// Copyright (C) 2022 Soni L. // SPDX-License-Identifier: MIT OR Apache-2.0 //! Deserialization-related parts of the VM. use std::borrow::Cow; use std::marker::PhantomData; use indexmap::IndexMap; use serde::Serialize; use serde::de::Error as _; use serde::de::IntoDeserializer as _; use smallvec::SmallVec; use super::Frame; use super::Interpreter; use super::Pack; use super::PatternConstants; use super::PatternElement; use super::SerdeObject; use super::Type; use super::Value; use crate::errors::MatchError; /// A `DeserializeSeed` for Datafu input. /// /// This converts from Serde to Datafu's internal representation (a "pack"). pub(crate) struct Packer<'pat, 'state, O: Serialize> { /// The global interpreter state. interp: Interpreter<'pat, 'state, O>, /// Current call limit. call_limit: usize, /// Whether we're collecting values. collecting: bool, } struct FramesMut<'packer, 'pat> { frames: &'packer mut Vec>, } struct Frames<'packer, 'pat> { frames: &'packer Vec>, } impl<'packer, 'pat> FramesMut<'packer, 'pat> { fn iter_mut<'a>( &'a mut self, ) -> impl Iterator> + DoubleEndedIterator where 'packer: 'a, { self.frames.iter_mut() } fn iter_active_mut<'a>( &'a mut self, ) -> impl Iterator> + DoubleEndedIterator where 'packer: 'a, { self.iter_mut().filter(|frame| { frame.active() }) } } impl<'packer, 'pat> Frames<'packer, 'pat> { fn iter<'a>( &'a self, ) -> impl Iterator> + DoubleEndedIterator where 'packer: 'a, { self.frames.iter() } fn iter_active<'a>( &'a self, ) -> impl Iterator> + DoubleEndedIterator where 'packer: 'a, { self.iter().filter(|frame| { frame.active() }) } } impl<'pat, 'state, 'de, O: Serialize> Packer<'pat, 'state, O> { /// Creates a new Packer. pub(crate) fn new( interp: Interpreter<'pat, 'state, O>, call_limit: usize, ) -> Self { Self { interp: interp, call_limit: call_limit, collecting: false, } } fn frames_mut(&mut self) -> FramesMut<'_, 'pat> { FramesMut { frames: &mut *self.interp.frames, } } fn frames(&mut self) -> Frames<'_, 'pat> { Frames { frames: &*self.interp.frames, } } /// Steps the VM into the next operation. fn step_in(&mut self) -> Result<(), E> { if self.call_limit > 0 { self.call_limit -= 1; } else { self.interp.error.insert(MatchError::StackOverflow); return Err(todo!()); } // iterate up to the *live* length (i.e. the loop is allowed to modify // the length). // NOTE: we need to use while-let so as to not borrow anything in an // iterator. filtering happens on the *result* of the iterator. let mut index_iter = 0..; while let Some(index) = index_iter.next().filter(|&i| { i < self.interp.frames.len() }) { let frame = &mut self.interp.frames[index]; if frame.overstep > 0 || !frame.matches { // overstepped and non-matching frames frame.overstep += 1; } else { if !frame.next() { // empty/end-of frames // 1 layer of step-in. // step-out will undo this. // this is correct because this branch implies overstep = 0 frame.overstep = 1; } else if matches!( frame.op(), PatternElement::SubtreeMarker, ) { // subtrees! // these are tricky, because the current frame can be moved // in memory. so we have to use indexing every time. // tho first we set it as overstep because it has special // handling. frame.overstep = 1; let mut at = index + 1; while self.interp.frames[index].next() { let op = self.interp.frames[index].op(); if let PatternElement::ValueSubtree { index: subtree, .. } = op { let new_frame = Frame { ops: &self.interp.pat.protos[subtree][..], iar: None, overstep: 0, matches: true, }; // we want the "newest" frame last, so it is // easier to unwind back. self.interp.frames.insert(at, new_frame); at += 1; } else { unreachable!