seeking and broken writing

7 files changed, 310 insertions, 111 deletions

diff --git a/ebml/src/bin/experiment.rs b/ebml/src/bin/experiment.rs
index c510bb5..c850753 100644
--- a/ebml/src/bin/experiment.rs
+++ b/ebml/src/bin/experiment.rs
@@ -1,12 +1,31 @@
-use ebml::read::EbmlReader;
-use std::{fs::File, io::BufReader};
+use ebml::{matroska::MatroskaTag, read::EbmlReader, write::EbmlWriter};
+use std::{
+    fs::File,
+    io::{stdin, stdout, BufReader, BufWriter},
+};
 
-fn main() {
+fn main() -> anyhow::Result<()> {
     env_logger::init_from_env("LOG");
-    let f = File::open(std::env::args().skip(1).next().unwrap()).unwrap();
-    let mut r = EbmlReader::new(BufReader::new(f));
+    let path = std::env::args().skip(1).next().unwrap();
+    let mut r = EbmlReader::new(BufReader::new(File::open(path)?));
+    let mut w = EbmlWriter::new(BufWriter::new(stdout()), 0);
 
-    for tag in r {
-        println!("{:?}", tag);
+    // r.seek(
+    //     631147167 + 52,
+    //     ebml::matroska::MatroskaTag::Cues(Master::Start),
+    // )
+    // .unwrap();
+
+    while let Some(tag) = r.next() {
+        let tag = tag?;
+        w.write_tag(&tag)?;
+        // println!("{} {tag:?}", r.position)
+        // match tag {
+        //     tag @ MatroskaTag::SeekHead(_) => {
+        //         println!("{:?}", r.collect_master(tag));
+        //     }
+        //     _ => (),
+        // }
     }
+    Ok(())
 }
diff --git a/ebml/src/bin/mkvdump.rs b/ebml/src/bin/mkvdump.rs
new file mode 100644
index 0000000..ae621e6
--- /dev/null
+++ b/ebml/src/bin/mkvdump.rs
@@ -0,0 +1,14 @@
+use ebml::read::EbmlReader;
+use std::{fs::File, io::BufReader};
+
+fn main() -> anyhow::Result<()> {
+    env_logger::init_from_env("LOG");
+    let path = std::env::args().skip(1).next().unwrap();
+    let mut r = EbmlReader::new(BufReader::new(File::open(path)?));
+
+    while let Some(tag) = r.next() {
+        let tag = tag?;
+        println!("{} {tag:?}", r.position)
+    }
+    Ok(())
+}
diff --git a/ebml/src/lib.rs b/ebml/src/lib.rs
index 22af473..a03ce0d 100644
--- a/ebml/src/lib.rs
+++ b/ebml/src/lib.rs
@@ -1,77 +1,18 @@
-use anyhow::bail;
-
 pub mod matroska;
 pub mod read;
 pub mod write;
+pub mod size;
+
+use size::EbmlSize;
+
+pub use read::ReadValue;
+pub use write::WriteValue;
 
-#[derive(Debug, Clone)]
+#[derive(Debug, Clone, PartialEq, Eq)]
 pub enum Master {
-    Start,
+    Start(EbmlSize),
     End,
 }
 
-pub trait ValueFromBuf: Sized {
-    fn from_buf(buf: &[u8]) -> anyhow::Result<Self>;
-}
 
-impl ValueFromBuf for u64 {
-    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
-        if buf.len() > 8 {
-            bail!("u64 too big")
-        }
-        let mut val = 0u64;
-        for byte in buf {
-            val <<= 8;
-            val |= *byte as u64;
-        }
-        Ok(val)
-    }
-}
-impl ValueFromBuf for i64 {
-    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
-        if buf.len() > 8 {
-            bail!("i64 too big")
-        }
-        Ok(if buf[0] > 127 {
-            if buf.len() == 8 {
-                i64::from_be_bytes(buf.try_into().unwrap())
-            } else {
-                -((1 << (buf.len() * 8)) - (u64::from_buf(buf)? as i64))
-            }
-        } else {
-            u64::from_buf(buf)? as i64
-        })
-    }
-}
-impl ValueFromBuf for f64 {
-    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
-        Ok(if buf.len() == 4 {
-            f32::from_be_bytes(buf.try_into().unwrap()) as f64
-        } else if buf.len() == 8 {
-            f64::from_be_bytes(buf.try_into().unwrap())
-        } else {
-            bail!("float is not 4 or 8 bytes long");
-        })
-    }
-}
 
