Add support for exploding graphics files

README.md

@@ -1,14 +1,6 @@
 # monokakido.rs
-A Rust library for parsing and interpreting the [Monokakido](https://www.monokakido.jp/en/dictionaries/app/) dictionary format.
-Aiming for full test coverage and efficient implementation with minimal dependencies.
 
-## Notice
-
-This library started as a personal project driven by curiosity.
-It is ABSOLUTELY NOT inteded to support piracy;
-I strongly condemn making unauthorized copies of Monokakido's dictionaries,
-and take no part or responsibility in that kind of activity.
-Please buy your own dictionaries directly from Monokakido to show your love and support.
+A Rust library for parsing and interpreting the [Monokakido](https://www.monokakido.jp/en/dictionaries/app/) dictionary format. Aiming for full test coverage and efficient implementation with minimal dependencies.
 
 ## TODO:
 - Add headline support
@@ -47,5 +39,12 @@ Please buy your own dictionaries directly from Monokakido to show your love and
 ## Planned to support:
 - WISDOM3
 - SMK8
+- RHEJ
+- OLT
+- OLEX
+- OLDAE
+- OCD
+- OALD10
 - NHKACCENT2
 - DAIJISEN2
+- CCCAD

src/bin/monokakido-explode.rs

@@ -33,6 +33,7 @@ fn explode() -> Result<(), Error> {
 
     let pages_dir = out_dir(&dict) + "pages/";
     let audio_dir = out_dir(&dict) + "audio/";
+    let graphics_dir = out_dir(&dict) + "graphics/";
 
     create_dir_all(&pages_dir)?;
     let mut path = String::from(&pages_dir);
@@ -56,6 +57,18 @@ fn explode() -> Result<(), Error> {
         }
     }
 
+    if let Some(graphics) = &mut dict.graphics {
+        create_dir_all(&graphics_dir)?;
+        let mut path = String::from(&graphics_dir);
+        for idx in graphics.idx_iter()? {
+            let (id, graphics) = graphics.get_by_idx(idx)?;
+            write!(&mut path, "{id}")?;
+            let mut file = File::create(&path)?;
+            path.truncate(graphics_dir.len());
+            file.write_all(graphics)?;
+        }
+    }
+
     write_index(&dict, &dict.keys.index_len, "index_len.tsv")?;
     write_index(&dict, &dict.keys.index_prefix, "index_prefix.tsv")?;
     write_index(&dict, &dict.keys.index_suffix, "index_suffix.tsv")?;

src/dict.rs

@@ -5,12 +5,13 @@ use std::{
     path::{Path, PathBuf},
 };
 
-use crate::{audio::Audio, key::Keys, pages::Pages, Error};
+use crate::{media::Media, key::Keys, pages::Pages, Error};
 
 pub struct MonokakidoDict {
     paths: Paths,
     pub pages: Pages,
-    pub audio: Option<Audio>,
+    pub audio: Option<Media>,
+    pub graphics: Option<Media>,
     pub keys: Keys,
 }
 
@@ -60,7 +61,9 @@ impl Paths {
     }
 
     pub(crate) fn key_path(&self) -> PathBuf {
-        let mut pb = self.contents_path();
+        let mut pb = PathBuf::from(&self.base_path);
+        pb.push("Contents");
+        pb.push(&self.contents_dir);
         pb.push("key");
         pb
     }
@@ -72,7 +75,9 @@ impl Paths {
     }
 
     pub(crate) fn headline_path(&self) -> PathBuf {
-        let mut pb = self.contents_path();
+        let mut pb = PathBuf::from(&self.base_path);
+        pb.push("Contents");
+        pb.push(&self.contents_dir);
         pb.push("headline");
         pb
     }
@@ -131,13 +136,15 @@ impl MonokakidoDict {
             contents_dir: contents.dir,
         };
         let pages = Pages::new(&paths)?;
-        let audio = Audio::new(&paths)?;
-        let keys = Keys::new(paths.key_headword_path())?;
+        let audio = Media::new(&paths)?;
+        let graphics = Media::new(&paths)?;
+        let keys = Keys::new(&paths)?;
 
