Merge pull request #257 from crunchy-labs/feature/relative_sequence_number

Add flags and option to control special episode behavior (#206, #241, #246)

crunchy-cli-core/src/archive/command.rs

@@ -47,11 +47,18 @@ pub struct Archive {
       {season_number}            → Number of the season\n  \
       {episode_number}           → Number of the episode\n  \
       {relative_episode_number}  → Number of the episode relative to its season\n  \
+      {sequence_number}          → Like '{episode_number}' but without possible non-number characters\n  \
+      {relative_sequence_number} → Like '{relative_episode_number}' but with support for episode 0's and .5's\n  \
       {series_id}                → ID of the series\n  \
       {season_id}                → ID of the season\n  \
       {episode_id}               → ID of the episode")]
     #[arg(short, long, default_value = "{title}.mkv")]
     pub(crate) output: String,
+    #[arg(help = "Name of the output file if the episode is a special")]
+    #[arg(long_help = "Name of the output file if the episode is a special. \
+    If not set, the '-o'/'--output' flag will be used as name template")]
+    #[arg(long)]
+    pub(crate) output_specials: Option<String>,
 
     #[arg(help = "Video resolution")]
     #[arg(long_help = "The video resolution.\
@@ -93,6 +100,9 @@ pub struct Archive {
     #[arg(help = "Skip files which are already existing")]
     #[arg(long, default_value_t = false)]
     pub(crate) skip_existing: bool,
+    #[arg(help = "Skip special episodes")]
+    #[arg(long, default_value_t = false)]
+    pub(crate) skip_specials: bool,
 
     #[arg(help = "Skip any interactive input")]
     #[arg(short, long, default_value_t = false)]
@@ -121,6 +131,17 @@ impl Execute for Archive {
             && self.output != "-"
         {
             bail!("File extension is not '.mkv'. Currently only matroska / '.mkv' files are supported")
+        } else if let Some(special_output) = &self.output_specials {
+            if PathBuf::from(special_output)
+                .extension()
+                .unwrap_or_default()
+                .to_string_lossy()
+                != "mkv"
+                && !is_special_file(special_output)
+                && special_output != "-"
+            {
+                bail!("File extension for special episodes is not '.mkv'. Currently only matroska / '.mkv' files are supported")
+            }
         }
 
         self.audio = all_locale_in_locales(self.audio.clone());
@@ -145,7 +166,8 @@ impl Execute for Archive {
 
         for (i, (media_collection, url_filter)) in parsed_urls.into_iter().enumerate() {
             let progress_handler = progress!("Fetching series details");
-            let single_format_collection = ArchiveFilter::new(url_filter, self.clone(), !self.yes)
+            let single_format_collection =
+                ArchiveFilter::new(url_filter, self.clone(), !self.yes, self.skip_specials)
                     .visit(media_collection)
                     .await?;
 
@@ -173,7 +195,15 @@ impl Execute for Archive {
                     downloader.add_format(download_format)
                 }
 
-                let formatted_path = format.format_path((&self.output).into());
+                let formatted_path = if format.is_special() {
+                    format.format_path(
+                        self.output_specials
+                            .as_ref()
+                            .map_or((&self.output).into(), |so| so.into()),
+                    )
+                } else {
+                    format.format_path((&self.output).into())
+                };
                 let (path, changed) = free_file(formatted_path.clone());
 
                 if changed && self.skip_existing {

crunchy-cli-core/src/archive/filter.rs

@@ -2,7 +2,7 @@ use crate::archive::command::Archive;
 use crate::utils::filter::{real_dedup_vec, Filter};
 use crate::utils::format::{Format, SingleFormat, SingleFormatCollection};
 use crate::utils::interactive_select::{check_for_duplicated_seasons, get_duplicated_seasons};
-use crate::utils::parse::UrlFilter;
+use crate::utils::parse::{fract, UrlFilter};
 use anyhow::Result;
 use crunchyroll_rs::{Concert, Episode, Locale, Movie, MovieListing, MusicVideo, Season, Series};
 use log::{info, warn};
@@ -18,19 +18,26 @@ pub(crate) struct ArchiveFilter {
     url_filter: UrlFilter,
     archive: Archive,
     interactive_input: bool,
-    season_episode_count: HashMap<String, Vec<String>>,
+    skip_special: bool,
+    season_episodes: HashMap<String, Vec<Episode>>,
     season_subtitles_missing: Vec<u32>,
     season_sorting: Vec<String>,
     visited: Visited,
 }
 
