1 files changed, 120 insertions, 0 deletions
diff --git a/import/src/matroska.rs b/import/src/matroska.rs
index e69de29..f9a59ab 100644
--- a/import/src/matroska.rs
+++ b/import/src/matroska.rs
@@ -0,0 +1,120 @@
+/*
+    This file is part of jellything (https://codeberg.org/metamuffin/jellything)
+    which is licensed under the GNU Affero General Public License (version 3); see /COPYING.
+    Copyright (C) 2025 metamuffin <metamuffin.org>
+*/
+use crate::infojson::{self, YVideo};
+use anyhow::{Context, Result};
+use bincode::{Decode, Encode};
+use ebml_struct::{
+    ids::*,
+    matroska::*,
+    read::{EbmlReadExt, TagRead},
+};
+use jellybase::{
+    assetfed::AssetInner,
+    cache::{cache_file, cache_memory},
+    common::Asset,
+};
+use log::info;
+use std::{
+    fs::File,
+    io::{BufReader, ErrorKind, Read, Write},
+    path::Path,
+    sync::Arc,
+};
+
+#[derive(Encode, Decode, Clone)]
+pub(crate) struct MatroskaMetadata {
+    pub info: Option<Info>,
+    pub tracks: Option<Tracks>,
+    pub cover: Option<Asset>,
+    pub chapters: Option<Chapters>,
+    pub tags: Option<Tags>,
+    pub infojson: Option<YVideo>,
+}
+pub(crate) fn matroska_metadata(path: &Path) -> Result<Arc<Option<MatroskaMetadata>>> {
+    cache_memory(&["mkmeta-v1", path.to_string_lossy().as_ref()], || {
+        let mut magic = [0; 4];
+        File::open(path)?.read_exact(&mut magic).ok();
+        if !matches!(magic, [0x1A, 0x45, 0xDF, 0xA3]) {
+            return Ok(None);
+        }
+
+        info!("reading media file {path:?}");
+        let mut file = BufReader::new(File::open(path)?);
+        let mut file = file.by_ref().take(u64::MAX);
+
+        let (x, mut ebml) = file.read_tag()?;
+        assert_eq!(x, EL_EBML);
+        let ebml = Ebml::read(&mut ebml).unwrap();
+        assert!(ebml.doc_type == "matroska" || ebml.doc_type == "webm");
+        let (x, mut segment) = file.read_tag()?;
+        assert_eq!(x, EL_SEGMENT);
+
+        let mut info = None;
+        let mut infojson = None;
+        let mut tracks = None;
+        let mut cover = None;
+        let mut chapters = None;
+        let mut tags = None;
+        loop {
+            let (x, mut seg) = match segment.read_tag() {
+                Ok(o) => o,
+                Err(e) if e.kind() == ErrorKind::UnexpectedEof => break,
+                Err(e) => return Err(e.into()),
+            };
+            match x {
+                EL_INFO => info = Some(Info::read(&mut seg).context("info")?),
+                EL_TRACKS => tracks = Some(Tracks::read(&mut seg).context("tracks")?),
+                EL_CHAPTERS => chapters = Some(Chapters::read(&mut seg).context("chapters")?),
+                EL_TAGS => tags = Some(Tags::read(&mut seg).context("tags")?),
+                EL_ATTACHMENTS => {
+                    let attachments = Attachments::read(&mut seg).context("attachments")?;
+                    for f in attachments.files {
+                        match f.name.as_str() {
+                            "info.json" => {
+                                infojson = Some(
+                                    serde_json::from_slice::<infojson::YVideo>(&f.data)
+                                        .context("infojson")?,
+                                );
+                            }
+                            "cover.webp" | "cover.png" | "cover.jpg" | "cover.jpeg"
+                            | "cover.avif" => {
+                                cover = Some(
+                                    AssetInner::Cache(cache_file(
+                                        &["att-cover", path.to_string_lossy().as_ref()],
+                                        move |mut file| {
+                                            file.write_all(&f.data)?;
+                                            Ok(())
+                                        },
+                                    )?)
+                                    .ser(),
+                                )
+                            }
+                            a => println!("{a:?}"),
+                        }
+                    }
+                }
+                EL_VOID | EL_CRC32 | EL_CUES | EL_SEEKHEAD => {
+                    seg.consume()?;
+                }
+                EL_CLUSTER => {
+                    break;
+                }
+                id => {
+                    eprintln!("unknown top-level element {id:x}");
+                    seg.consume()?;
+                }
+            }
+        }
+        Ok(Some(MatroskaMetadata {
+            chapters,
+            cover,
+            info,
+            infojson,
+            tags,
+            tracks,
+        }))
+    })
+}