clap::Arg::with_name("input")
                     .value_name("INFILE")
                     .help("Input file, leave out for stdin"),
            )
            .arg(clap::Arg::with_name("output")
                     .short("o")
                     .long("output")
                     .value_name("OUTFILE")
                     .help("Output file, defaults to stdout"),
            )
            .arg(clap::Arg::with_name("fromtime")
                     .short("f")
                     .long("from-time")
                     .value_name("TIME")
                     .help("Time of the first affected entry, defaults to 0. Use seconds, M:S or \
                     H:M:S, decimals are supported. Uses the original times before any shifts or \
                     scaling."),
            )
            .arg(clap::Arg::with_name("fromindex")
                     .short("F")
                     .long("from-index")
                     .value_name("INDEX")
                     .help("Time of the first affected entry, defaults to 0. Use seconds, M:S or \
                     H:M:S, decimals are supported. Uses the original times before any shifts or \
                     scaling."),
            )
            .arg(clap::Arg::with_name("move")
                     .short("m")
                     .long("move")
                     .value_name("TIME")
                     .help("Move subtitles in time. Use seconds, M:S or H:M:S, decimals and minus \
                     are supported."),
            )
            .arg(clap::Arg::with_name("scale")
                     .short("s")
                     .long("scale")
                     .value_name("RATIO")
                     .help("Scale subtitle times and durations to compensate for bitrate \
                     differences. 1 means identity, 1.1 makes all times 10% longer. If a start \
                     time was given, the scaling is relative to this point, otherwise to the first \
                     subtitle in the file. Has no effect if '--autoscale' is used."),
            )
            .arg(clap::Arg::with_name("autoscale")
                     .short("S")
                     .long("autoscale")
                     .value_name("SUBTIME=VIDEOTIME")
                     .help("Calculate scaling based on a perceived difference. The scaling is \
                     related to the first subtitle, so ensure it is aligned properly with '--move'."),
            )
            .arg(clap::Arg::with_name("durscale")
                     .short("d")
                     .long("durscale")
                     .value_name("RATIO")
                     .help("Scale durations, can be combined with '--scale' or '--autoscale'. The \
                     given value will always be multiplied by the absolute time scale. 1 means \
                     identity, 1.1 makes all times 10% longer."),
            )
            .arg(clap::Arg::with_name("renumber")
                     .short("r")
                     .long("renumber")
                     .help("Change all numbers to be sequential starting with 1"),
            )
            .arg(clap::Arg::with_name("v").short("v").multiple(true).help(
                "Sets the level of verbosity (adds to the default - info)",
            ))
            .get_matches();

    let mut log_level = "info".to_owned();

    if argv.is_present("v") {
        // bump verbosity if -v's are present
        let pos = LOG_LEVELS
            .iter()
            .position(|x| x == &log_level)
            .unwrap();

        log_level = match LOG_LEVELS
            .iter()
            .nth(pos + argv.occurrences_of("v") as usize)
            {
                Some(new_level) => new_level.to_string(),
                None => "trace".to_owned(),
            };
    }

    //println!("LEVEL={}", log_level);

    // init logging
    let env = env_logger::Env::default().default_filter_or(log_level);
    let mut builder = env_logger::Builder::from_env(env);
    let lib_level = log::LevelFilter::Info;
    for lib in &SPAMMY_LIBS {
        builder.filter_module(lib, lib_level);
    }
    builder.init();

    let from_time = match argv.value_of("fromtime") {
        Some(s) => {
            SubDuration::try_from(s).expect("Bad --from-time format").as_instant()
        }
        None => SubInstant(0f32)
    };

    let from_index = match argv.value_of("fromindex") {
        Some(s) => {
            s.parse().expect("Bad --from-index format")
        }
        None => 0u32
    };

    let shift = match argv.value_of("move") {
        Some(s) => {
            SubDuration::try_from(s).expect("Bad --move format")
        }
        None => SubDuration(0f32)
    };

    let scale = match argv.value_of("scale") {
        Some(s) => {
            s.parse().expect("Bad --scale format")
        }
        None => 1f32
    };

    let durscale = match argv.value_of("durscale") {
        Some(s) => {
            s.parse().expect("Bad --durscale format")
        }
        None => 1f32
    }; // always also shrink durations

    let autoscale = match argv.value_of("autoscale") {
        Some(s) => {
            let halves : Vec<&str> = s.split("=").collect();
            if halves.len() != 2 {
                panic!("Bad --autoscale format, should be SUBTIME=VIDEOTIME")
            }
            let (first, second) = (halves[0], halves[1]);
            let subtime = SubDuration::try_from(first).expect("Bad --autoscale format").as_instant();
            let vidtime = SubDuration::try_from(second).expect("Bad --autoscale format").as_instant();

