fish-shell/fish-rust/src/parse_tree.rs

200 lines
6.5 KiB
Rust
Raw Normal View History

Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
//! Programmatic representation of fish code.
use std::pin::Pin;
use std::rc::Rc;
use crate::ast::Ast;
use crate::parse_constants::{
token_type_user_presentable_description, ParseErrorCode, ParseErrorList, ParseErrorListFfi,
ParseKeyword, ParseTokenType, ParseTreeFlags, SourceOffset, SourceRange, SOURCE_OFFSET_INVALID,
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
};
use crate::tokenizer::TokenizerError;
use crate::wchar::{wstr, WString, L};
use crate::wchar_ffi::{WCharFromFFI, WCharToFFI};
use crate::wutil::sprintf;
use cxx::{CxxWString, UniquePtr};
/// A struct representing the token type that we use internally.
#[derive(Clone, Copy)]
pub struct ParseToken {
/// The type of the token as represented by the parser
pub typ: ParseTokenType,
/// Any keyword represented by this token
pub keyword: ParseKeyword,
/// Hackish: whether the source contains a dash prefix
pub has_dash_prefix: bool,
/// Hackish: whether the source looks like '-h' or '--help'
pub is_help_argument: bool,
/// Hackish: if TOK_END, whether the source is a newline.
pub is_newline: bool,
// Hackish: whether this token is a string like FOO=bar
pub may_be_variable_assignment: bool,
/// If this is a tokenizer error, that error.
pub tok_error: TokenizerError,
source_start: SourceOffset,
source_length: SourceOffset,
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
}
impl ParseToken {
pub fn new(typ: ParseTokenType) -> Self {
ParseToken {
typ,
keyword: ParseKeyword::none,
has_dash_prefix: false,
is_help_argument: false,
is_newline: false,
may_be_variable_assignment: false,
tok_error: TokenizerError::none,
source_start: SOURCE_OFFSET_INVALID.try_into().unwrap(),
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
source_length: 0,
}
}
pub fn set_source_start(&mut self, value: usize) {
self.source_start = value.try_into().unwrap();
}
pub fn source_start(&self) -> usize {
self.source_start.try_into().unwrap()
}
pub fn set_source_length(&mut self, value: usize) {
self.source_length = value.try_into().unwrap();
}
pub fn source_length(&self) -> usize {
self.source_length.try_into().unwrap()
}
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
/// \return the source range.
/// Note the start may be invalid.
pub fn range(&self) -> SourceRange {
SourceRange::new(self.source_start(), self.source_length())
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
}
/// \return whether we are a string with the dash prefix set.
pub fn is_dash_prefix_string(&self) -> bool {
self.typ == ParseTokenType::string && self.has_dash_prefix
}
/// Returns a string description of the given parse token.
pub fn describe(&self) -> WString {
let mut result = Into::<&'static wstr>::into(self.typ).to_owned();
if self.keyword != ParseKeyword::none {
result += &sprintf!(L!(" <%ls>"), Into::<&'static wstr>::into(self.keyword))[..]
}
result
}
pub fn user_presentable_description(&self) -> WString {
token_type_user_presentable_description(self.typ, self.keyword)
}
}
impl From<TokenizerError> for ParseErrorCode {
fn from(err: TokenizerError) -> Self {
match err {
TokenizerError::none => ParseErrorCode::none,
TokenizerError::unterminated_quote => ParseErrorCode::tokenizer_unterminated_quote,
TokenizerError::unterminated_subshell => {
ParseErrorCode::tokenizer_unterminated_subshell
}
TokenizerError::unterminated_slice => ParseErrorCode::tokenizer_unterminated_slice,
TokenizerError::unterminated_escape => ParseErrorCode::tokenizer_unterminated_escape,
_ => ParseErrorCode::tokenizer_other,
}
}
}
/// A type wrapping up a parse tree and the original source behind it.
pub struct ParsedSource {
2023-04-22 05:57:16 +08:00
pub src: WString,
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
src_ffi: UniquePtr<CxxWString>,
2023-04-22 05:57:16 +08:00
pub ast: Ast,
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
}
impl ParsedSource {
fn new(src: WString, ast: Ast) -> Self {
let src_ffi = src.to_ffi();
ParsedSource { src, src_ffi, ast }
}
}
pub type ParsedSourceRef = Option<Rc<ParsedSource>>;
/// Return a shared pointer to ParsedSource, or null on failure.
/// If parse_flag_continue_after_error is not set, this will return null on any error.