() } } } } } Ok(()) } /// Steps the VM back into the previous operation. fn step_out( &mut self, mut packs: Vec>, ) -> Vec> { // this code attempts to maintain the logical invariant of: // self.frames().iter_active().count() == packs.len() self.call_limit += 1; let mut index_iter = 0..; let mut pack_index = packs.len(); while let Some(index) = index_iter.next().filter(|&i| { i < self.interp.frames.len() }) { // iterate backwards let index = self.interp.frames.len() - index - 1; let frame = &mut self.interp.frames[index]; let has_pack = frame.matches; if frame.overstep > 0 { // handle overstep frame.overstep -= 1; } else { if has_pack { pack_index -= 1; } // unwind frame if frame.prev() { // successfully unwound. do nothing. } else { // find parent frame. let mut count = 1; let mut target = index; let mut target_pack = pack_index; let mut target_unwound = false; while count > 0 && target > 0 { target -= 1; match self.interp.frames[target].num_subtrees() { Some((num, _)) if num < count => { if has_pack { debug_assert!(target_pack > 0); target_pack -= 1; } count -= num; }, Some((num, unwound)) => { if has_pack { debug_assert!(target_pack > 0); target_pack -= 1; } target_unwound = unwound; count = 0; }, None => { if has_pack { if self.interp.frames[target].matches { debug_assert!(target_pack > 0); target_pack -= 1; } } count += 1; }, } } if count == 0 { let frame = self.interp.frames.remove(target); // TODO what to do with `frame`? if has_pack { // has parent frame let pack = packs.remove(pack_index); if !target_unwound { packs.insert(target_pack, pack); } else { packs[target_pack].merge_from(pack); } } } } } } packs } } impl<'pat, 'state, 'de, O> serde::de::DeserializeSeed<'de> for &mut Packer<'pat, 'state, O> where O: Serialize, { type Value = (Vec>, Option>); fn deserialize( mut self, deserializer: D, ) -> Result where D: serde::Deserializer<'de> { if let Err(e) = self.step_in() { return Err(e); } let pat = self.interp.pat; let target_type = self.frames().iter_active().try_fold( Type::IgnoredAny, |target_type, frame| { Ok(match (target_type, frame.get_type()) { // required type binds stronger than any/ignored_any (Type::IgnoredAny, Some((ty, true))) => ty, (Type::Any, Some((ty, true))) => ty, (ty, Some((Type::IgnoredAny, true))) => ty, (ty, Some((Type::Any, true))) => ty, // prefer owned if any branch prefers owned (Type::String, Some((Type::Str, true))) => { Type::String }, (Type::Str, Some((Type::String, true))) => { Type::String }, (Type::Bytes, Some((Type::ByteBuf, true))) => { Type::ByteBuf }, (Type::ByteBuf, Some((Type::Bytes, true))) => { Type::ByteBuf }, // types which are the same are okay (left, Some((right, _))) if left == right => { left }, // optional type vs Any/IgnoredAny prefers Any (Type::IgnoredAny, Some((_, false))) => Type::Any, (Type::Any, Some((_, false))) => Type::Any, // types which are not the same are an error because we // only request a specific type if it's actually required (left, Some((right, _))) => { return Err(todo!()); }, _ => Type::Any, }) }, ); let target_type = match target_type { Ok(target_type) => target_type, Err(e) => { self.interp.error.insert(e); return Err(D::Error::custom("type conflict")); }, }; match target_type { Type::Any => deserializer.deserialize_any(&mut *self), Type::IgnoredAny => { deserializer.deserialize_ignored_any(&mut *self) }, Type::Bool => deserializer.deserialize_bool(&mut *self), Type::I8 => deserializer.deserialize_i8(&mut *self), Type::I16 => deserializer.deserialize_i16(&mut *self), Type::I32 => deserializer.deserialize_i32(&mut *self), Type::I64 => deserializer.deserialize_i64(&mut *self), Type::I128 => deserializer.deserialize_i128(&mut *self), Type::U8 => deserializer.deserialize_u8(&mut *self), Type::U16 => deserializer.deserialize_u16(&mut *self), Type::U32 => deserializer.deserialize_u32(&mut *self), Type::U64 => deserializer.deserialize_u64(&mut *self), Type::U128 => deserializer.deserialize_u128(&mut *self), Type::F32 => deserializer.deserialize_f32(&mut *self), Type::F64 => deserializer.deserialize_f64(&mut *self), Type::Char => deserializer.deserialize_char(&mut *self), Type::Str if !self.collecting => { deserializer.deserialize_str(&mut *self) }, Type::Str | Type::String => { deserializer.deserialize_string(&mut *self) }, Type::Bytes if !self.collecting => { deserializer.deserialize_bytes(&mut *self) }, Type::Bytes | Type::ByteBuf => { deserializer.deserialize_byte_buf(&mut *self) }, Type::Option => deserializer.deserialize_option(&mut *self), Type::Unit => deserializer.deserialize_unit(&mut *self), Type::Seq => deserializer.deserialize_seq(&mut *self), Type::Map => deserializer.deserialize_map(&mut *self), Type::Identifier => { deserializer.deserialize_identifier(&mut *self) }, Type::Tuple(len) => { deserializer.deserialize_tuple(len, &mut *self) }, Type::UnitStruct(name) => { deserializer.deserialize_unit_struct(name, &mut *self) }, Type::NewtypeStruct(name) => { deserializer.deserialize_newtype_struct(name, &mut *self) }, Type::TupleStruct { name, len } => { deserializer.deserialize_tuple_struct(name, len, &mut *self) }, Type::Struct { name, fields } => { deserializer.deserialize_struct(name, fields, &mut *self) }, Type::Enum { name, variants } => { deserializer.deserialize_enum(name, variants, &mut *self) }, }.map(|(packs, obj)| (self.step_out(packs), obj)) } } /// visit method generator for simple values (primitives). /// /// can generate whole function or just the glue. macro_rules! vs { (fn $visit:ident $obj:ident ($data_type:pat) $rust_type:ty) => { fn $visit(mut self, v: $rust_type) -> Result where E: serde::de::Error, { vs!(self (v) $obj ($data_type)) } }; ($this:ident $v:tt $obj:ident ($data_type:pat)) => { { let pat = $this.interp.pat; let mut obj = None; if $this.collecting { obj = Some(SerdeObject::$obj$v); } let mut packs = Vec::new(); let result = { $this.frames_mut().iter_active_mut().try_for_each(|frame| { let ty = frame.get_type(); match ty { | Some(($data_type, _)) | Some((Type::Any, _)) | Some((Type::IgnoredAny, _)) | None => {}, Some((_, false)) => { frame.matches = false; return Ok(()); }, Some((_, true)) => { return Err(MatchError::ValidationError) }, } let mut pack = Pack::default(); if let Some(name) = frame.get_name(pat) { let mut map = IndexMap::new(); map.insert(name, (Pack::default(), SerdeObject::$obj$v)); pack.subpacks.push(map); } packs.push(pack); Ok(()) }) }; match result { Err(e) => { $this.interp.error.insert(e); return Err(todo!()); }, _ => (), } Ok((packs, obj)) } }; } impl<'pat, 'state, 'de, O> serde::de::Visitor<'de> for &mut Packer<'pat, 'state, O> where O: Serialize, { type Value = (Vec>, Option>); fn expecting(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { write!(f, "unsure") } vs!(fn visit_bool Bool (Type::Bool) bool); vs!(fn visit_i8 I8 (Type::I8) i8); vs!(fn visit_i16 I16 (Type::I16) i16); vs!(fn visit_i32 I32 (Type::I32) i32); vs!(fn visit_i64 I64 (Type::I64) i64); vs!(fn visit_i128 I128 (Type::I128) i128); vs!(fn visit_u8 U8 (Type::U8) u8); vs!(fn visit_u16 U16 (Type::U16) u16); vs!(fn visit_u32 U32 (Type::U32) u32); vs!(fn visit_u64 U64 (Type::U64) u64); vs!(fn visit_u128 U128 (Type::U128) u128); vs!(fn visit_f32 F32 (Type::F32) f32); vs!(fn visit_f64 F64 (Type::F64) f64); vs!(fn visit_char Char (Type::Char) char); fn visit_str(self, v: &str) -> Result where E: serde::de::Error, { // no real option but to clone. vs!(self (Cow::Owned(v.to_owned())) Str (Type::String | Type::Str)) } fn visit_borrowed_str(self, v: &'de str) -> Result where E: serde::de::Error, { vs!(self (Cow::Borrowed(v)) Str (Type::String | Type::Str)) } fn visit_string(self, v: String) -> Result where E: serde::de::Error, { // TODO try to avoid cloning self.visit_str(&*v) } fn visit_bytes(self, v: &[u8]) -> Result where E: serde::de::Error, { vs!