            Some((subtime, vidtime))
        }
        None => None
    };

    let inf = argv.value_of("input");
    let outf = argv.value_of("output");
    let stdin = io::stdin();
    let stdout = io::stdout();

    let mut lines_iterator: Box<dyn Iterator<Item=Result<String, io::Error>>> = match inf {
        None => {
            Box::new(stdin.lock().lines())
        }
        Some(f) => {
            let file = File::open(f).expect(&format!("Could not open file: {:?}", f));
            Box::new(BufReader::new(file).lines())
        }
    };

    let mut outfile: Box<dyn Write> = match outf {
        None => {
            Box::new(stdout.lock())
        }
        Some(f) => {
            let file = OpenOptions::new()
                .create(true)
                .truncate(true)
                .write(true)
                .open(f)
                .expect(&format!("Could not open file: {:?}", f));
            Box::new(file)
        }
    };

    let renumber = argv.is_present("renumber");

    transform_subtitles(&mut lines_iterator, &mut outfile, TransformOpts {
        renumber,
        autoscale,
        durscale,
        scale,
        shift,
        from_index,
        from_time,
    });
}

#[derive(Debug)]
struct TransformOpts {
    renumber: bool,
    autoscale: Option<(SubInstant, SubInstant)>,
    durscale: f32,
    scale: f32,
    shift: SubDuration,
    from_index: u32,
    from_time: SubInstant,
}

fn transform_subtitles<'a>(lines : &mut Box<dyn Iterator<Item=Result<String, io::Error>> + 'a>,
                       outfile : &mut Box<dyn Write + 'a>,
                       mut opts : TransformOpts) {
    debug!("Opts: {:#?}", opts);

    let mut start_time = SubInstant(0f32);
    let mut first_found = false;
    let mut renumber_i: u32 = 0;

    let mut linebuf = vec![];
    while let Some(Ok(x)) = lines.next() {
        let mut x = x.trim();
        if x.starts_with('\u{feff}') {
            debug!("Stripping BOM mark");
            x = &x[3..];
        }
        let x = x.trim();
        if x.is_empty() {
            continue;
        }

        // 236
        // 00:18:01,755 --> 00:18:03,774
        // （掃除機の音）
        // う～ん…。
        match u32::from_str(x) {
            Ok(num) => {
                // println!("Entry {}", num);
                let datesrow = lines.next().unwrap().unwrap();
                if datesrow.contains(" --> ") {
                    let mut halves = datesrow.split(" --> ");
                    let (first, second) = (halves.next().unwrap(), halves.next().unwrap());
                    let start = SubInstant::try_from(first).unwrap();
                    let end = SubInstant::try_from(second).unwrap();

                    linebuf.clear();
                    while let Some(Ok(x)) = lines.next() {
                        if x.is_empty() {
                            break; // space between the entries
                        }
                        linebuf.push(x);
                    }

                    let mut subtitle = Subtitle {
                        num,
                        start,
                        dur: SubDuration(end.0 - start.0),
                        text: linebuf.join("\n"),
                    };

                    if start >= opts.from_time && num >= opts.from_index {
                        if !first_found {
                            debug!("Scaling anchored at {} (#{}), editing starts", start, num);
                            debug!("Shifting by: {}", opts.shift);

                            start_time = start;
                            first_found = true;

                            if let Some((mut subt, mut vidt)) = opts.autoscale {
                                debug!("Autoscale: VT {} -> ST {}", vidt, subt);
                                subt -= start_time;
                                vidt -= start_time + opts.shift;
                                if subt.0 <= 0f32 {
                                    panic!("Error in autoscale, start time is negative or zero.");
                                }
                                if vidt.0 <= 0f32 {
                                    panic!("Error in autoscale, end time is negative or zero.");
                                }
                                debug!("  relative to #{}, after \"move\": VT {} -> ST {}", num, vidt, subt);
                                opts.scale = vidt.0 / subt.0;
                                debug!("Resolved scale as {}", opts.scale);
                            }

                            opts.durscale *= opts.scale;
                            debug!("Duration scaling is {}", opts.durscale);
                        }

                        if opts.scale != 1f32 {
                            subtitle.start = subtitle.start.scale(start_time, opts.scale);
                        }

                        subtitle.dur *= opts.durscale;
                        subtitle.start += opts.shift;
                    }

                    if subtitle.start.0 < 0f32 {
                        warn!("Discarding negative time entry #{} @ {:.3}s", subtitle.num, subtitle.start.0);
                        continue;
                    }

