Refactoring and bugfixes

This commit is contained in:
Martin Haug 2021-11-18 16:21:45 +01:00
parent 3162c6a83a
commit fdb9d0743d
8 changed files with 1101 additions and 565 deletions

420
Cargo.lock generated
View File

@ -38,6 +38,17 @@ version = "0.7.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8da52d66c7071e2e3fa2a1e5c6d088fec47b593032b254f5e980de8ea54454d6"
[[package]]
name = "atty"
version = "0.2.14"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d9b39be18770d11421cdb1b9947a45dd3f37e93092cbf377614828a319d5fee8"
dependencies = [
"hermit-abi",
"libc",
"winapi",
]
[[package]]
name = "autocfg"
version = "1.0.1"
@ -56,6 +67,24 @@ version = "1.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bef38d45163c2f1dde094a7dfd33ccf595c92905c8f8f4fdc18d06fb1037718a"
[[package]]
name = "bstr"
version = "0.2.17"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ba3569f383e8f1598449f1a423e72e99569137b47740b1da11ef19af3d5c3223"
dependencies = [
"lazy_static",
"memchr",
"regex-automata",
"serde",
]
[[package]]
name = "bumpalo"
version = "3.8.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8f1e260c3a9040a7c19a12468758f4c16f31a81a1fe087482be9570ec864bb6c"
[[package]]
name = "bytemuck"
version = "1.7.3"
@ -68,12 +97,32 @@ version = "1.4.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "14c189c53d098945499cdfa7ecc63567cf3886b3332b312a5b4585d8d3a6a610"
[[package]]
name = "cast"
version = "0.2.7"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4c24dab4283a142afa2fdca129b80ad2c6284e073930f964c3a1293c225ee39a"
dependencies = [
"rustc_version",
]
[[package]]
name = "cfg-if"
version = "1.0.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"
[[package]]
name = "clap"
version = "2.34.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a0610544180c38b88101fecf2dd634b174a62eef6946f84dfc6a7127512b381c"
dependencies = [
"bitflags",
"textwrap",
"unicode-width",
]
[[package]]
name = "codespan-reporting"
version = "0.11.1"
@ -99,6 +148,108 @@ dependencies = [
"cfg-if",
]
[[package]]
name = "criterion"
version = "0.3.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1604dafd25fba2fe2d5895a9da139f8dc9b319a5fe5354ca137cbbce4e178d10"
dependencies = [
"atty",
"cast",
"clap",
"criterion-plot",
"csv",
"itertools",
"lazy_static",
"num-traits",
"oorandom",
"plotters",
"rayon",
"regex",
"serde",
"serde_cbor",
"serde_derive",
"serde_json",
"tinytemplate",
"walkdir",
]
[[package]]
name = "criterion-plot"
version = "0.4.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d00996de9f2f7559f7f4dc286073197f83e92256a59ed395f9aac01fe717da57"
dependencies = [
"cast",
"itertools",
]
[[package]]
name = "crossbeam-channel"
version = "0.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "06ed27e177f16d65f0f0c22a213e17c696ace5dd64b14258b52f9417ccb52db4"
dependencies = [
"cfg-if",
"crossbeam-utils",
]
[[package]]
name = "crossbeam-deque"
version = "0.8.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6455c0ca19f0d2fbf751b908d5c55c1f5cbc65e03c4225427254b46890bdde1e"
dependencies = [
"cfg-if",
"crossbeam-epoch",
"crossbeam-utils",
]
[[package]]
name = "crossbeam-epoch"
version = "0.9.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "4ec02e091aa634e2c3ada4a392989e7c3116673ef0ac5b72232439094d73b7fd"
dependencies = [
"cfg-if",
"crossbeam-utils",
"lazy_static",
"memoffset",
"scopeguard",
]
[[package]]
name = "crossbeam-utils"
version = "0.8.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d82cfc11ce7f2c3faef78d8a684447b40d503d9681acebed6cb728d45940c4db"
dependencies = [
"cfg-if",
"lazy_static",
]
[[package]]
name = "csv"
version = "1.1.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "22813a6dc45b335f9bade10bf7271dc477e81113e89eb251a0bc2a8a81c536e1"
dependencies = [
"bstr",
"csv-core",
"itoa 0.4.8",
"ryu",
"serde",
]
[[package]]
name = "csv-core"
version = "0.1.10"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2b2466559f260f48ad25fe6317b3c8dac77b5bdb5763ac7d9d6103530663bc90"
dependencies = [
"memchr",
]
[[package]]
name = "data-url"
version = "0.1.1"
@ -276,6 +427,21 @@ dependencies = [
"wasi",
]
[[package]]
name = "half"
version = "1.8.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "eabb4a44450da02c90444cf74558da904edde8fb4e9035a9a6a4e15445af0bd7"