-impl ValueFromBuf for Vec<u8> {
-    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
-        Ok(buf.to_vec())
-    }
-}
-impl ValueFromBuf for String {
-    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
-        Ok(String::from_utf8(Vec::from(buf))?)
-    }
-}
-impl ValueFromBuf for Master {
-    /// this has an internal usage, where buf has len of 1, when an end should be constructed
-    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
-        Ok(match buf.len() {
-            0 => Master::Start,
-            1 => Master::End,
-            _ => unreachable!(),
-        })
-    }
-}
diff --git a/ebml/src/matroska.rs b/ebml/src/matroska.rs
index 9678e80..b4078ab 100644
--- a/ebml/src/matroska.rs
+++ b/ebml/src/matroska.rs
@@ -34,7 +34,7 @@ define_ebml! {
 
         Chapters[0x1043A770]: {
             EditionEntry[0x45B9]: {
-                ChapPterAtom[0xB6]: {
+                ChapterAtom[0xB6]: {
                     ChapProcess[0x6944]: {
                         ChapProcessCodecID[0x6955]: Uint,
                         ChapProcessCommand[0x6911]: {
diff --git a/ebml/src/read.rs b/ebml/src/read.rs
index 3e02e18..3853782 100644
--- a/ebml/src/read.rs
+++ b/ebml/src/read.rs
@@ -1,11 +1,11 @@
-use anyhow::{bail, Result};
-use log::warn;
+use anyhow::{anyhow, bail, Result};
+use log::{debug, warn};
 use std::{
     collections::VecDeque,
-    io::{Read, Seek},
+    io::{Read, Seek, SeekFrom},
 };
 
-use crate::matroska::MatroskaTag;
+use crate::{matroska::MatroskaTag, size::EbmlSize, Master};
 
 trait ReadAndSeek: Read + Seek {}
 impl<T: Read + Seek> ReadAndSeek for T {}
@@ -83,10 +83,11 @@ impl EbmlReader {
             if let Some(end) = e.end {
                 if self.position >= end {
                     if self.position != end {
-                        warn!("we missed the end")
+                        warn!("we missed the end by {} bytes", self.position - end)
                     }
                     self.stack.pop();
-                    self.queue.push_back(MatroskaTag::parse(e.id, &[0])?);
+                    self.queue
+                        .push_back(MatroskaTag::construct_master(e.id, Master::End)?);
                 }
             }
         }
@@ -95,7 +96,7 @@ impl EbmlReader {
         let size = self.read_tag_size()?;
         let is_master = MatroskaTag::is_master(id)?;
         let tag = if is_master {
-            MatroskaTag::parse(id, &[])?
+            MatroskaTag::construct_master(id, Master::Start(size))?
         } else {
             let data = self.read_buf(size)?;
             MatroskaTag::parse(id, &data)?
@@ -114,8 +115,11 @@ impl EbmlReader {
                 if valid_child {
                     break;
                 } else {
-                    self.queue
-                        .push_back(MatroskaTag::parse(self.stack.pop().unwrap().id, &[0])?);
+                    debug!("{id}");
+                    self.queue.push_back(MatroskaTag::construct_master(
+                        self.stack.pop().unwrap().id,
+                        Master::End,
+                    )?);
                 }
             }
         }
@@ -129,6 +133,39 @@ impl EbmlReader {
         self.queue.push_back(tag);
         Ok(())
     }
+
+    /// context should be the next expected tag, such that the stack can be derived from its path.
+    pub fn seek(&mut self, position: usize, context: MatroskaTag) -> Result<()> {
+        let path = context
+            .path()
+            .ok_or(anyhow!("global tags dont give context"))?;
+        debug!(
+            "seeking to {position} with a context restored from path {:x?}",
+            path
+        );
+        self.queue.clear();
+        self.position = position;
+        self.inner.seek(SeekFrom::Start(position as u64))?;
+        self.stack = path
+            .iter()
+            .map(|id| StackTag { id: *id, end: None })
+            .collect();
+        Ok(())
+    }
+
+    pub fn collect_master(&mut self, master: MatroskaTag) -> Result<Vec<MatroskaTag>> {
+        let end = MatroskaTag::construct_master(master.id(), Master::End)?;
+        let mut out = vec![];
+        while let Some(t) = self.next() {
+            let t = t?;
+            if t == end {
+                break;
+            } else {
+                out.push(t)
+            }
+        }
+        Ok(out)
+    }
 }
 
 impl Iterator for EbmlReader {
@@ -146,31 +183,63 @@ impl Iterator for EbmlReader {
     }
 }
 
