mkv: detect parent overrun after the header parsing

[sscobici]

Declaring a constant for EbmlElementHeader doesn't allow to use EbmlElementHeader::MIN_SIZE without specifying the scheme which I considered too complex for a constant. I can define it globally in the file or inside the Schema if you want. Currently it is a 2 with a comment

I added skipping of a byte to put reader position to the end of parent. Does it make sense, taking into consideration that it will be skipped later anyway?

For element header seeking I propose to rewind the stream to the parent_end vs header.pos, which should be similar to as everything was ok. Let me know your preference.

[pdeljanov]

Declaring a constant for EbmlElementHeader doesn't allow to use EbmlElementHeader::MIN_SIZE without specifying the scheme which I considered too complex for a constant. I can define it globally in the file or inside the Schema if you want. Currently it is a 2 with a comment

This should be pretty simple since everything is already parameterized by a schema, S. The min. size is really a property of EBML, and not the schema, however, we don't have any lower abstraction for an EBML header other than EbmlElementHeader<S>.

Did you try this?

impl<S: EbmlSchema> EbmlElementHeader<S> {
    /// Minimum EBML header size.
    pub const MIN_SIZE: u64 = 2;
    // ...

if parent_end - pos < EbmlElementHeader::<S>::MIN_SIZE {
    // ...

I added skipping of a byte to put reader position to the end of parent. Does it make sense, taking into consideration that it will be skipped later anyway?

The skip later will yield a debug message which is useful for indicating a strangely muxed file. I wouldn't call ignore_bytes so we can see this message.

For element header seeking I propose to rewind the stream to the parent_end vs header.pos, which should be similar to as everything was ok. Let me know your preference.

This is okay.

[pdeljanov]

Thanks again!