From 413135b9252de65c61045ade984de7b4d3319c2d Mon Sep 17 00:00:00 2001 From: Florian Klink Date: Tue, 20 Aug 2024 16:21:52 +0300 Subject: refactor(nix-compat/aterm): update function names a bit Don't call functions bstr or str when they return BString or String, it's confusing. Rename them to `string` and `bytes`. We might be able to generalize over this being BString or Vec later. Change-Id: I8198551ed3ba1cfc479bf7e3cbbc13a426faf4c0 Reviewed-on: https://cl.tvl.fyi/c/depot/+/12257 Autosubmit: flokli Tested-by: BuildkiteCI Reviewed-by: Connor Brewster --- tvix/nix-compat/src/aterm/mod.rs | 4 ++-- tvix/nix-compat/src/aterm/parser.rs | 28 +++++++++++++++------------- tvix/nix-compat/src/derivation/parser.rs | 20 +++++++++----------- 3 files changed, 26 insertions(+), 26 deletions(-) (limited to 'tvix/nix-compat/src') diff --git a/tvix/nix-compat/src/aterm/mod.rs b/tvix/nix-compat/src/aterm/mod.rs index 8806b6caf2e5..bb3b77bc7399 100644 --- a/tvix/nix-compat/src/aterm/mod.rs +++ b/tvix/nix-compat/src/aterm/mod.rs @@ -2,6 +2,6 @@ mod escape; mod parser; pub(crate) use escape::escape_bytes; -pub(crate) use parser::parse_bstr_field; -pub(crate) use parser::parse_str_list; +pub(crate) use parser::parse_bytes_field; pub(crate) use parser::parse_string_field; +pub(crate) use parser::parse_string_list; diff --git a/tvix/nix-compat/src/aterm/parser.rs b/tvix/nix-compat/src/aterm/parser.rs index a30cb40ab08d..a570573a8700 100644 --- a/tvix/nix-compat/src/aterm/parser.rs +++ b/tvix/nix-compat/src/aterm/parser.rs @@ -11,8 +11,10 @@ use nom::multi::separated_list0; use nom::sequence::delimited; use nom::IResult; -/// Parse a bstr and undo any escaping. -fn parse_escaped_bstr(i: &[u8]) -> IResult<&[u8], BString> { +/// Parse a bstr and undo any escaping (which is why this needs to allocate). +// FUTUREWORK: have a version for fields that are known to not need escaping +// (like store paths), and use &str. +fn parse_escaped_bytes(i: &[u8]) -> IResult<&[u8], BString> { escaped_transform( is_not("\"\\"), '\\', @@ -29,14 +31,14 @@ fn parse_escaped_bstr(i: &[u8]) -> IResult<&[u8], BString> { /// Parse a field in double quotes, undo any escaping, and return the unquoted /// and decoded `Vec`. -pub(crate) fn parse_bstr_field(i: &[u8]) -> IResult<&[u8], BString> { +pub(crate) fn parse_bytes_field(i: &[u8]) -> IResult<&[u8], BString> { // inside double quotes… delimited( nomchar('\"'), // There is alt(( // …either is a bstr after unescaping - parse_escaped_bstr, + parse_escaped_bytes, // …or an empty string. map(tag(b""), |_| BString::default()), )), @@ -45,8 +47,8 @@ pub(crate) fn parse_bstr_field(i: &[u8]) -> IResult<&[u8], BString> { } /// Parse a field in double quotes, undo any escaping, and return the unquoted -/// and decoded string, if it's a valid string. Or fail parsing if the bytes are -/// no valid UTF-8. +/// and decoded [String], if it's valid UTF-8. +/// Or fail parsing if the bytes are no valid UTF-8. pub(crate) fn parse_string_field(i: &[u8]) -> IResult<&[u8], String> { // inside double quotes… delimited( @@ -54,18 +56,18 @@ pub(crate) fn parse_string_field(i: &[u8]) -> IResult<&[u8], String> { // There is alt(( // either is a String after unescaping - nom::combinator::map_opt(parse_escaped_bstr, |escaped_bstr| { - String::from_utf8(escaped_bstr.into()).ok() + nom::combinator::map_opt(parse_escaped_bytes, |escaped_bytes| { + String::from_utf8(escaped_bytes.into()).ok() }), // or an empty string. - map(tag(b""), |_| String::new()), + map(tag(b""), |_| "".to_string()), )), nomchar('\"'), )(i) } -/// Parse a list of of string fields (enclosed in brackets) -pub(crate) fn parse_str_list(i: &[u8]) -> IResult<&[u8], Vec> { +/// Parse a list of string fields (enclosed in brackets) +pub(crate) fn parse_string_list(i: &[u8]) -> IResult<&[u8], Vec> { // inside brackets delimited( nomchar('['), @@ -89,7 +91,7 @@ mod tests { #[case] expected: &[u8], #[case] exp_rest: &[u8], ) { - let (rest, parsed) = super::parse_bstr_field(input).expect("must parse"); + let (rest, parsed) = super::parse_bytes_field(input).expect("must parse"); assert_eq!