[[package]]
name = "hermit-abi"
version = "0.1.19"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "62b467343b94ba476dcb2500d242dadbb39557df889310ac77c5d99100aaac33"
dependencies = [
"libc",
]
[[package]]
name = "iai"
version = "0.1.1"
@ -315,12 +481,27 @@ version = "0.4.8"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b71991ff56294aa922b450139ee08b3bfc70982c6b2c7562771375cf73542dd4"
[[package]]
name = "itoa"
version = "1.0.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1aab8fc367588b89dcee83ab0fd66b72b50b72fa1904d7095045ace2b0c81c35"
[[package]]
name = "jpeg-decoder"
version = "0.1.22"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "229d53d58899083193af11e15917b5640cd40b29ff475a1fe4ef725deb02d0f2"
[[package]]
name = "js-sys"
version = "0.3.55"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7cc9ffccd38c451a86bf13657df244e9c3f37493cce8e5e21e940963777acc84"
dependencies = [
"wasm-bindgen",
]
[[package]]
name = "kurbo"
version = "0.8.3"
@ -330,6 +511,12 @@ dependencies = [
"arrayvec 0.7.2",
]
[[package]]
name = "lazy_static"
version = "1.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646"
[[package]]
name = "libc"
version = "0.2.112"
@ -351,6 +538,12 @@ version = "0.1.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a3e378b66a060d48947b590737b30a1be76706c8dd7b8ba0f2fe3989c68a853f"
[[package]]
name = "memchr"
version = "2.4.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "308cc39be01b73d0d18f82a0e7b2a3df85245f84af96fdddc5d202d27e47b86a"
[[package]]
name = "memmap2"
version = "0.5.0"
@ -360,6 +553,15 @@ dependencies = [
"libc",
]
[[package]]
name = "memoffset"
version = "0.6.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5aa361d4faea93603064a027415f07bd8e1d5c88c9fbf68bf56a285428fd79ce"
dependencies = [
"autocfg",
]
[[package]]
name = "miniz_oxide"
version = "0.3.7"
@ -420,12 +622,28 @@ dependencies = [
"autocfg",
]
[[package]]
name = "num_cpus"
version = "1.13.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "19e64526ebdee182341572e50e9ad03965aa510cd94427a4549448f285e957a1"
dependencies = [
"hermit-abi",
"libc",
]
[[package]]
name = "once_cell"
version = "1.9.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "da32515d9f6e6e489d7bc9d84c71b060db7247dc035bbe44eac88cf87486d8d5"
[[package]]
name = "oorandom"
version = "11.1.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0ab1bc2a289d34bd04a330323ac98a1b4bc82c9d9fcb1e66b63caa84da26b575"
[[package]]
name = "pdf-writer"
version = "0.4.1"
@ -433,7 +651,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "36d760a6f2ac90811cba1006a298e8a7e5ce2c922bb5dc7f7000911a4a6b60f4"
dependencies = [
"bitflags",
"itoa",
"itoa 0.4.8",
"ryu",
]
@ -443,6 +661,34 @@ version = "0.4.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "db8bcd96cb740d03149cbad5518db9fd87126a10ab519c011893b1754134c468"
[[package]]
name = "plotters"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "32a3fd9ec30b9749ce28cd91f255d569591cdf937fe280c312143e3c4bad6f2a"
dependencies = [
"num-traits",
"plotters-backend",
"plotters-svg",
"wasm-bindgen",
"web-sys",
]
[[package]]
name = "plotters-backend"
version = "0.3.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d88417318da0eaf0fdcdb51a0ee6c3bed624333bff8f946733049380be67ac1c"
[[package]]
name = "plotters-svg"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "521fa9638fa597e1dc53e9412a4f9cefb01187ee1f7413076f9e6749e2885ba9"
dependencies = [
"plotters-backend",
]
[[package]]
name = "png"
version = "0.16.8"
@ -532,6 +778,31 @@ dependencies = [
"rand_core",
]
[[package]]
name = "rayon"
version = "1.5.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c06aca804d41dbc8ba42dfd964f0d01334eceb64314b9ecf7c5fad5188a06d90"
dependencies = [
"autocfg",
"crossbeam-deque",
"either",
"rayon-core",
]
[[package]]
name = "rayon-core"
version = "1.9.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d78120e2c850279833f1dd3582f730c4ab53ed95aeaaaa862a2a5c71b1656d8e"
dependencies = [
"crossbeam-channel",
"crossbeam-deque",
"crossbeam-utils",
"lazy_static",
"num_cpus",
]
[[package]]
name = "rctree"
version = "0.4.0"
@ -557,6 +828,27 @@ dependencies = [
"redox_syscall",
]
[[package]]
name = "regex"
