fix(event)!: use typed-path for filesystem paths

This should make them work consistently on Windows,
where e.g. PT_FSPATH still describes a Unix-style path.

Signed-off-by: Grzegorz Nosek <[email protected]>

Author: gnosek

Cargo.lock

@@ -24,6 +24,7 @@ bitflags = { version = "2.4.2" }
 anyhow = "1.0.81"
 chrono = "0.4.38"
 serde = { version = "1.0.210", features = ["derive"], optional = true }
+typed-path = "0.9.3"
 
 [target.'cfg(target_os = "linux")'.dependencies]
 nix = { version = "0.29.0", features = ["signal"] }

falco_event/Cargo.toml

@@ -24,9 +24,9 @@ use std::ffi::CStr;
 pub use std::net::IpAddr as PT_IPADDR;
 pub use std::net::Ipv4Addr as PT_IPV4ADDR;
 pub use std::net::Ipv6Addr as PT_IPV6ADDR;
-pub use std::path::Path as PT_FSPATH;
 pub use std::time::Duration as PT_RELTIME;
 pub use std::time::SystemTime as PT_ABSTIME;
+pub use typed_path::UnixPath as PT_FSPATH;
 
 /// Signed 8-bit value ([i8])
 pub type PT_INT8 = i8;
@@ -97,7 +97,7 @@ pub mod owned {
     pub use crate::types::OwnedSockTuple as PT_SOCKTUPLE;
     pub use std::ffi::CString as PT_CHARBUF;
     use std::ffi::CString;
-    pub use std::path::PathBuf as PT_FSPATH;
+    pub use typed_path::UnixPathBuf as PT_FSPATH;
 
     /// Arbitrary (owned) byte buffer (`Vec<u8>`)
     pub type PT_BYTEBUF = Vec<u8>;

falco_event/src/fields/types.rs

@@ -12,6 +12,7 @@ mod utf_chunked;
 #[cfg(feature = "serde")]
 pub mod serde {
     pub use super::bytebuf::serde::*;
+    pub use super::path::serde::*;
     pub use super::string::serde::*;
 }
 

falco_event/src/types/mod.rs

@@ -3,19 +3,19 @@ use crate::ffi::{PPM_AF_INET, PPM_AF_INET6, PPM_AF_LOCAL, PPM_AF_UNSPEC};
 use crate::types::format::Format;
 use crate::types::{Borrow, Borrowed, EndpointV4, EndpointV6};
 use byteorder::{ReadBytesExt, WriteBytesExt};
-use std::ffi::OsStr;
 use std::fmt::Formatter;
 use std::io::Write;
-use std::os::unix::ffi::OsStrExt;
-use std::path::{Path, PathBuf};
+use typed_path::{UnixPath, UnixPathBuf};
 
 /// A socket address
 #[cfg_attr(feature = "serde", derive(serde::Serialize))]
 #[cfg_attr(feature = "serde", serde(rename_all = "lowercase"))]
 #[derive(Debug)]
 pub enum SockAddr<'a> {
     /// Unix sockets
-    Unix(&'a Path),
+    Unix(
+        #[cfg_attr(feature = "serde", serde(with = "crate::types::serde::unix_path"))] &'a UnixPath,
+    ),
 
