From fc3d6745ff9e35d163d6c6ac7fbab217318f9d39 Mon Sep 17 00:00:00 2001
From: Gwen Lg <me@gwenlg.fr>
Date: Sat, 8 Mar 2025 22:42:18 +0100
Subject: [PATCH] feat(vobsub)!: make Index optional.

`sub` file parsing don't need `idx`, so separate `sub` data from `idx`.
Create a new Sub struct who can initialized from file, and contain the content of the `sub`.
Remove `sub_data` from `Index` struct.
---
 src/vobsub/idx.rs |  8 +-------
 src/vobsub/mod.rs |  2 +-
 src/vobsub/sub.rs | 49 ++++++++++++++++++++++++++++++++++++++++-------
 3 files changed, 44 insertions(+), 15 deletions(-)

diff --git a/src/vobsub/idx.rs b/src/vobsub/idx.rs
index f32b6de..557f56d 100644
--- a/src/vobsub/idx.rs
+++ b/src/vobsub/idx.rs
@@ -13,7 +13,7 @@ use crate::vobsub::IResultExt;
 
 use super::{
     palette::{palette, DEFAULT_PALETTE},
-    sub, Palette, VobSubError,
+    Palette, VobSubError,
 };
 
 /// A `*.idx` file describing the subtitles in a `*.sub` file.
@@ -81,12 +81,6 @@ impl Index {
     pub const fn palette(&self) -> &Palette {
         &self.palette
     }
-
-    /// Iterate over the subtitles associated with this `*.idx` file.
-    #[must_use]
-    pub fn subtitles<D>(&self) -> sub::VobsubParser<D> {
-        sub::VobsubParser::new(&self.sub_data)
-    }
 }
 
 /// Read the palette in `*.idx` file content
diff --git a/src/vobsub/mod.rs b/src/vobsub/mod.rs
index 3ba6b3b..433f85c 100644
--- a/src/vobsub/mod.rs
+++ b/src/vobsub/mod.rs
@@ -79,7 +79,7 @@ pub use self::{
     img::{conv_to_rgba, VobSubIndexedImage, VobSubOcrImage, VobSubToImage},
     palette::{palette, palette_rgb_to_luminance, Palette},
     probe::{is_idx_file, is_sub_file},
-    sub::ErrorMissing,
+    sub::{ErrorMissing, Sub},
 };
 
 use crate::content::ContentError;
diff --git a/src/vobsub/sub.rs b/src/vobsub/sub.rs
index d3cb5ad..7857017 100644
--- a/src/vobsub/sub.rs
+++ b/src/vobsub/sub.rs
@@ -27,7 +27,9 @@ use nom::{
     sequence::{preceded, Tuple},
     IResult,
 };
-use std::{cmp::Ordering, fmt::Debug, iter::FusedIterator, marker::PhantomData};
+use std::{
+    cmp::Ordering, fmt::Debug, fs, io::Read, iter::FusedIterator, marker::PhantomData, path::Path,
+};
 use thiserror::Error;
 
 /// Parse four 4-bit palette entries.
@@ -319,6 +321,41 @@ macro_rules! try_iter {
     };
 }
 
+/// Store the content of a `*.sub` file.
+pub struct Sub {
+    /// Our compressed subtitle data.
+    data: Vec<u8>,
+}
+impl Sub {
+    /// Init a `Sub` from a file path.
+    ///
+    /// # Errors
+    ///
+    /// Will return `VobSubError::Io` if not able to read a file from `path`.
+    pub fn open<P>(path: P) -> Result<Self, VobSubError>
+    where
+        P: AsRef<Path> + Clone,
+    {
+        let mut sub = fs::File::open(path.as_ref()).map_err(|source| VobSubError::Io {
+            source,
+            path: path.as_ref().to_path_buf(),
+        })?;
+        let mut data = vec![];
+        sub.read_to_end(&mut data)
+            .map_err(|source| VobSubError::Io {
+                source,
+                path: path.as_ref().to_path_buf(),
+            })?;
+        Ok(Self { data })
+    }
+
+    /// Iterate over the subtitles associated with this `*.idx` file.
+    #[must_use]
+    pub fn subtitles<D>(&self) -> VobsubParser<D> {
+        VobsubParser::new(&self.data)
+    }
+}
+
 /// An internal iterator over subtitles.  These subtitles may not have a
 /// valid `end_time`, so we'll try to fix them up before letting the user
 /// see them.
@@ -414,7 +451,6 @@ impl<D> FusedIterator for VobsubParser<'_, D> {}
 #[cfg(test)]
 mod tests {
     use super::*;
-    use crate::vobsub::idx;
 
     #[test]
     fn parse_palette_entries() {
@@ -549,9 +585,9 @@ mod tests {
     #[test]
     fn parse_subtitles_from_subtitle_edit() {
         //use env_logger;
-        use idx::Index;
+
         //let _ = env_logger::init();
-        let idx = Index::open("./fixtures/tiny.idx").unwrap();
+        let idx = Sub::open("./fixtures/tiny.sub").unwrap();
         let mut subs = idx.subtitles::<TimeSpan>();
         subs.next().expect("missing sub").unwrap();
         assert!(subs.next().is_none());
@@ -560,18 +596,17 @@ mod tests {
     #[test]
     fn parse_fuzz_corpus_seeds() {
         //use env_logger;
-        use idx::Index;
         //let _ = env_logger::init();
 
         // Make sure these two fuzz corpus inputs still work, and that they
         // return the same subtitle data.
-        let tiny = Index::open("./fixtures/tiny.idx")
+        let tiny = Sub::open("./fixtures/tiny.sub")
             .unwrap()
             .subtitles::<TimeSpan>()
             .next()
             .unwrap()
             .unwrap();
-        let split = Index::open("./fixtures/tiny-split.idx")
+        let split = Sub::open("./fixtures/tiny-split.sub")
             .unwrap()
             .subtitles::<TimeSpan>()
             .next()