         Ok(MonokakidoDict {
             paths,
             pages,
             audio,
+            graphics,
             keys,
         })
     }

src/key.rs

@@ -4,12 +4,12 @@ use std::{
     fs::File,
     io::{Read, Seek},
     mem::size_of,
-    path::Path,
     str::from_utf8,
 };
 
 use crate::{
-    abi_utils::{read_vec, TransmuteSafe, LE32},
+    abi_utils::{TransmuteSafe, LE32, read_vec},
+    dict::Paths,
     Error,
 };
 
@@ -19,42 +19,27 @@ mod abi {
     #[repr(C)]
     #[derive(Debug, Clone, Copy, Default)]
     pub(super) struct FileHeader {
-        pub ver: LE32,
         magic1: LE32,
+        magic2: LE32,
         pub words_offset: LE32,
         pub idx_offset: LE32,
-        // Jimmy-Z: no idea what this is
-        // present on (not limited to) OALD10, SANKOKU8
-        pub next_offset: LE32,
+        magic3: LE32,
+        magic4: LE32,
         magic5: LE32,
         magic6: LE32,
-        magic7: LE32,
     }
 
     impl FileHeader {
-        pub(super) fn from(r: &mut impl Read) -> Result<Self, Error> {
-            let mut h = FileHeader::default();
-            r.read_exact(&mut h.as_bytes_mut()[..0x10])?;
-            if h.ver.read() == 0x10000 && h.words_offset.read() == 0x10{
-            } else if h.ver.read() == 0x20000 && h.words_offset.read() == 0x20 {
-                r.read_exact(&mut h.as_bytes_mut()[0x10..])?;
-            } else {
-                return Err(Error::KeyFileHeaderValidate)
-            }
-            if h.ver.read() == 0x10000
-                && h.magic1.read() == 0
-                && h.words_offset.us() < h.idx_offset.us()
+        pub(super) fn validate(&self) -> Result<(), Error> {
+            if self.magic1.read() == 0x20000
+                && self.magic2.read() == 0
+                && self.magic3.read() == 0
+                && self.magic4.read() == 0
+                && self.magic5.read() == 0
+                && self.magic6.read() == 0
+                && self.words_offset.us() < self.idx_offset.us()
             {
-                Ok(h)
-            } else if h.ver.read() == 0x20000
-                && h.magic1.read() == 0
-                && h.magic5.read() == 0
-                && h.magic6.read() == 0
-                && h.magic7.read() == 0
-                && h.words_offset.us() < h.idx_offset.us()
-                && (h.next_offset.read() == 0 || h.idx_offset.us() < h.next_offset.us())
-            {
-                Ok(h)
+                Ok(())
             } else {
                 Err(Error::KeyFileHeaderValidate)
             }
@@ -134,20 +119,18 @@ impl Keys {
         Ok(())
     }
 
-    pub fn new<P: AsRef<Path>>(path: P) -> Result<Keys, Error> {
-        let mut file = File::open(path)?;
+    pub fn new(paths: &Paths) -> Result<Keys, Error> {
+        let mut file = File::open(paths.key_headword_path())?;
         let file_size = file.metadata()?.len() as usize;
-        let hdr = FileHeader::from(&mut file)?;
+        let mut hdr = FileHeader::default();
+        file.read_exact(hdr.as_bytes_mut())?;
+        hdr.validate()?;
 
         file.seek(std::io::SeekFrom::Start(hdr.words_offset.read() as u64))?;
         let words = read_vec(&mut file, hdr.words_offset.us(), hdr.idx_offset.us())?;
         let Some(words) = words else { return Err(Error::InvalidIndex); };
 
-        let idx_end = (if hdr.next_offset.us() == 0 {
-            file_size
-        } else {
-            hdr.next_offset.us()
-        }) - hdr.idx_offset.us();
+        let idx_end = file_size - hdr.idx_offset.us();
         let mut ihdr = IndexHeader::default();
         file.seek(std::io::SeekFrom::Start(hdr.idx_offset.read() as u64))?;
         file.read_exact(ihdr.as_bytes_mut())?;

src/lib.rs

@@ -1,13 +1,13 @@
 mod abi_utils;
-mod audio;
+mod media;
 mod dict;
 mod error;
 mod key;
 mod pages;
-pub mod resource;
+mod resource;
 mod headline;
 
-pub use audio::Audio;
+pub use media::Media;
 pub use dict::MonokakidoDict;
 pub use error::Error;
 pub use key::{KeyIndex, Keys, PageItemId};

src/media.rs

@@ -8,22 +8,22 @@ use crate::{
 
 const RSC_NAME: &str = "audio";
 
