From 6dd30531ac62f6a3a564b7341d43f6cd71b90794 Mon Sep 17 00:00:00 2001 From: SoniEx2 Date: Tue, 15 Nov 2022 08:23:02 -0300 Subject: Rework subtrees again and update docs --- src/vm/de.rs | 295 ++++++++++++++++++++++++++++++++++++++++++++++++++++++---- src/vm/mod.rs | 4 +- 2 files changed, 279 insertions(+), 20 deletions(-) (limited to 'src/vm') diff --git a/src/vm/de.rs b/src/vm/de.rs index 493b658..8f2aa8d 100644 --- a/src/vm/de.rs +++ b/src/vm/de.rs @@ -169,6 +169,7 @@ impl<'pat, 'state, 'de, O: Serialize> Packer<'pat, 'state, O> { matches: true, poison: false, }; + debug_assert!(!new_frame.ops.is_empty()); // we want the "newest" frame last, so it is // easier to unwind back. self.interp.frames.insert(at, new_frame); @@ -209,12 +210,13 @@ impl<'pat, 'state, 'de, O: Serialize> Packer<'pat, 'state, O> { pack_index -= 1; } if frame.poison { - if has_pack { - packs.remove(pack_index); - } - frame.matches = false; - has_pack = false; + // FIXME should frame.poison always remove the pack? if frame.is_value() { + if has_pack { + packs.remove(pack_index); + } + frame.matches = false; + has_pack = false; frame.poison = false; } } @@ -260,8 +262,14 @@ impl<'pat, 'state, 'de, O: Serialize> Packer<'pat, 'state, O> { packs[target_pack].merge_breadth(pack); } } else { + //if frame.poison { + // target_frame.poison = true; + //} if !optional { - target_frame.poison = true; + self.interp.error.insert({ + MatchError::ValidationError + }); + return Err(E::custom("subtree failed")); } } if let Some((0, _)) = target_frame.num_subtrees() { @@ -864,10 +872,14 @@ where } } } + let mut poison = false; for (f, m) in self.frames_mut().iter_active_mut().zip(output_matches) { - // FIXME inspect frame.key_subtree() for optional? - // what is this even supposed to do again? f.matches = m; + if !m { + if let Some((_, false)) = f.key_subtree() { + poison = true; + } + } } let obj = SerdeObject::Map(obj_inner); let mut final_packs = self.step_out(output_packs)?; @@ -880,6 +892,7 @@ where | Some((Type::IgnoredAny, _)) | None => { + frame.poison = poison; let matched = std::mem::replace(&mut frame.matches, true); if !matched { final_packs.insert( @@ -1360,6 +1373,36 @@ mod tests { assert_eq!(pack.subpacks.len(), 2); } + #[test] + fn test_empty_subtrees() { + // tests empty subtrees + let consts = crate::parser::parse::<&'static str, &'static str, ()>( + ":map()(())", + None, + None, + ).unwrap(); + dbg!(&consts); + let data = r#"{"hello": "world"}"#; + let mut der = JsonDeserializer::from_str(data); + let mut err = Default::default(); + let mut frames = Default::default(); + let interp = Interpreter::new( + &consts, + &mut err, + &mut frames, + //&mut output, + ); + let result = Packer::new( + interp, + MAX_CALLS, + ).deserialize(&mut der); + let (mut packs, obj) = result.unwrap(); + assert!(obj.is_none()); + assert_eq!(packs.len(), 1); + let pack = packs.pop().unwrap(); + assert_eq!(pack.subpacks.len(), 0); + } + #[test] fn test_parser_subtrees() { // use a parsed pattern with subtrees to test Packer @@ -1402,15 +1445,165 @@ mod tests { } #[test] - fn test_subtree_missing() { - // use a parsed pattern that might actually be used in the real world. + fn test_parser_subtrees_validation() { + // checks that the Packer can validate multiple keys. + let consts = crate::parser::parse::<&'static str, &'static str, ()>( + ":map(->['name'?]name:str)(->['value'?]value:u32)(->[:str]:?ignored_any)", + None, + None, + ).unwrap(); + let data = &[ + 0x02, // map length (2) + 0x04, // string length (4) + 0x6E, 0x61, 0x6D, 0x65, // b'name' + 0x01, // string length (1) + 0x61, // b'a' + 0x05, // string length (5) + 0x76, 0x61, 0x6C, 0x75, 0x65, // b'value' + 0x01, // 1 + ]; + let mut der = PostcardDeserializer::from_bytes(data); + let mut err = Default::default(); + let mut frames = Default::default(); + let interp = Interpreter::new( + &consts, + &mut err, + &mut frames, + //&mut output, + ); + let result = Packer::new( + interp, + MAX_CALLS, + ).deserialize(&mut der); + let (mut packs, obj) = result.unwrap(); + assert!