tvl-depot/tvix/serde/src/de.rs

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

476 lines
13 KiB
Rust
Raw Normal View History

//! Deserialisation from Nix to Rust values.
fix(tvix): Represent strings as byte arrays C++ nix uses C-style zero-terminated char pointers to represent strings internally - however, up to this point, tvix has used Rust `String` and `str` for string values. Since those are required to be valid utf-8, we haven't been able to properly represent all the string values that Nix supports. To fix that, this change converts the internal representation of the NixString struct from `Box<str>` to `BString`, from the `bstr` crate - this is a wrapper around a `Vec<u8>` with extra functions for treating that byte vector as a "morally string-like" value, which is basically exactly what we need. Since this changes a pretty fundamental assumption about a pretty core type, there are a *lot* of changes in a lot of places to make this work, but I've tried to keep the general philosophy and intent of most of the code in most places intact. Most notably, there's nothing that's been done to make the derivation stuff in //tvix/glue work with non-utf8 strings everywhere, instead opting to just convert to String/str when passing things into that - there *might* be something to be done there, but I don't know what the rules should be and I don't want to figure them out in this change. To deal with OS-native paths in a way that also works in WASM for tvixbolt, this also adds a dependency on the "os_str_bytes" crate. Fixes: b/189 Fixes: b/337 Change-Id: I5e6eb29c62f47dd91af954f5e12bfc3d186f5526 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10200 Reviewed-by: tazjin <tazjin@tvl.su> Reviewed-by: flokli <flokli@flokli.de> Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: aspen <root@gws.fyi> Tested-by: BuildkiteCI
2023-12-05 17:25:52 -05:00
use bstr::ByteSlice;
use serde::de::value::{MapDeserializer, SeqDeserializer};
use serde::de::{self, EnumAccess, VariantAccess};
pub use tvix_eval::Evaluation;
use tvix_eval::{EvalIO, Value};
use crate::error::Error;
struct NixDeserializer {
value: tvix_eval::Value,
}
impl NixDeserializer {
fn new(value: Value) -> Self {
if let Value::Thunk(thunk) = value {
Self::new(thunk.value().clone())
} else {
Self { value }
}
}
}
impl de::IntoDeserializer<'_, Error> for NixDeserializer {
type Deserializer = Self;
fn into_deserializer(self) -> Self::Deserializer {
self
}
}
/// Evaluate the Nix code in `src` and attempt to deserialise the
/// value it returns to `T`.
pub fn from_str<'code, T>(src: &'code str) -> Result<T, Error>
where
T: serde::Deserialize<'code>,
{
from_str_with_config(src, |_| /* no extra config */ ())
}
/// Evaluate the Nix code in `src`, with extra configuration for the
/// `tvix_eval::Evaluation` provided by the given closure.
pub fn from_str_with_config<'code, T, F>(src: &'code str, config: F) -> Result<T, Error>
where
T: serde::Deserialize<'code>,
F: FnOnce(&mut Evaluation<Box<dyn EvalIO>>),
{
// First step is to evaluate the Nix code ...
let mut eval = Evaluation::new_pure();
config(&mut eval);
eval.strict = true;
let result = eval.evaluate(src, None);
if !result.errors.is_empty() {
return Err(Error::NixErrors {
errors: result.errors,
});
}
let de = NixDeserializer::new(result.value.expect("value should be present on success"));
T::deserialize(de)
}
fn unexpected(expected: &'static str, got: &Value) -> Error {
Error::UnexpectedType {
expected,
got: got.type_of(),
}
}
fn visit_integer<I: TryFrom<i64>>(v: &Value) -> Result<I, Error> {
match v {
Value::Integer(i) => I::try_from(*i).map_err(|_| Error::IntegerConversion {
got: *i,
need: std::any::type_name::<I>(),
}),
_ => Err(unexpected("integer", v)),
}
}
impl<'de> de::Deserializer<'de> for NixDeserializer {
type Error = Error;
fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
match self.value {
Value::Null => visitor.visit_unit(),
Value::Bool(b) => visitor.visit_bool(b),
Value::Integer(i) => visitor.visit_i64(i),
Value::Float(f) => visitor.visit_f64(f),
Value::String(s) => visitor.visit_string(s.to_string()),
Value::Path(p) => visitor.visit_string(p.to_string_lossy().into()), // TODO: hmm
Value::Attrs(_) => self.deserialize_map(visitor),
Value::List(_) => self.deserialize_seq(visitor),
// tvix-eval types that can not be deserialized through serde.