(self (Cow::Owned(v.to_owned())) Bytes (Type::Bytes | Type::ByteBuf)) } fn visit_borrowed_bytes(self, v: &'de [u8]) -> Result where E: serde::de::Error, { vs!(self (Cow::Borrowed(v)) Bytes (Type::Bytes | Type::ByteBuf)) } fn visit_byte_buf(self, v: Vec) -> Result where E: serde::de::Error, { // TODO try to avoid cloning self.visit_bytes(&*v) } fn visit_none(self) -> Result where E: serde::de::Error, { vs!(self {} None (Type::Option)) } fn visit_some(self, deserializer: D) -> Result where D: serde::de::Deserializer<'de>, { todo!() } fn visit_unit(self) -> Result where E: serde::de::Error, { vs!(self {} Unit (Type::Unit)) } fn visit_newtype_struct( self, deserializer: D ) -> Result where D: serde::de::Deserializer<'de>, { todo!() } fn visit_seq(self, seq: A) -> Result where A: serde::de::SeqAccess<'de>, { let mut obj = None; if self.collecting { obj = Some(SerdeObject::Seq(Vec::new())); } todo!() } fn visit_map(self, mut map: A) -> Result where A: serde::de::MapAccess<'de>, { let old_collecting = self.collecting; let pat = self.interp.pat; let mut collecting = old_collecting; let typeck = self.frames_mut().iter_active_mut().try_for_each(|frame| { let ty = frame.get_type(); match ty { | Some((Type::Map, _)) | Some((Type::Any, _)) | Some((Type::IgnoredAny, _)) | None => {}, Some((_, false)) => { frame.matches = false; return Ok(()); }, Some((_, true)) => { return Err(MatchError::ValidationError) }, } if frame.get_name(pat).is_some() { collecting = true; } Ok(()) }); match typeck { Err(e) => { self.interp.error.insert(e); return Err(todo!()); }, _ => (), } if let Err(e) = self.step_in() { return Err(e); } self.collecting = collecting; let mut subframes = Vec::new(); self.frames().iter_active().for_each(|frame| { if let Some(key_subtree) = frame.key_subtree() { subframes.push(Frame { ops: &pat.protos[key_subtree], iar: None, overstep: 0, matches: true, }); } }); let mut obj_inner = Vec::new(); let mut output_packs = Vec::new(); while let Some(packed_key) = { let subinterp = Interpreter { pat: pat, frames: &mut subframes, error: self.interp.error, }; let mut subpacker = Packer { interp: subinterp, collecting: self.collecting, call_limit: self.call_limit, }; map.next_key_seed(&mut subpacker)? } { self.frames_mut().iter_active_mut().filter(|frame| { frame.key_subtree().is_some() }).zip(&mut subframes).for_each(|(frame, subframe)| { frame.matches = subframe.matches; // reset subframe for next iteration // NOTE wait to reset subframe.matches when merging packs!!! subframe.iar = None; }); self.frames_mut().iter_active_mut().for_each(|frame| { // mark every non-subtree key as matching. if frame.key_subtree().is_none() { frame.matches = true; } }); let packed_value = map.next_value_seed(&mut *self)?; if self.collecting { obj_inner.push( (packed_key.1.unwrap(), packed_value.1.unwrap()), ); } let mut key_packs_per_frame = packed_key.0.into_iter(); let mut value_packs_per_frame = packed_value.0; // whatever is active in self.frames(), if matches, has a pack // whatever is in subframes, if matches, has a pack // count(active self.frames() with subtree which match) is always // smaller than count(subframes which match) because the former // gets updated by next_value_seed // count(active self.frames() with subtree) == count(subframes) // tl;dr: need to figure out which packs produced by subframes line // up with which packs produced by self, discarding extra subframes // (where the corresponding self frame doesn't match) and accepting // extra packs produced by self. // NOTE: key_packs_per_frame ~ subframes // value_packs_per_frame ~ self // keys come first tho (key.merge_from(value)) let mut iter_subframes = subframes.iter_mut(); // related to value_packs_per_frame let mut pack_index = 0; for frame in self.frames().iter_active() { // check if this frame has an associated subframe let