     /// IPv4 sockets
     V4(EndpointV4),
@@ -71,9 +71,9 @@ impl<'a> FromBytes<'a> for SockAddr<'a> {
         let variant = buf.read_u8()?;
         match variant as u32 {
             PPM_AF_LOCAL => {
-                let path = <OsStr as OsStrExt>::from_bytes(buf);
-                *buf = &[];
-                Ok(Self::Unix(Path::new(path)))
+                // TODO embedded NULs
+                let path = std::mem::take(buf);
+                Ok(Self::Unix(UnixPath::new(path)))
             }
             PPM_AF_INET => {
                 let addr = EndpointV4::from_bytes(buf)?;
@@ -97,7 +97,7 @@ where
     fn format(&self, fmt: &mut Formatter) -> std::fmt::Result {
         match self {
             SockAddr::Unix(u) => {
-                let bytes = u.as_os_str().as_bytes();
+                let bytes = u.as_bytes();
                 fmt.write_str("unix://")?;
                 bytes.format(fmt)
             }
@@ -114,7 +114,9 @@ where
 #[derive(Debug)]
 pub enum OwnedSockAddr {
     /// Unix sockets
-    Unix(PathBuf),
+    Unix(
+        #[cfg_attr(feature = "serde", serde(with = "crate::types::serde::unix_path"))] UnixPathBuf,
+    ),
 
     /// IPv4 sockets
     V4(EndpointV4),
@@ -150,11 +152,11 @@ impl Borrow for OwnedSockAddr {
 mod tests {
     use crate::types::{OwnedSockAddr, Port, SockAddr};
     use std::net::{Ipv4Addr, Ipv6Addr};
-    use std::path::Path;
+    use typed_path::UnixPath;
 
     #[test]
     fn test_serde_sockaddr_unix() {
-        let path = Path::new("/path/to/unix");
+        let path = UnixPath::new("/path/to/unix");
         let sockaddr = SockAddr::Unix(path);
 
         let json = serde_json::to_string(&sockaddr).unwrap();

falco_event/src/types/net/sockaddr.rs

@@ -1,13 +1,13 @@
 use std::fmt::{Debug, Display, Formatter};
 use std::io::Write;
-use std::path::{Path, PathBuf};
 
 use crate::ffi::{PPM_AF_INET, PPM_AF_INET6, PPM_AF_LOCAL};
 use crate::fields::{FromBytes, FromBytesResult, ToBytes};
 use crate::types::format::Format;
 use crate::types::net::endpoint::{EndpointV4, EndpointV6};
 use crate::types::{Borrow, Borrowed};
 use byteorder::{ReadBytesExt, WriteBytesExt};
+use typed_path::{UnixPath, UnixPathBuf};
 
 /// Socket tuple: describing both endpoints of a connection
 #[cfg_attr(feature = "serde", derive(serde::Serialize))]
@@ -21,7 +21,8 @@ pub enum SockTuple<'a> {
         /// destination socket kernel pointer
         dest_ptr: u64,
         /// filesystem path to the socket
-        path: &'a Path,
+        #[cfg_attr(feature = "serde", serde(with = "crate::types::serde::unix_path"))]
+        path: &'a UnixPath,
     },
 
     /// IPv4 connection
@@ -146,7 +147,7 @@ impl<'a> FromBytes<'a> for SockTuple<'a> {
 
 impl<F> Format<F> for SockTuple<'_>
 where
-    for<'a> &'a Path: Format<F>,
+    for<'a> &'a UnixPath: Format<F>,
     EndpointV4: Format<F>,
     EndpointV6: Format<F>,
 {
@@ -187,7 +188,8 @@ pub enum OwnedSockTuple {
         /// destination socket kernel pointer
         dest_ptr: u64,
         /// filesystem path to the socket
-        path: PathBuf,
+        #[cfg_attr(feature = "serde", serde(with = "crate::types::serde::unix_path"))]
+        path: UnixPathBuf,
     },
 
     /// IPv4 connection
@@ -249,7 +251,6 @@ mod tests {
     use super::*;
     use crate::types::Port;
     use std::net::{Ipv4Addr, Ipv6Addr};
-    use std::os::unix::ffi::OsStrExt;
     use std::str::FromStr;
 