-pub struct Audio {
+pub struct Media {
     path: PathBuf,
-    res: Option<AudioResource>,
+    res: Option<MediaResource>,
 }
 
-enum AudioResource {
+enum MediaResource {
     Rsc(Rsc),
     Nrsc(Nrsc),
 }
 
-impl Audio {
+impl Media {
     pub fn new(paths: &Paths) -> Result<Option<Self>, Error> {
         let mut path = paths.contents_path();
         path.push(RSC_NAME);
         Ok(if path.exists() {
-            Some(Audio { path, res: None })
+            Some(Media { path, res: None })
         } else {
             None
         })
@@ -35,9 +35,9 @@ impl Audio {
             let nrsc_index_exists = self.path.exists();
             self.path.pop();
             self.res = Some(if nrsc_index_exists {
-                AudioResource::Nrsc(Nrsc::new(&self.path)?)
+                MediaResource::Nrsc(Nrsc::new(&self.path)?)
             } else {
-                AudioResource::Rsc(Rsc::new(&self.path, RSC_NAME)?)
+                MediaResource::Rsc(Rsc::new(&self.path, RSC_NAME)?)
             });
         }
         Ok(())
@@ -47,22 +47,22 @@ impl Audio {
         self.init()?;
         let Some(res) = self.res.as_mut() else { unreachable!() };
         match res {
-            AudioResource::Rsc(rsc) => rsc.get(id.parse::<u32>().map_err(|_| Error::InvalidIndex)?),
-            AudioResource::Nrsc(nrsc) => nrsc.get(id),
+            MediaResource::Rsc(rsc) => rsc.get(id.parse::<u32>().map_err(|_| Error::InvalidIndex)?),
+            MediaResource::Nrsc(nrsc) => nrsc.get(id),
         }
     }
 
-    pub fn get_by_idx(&mut self, idx: usize) -> Result<(AudioId, &[u8]), Error> {
+    pub fn get_by_idx(&mut self, idx: usize) -> Result<(MediaId, &[u8]), Error> {
         self.init()?;
         let Some(res) = self.res.as_mut() else { unreachable!() };
         Ok(match res {
-            AudioResource::Rsc(rsc) => {
+            MediaResource::Rsc(rsc) => {
                 let (id, page) = rsc.get_by_idx(idx)?;
-                (AudioId::Num(id), page)
+                (MediaId::Num(id), page)
             }
-            AudioResource::Nrsc(nrsc) => {
+            MediaResource::Nrsc(nrsc) => {
                 let (id, page) = nrsc.get_by_idx(idx)?;
-                (AudioId::Str(id), page)
+                (MediaId::Str(id), page)
             }
         })
     }
@@ -71,19 +71,19 @@ impl Audio {
         self.init()?;
         let Some(res) = self.res.as_ref() else { unreachable!() };
         Ok(0..match res {
-            AudioResource::Rsc(rsc) => rsc.len(),
-            AudioResource::Nrsc(nrsc) => nrsc.len(),
+            MediaResource::Rsc(rsc) => rsc.len(),
+            MediaResource::Nrsc(nrsc) => nrsc.len(),
         })
     }
 }
 
 #[derive(Debug)]
-pub enum AudioId<'a> {
+pub enum MediaId<'a> {
     Str(&'a str),
     Num(u32),
 }
 
-impl Display for AudioId<'_> {
+impl Display for MediaId<'_> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         match self {
             Self::Str(str) => f.write_str(str),

src/resource/nrsc.rs

@@ -210,7 +210,7 @@ impl Nrsc {
         Ok(files)
     }
 
-    pub fn new(path: &Path) -> Result<Self, Error> {
+    pub(crate) fn new(path: &Path) -> Result<Self, Error> {
         let files = Nrsc::files(path)?;
         let index = NrscIndex::new(path)?;
         Ok(Nrsc {

src/resource/rsc.rs

@@ -280,7 +280,7 @@ impl Rsc {
         Ok(files)
     }
 
-    pub fn new(path: &Path, rsc_name: &str) -> Result<Self, Error> {
+    pub(crate) fn new(path: &Path, rsc_name: &str) -> Result<Self, Error> {
         let files = Rsc::files(path, rsc_name)?;
         let index = RscIndex::new(path, rsc_name)?;
         Ok(Self {