                    // advance numbering only for the really emitted entries
                    if opts.renumber {
                        renumber_i += 1;
                        subtitle.num = renumber_i;
                    }

                    outfile.write(subtitle.to_string().as_bytes()).expect("failed to write");
                }
            }
            Err(e) => {
                error!("couldnt parse >{}<: {}", x, e);
                for b in x.as_bytes() {
                    error!("{:#02x} - {}", b, b);
                }
                error!("\n");
            }
        }
    }

    outfile.flush().unwrap();
}

#[derive(Copy, Clone, PartialEq, PartialOrd)]
struct SubInstant(f32);

impl Debug for SubInstant {
    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
        write!(f, "Time({})", self)
    }
}

#[derive(Copy, Clone, PartialEq, PartialOrd)]
struct SubDuration(f32);

impl Debug for SubDuration {
    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
        write!(f, "Duration({})", self)
    }
}

impl Display for SubDuration {
    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
        write!(f, "{}", SubInstant(self.0))
    }
}

impl Add<SubDuration> for SubInstant {
    type Output = SubInstant;

    fn add(self, rhs: SubDuration) -> Self::Output {
        SubInstant(self.0 + rhs.0)
    }
}

impl Sub<SubDuration> for SubInstant {
    type Output = SubInstant;

    fn sub(self, rhs: SubDuration) -> Self::Output {
        SubInstant(self.0 - rhs.0)
    }
}

impl Mul<f32> for SubDuration {
    type Output = SubDuration;

    fn mul(self, rhs: f32) -> Self::Output {
        SubDuration(self.0 * rhs)
    }
}

impl MulAssign<f32> for SubDuration {
    fn mul_assign(&mut self, rhs: f32) {
        self.0 *= rhs;
    }
}

impl AddAssign<f32> for SubDuration {
    fn add_assign(&mut self, rhs: f32) {
        self.0 += rhs;
    }
}

impl SubInstant {
    /// Scale by a factor with a custom start time
    pub fn scale(&self, start: SubInstant, factor: f32) -> SubInstant {
        SubInstant(start.0 + (self.0 - start.0) * factor)
    }
}

impl AddAssign<f32> for SubInstant {
    fn add_assign(&mut self, rhs: f32) {
        self.0 += rhs;
    }
}

impl AddAssign<SubInstant> for SubInstant {
    fn add_assign(&mut self, rhs: SubInstant) {
        self.0 += rhs.0;
    }
}

impl SubAssign<SubInstant> for SubInstant {
    fn sub_assign(&mut self, rhs: SubInstant) {
        self.0 -= rhs.0;
    }
}

impl SubAssign<SubDuration> for SubInstant {
    fn sub_assign(&mut self, rhs: SubDuration) {
        self.0 -= rhs.0;
    }
}

impl AddAssign<SubDuration> for SubInstant {
    fn add_assign(&mut self, rhs: SubDuration) {
        self.0 += rhs.0;
    }
}

impl Display for SubInstant {
    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
        // TODO optimize this
        let sign = self.0.signum();
        let mut secs = self.0.abs();
        let hours = (secs / 3600f32).floor();
        secs -= hours * 3600f32;
        let minutes = (secs / 60f32).floor();
        secs -= minutes * 60f32;
        let msecs = ((secs % 1f32) * 1000f32).round();
        write!(f, "{}{:02}:{:02}:{:02},{:03}",
               if sign.is_sign_negative() { "-" } else { "" },
               hours, minutes, secs.floor(), msecs)
    }
}

#[derive(Clone, Debug)]
struct Subtitle {
    num: u32,
    start: SubInstant,
    dur: SubDuration,
    text: String,
}

impl Display for Subtitle {
    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
        write!(f, "{}\n{} --> {}\n{}\n\n",
               self.num,
               self.start, self.start + self.dur,
               self.text
        )
    }
}


impl TryFrom<&str> for SubInstant {
    type Error = failure::Error;

    fn try_from(value: &str) -> Result<Self, Self::Error> {
        lazy_static! {
            static ref DATE_RE: Regex = Regex::new(r"^(-)?(?P<h>\d+):(?P<m>\d+):(?P<s>\d+(:?[,.]\d+)?)$").unwrap();
        }

        match DATE_RE.captures(value) {
            Some(caps) => {
                let minus = if caps.get(1).is_some() { -1f32 } else { 1f32 };

                let h = &caps["h"];
                let m = &caps["m"];
                let s = caps["s"].replace(",", ".");