Value::Closure(_)
| Value::Builtin(_)
| Value::Thunk(_)
| Value::AttrNotFound
| Value::Blueprint(_)
| Value::DeferredUpvalue(_)
| Value::UnresolvedPath(_)
| Value::Json(..)
| Value::Catchable(_)
fix(tvix/eval): only finalise formal arguments if defaulting When dealing with a formal argument in a function argument pattern that has a default expression, there are two different things that can happen at runtime: Either we select its value from the passed attribute successfully or we need to use the default expression. Both of these may be thunks and both of these may need finalisers. However, in the former case this is taken care of elsewhere, the value will always be finalised already if necessary. In the latter case we may need to finalise the thunk resulting from the default expression. However, the thunk corresponding to the expression may never end up in the local's stack slot. Since finalisation goes by stack slot (and not constants), we need to prevent a case where we don't fall back to the default expression, but finalise anyways. Previously, we worked around this by making `OpFinalise` ignore non-thunks. Since finalisation of already evaluated thunks still crashed, the faulty compilation of function pattern arguments could still cause a crash. As a new approach, we reinstate the old behavior of `OpFinalise` to crash whenever encountering something that is either not a thunk or doesn't need finalisation. This can also help catching (similar) miscompilations in the future. To then prevent the crash, we need to track whether we have fallen back or not at runtime. This is done using an additional phantom on the stack that holds a new `FinaliseRequest` value. When it comes to finalisation we check this value and conditionally execute `OpFinalise` based on its value. Resolves b/261 and b/265 (partially). Change-Id: Ic04fb80ec671a2ba11fa645090769c335fb7f58b Reviewed-on: https://cl.tvl.fyi/c/depot/+/8705 Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI Autosubmit: sterni <sternenseemann@systemli.org>
2023-06-03 02:10:31 +02:00
| Value::FinaliseRequest(_) => Err(Error::Unserializable {
value_type: self.value.type_of(),
}),
}
}
fn deserialize_bool<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
match self.value {
Value::Bool(b) => visitor.visit_bool(b),
_ => Err(unexpected("bool", &self.value)),
}
}
fn deserialize_i8<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_i8(visit_integer(&self.value)?)
}
fn deserialize_i16<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_i16(visit_integer(&self.value)?)
}
fn deserialize_i32<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_i32(visit_integer(&self.value)?)
}
fn deserialize_i64<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_i64(visit_integer(&self.value)?)
}
fn deserialize_u8<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_u8(visit_integer(&self.value)?)
}
fn deserialize_u16<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_u16(visit_integer(&self.value)?)
}
fn deserialize_u32<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_u32(visit_integer(&self.value)?)
}
fn deserialize_u64<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_u64(visit_integer(&self.value)?)