 impl ArchiveFilter {
-    pub(crate) fn new(url_filter: UrlFilter, archive: Archive, interactive_input: bool) -> Self {
+    pub(crate) fn new(
+        url_filter: UrlFilter,
+        archive: Archive,
+        interactive_input: bool,
+        skip_special: bool,
+    ) -> Self {
         Self {
             url_filter,
             archive,
             interactive_input,
-            season_episode_count: HashMap::new(),
+            skip_special,
+            season_episodes: HashMap::new(),
             season_subtitles_missing: vec![],
             season_sorting: vec![],
             visited: Visited::None,
@@ -226,12 +233,12 @@ impl Filter for ArchiveFilter {
             episodes.extend(eps)
         }
 
-        if Format::has_relative_episodes_fmt(&self.archive.output) {
+        if Format::has_relative_fmt(&self.archive.output) {
             for episode in episodes.iter() {
-                self.season_episode_count
+                self.season_episodes
                     .entry(episode.season_id.clone())
                     .or_insert(vec![])
-                    .push(episode.id.clone())
+                    .push(episode.clone())
             }
         }
 
@@ -241,7 +248,14 @@ impl Filter for ArchiveFilter {
     async fn visit_episode(&mut self, mut episode: Episode) -> Result<Option<Self::T>> {
         if !self
             .url_filter
-            .is_episode_valid(episode.episode_number, episode.season_number)
+            .is_episode_valid(episode.sequence_number, episode.season_number)
+        {
+            return Ok(None);
+        }
+
+        // skip the episode if it's a special
+        if self.skip_special
+            && (episode.sequence_number == 0.0 || episode.sequence_number.fract() != 0.0)
         {
             return Ok(None);
         }
@@ -299,22 +313,36 @@ impl Filter for ArchiveFilter {
             episodes.push((episode.clone(), episode.subtitle_locales.clone()))
         }
 
-        let relative_episode_number = if Format::has_relative_episodes_fmt(&self.archive.output) {
+        let mut relative_episode_number = None;
-            if self.season_episode_count.get(&episode.season_id).is_none() {
+        let mut relative_sequence_number = None;
-                let season_episodes = episode.season().await?.episodes().await?;
+        // get the relative episode number. only done if the output string has the pattern to include
-                self.season_episode_count.insert(
+        // the relative episode number as this requires some extra fetching
+        if Format::has_relative_fmt(&self.archive.output) {
+            let season_eps = match self.season_episodes.get(&episode.season_id) {
+                Some(eps) => eps,
+                None => {
+                    self.season_episodes.insert(
                         episode.season_id.clone(),
-                    season_episodes.into_iter().map(|e| e.id).collect(),
+                        episode.season().await?.episodes().await?,
                     );
+                    self.season_episodes.get(&episode.season_id).unwrap()
                 }
-            let relative_episode_number = self
+            };
-                .season_episode_count
+            let mut non_integer_sequence_number_count = 0;
-                .get(&episode.season_id)
+            for (i, ep) in season_eps.iter().enumerate() {
-                .unwrap()
+                if ep.sequence_number.fract() != 0.0 || ep.sequence_number == 0.0 {
-                .iter()
+                    non_integer_sequence_number_count += 1;
-                .position(|id| id == &episode.id)
+                }
-                .map(|index| index + 1);
+                if ep.id == episode.id {
-            if relative_episode_number.is_none() {
+                    relative_episode_number = Some(i + 1);
+                    relative_sequence_number = Some(
+                        (i + 1 - non_integer_sequence_number_count) as f32
+                            + fract(ep.sequence_number),
+                    );
+                    break;
+                }
+            }
+            if relative_episode_number.is_none() || relative_sequence_number.is_none() {
                 warn!(
                     "Failed to get relative episode number for episode {} ({}) of {} season {}",
                     episode.episode_number,
@@ -323,16 +351,18 @@ impl Filter for ArchiveFilter {
                     episode.season_number,
                 )
             }
-            relative_episode_number
+        }
-        } else {
-            None
-        };
 