(obj.is_none()); + assert_eq!(packs.len(), 1); + let pack = packs.pop().unwrap(); + assert_eq!(pack.subpacks.len(), 1); + assert_eq!(pack.subpacks[0]["name"].1, SerdeObject::Str(From::from("a"))); + assert_eq!(pack.subpacks[0]["value"].1, SerdeObject::U32(1)); + } + + #[test] + fn test_one_or_more_subtrees_validation() { + // checks that the Packer supports OR-like validation semantics + let consts = crate::parser::parse::<&'static str, &'static str, ()>( + ":map((->['name'?]?name)?(->['value'?]?value)?)(->[:str]:u32)", + None, + None, + ).unwrap(); + let data = &[ + 0x01, // map length (1) + 0x04, // string length (4) + 0x6E, 0x61, 0x6D, 0x65, // b'name' + 0x01, // 1 + ]; + let mut der = PostcardDeserializer::from_bytes(data); + let mut err = Default::default(); + let mut frames = Default::default(); + let interp = Interpreter::new( + &consts, + &mut err, + &mut frames, + //&mut output, + ); + let result = Packer::new( + interp, + MAX_CALLS, + ).deserialize(&mut der); + let (mut packs, obj) = result.unwrap(); + assert!(obj.is_none()); + assert_eq!(packs.len(), 1); + let pack = packs.pop().unwrap(); + assert_eq!(pack.subpacks.len(), 1); + assert_eq!(pack.subpacks[0]["name"].1, SerdeObject::U32(1)); + } + + #[test] + fn test_one_or_more_subtrees_validation_failure() { + // checks that the Packer supports OR-like validation semantics + // (failure) + let consts = crate::parser::parse::<&'static str, &'static str, ()>( + ":map((->['name'?]?name)?(->['value'?]?value)?)(->[:str]:u32)", + None, + None, + ).unwrap(); + let data = &[ + 0x01, // map length (1) + 0x04, // string length (4) + 0x6E, 0x65, 0x6D, 0x65, // b'neme' + 0x01, // 1 + ]; + let mut der = PostcardDeserializer::from_bytes(data); + let mut err = Default::default(); + let mut frames = Default::default(); + let interp = Interpreter::new( + &consts, + &mut err, + &mut frames, + //&mut output, + ); + let result = Packer::new( + interp, + MAX_CALLS, + ).deserialize(&mut der); + assert!(matches!(err, Some(MatchError::ValidationError))); + assert!(result.is_err()); + } + + // FIXME this should test that the map doesn't contain other keys aside + // from the ones requested by the pattern, but this doesn't match existing + // datafu semantics at all. we need new syntax for it. + //#[test] + //fn test_parser_subtrees_no_additional_keys() { + // // use a parsed pattern with subtrees to test Packer + // // also test a non-self-describing format (postcard) + // // also require at least one subtree to match on every iteration. + // // also this checks for validation error + // let consts = crate::parser::parse::<&'static str, &'static str, ()>( + // ":map((->['name'?]name)?(->['value'?]value)?)(->[:str]:u32)", + // None, + // None, + // ).unwrap(); + // let data = &[ + // 0x03, // map length (3) + // 0x04, // string length (4) + // 0x6E, 0x61, 0x6D, 0x65, // b'name' + // 0x01, // 1 + // 0x05, // string length (5) + // 0x76, 0x61, 0x6C, 0x75, 0x65, // b'value' + // 0x01, // 1 + // 0x05, // string length (5) + // 0x76, 0x65, 0x6C, 0x75, 0x65, // b'velue' + // 0x01, // 1 + // ]; + // let mut der = PostcardDeserializer::from_bytes(data); + // let mut err = Default::default(); + // let mut frames = Default::default(); + // let interp = Interpreter::new( + // &consts, + // &mut err, + // &mut frames, + // //&mut output, + // ); + // let result = Packer::new( + // interp, + // MAX_CALLS, + // ).deserialize(&mut der); + // assert!