}
fn deserialize_f32<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
if let Value::Float(f) = self.value {
return visitor.visit_f32(f as f32);
}
Err(unexpected("float", &self.value))
}
fn deserialize_f64<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
if let Value::Float(f) = self.value {
return visitor.visit_f64(f);
}
Err(unexpected("float", &self.value))
}
fn deserialize_char<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
if let Value::String(s) = &self.value {
fix(tvix): Represent strings as byte arrays C++ nix uses C-style zero-terminated char pointers to represent strings internally - however, up to this point, tvix has used Rust `String` and `str` for string values. Since those are required to be valid utf-8, we haven't been able to properly represent all the string values that Nix supports. To fix that, this change converts the internal representation of the NixString struct from `Box<str>` to `BString`, from the `bstr` crate - this is a wrapper around a `Vec<u8>` with extra functions for treating that byte vector as a "morally string-like" value, which is basically exactly what we need. Since this changes a pretty fundamental assumption about a pretty core type, there are a *lot* of changes in a lot of places to make this work, but I've tried to keep the general philosophy and intent of most of the code in most places intact. Most notably, there's nothing that's been done to make the derivation stuff in //tvix/glue work with non-utf8 strings everywhere, instead opting to just convert to String/str when passing things into that - there *might* be something to be done there, but I don't know what the rules should be and I don't want to figure them out in this change. To deal with OS-native paths in a way that also works in WASM for tvixbolt, this also adds a dependency on the "os_str_bytes" crate. Fixes: b/189 Fixes: b/337 Change-Id: I5e6eb29c62f47dd91af954f5e12bfc3d186f5526 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10200 Reviewed-by: tazjin <tazjin@tvl.su> Reviewed-by: flokli <flokli@flokli.de> Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: aspen <root@gws.fyi> Tested-by: BuildkiteCI
2023-12-05 17:25:52 -05:00
let chars = s.chars().collect::<Vec<_>>();
if chars.len() == 1 {
return visitor.visit_char(chars[0]);
}
}
Err(unexpected("char", &self.value))
}
fn deserialize_str<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
if let Value::String(s) = &self.value {
fix(tvix): Represent strings as byte arrays C++ nix uses C-style zero-terminated char pointers to represent strings internally - however, up to this point, tvix has used Rust `String` and `str` for string values. Since those are required to be valid utf-8, we haven't been able to properly represent all the string values that Nix supports. To fix that, this change converts the internal representation of the NixString struct from `Box<str>` to `BString`, from the `bstr` crate - this is a wrapper around a `Vec<u8>` with extra functions for treating that byte vector as a "morally string-like" value, which is basically exactly what we need. Since this changes a pretty fundamental assumption about a pretty core type, there are a *lot* of changes in a lot of places to make this work, but I've tried to keep the general philosophy and intent of most of the code in most places intact. Most notably, there's nothing that's been done to make the derivation stuff in //tvix/glue work with non-utf8 strings everywhere, instead opting to just convert to String/str when passing things into that - there *might* be something to be done there, but I don't know what the rules should be and I don't want to figure them out in this change. To deal with OS-native paths in a way that also works in WASM for tvixbolt, this also adds a dependency on the "os_str_bytes" crate. Fixes: b/189 Fixes: b/337 Change-Id: I5e6eb29c62f47dd91af954f5e12bfc3d186f5526 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10200 Reviewed-by: tazjin <tazjin@tvl.su> Reviewed-by: flokli <flokli@flokli.de> Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: aspen <root@gws.fyi> Tested-by: BuildkiteCI
2023-12-05 17:25:52 -05:00
if let Ok(s) = s.to_str() {
return visitor.visit_str(s);
}
}
Err(unexpected("string", &self.value))
}
fn deserialize_string<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
if let Value::String(s) = &self.value {
fix(tvix): Represent strings as byte arrays C++ nix uses C-style zero-terminated char pointers to represent strings internally - however, up to this point, tvix has used Rust `String` and `str` for string values. Since those are required to be valid utf-8, we haven't been able to properly represent all the string values that Nix supports. To fix that, this change converts the internal representation of the NixString struct from `Box<str>` to `BString`, from the `bstr` crate - this is a wrapper around a `Vec<u8>` with extra functions for treating that byte vector as a "morally string-like" value, which is basically exactly what we need. Since this changes a pretty fundamental assumption about a pretty core type, there are a *lot* of changes in a lot of places to make this work, but I've tried to keep the general philosophy and intent of most of the code in most places intact. Most notably, there's nothing that's been done to make the derivation stuff in //tvix/glue work with non-utf8 strings everywhere, instead opting to just convert to String/str when passing things into that - there *might* be something to be done there, but I don't know what the rules should be and I don't want to figure them out in this change. To deal with OS-native paths in a way that also works in WASM for tvixbolt, this also adds a dependency on the "os_str_bytes" crate. Fixes: b/189 Fixes: b/337 Change-Id: I5e6eb29c62f47dd91af954f5e12bfc3d186f5526 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10200 Reviewed-by: tazjin <tazjin@tvl.su> Reviewed-by: flokli <flokli@flokli.de> Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: aspen <root@gws.fyi> Tested-by: BuildkiteCI
2023-12-05 17:25:52 -05:00
if let Ok(s) = s.to_str() {
return visitor.visit_str(s);
}
}
Err(unexpected("string", &self.value))
}
fn deserialize_bytes<V>(self, _visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
unimplemented!()
}
fn deserialize_byte_buf<V>(self, _visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
unimplemented!()
}
// Note that this can not distinguish between a serialisation of
// `Some(())` and `None`.