                Ok(SubInstant(minus * (f32::from_str(h).unwrap() * 3600f32 +
                    f32::from_str(m).unwrap() * 60f32 +
                    f32::from_str(&s).unwrap())))
            }
            None => Err(failure::err_msg("Error parsing time."))
        }
    }
}

impl SubDuration {
    pub fn as_instant(&self) -> SubInstant {
        SubInstant(self.0)
    }
}

impl TryFrom<&str> for SubDuration {
    type Error = failure::Error;

    fn try_from(value: &str) -> Result<Self, Self::Error> {
        lazy_static! {
            static ref TIME_RE: Regex = Regex::new(r"^(-)?(?:(\d+):)?(?:(\d+):)?(\d+(?:[,.]\d+)?)$").unwrap();
        }

        match TIME_RE.captures(value) {
            Some(caps) => {
                let minus = if caps.get(1).is_some() { -1f32 } else { 1f32 };
                let a = caps.get(2).map(|m| m.as_str()).unwrap_or("");
                let b = caps.get(3).map(|m| m.as_str()).unwrap_or("");
                let s = caps.get(4).map(|m| m.as_str()).unwrap_or("0").replace(",", ".");

                let (h, m) = if b.is_empty() {
                    ("0", if a.is_empty() { "0" } else { a })
                } else {
                    (a, b)
                };

                Ok(SubDuration(minus * (f32::from_str(h).unwrap() * 3600f32 +
                    f32::from_str(m).unwrap() * 60f32 +
                    f32::from_str(&s).unwrap())))
            }
            None => Err(failure::err_msg("Error parsing time: No match"))
        }
    }
}

#[test]
fn test_parse_duration() {
    // this is used for user input on the command line
    let bad = SubDuration(-1f32);
    assert_eq!(SubDuration(45678f32), SubDuration::try_from("45678").unwrap_or(bad), "integer secs");
    assert_eq!(SubDuration(1.23f32), SubDuration::try_from("1.23").unwrap_or(bad), "float secs with period");
    assert_eq!(SubDuration(-1.23f32), SubDuration::try_from("-1.23").unwrap_or(bad), "MINUS float secs with period");
    assert_eq!(SubDuration(1.23f32), SubDuration::try_from("1,23").unwrap_or(bad), "float secs with comma");
    assert_eq!(SubDuration(121.15f32), SubDuration::try_from("2:1.15").unwrap_or(bad), "m:s.frac");
    assert_eq!(SubDuration(121.15f32), SubDuration::try_from("2:01.15").unwrap_or(bad), "m:0s.frac");
    assert_eq!(SubDuration(121.15f32), SubDuration::try_from("02:01.15").unwrap_or(bad), "0m:0s.frac");
    assert_eq!(SubDuration(121.15f32), SubDuration::try_from("02:01,15").unwrap_or(bad), "0m:0s,frac");

    assert_eq!(SubDuration(3721.15f32), SubDuration::try_from("1:02:01,15").unwrap_or(bad), "h:0m:0s,frac");
    assert_eq!(SubDuration(3721.15f32), SubDuration::try_from("1:02:01,15").unwrap_or(bad), "h:0m:0s.frac");
    assert_eq!(SubDuration(3721.15f32), SubDuration::try_from("01:02:01,15").unwrap_or(bad), "0h:0m:0s,frac");
    assert_eq!(SubDuration(-3721.15f32), SubDuration::try_from("-01:02:01,15").unwrap_or(bad), "-0h:0m:0s,frac");
}

#[test]
fn test_parse_instant() {
    let bad = SubInstant(-1f32);
    assert_eq!(SubInstant(1081.755f32), SubInstant::try_from("00:18:01,755").unwrap_or(bad));
    assert_eq!(SubInstant(1081.755f32), SubInstant::try_from("00:18:01.755").unwrap_or(bad));
    assert_eq!(SubInstant(1081.7f32), SubInstant::try_from("00:18:01.7").unwrap_or(bad));
    assert_eq!(SubInstant(1081.7f32), SubInstant::try_from("0:18:1.7").unwrap_or(bad));
    assert_eq!(SubInstant(0f32), SubInstant::try_from("00:00:00,000").unwrap_or(bad));
    assert_eq!(SubInstant(-3600f32), SubInstant::try_from("-01:00:00,000").unwrap_or(bad));
}

#[test]
fn test_stringify_instant() {
    assert_eq!("00:18:01,755", SubInstant::try_from("00:18:01,755").unwrap().to_string());
    assert_eq!("-00:18:01,755", SubInstant::try_from("-00:18:01,755").unwrap().to_string());
    assert_eq!("-00:18:01,700", SubInstant::try_from("-00:18:01.7").unwrap().to_string());
    assert_eq!("00:00:00,000", SubInstant::try_from("00:00:00,000").unwrap().to_string());
    assert_eq!("-00:00:00,000", SubInstant::try_from("-00:00:00,000").unwrap().to_string());
}