version = "1.5.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d07a8629359eb56f1e2fb1652bb04212c072a87ba68546a04065d525673ac461"
dependencies = [
"regex-syntax",
]
[[package]]
name = "regex-automata"
version = "0.1.10"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6c230d73fb8d8c1b9c0b3135c5142a8acee3a0558fb8db5cf1cb65f8d7862132"
[[package]]
name = "regex-syntax"
version = "0.6.25"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f497285884f3fcff424ffc933e56d7cbca511def0c9831a7f9b5f6153e3cc89b"
[[package]]
name = "resvg"
version = "0.19.0"
@ -590,6 +882,15 @@ dependencies = [
"xmlparser",
]
[[package]]
name = "rustc_version"
version = "0.4.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "bfa0f585226d2e68097d4f95d113b15b83a82e819ab25717ec0590d9584ef366"
dependencies = [
"semver",
]
[[package]]
name = "rustybuzz"
version = "0.4.0"
@ -630,6 +931,18 @@ dependencies = [
"winapi-util",
]
[[package]]
name = "scopeguard"
version = "1.1.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d29ab0c6d3fc0ee92fe66e2d99f700eab17a8d57d1c1d3b748380fb20baa78cd"
[[package]]
name = "semver"
version = "1.0.4"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "568a8e6258aa33c13358f81fd834adb854c6f7c9468520910a9b1e8fac068012"
[[package]]
name = "serde"
version = "1.0.132"
@ -639,6 +952,16 @@ dependencies = [
"serde_derive",
]
[[package]]
name = "serde_cbor"
version = "0.11.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2bef2ebfde456fb76bbcf9f59315333decc4fda0b2b44b420243c11e0f5ec1f5"
dependencies = [
"half",
"serde",
]
[[package]]
name = "serde_derive"
version = "1.0.132"
@ -650,6 +973,17 @@ dependencies = [
"syn",
]
[[package]]
name = "serde_json"
version = "1.0.74"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ee2bb9cd061c5865d345bb02ca49fcef1391741b672b54a0bf7b679badec3142"
dependencies = [
"itoa 1.0.1",
"ryu",
"serde",
]
[[package]]
name = "simplecss"
version = "0.2.1"
@ -712,6 +1046,15 @@ dependencies = [
"winapi-util",
]
[[package]]
name = "textwrap"
version = "0.11.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d326610f408c7a4eb6f51c37c330e496b08506c9457c9d34287ecc38809fb060"
dependencies = [
"unicode-width",
]
[[package]]
name = "thiserror"
version = "1.0.30"
@ -746,6 +1089,16 @@ dependencies = [
"safe_arch",
]
[[package]]
name = "tinytemplate"
version = "1.2.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "be4d6b5f19ff7664e8c98d03e2139cb510db9b0a60b55f8e8709b689d939b6bc"
dependencies = [
"serde",
"serde_json",
]
[[package]]
name = "ttf-parser"
version = "0.12.3"
@ -758,6 +1111,7 @@ version = "0.1.0"
dependencies = [
"anyhow",
"codespan-reporting",
"criterion",
"dirs",
"filedescriptor",
"fxhash",
@ -892,6 +1246,70 @@ version = "0.10.2+wasi-snapshot-preview1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "fd6fbd9a79829dd1ad0cc20627bf1ed606756a7f77edff7b66b7064f9cb327c6"
[[package]]
name = "wasm-bindgen"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "632f73e236b219150ea279196e54e610f5dbafa5d61786303d4da54f84e47fce"
dependencies = [
"cfg-if",
"wasm-bindgen-macro",
]
[[package]]
name = "wasm-bindgen-backend"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "a317bf8f9fba2476b4b2c85ef4c4af8ff39c3c7f0cdfeed4f82c34a880aa837b"
dependencies = [
"bumpalo",
"lazy_static",
"log",
"proc-macro2",
"quote",
"syn",
"wasm-bindgen-shared",
]
[[package]]
name = "wasm-bindgen-macro"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d56146e7c495528bf6587663bea13a8eb588d39b36b679d83972e1a2dbbdacf9"
dependencies = [
"quote",
"wasm-bindgen-macro-support",
]
[[package]]
name = "wasm-bindgen-macro-support"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "7803e0eea25835f8abdc585cd3021b3deb11543c6fe226dcd30b228857c5c5ab"
dependencies = [
"proc-macro2",
"quote",
"syn",
"wasm-bindgen-backend",
"wasm-bindgen-shared",
]
[[package]]
name = "wasm-bindgen-shared"
version = "0.2.78"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0237232789cf037d5480773fe568aac745bfe2afbc11a863e97901780a6b47cc"
[[package]]
name = "web-sys"
version = "0.3.55"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "38eb105f1c59d9eaa6b5cdc92b859d85b926e82cb2e0945cd0c9259faa6fe9fb"
dependencies = [
"js-sys",
"wasm-bindgen",
]
[[package]]
name = "winapi"
version = "0.3.9"