fn deserialize_option<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
if let Value::Null = self.value {
visitor.visit_none()
} else {
visitor.visit_some(self)
}
}
fn deserialize_unit<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
if let Value::Null = self.value {
return visitor.visit_unit();
}
Err(unexpected("null", &self.value))
}
fn deserialize_unit_struct<V>(
self,
_name: &'static str,
visitor: V,
) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
self.deserialize_unit(visitor)
}
fn deserialize_newtype_struct<V>(
self,
_name: &'static str,
visitor: V,
) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_newtype_struct(self)
}
fn deserialize_seq<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
if let Value::List(list) = self.value {
let mut seq = SeqDeserializer::new(list.into_iter().map(NixDeserializer::new));
let result = visitor.visit_seq(&mut seq)?;
seq.end()?;
return Ok(result);
}
Err(unexpected("list", &self.value))
}
fn deserialize_tuple<V>(self, _len: usize, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
// just represent tuples as lists ...
self.deserialize_seq(visitor)
}
fn deserialize_tuple_struct<V>(
self,
_name: &'static str,
_len: usize,
visitor: V,
) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
// same as above
self.deserialize_seq(visitor)
}
fn deserialize_map<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
if let Value::Attrs(attrs) = self.value {
let mut map = MapDeserializer::new(attrs.into_iter().map(|(k, v)| {
(
NixDeserializer::new(Value::from(k)),
NixDeserializer::new(v),
)
}));
let result = visitor.visit_map(&mut map)?;
map.end()?;
return Ok(result);
}
Err(unexpected("map", &self.value))
}
fn deserialize_struct<V>(
self,
_name: &'static str,
_fields: &'static [&'static str],
visitor: V,
) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
self.deserialize_map(visitor)
}
// This method is responsible for deserializing the externally
// tagged enum variant serialisation.
fn deserialize_enum<V>(
self,
name: &'static str,
_variants: &'static [&'static str],
visitor: V,
) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
match self.value {
// a string represents a unit variant
fix(tvix): Represent strings as byte arrays C++ nix uses C-style zero-terminated char pointers to represent strings internally - however, up to this point, tvix has used Rust `String` and `str` for string values. Since those are required to be valid utf-8, we haven't been able to properly represent all the string values that Nix supports. To fix that, this change converts the internal representation of the NixString struct from `Box<str>` to `BString`, from the `bstr` crate - this is a wrapper around a `Vec<u8>` with extra functions for treating that byte vector as a "morally string-like" value, which is basically exactly what we need. Since this changes a pretty fundamental assumption about a pretty core type, there are a *lot* of changes in a lot of places to make this work, but I've tried to keep the general philosophy and intent of most of the code in most places intact. Most notably, there's nothing that's been done to make the derivation stuff in //tvix/glue work with non-utf8 strings everywhere, instead opting to just convert to String/str when passing things into that - there *might* be something to be done there, but I don't know what the rules should be and I don't want to figure them out in this change. To deal with OS-native paths in a way that also works in WASM for tvixbolt, this also adds a dependency on the "os_str_bytes" crate. Fixes: b/189 Fixes: b/337 Change-Id: I5e6eb29c62f47dd91af954f5e12bfc3d186f5526 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10200 Reviewed-by: tazjin <tazjin@tvl.su> Reviewed-by: flokli <flokli@flokli.de> Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: aspen <root@gws.fyi> Tested-by: BuildkiteCI