(matches!(err, Some(MatchError::ValidationError))); + // assert!(result.is_err()); + //} + + #[test] + fn test_key_optionally_missing() { + // test a pattern with a missing key let consts = crate::parser::parse::<&'static str, &'static str, ()>( " :map ->['a'?]:map ->[b:?str]:?map - (->['x'?]x:?bool) - (->['y'?]y:?bool)? + (->['x'?]x:?bool)? + (->['y'?]?y:?bool)? ", None, None @@ -1433,7 +1626,6 @@ mod tests { assert!(obj.is_none()); assert_eq!(packs.len(), 1); let pack = &packs[0]; - dbg!(pack); assert_eq!(pack.subpacks.len(), 1); let b = &pack.subpacks[0]["b"]; assert_eq!(b.1, SerdeObject::Str(From::from("2"))); @@ -1442,17 +1634,84 @@ mod tests { assert_eq!(b.0.subpacks[0]["y"].1, SerdeObject::Bool(true)); } + #[test] + fn test_map_without_values() { + // test that the map without values still collects the key + let consts = crate::parser::parse::<&'static str, &'static str, ()>( + " + :map + ->[a:?str]?:?map + ->[b:?str]?:?map + ", + None, + None + ).unwrap(); + let data = r#"{"a": {}}"#; + let mut der = JsonDeserializer::from_str(data); + let mut err = Default::default(); + let mut frames = Default::default(); + let interp = Interpreter::new( + &consts, + &mut err, + &mut frames, + //&mut output, + ); + let result = Packer::new( + interp, + MAX_CALLS, + ).deserialize(&mut der); + let (mut packs, obj) = result.unwrap(); + assert!(obj.is_none()); + assert_eq!(packs.len(), 1); + let pack = &packs[0]; + assert_eq!(pack.subpacks.len(), 1); + let a = &pack.subpacks[0]["a"]; + assert_eq!(a.1, SerdeObject::Str(From::from("a"))); + assert_eq!(a.0.subpacks.len(), 0); + } + + #[test] + fn test_guaranteed_at_least_once() { + // test that at least one match is required before collecting the key + let consts = crate::parser::parse::<&'static str, &'static str, ()>( + " + :map + ->[a:?str]:?map + ->[b:?str]:?map + ", + None, + None + ).unwrap(); + let data = r#"{"a": {}}"#; + let mut der = JsonDeserializer::from_str(data); + let mut err = Default::default(); + let mut frames = Default::default(); + let interp = Interpreter::new( + &consts, + &mut err, + &mut frames, + //&mut output, + ); + let result = Packer::new( + interp, + MAX_CALLS, + ).deserialize(&mut der); + let (mut packs, obj) = result.unwrap(); + assert!(obj.is_none()); + assert_eq!(packs.len(), 0); + } + #[test] fn test_realish_use_case() { // use a parsed pattern that might actually be used in the real world. let consts = crate::parser::parse::<&'static str, &'static str, ()>( " :map - ->['projects'?]?:map - ->[commit:?str]?:?map - ->[url:?str]?:?map - ->[branch:?str]?:?map - (->['active'?]?active:?bool) + ->['projects'?]:map + ->[commit:?str]:?map + ->[url:?str]:?map + ->[branch:?str]:?map + (->['active'?]active:?bool)? (->['federate'?]?federate:?bool)? ", None, diff --git a/src/vm/mod.rs b/src/vm/mod.rs index 81131c0..21fec73 100644 --- a/src/vm/mod.rs +++ b/src/vm/mod.rs @@ -55,7 +55,7 @@ impl Default for PatternConstants { } } -impl std::fmt::Debug for PatternConstants { +impl std::fmt::Debug for PatternConstants { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { f.debug_struct( "PatternConstants" @@ -70,7 +70,7 @@ impl std::fmt::Debug for PatternConstants { &format_args!("({} predicates)", self.predicates.len()), ).field( "defs", - &format_args!("FIXME"), + &self.defs, ).finish() } } -- cgit 1.4.1