View File

@ -20,14 +20,21 @@ opt-level = 2
[dependencies]
fxhash = "0.2"
image = { version = "0.23", default-features = false, features = ["png", "jpeg"] }
image = { version = "0.23", default-features = false, features = [
"png",
"jpeg",
] }
itertools = "0.10"
miniz_oxide = "0.4"
once_cell = "1"
pdf-writer = "0.4"
rustybuzz = "0.4"
serde = { version = "1", features = ["derive", "rc"] }
svg2pdf = { version = "0.1", default-features = false, features = ["text", "png", "jpeg"] }
svg2pdf = { version = "0.1", default-features = false, features = [
"text",
"png",
"jpeg",
] }
ttf-parser = "0.12"
typst-macros = { path = "./macros" }
unicode-bidi = "0.3.5"
@ -54,6 +61,7 @@ walkdir = "2"
# Dependencies updates:
# - Bump ttf-parser when rustybuzz is updated
# - Bump usvg and resvg in conjunction with svg2pdf
criterion = "0.3"
[[bin]]
name = "typst"
@ -68,3 +76,8 @@ harness = false
name = "oneshot"
path = "benches/oneshot.rs"
harness = false
[[bench]]
name = "timed"
path = "benches/timed.rs"
harness = false

98
benches/timed.rs Normal file
View File

@ -0,0 +1,98 @@
use std::path::Path;
use criterion::{black_box, criterion_group, criterion_main, Criterion};
use typst::eval::eval;
use typst::layout::layout;
use typst::loading::MemLoader;
use typst::parse::{parse, Scanner, TokenMode, Tokens};
use typst::source::SourceId;
use typst::Context;
const SRC: &str = include_str!("bench.typ");
const FONT: &[u8] = include_bytes!("../fonts/IBMPlexSans-Regular.ttf");
fn context() -> (Context, SourceId) {
let loader = MemLoader::new().with(Path::new("font.ttf"), FONT).wrap();
let mut ctx = Context::new(loader);
let id = ctx.sources.provide(Path::new("src.typ"), SRC.to_string());
(ctx, id)
}
fn bench_decode(c: &mut Criterion) {
c.bench_function("decode", |b| {
b.iter(|| {
// We don't use chars().count() because that has a special
// superfast implementation.
let mut count = 0;
let mut chars = black_box(SRC).chars();
while let Some(_) = chars.next() {
count += 1;
}
count
})
});
}
fn bench_scan(c: &mut Criterion) {
c.bench_function("scan", |b| {
b.iter(|| {
let mut count = 0;
let mut scanner = Scanner::new(black_box(SRC));
while let Some(_) = scanner.eat() {
count += 1;
}
count
})
});
}
fn bench_tokenize(c: &mut Criterion) {
c.bench_function("tokenize", |b| {
b.iter(|| Tokens::new(black_box(SRC), black_box(TokenMode::Markup)).count())
});
}
fn bench_parse(c: &mut Criterion) {
c.bench_function("parse", |b| b.iter(|| parse(SRC)));
}
fn bench_edit(c: &mut Criterion) {