(exp_rest, rest, "expected remainder"); assert_eq!(expected, parsed); } @@ -118,7 +120,7 @@ mod tests { #[case::empty_list(b"[]", vec![], b"")] #[case::empty_list_with_rest(b"[]blub", vec![], b"blub")] fn parse_list(#[case] input: &[u8], #[case] expected: Vec, #[case] exp_rest: &[u8]) { - let (rest, parsed) = super::parse_str_list(input).expect("must parse"); + let (rest, parsed) = super::parse_string_list(input).expect("must parse"); assert_eq!(exp_rest, rest, "expected remainder"); assert_eq!(expected, parsed); } diff --git a/tvix/nix-compat/src/derivation/parser.rs b/tvix/nix-compat/src/derivation/parser.rs index 2775294960fe..7fc56389d17a 100644 --- a/tvix/nix-compat/src/derivation/parser.rs +++ b/tvix/nix-compat/src/derivation/parser.rs @@ -3,7 +3,6 @@ //! //! [ATerm]: http://program-transformation.org/Tools/ATermFormat.html -use bstr::BString; use nom::bytes::complete::tag; use nom::character::complete::char as nomchar; use nom::combinator::{all_consuming, map_res}; @@ -73,7 +72,7 @@ fn parse_output(i: &[u8]) -> NomResult<&[u8], (String, Output)> { terminated(aterm::parse_string_field, nomchar(',')), terminated(aterm::parse_string_field, nomchar(',')), terminated(aterm::parse_string_field, nomchar(',')), - aterm::parse_bstr_field, + aterm::parse_bytes_field, ))(i) .map_err(into_nomerror) }, @@ -150,10 +149,10 @@ fn parse_outputs(i: &[u8]) -> NomResult<&[u8], BTreeMap> { } fn parse_input_derivations(i: &[u8]) -> NomResult<&[u8], BTreeMap>> { - let (i, input_derivations_list) = parse_kv::, _>(aterm::parse_str_list)(i)?; + let (i, input_derivations_list) = parse_kv(aterm::parse_string_list)(i)?; // This is a HashMap of drv paths to a list of output names. - let mut input_derivations: BTreeMap> = BTreeMap::new(); + let mut input_derivations: BTreeMap> = BTreeMap::new(); for (input_derivation, output_names) in input_derivations_list { let mut new_output_names = BTreeSet::new(); @@ -179,7 +178,7 @@ fn parse_input_derivations(i: &[u8]) -> NomResult<&[u8], BTreeMap NomResult<&[u8], BTreeSet> { - let (i, input_sources_lst) = aterm::parse_str_list(i).map_err(into_nomerror)?; + let (i, input_sources_lst) = aterm::parse_string_list(i).map_err(into_nomerror)?; let mut input_sources: BTreeSet<_> = BTreeSet::new(); for input_source in input_sources_lst.into_iter() { @@ -240,9 +239,9 @@ pub fn parse_derivation(i: &[u8]) -> NomResult<&[u8], Derivation> { // // parse builder |i| terminated(aterm::parse_string_field, nomchar(','))(i).map_err(into_nomerror), // // parse arguments - |i| terminated(aterm::parse_str_list, nomchar(','))(i).map_err(into_nomerror), + |i| terminated(aterm::parse_string_list, nomchar(','))(i).map_err(into_nomerror), // parse environment - parse_kv::(aterm::parse_bstr_field), + parse_kv(aterm::parse_bytes_field), )), nomchar(')'), ) @@ -427,8 +426,8 @@ mod tests { #[case] expected: &BTreeMap, #[case] exp_rest: &[u8], ) { - let (rest, parsed) = super::parse_kv::(crate::aterm::parse_bstr_field)(input) - .expect("must parse"); + let (rest, parsed) = + super::parse_kv(crate::aterm::parse_bytes_field)(input).expect("must parse"); assert_eq!(exp_rest, rest, "expected remainder"); assert_eq!(*expected, parsed); } @@ -437,8 +436,7 @@ mod tests { #[test] fn parse_kv_fail_dup_keys() { let input: &'static [u8] = b"[(\"a\",\"1\"),(\"a\",\"2\")]"; - let e = super::parse_kv::(crate::aterm::parse_bstr_field)(input) - .expect_err("must fail"); + let e = super::parse_kv(crate::aterm::parse_bytes_field)(input).expect_err("must fail"); match e { nom::Err::Failure(e) => { -- cgit 1.4.1