     #[test]
@@ -324,10 +325,7 @@ mod tests {
 
         assert_eq!(source_addr, 0);
         assert_eq!(dest_addr, 0xffff98bc4ecf2000);
-        assert_eq!(
-            path.as_os_str().as_bytes(),
-            b"/var/run/nscd/socket".as_slice()
-        );
+        assert_eq!(path.as_bytes(), b"/var/run/nscd/socket".as_slice());
 
         assert_eq!(binary, binary2.as_slice(),);
     }
@@ -337,12 +335,12 @@ mod tests {
 mod serde_tests {
     use crate::types::{OwnedSockTuple, Port, SockTuple};
     use std::net::{Ipv4Addr, Ipv6Addr};
-    use std::path::Path;
     use std::str::FromStr;
+    use typed_path::UnixPath;
 
     #[test]
     fn test_serde_socktuple_unix() {
-        let path = Path::new("/path/to/unix");
+        let path = UnixPath::new("/path/to/unix");
         let sockaddr = SockTuple::Unix {
             source_ptr: 1,
             dest_ptr: 2,

falco_event/src/types/net/socktuple.rs

@@ -1,28 +1,25 @@
-use std::ffi::{CStr, OsStr};
-use std::fmt::Formatter;
-use std::io::Write;
-use std::os::unix::ffi::OsStrExt;
-use std::path::{Path, PathBuf};
-
 use crate::event_derive::{FromBytes, FromBytesResult, ToBytes};
 use crate::types::format::Format;
 use crate::types::{Borrow, Borrowed};
+use std::ffi::CStr;
+use std::fmt::Formatter;
+use std::io::Write;
+use typed_path::{UnixPath, UnixPathBuf};
 
-impl<'a> FromBytes<'a> for &'a Path {
+impl<'a> FromBytes<'a> for &'a UnixPath {
     fn from_bytes(buf: &mut &'a [u8]) -> FromBytesResult<Self> {
         let buf = <&CStr>::from_bytes(buf)?;
-        let osstr = OsStr::from_bytes(buf.to_bytes());
-        Ok(Path::new(osstr))
+        Ok(UnixPath::new(buf.to_bytes()))
     }
 }
 
-impl ToBytes for &Path {
+impl ToBytes for &UnixPath {
     fn binary_size(&self) -> usize {
-        self.as_os_str().len()
+        self.as_bytes().len() + 1
     }
 
     fn write<W: Write>(&self, mut writer: W) -> std::io::Result<()> {
-        self.as_os_str().as_bytes().write(&mut writer)?;
+        self.as_bytes().write(&mut writer)?;
         0u8.write(writer)
     }
 
@@ -31,59 +28,127 @@ impl ToBytes for &Path {
     }
 }
 
-impl<F> Format<F> for &Path
+impl<F> Format<F> for &UnixPath
 where
     for<'a> &'a [u8]: Format<F>,
 {
     fn format(&self, fmt: &mut Formatter) -> std::fmt::Result {
-        let bytes = self.as_os_str().as_bytes();
+        let bytes = self.as_bytes();
         bytes.format(fmt)
     }
 }
 
-impl Borrowed for Path {
-    type Owned = PathBuf;
+impl Borrowed for UnixPath {
+    type Owned = UnixPathBuf;
 }
 
-impl Borrow for PathBuf {
-    type Borrowed<'a> = &'a Path;
+impl Borrow for UnixPathBuf {
+    type Borrowed<'a> = &'a UnixPath;
 