let (mut ctx, id) = context();
c.bench_function("edit", |b| {
b.iter(|| black_box(ctx.sources.edit(id, 1168 .. 1171, "_Uhr_")))
});
}
fn bench_eval(c: &mut Criterion) {
let (mut ctx, id) = context();
let ast = ctx.sources.get(id).ast().unwrap();
c.bench_function("eval", |b| b.iter(|| eval(&mut ctx, id, &ast).unwrap()));
}
fn bench_to_tree(c: &mut Criterion) {
let (mut ctx, id) = context();
let module = ctx.evaluate(id).unwrap();
c.bench_function("to_tree", |b| {
b.iter(|| module.template.to_pages(ctx.style()))
});
}
fn bench_layout(c: &mut Criterion) {
let (mut ctx, id) = context();
let tree = ctx.execute(id).unwrap();
c.bench_function("layout", |b| b.iter(|| layout(&mut ctx, &tree)));
}
criterion_group!(
benches,
bench_decode,
bench_scan,
bench_tokenize,
bench_parse,
bench_edit,
bench_eval,
bench_to_tree,
bench_layout
);
criterion_main!(benches);

View File

@ -32,6 +32,13 @@ pub fn parse_atomic(src: &str, _: bool) -> Option<(Vec<Green>, bool)> {
p.eject_partial()
}
/// Parse an atomic primary. Returns `Some` if all of the input was consumed.
pub fn parse_atomic_markup(src: &str, _: bool) -> Option<(Vec<Green>, bool)> {
let mut p = Parser::new(src, TokenMode::Markup);
markup_expr(&mut p);
p.eject_partial()
}
/// Parse some markup. Returns `Some` if all of the input was consumed.
pub fn parse_markup(src: &str, _: bool) -> Option<(Vec<Green>, bool)> {
let mut p = Parser::new(src, TokenMode::Markup);
@ -171,17 +178,7 @@ fn markup_node(p: &mut Parser, at_start: &mut bool) {
| NodeKind::While
| NodeKind::For
| NodeKind::Import
| NodeKind::Include => {
let stmt = matches!(token, NodeKind::Let | NodeKind::Set | NodeKind::Import);
let group = if stmt { Group::Stmt } else { Group::Expr };
p.start_group(group);
let res = expr_prec(p, true, 0);
if stmt && res.is_ok() && !p.eof() {
p.expected_at("semicolon or line break");
}
p.end_group();
}
| NodeKind::Include => markup_expr(p),
// Block and template.
NodeKind::LeftBrace => block(p),
@ -222,6 +219,21 @@ fn enum_node(p: &mut Parser) {
});
}
/// Parse an expression within markup mode.
fn markup_expr(p: &mut Parser) {
if let Some(token) = p.peek() {
let stmt = matches!(token, NodeKind::Let | NodeKind::Set | NodeKind::Import);
let group = if stmt { Group::Stmt } else { Group::Expr };
p.start_group(group);
let res = expr_prec(p, true, 0);
if stmt && res.is_ok() && !p.eof() {
p.expected_at("semicolon or line break");
}
p.end_group();
}
}
/// Parse an expression.
fn expr(p: &mut Parser) -> ParseResult {
expr_prec(p, false, 0)