subframe = if frame.key_subtree().is_some() { // if there are more frames with associated subframes // than there are subframes, panic Some(iter_subframes.next().unwrap()) } else { None }; if frame.matches && subframe.is_some() { // this already implies subframe.matches let key_pack = key_packs_per_frame.next().unwrap(); let value_pack = &mut value_packs_per_frame[pack_index]; key_pack.merge_into(value_pack); pack_index += 1; } else if frame.matches { // value matches but there's no subframe, carry on pack_index += 1; } else if !frame.matches && subframe.is_some() { // frame didn't match but there was a subframe let subframe = subframe.unwrap(); if subframe.matches { // subframe matched, remove key pack let _ = key_packs_per_frame.next().unwrap(); } else { // neither matched, no relevant packs // do reset subframe for next_key_seed tho! subframe.matches = true; } } else { // no relevant packs } } if output_packs.is_empty() { output_packs = value_packs_per_frame; } else { for (left, right) in output_packs.iter_mut().zip( value_packs_per_frame, ) { left.subpacks.extend(right.subpacks) } } } let obj = SerdeObject::Map(obj_inner); let mut final_packs = self.step_out(output_packs); let mut iter_final_packs = final_packs.iter_mut(); self.frames_mut().iter_active_mut().zip({ final_packs.iter_mut() }).for_each(|(frame, pack)| { let ty = frame.get_type(); match ty { | Some((Type::Map, _)) | Some((Type::Any, _)) | Some((Type::IgnoredAny, _)) | None => { frame.matches = true; }, _ => return, } if let Some(name) = frame.get_name(pat) { // we can assume collecting == true let old_pack = std::mem::take(pack); let mut map = IndexMap::new(); map.insert(name, (old_pack, obj.clone())); pack.subpacks.push(map); } }); self.collecting = old_collecting; Ok((final_packs, collecting.then(|| obj))) } fn visit_enum(self, data: A) -> Result where A: serde::de::EnumAccess<'de>, { let mut obj = None; if self.collecting { obj = Some(SerdeObject::Enum { variant: todo!(), data: todo!(), }); } todo!() } } /// A `Deserializer` for Datafu output. /// /// This converts from Datafu's internal representation (a "pack") into the /// desired output type. pub struct Unpacker<'pat, 'de> { pack: Pack<'pat, 'de>, call_limit: usize, } impl<'pat, 'de> Unpacker<'pat, 'de> { /// Unpacks a Datafu "pack". pub fn new(pack: Pack<'pat, 'de>, call_limit: usize) -> Self { Self { pack, call_limit, } } } impl<'pat, 'de> serde::Deserializer<'de> for Unpacker<'pat, 'de> { // TODO datafu errors type Error = serde::de::value::Error; fn deserialize_any(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_bool(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_i8(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_i16(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_i32(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_i64(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_u8(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_u16(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_u32(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_u64(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_f32(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_f64(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_char(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_str(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_string(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_bytes(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_byte_buf(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_option(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_unit(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_unit_struct(self, _: &'static str, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_newtype_struct(self, _: &'static str, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_seq(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_tuple(self, _: usize, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_tuple_struct(self, _: &'static str, _: usize, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_map(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_struct( self, _: &'static str, fields: &'static [&'static str], visitor: V, ) -> Result where V: serde::de::Visitor<'de>, { todo!