pub fn parse_source(
src: WString,
flags: ParseTreeFlags,
errors: Option<&mut ParseErrorList>,
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
) -> ParsedSourceRef {
let ast = Ast::parse(&src, flags, errors);
if ast.errored() && !flags.contains(ParseTreeFlags::CONTINUE_AFTER_ERROR) {
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
None
} else {
Some(Rc::new(ParsedSource::new(src, ast)))
}
}
struct ParsedSourceRefFFI(pub ParsedSourceRef);
#[cxx::bridge]
mod parse_tree_ffi {
extern "C++" {
include!("ast.h");
pub type Ast = crate::ast::Ast;
pub type ParseErrorListFfi = crate::parse_constants::ParseErrorListFfi;
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
}
extern "Rust" {
type ParsedSourceRefFFI;
fn empty_parsed_source_ref() -> Box<ParsedSourceRefFFI>;
fn has_value(&self) -> bool;
fn new_parsed_source_ref(src: &CxxWString, ast: Pin<&mut Ast>) -> Box<ParsedSourceRefFFI>;
#[cxx_name = "parse_source"]
fn parse_source_ffi(
src: &CxxWString,
flags: u8,
errors: *mut ParseErrorListFfi,
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
) -> Box<ParsedSourceRefFFI>;
fn clone(self: &ParsedSourceRefFFI) -> Box<ParsedSourceRefFFI>;
fn src(self: &ParsedSourceRefFFI) -> &CxxWString;
fn ast(self: &ParsedSourceRefFFI) -> &Ast;
}
}
impl ParsedSourceRefFFI {
fn has_value(&self) -> bool {
self.0.is_some()
}
}
fn empty_parsed_source_ref() -> Box<ParsedSourceRefFFI> {
Box::new(ParsedSourceRefFFI(None))
}
fn new_parsed_source_ref(src: &CxxWString, ast: Pin<&mut Ast>) -> Box<ParsedSourceRefFFI> {
let mut stolen_ast = Ast::default();
std::mem::swap(&mut stolen_ast, ast.get_mut());
Box::new(ParsedSourceRefFFI(Some(Rc::new(ParsedSource::new(
src.from_ffi(),
stolen_ast,
)))))
}
fn parse_source_ffi(
src: &CxxWString,
flags: u8,
errors: *mut ParseErrorListFfi,
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
) -> Box<ParsedSourceRefFFI> {
Box::new(ParsedSourceRefFFI(parse_source(
src.from_ffi(),
ParseTreeFlags::from_bits(flags).unwrap(),
if errors.is_null() {
None
} else {
Some(unsafe { &mut (*errors).0 })
},
)))
Port AST to Rust The translation is fairly direct though it adds some duplication, for example there are multiple "match" statements that mimic function overloading. Rust has no overloading, and we cannot have generic methods in the Node trait (due to a Rust limitation, the error is like "cannot be made into an object") so we include the type name in method names. Give clients like "indent_visitor_t" a Rust companion ("IndentVisitor") that takes care of the AST traversal while the AST consumption remains in C++ for now. In future, "IndentVisitor" should absorb the entirety of "indent_visitor_t". This pattern requires that "fish_indent" be exposed includable header to the CXX bridge. Alternatively, we could define FFI wrappers for recursive AST traversal. Rust requires we separate the AST visitors for "mut" and "const" scenarios. Take this opportunity to concretize both visitors: The only client that requires mutable access is the populator. To match the structure of the C++ populator which makes heavy use of function overloading, we need to add a bunch of functions to the trait. Since there is no other mutable visit, this seems acceptable. The "const" visitors never use "will_visit_fields_of()" or "did_visit_fields_of()", so remove them (though this is debatable). Like in the C++ implementation, the AST nodes themselves are largely defined via macros. Union fields like "Statement" and "ArgumentOrRedirection" do currently not use macros but may in future. This commit also introduces a precedent for a type that is defined in one CXX bridge and used in another one - "ParseErrorList". To make this work we need to manually define "ExternType". There is one annoyance with CXX: functions that take explicit lifetime parameters require to be marked as unsafe. This makes little sense because functions that return `&Foo` with implicit lifetime can be misused the same way on the C++ side. One notable change is that we cannot directly port "find_block_open_keyword()" (which is used to compute an error) because it relies on the stack of visited nodes. We cannot modify a stack of node references while we do the "mut" walk. Happily, an idiomatic solution is easy: we can tell the AST visitor to backtrack to the parent node and create the error there. Since "node_t::accept_base" is no longer a template we don't need the "node_visitation_t" trampoline anymore. The added copying at the FFI boundary makes things slower (memcpy dominates the profile) but it's not unusable, which is good news: $ hyperfine ./fish.{old,new}" -c 'source ../share/completions/git.fish'" Benchmark 1: ./fish.old -c 'source ../share/completions/git.fish' Time (mean ± σ): 195.5 ms ± 2.9 ms [User: 190.1 ms, System: 4.4 ms] Range (min … max): 193.2 ms … 205.1 ms 15 runs Benchmark 2: ./fish.new -c 'source ../share/completions/git.fish' Time (mean ± σ): 677.5 ms ± 62.0 ms [User: 665.4 ms, System: 10.0 ms] Range (min … max): 611.7 ms … 805.5 ms 10 runs Summary './fish.old -c 'source ../share/completions/git.fish'' ran 3.47 ± 0.32 times faster than './fish.new -c 'source ../share/completions/git.fish'' Leftovers: - Enum variants are still snakecase; I didn't get around to changing this yet. - "ast_type_to_string()" still returns a snakecase name. This could be changed since it's not user visible.
2023-04-02 22:42:59 +08:00
}
impl ParsedSourceRefFFI {
fn clone(&self) -> Box<ParsedSourceRefFFI> {
Box::new(ParsedSourceRefFFI(self.0.clone()))
}
fn src(&self) -> &CxxWString {
&self.0.as_ref().unwrap().src_ffi
}
fn ast(&self) -> &Ast {
&self.0.as_ref().unwrap().ast
}
}