View File

@ -48,7 +48,7 @@ impl<'s> Parser<'s> {
}
/// End the parsing process and return multiple children.
pub fn eject(self) -> Option<(Vec<Green>, bool)>{
pub fn eject(self) -> Option<(Vec<Green>, bool)> {
if self.eof() && self.group_success() {
Some((self.children, self.tokens.was_unterminated()))
} else {

View File

@ -12,7 +12,7 @@ use crate::diag::TypResult;
use crate::loading::{FileHash, Loader};
use crate::parse::{is_newline, parse, Scanner};
use crate::syntax::ast::Markup;
use crate::syntax::{self, Category, GreenNode, RedNode, Span};
use crate::syntax::{self, Category, GreenNode, RedNode, Reparser, Span};
use crate::util::PathExt;
#[cfg(feature = "codespan-reporting")]
@ -285,9 +285,8 @@ impl SourceFile {
// Update the root node.
let span = Span::new(self.id, replace.start, replace.end);
if let Ok(range) =
Rc::make_mut(&mut self.root).incremental(&self.src, span, with.len())
{
let reparser = Reparser::new(&self.src, span, with.len());
if let Ok(range) = reparser.incremental(Rc::make_mut(&mut self.root)) {
range
} else {
self.root = parse(&self.src);
@ -502,6 +501,14 @@ mod tests {
test("= A heading", 3 .. 3, "n evocative", 2 .. 15);
test("your thing", 5 .. 5, "a", 4 .. 11);
test("a your thing a", 6 .. 7, "a", 2 .. 12);
test("{call(); abc}", 7 .. 7, "[]", 0 .. 15);
test("#call() abc", 7 .. 7, "[]", 0 .. 13);
// test(
// "hi\n- item\n- item 2\n - item 3",
// 10 .. 10,
// " ",
// 9 .. 33,
// );
test(
"#grid(columns: (auto, 1fr, 40%), [*plonk*], rect(width: 100%, height: 1pt, fill: conifer), [thing])",
16 .. 20,
@ -535,7 +542,7 @@ mod tests {
"C ",
14 .. 22,
);
test("{ let x = g() }", 10 .. 12, "f(54", 2 .. 15);
test("{ let x = g() }", 10 .. 12, "f(54", 0 .. 17);
test(
"a #let rect with (fill: eastern)\nb",
16 .. 31,