-#[derive(Debug, Clone, Copy)]
-pub enum EbmlSize {
-    Exact(usize),
-    Unknown,
+pub trait ReadValue: Sized {
+    fn from_buf(buf: &[u8]) -> anyhow::Result<Self>;
 }
-impl EbmlSize {
-    pub fn from_vint((value, len): (u64, usize)) -> EbmlSize {
-        if value == ((1 << (7 * len)) - 1) {
-            Self::Unknown
-        } else {
-            Self::Exact(value as usize)
+
+impl ReadValue for u64 {
+    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
+        if buf.len() > 8 {
+            bail!("u64 too big")
         }
-    }
-    pub fn some(self) -> Option<usize> {
-        match self {
-            EbmlSize::Exact(s) => Some(s),
-            EbmlSize::Unknown => None,
+        let mut val = 0u64;
+        for byte in buf {
+            val <<= 8;
+            val |= *byte as u64;
         }
+        Ok(val)
     }
 }
-impl Into<usize> for EbmlSize {
-    fn into(self) -> usize {
-        match self {
-            EbmlSize::Exact(s) => s,
-            EbmlSize::Unknown => panic!("unknown size, where it should have been known"),
+impl ReadValue for i64 {
+    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
+        if buf.len() > 8 {
+            bail!("i64 too big")
         }
+        Ok(if buf[0] > 127 {
+            if buf.len() == 8 {
+                i64::from_be_bytes(buf.try_into().unwrap())
+            } else {
+                -((1 << (buf.len() * 8)) - (u64::from_buf(buf)? as i64))
+            }
+        } else {
+            u64::from_buf(buf)? as i64
+        })
+    }
+}
+impl ReadValue for f64 {
+    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
+        Ok(if buf.len() == 4 {
+            f32::from_be_bytes(buf.try_into().unwrap()) as f64
+        } else if buf.len() == 8 {
+            f64::from_be_bytes(buf.try_into().unwrap())
+        } else {
+            bail!("float is not 4 or 8 bytes long");
+        })
+    }
+}
+
+impl ReadValue for Vec<u8> {
+    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
+        Ok(buf.to_vec())
+    }
+}
+impl ReadValue for String {
+    fn from_buf(buf: &[u8]) -> anyhow::Result<Self> {
+        Ok(String::from_utf8(Vec::from(buf))?)
+    }
+}
+impl ReadValue for Master {
+    fn from_buf(_: &[u8]) -> anyhow::Result<Self> {
+        bail!("master cannot be parsed")
     }
 }
diff --git a/ebml/src/size.rs b/ebml/src/size.rs
new file mode 100644
index 0000000..06d6ba8
--- /dev/null
+++ b/ebml/src/size.rs
@@ -0,0 +1,29 @@
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum EbmlSize {
+    Exact(usize),
+    Unknown,
+}
+impl EbmlSize {
+    pub fn from_vint((value, len): (u64, usize)) -> EbmlSize {
+        if value == ((1 << (7 * len)) - 1) {
+            Self::Unknown
+        } else {
+            Self::Exact(value as usize)
+        }
+    }
+    pub fn some(self) -> Option<usize> {
+        match self {
+            EbmlSize::Exact(s) => Some(s),
+            EbmlSize::Unknown => None,
+        }
+    }
+}
+impl Into<usize> for EbmlSize {
+    fn into(self) -> usize {
+        match self {
+            EbmlSize::Exact(s) => s,
+            EbmlSize::Unknown => panic!("unknown size, where it should have been known"),
+        }
+    }
+}
diff --git a/ebml/src/write.rs b/ebml/src/write.rs
index b294d44..f57a5c9 100644
--- a/ebml/src/write.rs
+++ b/ebml/src/write.rs
@@ -1,14 +1,19 @@
 use anyhow::{bail, Result};
 use std::io::Write;
 
-pub struct EbmlWriter<T> {
-    inner: T,
+use crate::{matroska::MatroskaTag, size::EbmlSize, Master};
+
+pub struct EbmlWriter {
+    inner: Box<dyn Write>,
     position: usize,
 }
 
-impl<T: Write> EbmlWriter<T> {
-    pub fn new(inner: T, position: usize) -> Self {
-        Self { inner, position }
+impl EbmlWriter {
+    pub fn new<T: Write + 'static>(inner: T, position: usize) -> Self {
+        Self {
+            inner: Box::new(inner),
+            position,
+        }
     }
 
     pub fn write(&mut self, data: &[u8]) -> Result<()> {
@@ -17,6 +22,21 @@ impl<T: Write> EbmlWriter<T> {
         Ok(())
     }
 