         Ok(Some(
             episodes
                 .into_iter()
                 .map(|(e, s)| {
-                    SingleFormat::new_from_episode(e, s, relative_episode_number.map(|n| n as u32))
+                    SingleFormat::new_from_episode(
+                        e,
+                        s,
+                        relative_episode_number.map(|n| n as u32),
+                        relative_sequence_number,
+                    )
                 })
                 .collect(),
         ))

crunchy-cli-core/src/download/command.rs

@@ -43,11 +43,18 @@ pub struct Download {
       {season_number}            → Number of the season\n  \
       {episode_number}           → Number of the episode\n  \
       {relative_episode_number}  → Number of the episode relative to its season\n  \
+      {sequence_number}          → Like '{episode_number}' but without possible non-number characters\n  \
+      {relative_sequence_number} → Like '{relative_episode_number}' but with support for episode 0's and .5's\n  \
       {series_id}                → ID of the series\n  \
       {season_id}                → ID of the season\n  \
       {episode_id}               → ID of the episode")]
     #[arg(short, long, default_value = "{title}.mp4")]
     pub(crate) output: String,
+    #[arg(help = "Name of the output file if the episode is a special")]
+    #[arg(long_help = "Name of the output file if the episode is a special. \
+    If not set, the '-o'/'--output' flag will be used as name template")]
+    #[arg(long)]
+    pub(crate) output_specials: Option<String>,
 
     #[arg(help = "Video resolution")]
     #[arg(long_help = "The video resolution.\
@@ -71,6 +78,9 @@ pub struct Download {
     #[arg(help = "Skip files which are already existing")]
     #[arg(long, default_value_t = false)]
     pub(crate) skip_existing: bool,
+    #[arg(help = "Skip special episodes")]
+    #[arg(long, default_value_t = false)]
+    pub(crate) skip_specials: bool,
 
     #[arg(help = "Skip any interactive input")]
     #[arg(short, long, default_value_t = false)]
@@ -114,6 +124,25 @@ impl Execute for Download {
             }
         }
 
+        if let Some(special_output) = &self.output_specials {
+            if Path::new(special_output)
+                .extension()
+                .unwrap_or_default()
+                .is_empty()
+                && !is_special_file(special_output)
+                && special_output != "-"
+            {
+                bail!("No file extension found. Please specify a file extension (via `--output-specials`) for the output file")
+            }
+            if let Some(ext) = Path::new(special_output).extension() {
+                if self.force_hardsub {
+                    warn!("Hardsubs are forced for special episodes. Adding subtitles may take a while")
+                } else if !["mkv", "mov", "mp4"].contains(&ext.to_string_lossy().as_ref()) {
+                    warn!("Detected a container which does not support softsubs. Adding subtitles for special episodes may take a while")
+                }
+            }
+        }
+
         Ok(())
     }
 
@@ -133,7 +162,8 @@ impl Execute for Download {
 
         for (i, (media_collection, url_filter)) in parsed_urls.into_iter().enumerate() {
             let progress_handler = progress!("Fetching series details");
-            let single_format_collection = DownloadFilter::new(url_filter, self.clone(), !self.yes)
+            let single_format_collection =
+                DownloadFilter::new(url_filter, self.clone(), !self.yes, self.skip_specials)
                     .visit(media_collection)
                     .await?;
 
@@ -165,7 +195,15 @@ impl Execute for Download {
                 let mut downloader = download_builder.clone().build();
                 downloader.add_format(download_format);
 