515
src/syntax/incremental.rs Normal file
View File

@ -0,0 +1,515 @@
use std::ops::Range;
use std::rc::Rc;
use super::{Green, GreenNode, NodeKind, Span};
use crate::parse::{
parse_atomic, parse_atomic_markup, parse_block, parse_comment, parse_markup,
parse_markup_elements, parse_template, TokenMode,
};
pub struct Reparser<'a> {
src: &'a str,
replace_range: Span,
replace_len: usize,
}
impl<'a> Reparser<'a> {
pub fn new(src: &'a str, replace_range: Span, replace_len: usize) -> Self {
Self { src, replace_range, replace_len }
}
}
impl Reparser<'_> {
/// Find the innermost child that is incremental safe.
pub fn incremental(&self, green: &mut GreenNode) -> Result<Range<usize>, ()> {
self.incremental_int(green, 0, TokenMode::Markup, true)
}
fn incremental_int(
&self,
green: &mut GreenNode,
mut offset: usize,
parent_mode: TokenMode,
outermost: bool,
) -> Result<Range<usize>, ()> {
let kind = green.kind().clone();
let mode = kind.mode().contextualize(parent_mode);
let mut loop_result = None;
let mut child_at_start = true;
let last = green.children.len() - 1;
let mut start = None;
for (i, child) in green.children.iter_mut().enumerate() {
let child_span =
Span::new(self.replace_range.source, offset, offset + child.len());
if child_span.surrounds(self.replace_range)
&& start.is_none()
&& ((self.replace_range.start != child_span.end
&& self.replace_range.end != child_span.start)
|| mode == TokenMode::Code
|| i == last)
{
let old_len = child.len();
// First, we try if the child has another, more specific applicable child.
if !kind.post().unsafe_interior() {
if let Ok(range) = match child {
Green::Node(n) => self.incremental_int(
Rc::make_mut(n),
offset,
kind.mode().child_mode(),
i == last && outermost,
),
Green::Token(_) => Err(()),
} {
let new_len = child.len();
green.update_child_len(new_len, old_len);
return Ok(range);
}
}
// This didn't work, so we try to self.replace_range the child at this
// level.
loop_result =
Some((i .. i + 1, child_span, i == last && outermost, child.kind()));
break;
} else if start.is_none()
&& child_span.contains(self.replace_range.start)
&& mode == TokenMode::Markup
&& child.kind().post().markup_safe()
{
start = Some((i, offset));
} else if child_span.contains(self.replace_range.end)
&& (self.replace_range.end != child_span.end || i == last)
&& mode == TokenMode::Markup
&& child.kind().post().markup_safe()
{
if let Some((start, start_offset)) = start {
loop_result = Some((
start .. i + 1,
Span::new(
self.replace_range.source,
start_offset,
offset + child.len(),
),
i == last && outermost,
child.kind(),
));
}
break;
} else if start.is_some()
&& (mode != TokenMode::Markup || !child.kind().post().markup_safe())
{
break;
}
offset += child.len();
child_at_start = child.kind().is_at_start(child_at_start);
}
// We now have a child that we can self.replace_range and a function to do so if
// the loop found any results at all.
let (child_idx_range, child_span, child_outermost, func, policy) =
loop_result.ok_or(()).and_then(|(a, b, c, child_kind)| {
let (func, policy) =
child_kind.reparsing_function(kind.mode().child_mode());
Ok((a, b, c, func?, policy))
})?;
let src_span = child_span.inserted(self.replace_range, self.replace_len);
let recompile_range = if policy == Postcondition::AtomicPrimary {
src_span.start .. self.src.len()
} else {
src_span.to_range()
};
let (mut new_children, unterminated) =
func(&self.src[recompile_range], child_at_start).ok_or(())?;
// Do not accept unclosed nodes if the old node did not use to be at the
// right edge of the tree.
if !child_outermost && unterminated {
return Err(());
}
let insertion = match check_invariants(
&new_children,
green.children(),
child_idx_range.clone(),
child_at_start,
mode,
src_span,
policy,
) {
InvariantResult::Ok => Ok(new_children),
InvariantResult::UseFirst => Ok(vec![std::mem::take(&mut new_children[0])]),
InvariantResult::Error => Err(()),
}?;
green.replace_child_range(child_idx_range, insertion);
Ok(src_span.to_range())
}
}
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
enum InvariantResult {
Ok,
UseFirst,
Error,
}
fn check_invariants(
use_children: &[Green],
old_children: &[Green],
child_idx_range: Range<usize>,
child_at_start: bool,
mode: TokenMode,
src_span: Span,
policy: Postcondition,
) -> InvariantResult {
let (new_children, ok) = if policy == Postcondition::AtomicPrimary {
if use_children.iter().map(Green::len).sum::<usize>() == src_span.len() {
(use_children, InvariantResult::Ok)