     fn borrow(&self) -> Self::Borrowed<'_> {
         self.as_path()
     }
 }
 
+#[cfg(feature = "serde")]
+pub mod serde {
+    pub mod unix_path {
+        use crate::types::utf_chunked::{OwnedUtfChunked, UtfChunked};
+        use serde::{Deserialize, Deserializer, Serialize, Serializer};
+        use typed_path::{UnixPath, UnixPathBuf};
+
+        pub fn serialize<S: Serializer>(path: &UnixPath, ser: S) -> Result<S::Ok, S::Error> {
+            if ser.is_human_readable() {
+                let chunks = UtfChunked::from(path.as_bytes());
+                chunks.serialize(ser)
+            } else {
+                path.as_bytes().serialize(ser)
+            }
+        }
+
+        pub fn deserialize<'de, D: Deserializer<'de>>(de: D) -> Result<UnixPathBuf, D::Error> {
+            let chunks: OwnedUtfChunked = Deserialize::deserialize(de)?;
+            Ok(UnixPathBuf::from(chunks.into_vec()))
+        }
+    }
+
+    pub mod unix_path_option {
+        use crate::types::utf_chunked::UtfChunked;
+        use serde::{Serialize, Serializer};
+        use typed_path::UnixPath;
+
+        pub fn serialize<S: Serializer>(
+            path: &Option<&UnixPath>,
+            ser: S,
+        ) -> Result<S::Ok, S::Error> {
+            if ser.is_human_readable() {
+                let chunks = path.as_ref().map(|path| UtfChunked::from(path.as_bytes()));
+                chunks.serialize(ser)
+            } else {
+                path.map(|p| p.as_bytes()).serialize(ser)
+            }
+        }
+    }
+
+    pub mod unix_path_option_owned {
+        use crate::types::utf_chunked::{OwnedUtfChunked, UtfChunked};
+        use serde::{Deserialize, Deserializer, Serialize, Serializer};
+        use typed_path::UnixPathBuf;
+
+        pub fn serialize<S: Serializer>(
+            path: &Option<UnixPathBuf>,
+            ser: S,
+        ) -> Result<S::Ok, S::Error> {
+            let chunks = path.as_ref().map(|path| UtfChunked::from(path.as_bytes()));
+            chunks.serialize(ser)
+        }
+
+        pub fn deserialize<'de, D: Deserializer<'de>>(
+            de: D,
+        ) -> Result<Option<UnixPathBuf>, D::Error> {
+            let chunks: Option<OwnedUtfChunked> = Deserialize::deserialize(de)?;
+            Ok(chunks.map(|c| UnixPathBuf::from(c.into_vec())))
+        }
+    }
+}
+
 #[cfg(test)]
 mod tests {
-    use std::path::{Path, PathBuf};
-    use std::str::FromStr;
-
     use crate::event_derive::{FromBytes, ToBytes};
+    use std::str::FromStr;
+    use typed_path::{UnixPath, UnixPathBuf};
 
     #[test]
     fn test_absolute_path() {
-        let path = PathBuf::from_str("/foo").unwrap();
+        let path = UnixPathBuf::from_str("/foo").unwrap();
         let mut binary = Vec::new();
 
+        assert_eq!(path.as_path().binary_size(), 5);
+
         path.as_path().write(&mut binary).unwrap();
         hexdump::hexdump(binary.as_slice());
 
         assert_eq!(binary.as_slice(), "/foo\0".as_bytes());
 
         let mut buf = binary.as_slice();
-        let path = <&Path>::from_bytes(&mut buf).unwrap();
+        let path = <&UnixPath>::from_bytes(&mut buf).unwrap();
         assert_eq!(path.to_str().unwrap(), "/foo");
     }
 
+    #[cfg(feature = "serde")]
     #[test]
     fn test_serde_absolute_path() {
-        let path = Path::new("/foo");
+        #[derive(serde::Deserialize, serde::Serialize)]
+        #[serde(transparent)]
+        struct SerPathBuf(#[serde(with = "super::serde::unix_path")] UnixPathBuf);
+
+        let path = SerPathBuf(UnixPathBuf::from("/foo"));
 
         let json = serde_json::to_string(&path).unwrap();
         assert_eq!(json, "\"/foo\"");
 
-        let path2: PathBuf = serde_json::from_str(&json).unwrap();
-        assert_eq!(path2, path);
+        let path2: SerPathBuf = serde_json::from_str(&json).unwrap();
+        assert_eq!(path2.0, path.0);
 
         let json2 = serde_json::to_string(&path2).unwrap();
         assert_eq!(json, json2);