-                let formatted_path = format.format_path((&self.output).into());
+                let formatted_path = if format.is_special() {
+                    format.format_path(
+                        self.output_specials
+                            .as_ref()
+                            .map_or((&self.output).into(), |so| so.into()),
+                    )
+                } else {
+                    format.format_path((&self.output).into())
+                };
                 let (path, changed) = free_file(formatted_path.clone());
 
                 if changed && self.skip_existing {

crunchy-cli-core/src/download/filter.rs

@@ -2,7 +2,7 @@ use crate::download::Download;
 use crate::utils::filter::Filter;
 use crate::utils::format::{Format, SingleFormat, SingleFormatCollection};
 use crate::utils::interactive_select::{check_for_duplicated_seasons, get_duplicated_seasons};
-use crate::utils::parse::UrlFilter;
+use crate::utils::parse::{fract, UrlFilter};
 use anyhow::{bail, Result};
 use crunchyroll_rs::{Concert, Episode, Movie, MovieListing, MusicVideo, Season, Series};
 use log::{error, info, warn};
@@ -12,18 +12,25 @@ pub(crate) struct DownloadFilter {
     url_filter: UrlFilter,
     download: Download,
     interactive_input: bool,
-    season_episode_count: HashMap<u32, Vec<String>>,
+    skip_special: bool,
+    season_episodes: HashMap<u32, Vec<Episode>>,
     season_subtitles_missing: Vec<u32>,
     season_visited: bool,
 }
 
 impl DownloadFilter {
-    pub(crate) fn new(url_filter: UrlFilter, download: Download, interactive_input: bool) -> Self {
+    pub(crate) fn new(
+        url_filter: UrlFilter,
+        download: Download,
+        interactive_input: bool,
+        skip_special: bool,
+    ) -> Self {
         Self {
             url_filter,
             download,
             interactive_input,
-            season_episode_count: HashMap::new(),
+            skip_special,
+            season_episodes: HashMap::new(),
             season_subtitles_missing: vec![],
             season_visited: false,
         }
@@ -107,18 +114,18 @@ impl Filter for DownloadFilter {
 
         let mut episodes = season.episodes().await?;
 
-        if Format::has_relative_episodes_fmt(&self.download.output) {
+        if Format::has_relative_fmt(&self.download.output) {
             for episode in episodes.iter() {
-                self.season_episode_count
+                self.season_episodes
                     .entry(episode.season_number)
                     .or_insert(vec![])
-                    .push(episode.id.clone())
+                    .push(episode.clone())
             }
         }
 
         episodes.retain(|e| {
             self.url_filter
-                .is_episode_valid(e.episode_number, season.season_number)
+                .is_episode_valid(e.sequence_number, season.season_number)
         });
 
         Ok(episodes)
@@ -127,7 +134,14 @@ impl Filter for DownloadFilter {
     async fn visit_episode(&mut self, mut episode: Episode) -> Result<Option<Self::T>> {
         if !self
             .url_filter
-            .is_episode_valid(episode.episode_number, episode.season_number)
+            .is_episode_valid(episode.sequence_number, episode.season_number)
+        {
+            return Ok(None);
+        }
+
+        // skip the episode if it's a special
+        if self.skip_special
+            && (episode.sequence_number == 0.0 || episode.sequence_number.fract() != 0.0)
         {
             return Ok(None);
         }
@@ -189,28 +203,36 @@ impl Filter for DownloadFilter {
             }
         }
 