+    pub fn write_tag(&mut self, tag: &MatroskaTag) -> Result<()> {
+        self.write_tag_id(tag.id())?;
+        let mut buf = vec![];
+        tag.write(&mut buf)?;
+        self.write(&buf)?;
+        Ok(())
+    }
+
+    pub fn write_tag_id(&mut self, id: u64) -> Result<()> {
+        for n in id.to_be_bytes().iter().skip_while(|&v| *v == 0u8) {
+            self.write(&[*n])?;
+        }
+        Ok(())
+    }
+
     pub fn write_vint(&mut self, i: u64) -> Result<()> {
         if i > (1 << 56) - 1 {
             bail!("vint does not fit");
@@ -29,8 +49,115 @@ impl<T: Write> EbmlWriter<T> {
             len += 1;
         }
         let mut bytes = i.to_be_bytes();
-        bytes[0] |= 1 << (8 - len);
-        self.write(&bytes)
+        let trunc = &mut bytes[(8 - len)..];
+        trunc[0] |= 1 << (8 - len);
+        self.write(&trunc)
     }
 }
 
+pub trait WriteValue {
+    fn write_to(&self, w: &mut Vec<u8>) -> Result<()>;
+}
+
+impl WriteValue for i64 {
+    fn write_to(&self, w: &mut Vec<u8>) -> Result<()> {
+        Ok(match 64 - self.leading_zeros() {
+            x if x <= 8 => {
+                w.push(0x81);
+                w.extend_from_slice(&(*self as i8).to_be_bytes());
+            }
+            x if x <= 16 => {
+                w.push(0x82);
+                w.extend_from_slice(&(*self as i16).to_be_bytes());
+            }
+            x if x <= 32 => {
+                w.push(0x84);
+                w.extend_from_slice(&(*self as i32).to_be_bytes());
+            }
+            _ => {
+                w.push(0x88);
+                w.extend_from_slice(&self.to_be_bytes());
+            }
+        })
+    }
+}
+impl WriteValue for u64 {
+    fn write_to(&self, w: &mut Vec<u8>) -> Result<()> {
+        Ok(match 64 - self.leading_zeros() {
+            x if x <= 8 => {
+                w.push(0x81);
+                w.extend_from_slice(&(*self as u8).to_be_bytes());
+            }
+            x if x <= 16 => {
+                w.push(0x82);
+                w.extend_from_slice(&(*self as u16).to_be_bytes());
+            }
+            x if x <= 32 => {
+                w.push(0x84);
+                w.extend_from_slice(&(*self as u32).to_be_bytes());
+            }
+            _ => {
+                w.push(0x88);
+                w.extend_from_slice(&self.to_be_bytes());
+            }
+        })
+    }
+}
+impl WriteValue for f64 {
+    fn write_to(&self, w: &mut Vec<u8>) -> Result<(), anyhow::Error> {
+        w.push(0x88);
+        w.extend_from_slice(&self.to_be_bytes());
+        Ok(())
+    }
+}
+impl WriteValue for Vec<u8> {
+    fn write_to(&self, w: &mut Vec<u8>) -> Result<(), anyhow::Error> {
+        write_vint(w, self.len() as u64)?;
+        w.extend_from_slice(&self);
+        Ok(())
+    }
+}
+impl WriteValue for String {
+    fn write_to(&self, w: &mut Vec<u8>) -> Result<(), anyhow::Error> {
+        let sl = self.as_bytes();
+        write_vint(w, sl.len() as u64)?;
+        w.extend_from_slice(sl);
+        Ok(())
+    }
+}
+impl WriteValue for EbmlSize {
+    fn write_to(&self, w: &mut Vec<u8>) -> Result<()> {
+        match self {
+            EbmlSize::Exact(s) => write_vint(w, *s as u64)?,
+            EbmlSize::Unknown => w.extend_from_slice(&(u64::MAX >> 7).to_be_bytes()),
+        }
+        Ok(())
+    }
+}
+
+impl WriteValue for Master {
+    fn write_to(&self, w: &mut Vec<u8>) -> Result<()> {
+        match self {
+            Master::Start(size) => size.write_to(w),
+            Master::End => Ok(()),
+        }
+    }
+}
+
+pub fn write_vint(w: &mut Vec<u8>, i: u64) -> Result<()> {
+    if i > (1 << 56) - 1 {
+        bail!("vint does not fit");
+    }
+    let mut len = 1;
+    while len <= 8 {
+        if i < (1 << ((7 * len) - 1)) {
+            break;
+        }
+        len += 1;
+    }
+    let mut bytes = i.to_be_bytes();
+    let trunc = &mut bytes[(8 - len)..];
+    trunc[0] |= 1 << (8 - len);
+    w.extend_from_slice(&trunc);
+    Ok(())
+}