2023-12-05 17:25:52 -05:00
Value::String(ref s) => {
if let Ok(s) = s.to_str() {
visitor.visit_enum(de::value::StrDeserializer::new(s))
} else {
Err(unexpected(name, &self.value))
}
}
// an attribute set however represents an externally
// tagged enum with content
Value::Attrs(attrs) => visitor.visit_enum(Enum(*attrs)),
_ => Err(unexpected(name, &self.value)),
}
}
fn deserialize_identifier<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
self.deserialize_str(visitor)
}
fn deserialize_ignored_any<V>(self, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
visitor.visit_unit()
}
}
struct Enum(tvix_eval::NixAttrs);
impl<'de> EnumAccess<'de> for Enum {
type Error = Error;
type Variant = NixDeserializer;
// TODO: pass the known variants down here and check against them
fn variant_seed<V>(self, seed: V) -> Result<(V::Value, Self::Variant), Self::Error>
where
V: de::DeserializeSeed<'de>,
{
if self.0.len() != 1 {
return Err(Error::AmbiguousEnum);
}
let (key, value) = self.0.into_iter().next().expect("length asserted above");
fix(tvix): Represent strings as byte arrays C++ nix uses C-style zero-terminated char pointers to represent strings internally - however, up to this point, tvix has used Rust `String` and `str` for string values. Since those are required to be valid utf-8, we haven't been able to properly represent all the string values that Nix supports. To fix that, this change converts the internal representation of the NixString struct from `Box<str>` to `BString`, from the `bstr` crate - this is a wrapper around a `Vec<u8>` with extra functions for treating that byte vector as a "morally string-like" value, which is basically exactly what we need. Since this changes a pretty fundamental assumption about a pretty core type, there are a *lot* of changes in a lot of places to make this work, but I've tried to keep the general philosophy and intent of most of the code in most places intact. Most notably, there's nothing that's been done to make the derivation stuff in //tvix/glue work with non-utf8 strings everywhere, instead opting to just convert to String/str when passing things into that - there *might* be something to be done there, but I don't know what the rules should be and I don't want to figure them out in this change. To deal with OS-native paths in a way that also works in WASM for tvixbolt, this also adds a dependency on the "os_str_bytes" crate. Fixes: b/189 Fixes: b/337 Change-Id: I5e6eb29c62f47dd91af954f5e12bfc3d186f5526 Reviewed-on: https://cl.tvl.fyi/c/depot/+/10200 Reviewed-by: tazjin <tazjin@tvl.su> Reviewed-by: flokli <flokli@flokli.de> Reviewed-by: sterni <sternenseemann@systemli.org> Autosubmit: aspen <root@gws.fyi> Tested-by: BuildkiteCI
2023-12-05 17:25:52 -05:00
if let Ok(k) = key.to_str() {
let val = seed.deserialize(de::value::StrDeserializer::<Error>::new(k))?;
Ok((val, NixDeserializer::new(value)))
} else {
Err(unexpected("string", &key.clone().into()))
}
}
}
impl<'de> VariantAccess<'de> for NixDeserializer {
type Error = Error;
fn unit_variant(self) -> Result<(), Self::Error> {
// If this case is hit, a user specified the name of a unit
// enum variant but gave it content. Unit enum deserialisation
// is handled in `deserialize_enum` above.
Err(Error::UnitEnumContent)
}
fn newtype_variant_seed<T>(self, seed: T) -> Result<T::Value, Self::Error>
where
T: de::DeserializeSeed<'de>,
{
seed.deserialize(self)
}
fn tuple_variant<V>(self, _len: usize, visitor: V) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
de::Deserializer::deserialize_seq(self, visitor)
}
fn struct_variant<V>(
self,
_fields: &'static [&'static str],
visitor: V,
) -> Result<V::Value, Self::Error>
where
V: de::Visitor<'de>,
{
de::Deserializer::deserialize_map(self, visitor)
}
}