+        let mut relative_episode_number = None;
+        let mut relative_sequence_number = None;
         // get the relative episode number. only done if the output string has the pattern to include
         // the relative episode number as this requires some extra fetching
-        let relative_episode_number = if Format::has_relative_episodes_fmt(&self.download.output) {
+        if Format::has_relative_fmt(&self.download.output) {
-            if self
+            let season_eps = match self.season_episodes.get(&episode.season_number) {
-                .season_episode_count
+                Some(eps) => eps,
-                .get(&episode.season_number)
+                None => {
-                .is_none()
+                    self.season_episodes.insert(
-            {
-                let season_episodes = episode.season().await?.episodes().await?;
-                self.season_episode_count.insert(
                         episode.season_number,
-                    season_episodes.into_iter().map(|e| e.id).collect(),
+                        episode.season().await?.episodes().await?,
                     );
+                    self.season_episodes.get(&episode.season_number).unwrap()
                 }
-            let relative_episode_number = self
+            };
-                .season_episode_count
+            let mut non_integer_sequence_number_count = 0;
-                .get(&episode.season_number)
+            for (i, ep) in season_eps.iter().enumerate() {
-                .unwrap()
+                if ep.sequence_number.fract() != 0.0 || ep.sequence_number == 0.0 {
-                .iter()
+                    non_integer_sequence_number_count += 1;
-                .position(|id| id == &episode.id)
+                }
-                .map(|index| index + 1);
+                if ep.id == episode.id {
-            if relative_episode_number.is_none() {
+                    relative_episode_number = Some(i + 1);
+                    relative_sequence_number = Some(
+                        (i + 1 - non_integer_sequence_number_count) as f32
+                            + fract(ep.sequence_number),
+                    );
+                    break;
+                }
+            }
+            if relative_episode_number.is_none() || relative_sequence_number.is_none() {
                 warn!(
                     "Failed to get relative episode number for episode {} ({}) of {} season {}",
                     episode.episode_number,
@@ -219,10 +241,7 @@ impl Filter for DownloadFilter {
                     episode.season_number,
                 )
             }
-            relative_episode_number
+        }
-        } else {
-            None
-        };
 
         Ok(Some(SingleFormat::new_from_episode(
             episode.clone(),
@@ -234,6 +253,7 @@ impl Filter for DownloadFilter {
                 }
             }),
             relative_episode_number.map(|n| n as u32),
+            relative_sequence_number,
         )))
     }
 

crunchy-cli-core/src/search/filter.rs

@@ -24,7 +24,7 @@ impl FilterOptions {
             self.check_audio_language(&vec![e.audio_locale.clone()])
                 && self
                     .url_filter
-                    .is_episode_valid(e.episode_number, e.season_number)
+                    .is_episode_valid(e.sequence_number, e.season_number)
         });
         episodes
     }

crunchy-cli-core/src/utils/format.rs

@@ -29,8 +29,9 @@ pub struct SingleFormat {
 
     pub episode_id: String,
     pub episode_number: String,
-    pub sequence_number: f32,
     pub relative_episode_number: Option<u32>,
+    pub sequence_number: f32,
+    pub relative_sequence_number: Option<f32>,
 
     pub duration: Duration,
 
@@ -42,6 +43,7 @@ impl SingleFormat {
         episode: Episode,
         subtitles: Vec<Locale>,
         relative_episode_number: Option<u32>,
+        relative_sequence_number: Option<f32>,
     ) -> Self {
         Self {
             identifier: if episode.identifier.is_empty() {
@@ -73,6 +75,7 @@ impl SingleFormat {
             },
             sequence_number: episode.sequence_number,
             relative_episode_number,
+            relative_sequence_number,
             duration: episode.duration,
             source: episode.into(),
         }
@@ -92,8 +95,9 @@ impl SingleFormat {
             season_number: 1,
             episode_id: movie.id.clone(),
             episode_number: "1".to_string(),
-            sequence_number: 1.0,
             relative_episode_number: Some(1),
+            sequence_number: 1.0,
+            relative_sequence_number: Some(1.0),
             duration: movie.duration,
             source: movie.into(),
         }
@@ -113,8 +117,9 @@ impl SingleFormat {
             season_number: 1,
             episode_id: music_video.id.clone(),
             episode_number: "1".to_string(),
-            sequence_number: 1.0,
             relative_episode_number: Some(1),
+            sequence_number: 1.0,
+            relative_sequence_number: Some(1.0),
             duration: music_video.duration,
             source: music_video.into(),
         }
@@ -134,8 +139,9 @@ impl SingleFormat {
             season_number: 1,
             episode_id: concert.id.clone(),
             episode_number: "1".to_string(),
-            sequence_number: 1.0,
             relative_episode_number: Some(1),
+            sequence_number: 1.0,
+            relative_sequence_number: Some(1.0),
             duration: concert.duration,
             source: concert.into(),
         }
@@ -328,8 +334,9 @@ pub struct Format {
 