() } fn deserialize_enum(self, _: &'static str, _: &'static [&'static str], _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_identifier(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } fn deserialize_ignored_any(self, _: V) -> Result where V: serde::de::Visitor<'de> { todo!() } } /// Deserializes a SerdeObject pub(crate) struct SerdeObjectDeserializer<'de, E> { pub(crate) obj: SerdeObject<'de>, pub(crate) value: Option>, pub(crate) _e: PhantomData E>, } impl<'de, E> serde::de::Deserializer<'de> for SerdeObjectDeserializer<'de, E> where E: serde::de::Error, { type Error = E; fn deserialize_any(self, v: V) -> Result where V: serde::de::Visitor<'de>, { match self.obj { SerdeObject::Bool(x) => v.visit_bool(x), SerdeObject::I8(x) => v.visit_i8(x), SerdeObject::I16(x) => v.visit_i16(x), SerdeObject::I32(x) => v.visit_i32(x), SerdeObject::I64(x) => v.visit_i64(x), SerdeObject::I128(x) => v.visit_i128(x), SerdeObject::U8(x) => v.visit_u8(x), SerdeObject::U16(x) => v.visit_u16(x), SerdeObject::U32(x) => v.visit_u32(x), SerdeObject::U64(x) => v.visit_u64(x), SerdeObject::U128(x) => v.visit_u128(x), SerdeObject::F32(x) => v.visit_f32(x), SerdeObject::F64(x) => v.visit_f64(x), SerdeObject::Char(x) => v.visit_char(x), SerdeObject::Str(Cow::Owned(x)) => v.visit_string(x), SerdeObject::Str(Cow::Borrowed(x)) => v.visit_borrowed_str(x), SerdeObject::Bytes(Cow::Owned(x)) => v.visit_byte_buf(x), SerdeObject::Bytes(Cow::Borrowed(x)) => v.visit_borrowed_bytes(x), SerdeObject::Some(x) => v.visit_some(x.into_deserializer()), SerdeObject::None => v.visit_none(), SerdeObject::Unit => v.visit_unit(), SerdeObject::Seq(x) => todo!(), SerdeObject::Map(x) => todo!(), SerdeObject::NewtypeStruct(x) => { v.visit_newtype_struct(x.into_deserializer()) }, SerdeObject::Enum { variant, data } => todo!(), } } fn deserialize_ignored_any(self, v: V) -> Result where V: serde::de::Visitor<'de>, { drop(self); v.visit_unit() } serde::forward_to_deserialize_any! { bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string bytes byte_buf option unit unit_struct newtype_struct seq tuple tuple_struct map struct enum identifier } } #[cfg(test)] mod tests { use super::Packer; use super::super::PatternConstants; use crate::vm::MAX_CALLS; use crate::vm::Interpreter; use crate::vm::Type; use crate::vm::Value; use crate::vm::PatternElement; use crate::vm::SerdeObject; use crate::vm::Frame; use serde_json::Deserializer as JsonDeserializer; use serde::de::DeserializeSeed as _; #[test] #[should_panic] fn test_broken() { // broken pattern, should never be emitted by parser. make sure it's // not accepted. let consts = PatternConstants::<()>::default(); let mut err = Default::default(); let mut frames = Default::default(); let interp = Interpreter::new(&consts, &mut err, &mut frames); let _ = Packer::new(interp, MAX_CALLS); } #[test] fn test_empty_create() { // test creating the parser with an empty pattern. let mut consts = PatternConstants::<()>::default(); consts.protos.push(Vec::new()); let mut err = Default::default(); let mut frames = Default::default(); let interp = Interpreter::new(&consts, &mut err, &mut frames); let _ = Packer::new(interp, MAX_CALLS); } #[test] fn test_empty_match() { // test matching something with an empty pattern. let mut consts = PatternConstants::<()>::default(); consts.protos.push(Vec::new()); let mut der = JsonDeserializer::from_str("{}"); let