     pub episode_id: String,
     pub episode_number: String,
-    pub sequence_number: f32,
     pub relative_episode_number: Option<u32>,
+    pub sequence_number: f32,
+    pub relative_sequence_number: Option<f32>,
 }
 
 impl Format {
@@ -363,8 +370,9 @@ impl Format {
             season_number: first_format.season_number,
             episode_id: first_format.episode_id,
             episode_number: first_format.episode_number,
-            sequence_number: first_format.sequence_number,
             relative_episode_number: first_format.relative_episode_number,
+            sequence_number: first_format.sequence_number,
+            relative_sequence_number: first_format.relative_sequence_number,
         }
     }
 
@@ -400,9 +408,28 @@ impl Format {
             )
             .replace(
                 "{relative_episode_number}",
-                &sanitize(
+                &format!(
+                    "{:0>2}",
+                    sanitize(
                         self.relative_episode_number.unwrap_or_default().to_string(),
                         true,
+                    )
+                ),
+            )
+            .replace(
+                "{sequence_number}",
+                &format!("{:0>2}", sanitize(self.sequence_number.to_string(), true)),
+            )
+            .replace(
+                "{relative_sequence_number}",
+                &format!(
+                    "{:0>2}",
+                    sanitize(
+                        self.relative_sequence_number
+                            .unwrap_or_default()
+                            .to_string(),
+                        true,
+                    )
                 ),
             );
 
@@ -446,7 +473,12 @@ impl Format {
         tab_info!("FPS: {:.2}", self.fps)
     }
 
-    pub fn has_relative_episodes_fmt<S: AsRef<str>>(s: S) -> bool {
+    pub fn is_special(&self) -> bool {
-        return s.as_ref().contains("{relative_episode_number}");
+        self.sequence_number == 0.0 || self.sequence_number.fract() != 0.0
+    }
+
+    pub fn has_relative_fmt<S: AsRef<str>>(s: S) -> bool {
+        return s.as_ref().contains("{relative_episode_number}")
+            || s.as_ref().contains("{relative_sequence_number}");
     }
 }

crunchy-cli-core/src/utils/parse.rs

@@ -10,8 +10,8 @@ use regex::Regex;
 /// If `to_*` is [`None`] they're set to [`u32::MAX`].
 #[derive(Debug, Default)]
 pub struct InnerUrlFilter {
-    from_episode: Option<u32>,
+    from_episode: Option<f32>,
-    to_episode: Option<u32>,
+    to_episode: Option<f32>,
     from_season: Option<u32>,
     to_season: Option<u32>,
 }
@@ -39,10 +39,10 @@ impl UrlFilter {
         })
     }
 
-    pub fn is_episode_valid(&self, episode: u32, season: u32) -> bool {
+    pub fn is_episode_valid(&self, episode: f32, season: u32) -> bool {
         self.inner.iter().any(|f| {
-            let from_episode = f.from_episode.unwrap_or(u32::MIN);
+            let from_episode = f.from_episode.unwrap_or(f32::MIN);
-            let to_episode = f.to_episode.unwrap_or(u32::MAX);
+            let to_episode = f.to_episode.unwrap_or(f32::MAX);
             let from_season = f.from_season.unwrap_or(u32::MIN);
             let to_season = f.to_season.unwrap_or(u32::MAX);
 
@@ -192,3 +192,13 @@ pub fn parse_resolution(mut resolution: String) -> Result<Resolution> {
         bail!("Could not find resolution")
     }
 }
+
+/// Dirty implementation of [`f32::fract`] with more accuracy.
+pub fn fract(input: f32) -> f32 {
+    if input.fract() == 0.0 {
+        return 0.0;
+    }
+    format!("0.{}", input.to_string().split('.').last().unwrap())
+        .parse::<f32>()
+        .unwrap()
+}