mut err = Default::default(); let mut frames = Default::default(); let interp = Interpreter::new(&consts, &mut err, &mut frames); let pack = Packer::new(interp, MAX_CALLS).deserialize(&mut der).unwrap(); } #[test] fn test_simple_match() { // test matching a simple value let mut consts = PatternConstants::<()>::default(); consts.strings.push("hello".into()); consts.protos.push(vec![ PatternElement::Value { name: Some(0), value: Some(Value::Type { ty: Type::U64, skippable: false, }), }, ]); let mut der = JsonDeserializer::from_str("3"); let mut err = Default::default(); let mut frames = Default::default(); let interp = Interpreter::new(&consts, &mut err, &mut frames); let packed = Packer::new(interp, MAX_CALLS).deserialize(&mut der); let (packs, obj) = packed.unwrap(); assert!(obj.is_none()); assert_eq!(packs[0].subpacks[0]["hello"].1, SerdeObject::U64(3)); } #[test] fn test_simple_error() { // test a value that doesn't match (serde_json error) let mut consts = PatternConstants::<()>::default(); consts.strings.push("hello".into()); consts.protos.push(vec![ PatternElement::Value { name: Some(0), value: Some(Value::Type { ty: Type::U64, skippable: false, }), }, ]); let mut der = JsonDeserializer::from_str("\"hello\""); let mut err = Default::default(); let mut frames = Default::default(); let interp = Interpreter::new(&consts, &mut err, &mut frames); let packed = Packer::new(interp, MAX_CALLS).deserialize(&mut der); dbg!(&packed); // error produced by serde_json assert!(packed.is_err()); } #[test] fn test_basic_multiframe() { // test multiple frames (matching and non-matching) let mut consts = PatternConstants::<()>::default(); consts.strings.push("a".into()); consts.strings.push("b".into()); consts.protos.push(vec![ PatternElement::Value { name: Some(0), value: Some(Value::Type { ty: Type::U64, skippable: true, }), }, ]); consts.protos.push(vec![ PatternElement::Value { name: Some(1), value: Some(Value::Type { ty: Type::Bool, skippable: true, }), }, ]); let mut der = JsonDeserializer::from_str(r#"10"#); let mut err = Default::default(); let mut frames: Vec<_> = Default::default(); frames.push(Frame { ops: &consts.protos[0], iar: None, matches: true, overstep: 0, }); frames.push(Frame { ops: &consts.protos[1], iar: None, matches: true, overstep: 0, }); let interp = Interpreter { pat: &consts, error: &mut err, frames: &mut frames, }; let packed = Packer::new(interp, MAX_CALLS).deserialize(&mut der); let (packs, obj) = packed.unwrap(); assert!(obj.is_none()); assert_eq!( packs[0].subpacks[0]["a"].1, SerdeObject::U64(10), ); assert_eq!(packs.len(), 1); assert!(frames[0].matches); assert!(!frames[1].matches); } #[test] fn test_map() { let mut consts = PatternConstants::<()>::default(); consts.strings.push("key".into()); consts.strings.push("value".into()); consts.protos.push(vec![ PatternElement::Value { name: Some(0), value: None, }, ]); consts.protos.push(vec![ PatternElement::Value { name: None, value: Some(Value::Type { ty: Type::Map, skippable: false, }), }, PatternElement::Tag { key_subtree: Some(0), }, PatternElement::Value { name: Some(1), value: Some(Value::Type { ty: Type::U64, skippable: false, }), }, ]); let mut der = JsonDeserializer::from_str(r#"{"hello": 0, "world": 1}"#); let mut err = Default::default(); let mut frames = Default::default(); let interp = Interpreter::new(&consts, &mut err, &mut frames); let packed = Packer::new(interp, MAX_CALLS).deserialize(&mut der); let (packs, obj) = packed.unwrap(); assert!(obj.is_none()); assert_eq!( packs[0].subpacks[0]["key"].1, SerdeObject::Str("hello".into()), ); assert_eq!( packs[0].subpacks[0]["value"].1, SerdeObject::U64(0), ); assert_eq!( packs[0].subpacks[1]["key"].1, SerdeObject::Str("world".into()), ); assert_eq!( packs[0].subpacks[1]["value"].1, SerdeObject::U64(1), ); } }