} else if use_children.len() == 1 && use_children[0].len() == src_span.len() {
(&use_children[0 .. 1], InvariantResult::UseFirst)
} else {
return InvariantResult::Error;
}
} else {
(use_children, InvariantResult::Ok)
};
let child_mode = old_children[child_idx_range.start].kind().mode().child_mode();
// Check if the children / child has the right type.
let same_kind = match policy {
Postcondition::SameKind(x) => x.map_or(true, |x| x == child_mode),
_ => false,
};
if same_kind || policy == Postcondition::AtomicPrimary {
if new_children.len() != 1 {
return InvariantResult::Error;
}
if same_kind {
if old_children[child_idx_range.start].kind() != new_children[0].kind() {
return InvariantResult::Error;
}
}
}
// Check if the neighbor invariants are still true.
if mode == TokenMode::Markup {
if child_idx_range.start > 0 {
if old_children[child_idx_range.start - 1].kind().pre()
== Precondition::RightWhitespace
&& !new_children[0].kind().is_whitespace()
{
return InvariantResult::Error;
}
}
if new_children.last().map(|x| x.kind().pre())
== Some(Precondition::RightWhitespace)
&& old_children.len() > child_idx_range.end
{
if !old_children[child_idx_range.end].kind().is_whitespace() {
return InvariantResult::Error;
}
}
let mut new_at_start = child_at_start;
for child in new_children {
new_at_start = child.kind().is_at_start(new_at_start);
}
for child in &old_children[child_idx_range.end ..] {
if child.kind().is_trivia() {
new_at_start = child.kind().is_at_start(new_at_start);
continue;
}
match child.kind().pre() {
Precondition::AtStart if !new_at_start => {
return InvariantResult::Error;
}
Precondition::NotAtStart if new_at_start => {
return InvariantResult::Error;
}
_ => {}
}
break;
}
}
ok
}
impl NodeKind {
pub fn reparsing_function(
&self,
parent_mode: TokenMode,
) -> (
Result<fn(&str, bool) -> Option<(Vec<Green>, bool)>, ()>,
Postcondition,
) {
let policy = self.post();
let mode = self.mode().contextualize(parent_mode);
match policy {
Postcondition::Unsafe | Postcondition::UnsafeLayer => (Err(()), policy),
Postcondition::AtomicPrimary if mode == TokenMode::Code => {
(Ok(parse_atomic), policy)
}
Postcondition::AtomicPrimary => (Ok(parse_atomic_markup), policy),
Postcondition::SameKind(x) if x == None || x == Some(mode) => {
let parser: fn(&str, bool) -> _ = match self {
NodeKind::Template => parse_template,
NodeKind::Block => parse_block,
NodeKind::LineComment | NodeKind::BlockComment => parse_comment,
_ => return (Err(()), policy),
};
(Ok(parser), policy)
}
_ => {
let parser: fn(&str, bool) -> _ = match mode {
TokenMode::Markup if self == &Self::Markup => parse_markup,
TokenMode::Markup => parse_markup_elements,
_ => return (Err(()), policy),
};
(Ok(parser), policy)
}
}
}
/// Whether it is safe to do incremental parsing on this node. Never allow
/// non-termination errors if this is not already the last leaf node.
pub fn post(&self) -> Postcondition {
match self {
// Replacing parenthesis changes if the expression is balanced and
// is therefore not safe.
Self::LeftBracket
| Self::RightBracket
| Self::LeftBrace
| Self::RightBrace
| Self::LeftParen
| Self::RightParen => Postcondition::Unsafe,
// Replacing an operator can change whether the parent is an
// operation which makes it unsafe. The star can appear in markup.
Self::Star
| Self::Comma
| Self::Semicolon
| Self::Colon
| Self::Plus
| Self::Minus
| Self::Slash
| Self::Eq
| Self::EqEq
| Self::ExclEq
| Self::Lt
| Self::LtEq
| Self::Gt
| Self::GtEq
| Self::PlusEq
| Self::HyphEq
| Self::StarEq
| Self::SlashEq
| Self::Not
| Self::And
| Self::Or
| Self::With
| Self::Dots
| Self::Arrow => Postcondition::Unsafe,
// These keywords are literals and can be safely be substituted with
// other expressions.
Self::None | Self::Auto => Postcondition::AtomicPrimary,
// These keywords change what kind of expression the parent is and
// how far the expression would go.
Self::Let
| Self::Set
| Self::If
| Self::Else
| Self::For
| Self::In
| Self::While
| Self::Break
| Self::Continue
| Self::Return
| Self::Import
| Self::Include
| Self::From => Postcondition::Unsafe,
Self::Markup => Postcondition::SameKind(None),
Self::Space(_) => Postcondition::SameKind(Some(TokenMode::Code)),
// These are all replaceable by other tokens.
Self::Parbreak
| Self::Linebreak
| Self::Text(_)
| Self::TextInLine(_)
| Self::NonBreakingSpace
| Self::EnDash
| Self::EmDash
| Self::Escape(_)
| Self::Strong
| Self::Emph
| Self::Heading
| Self::Enum
| Self::List
| Self::Raw(_)
| Self::Math(_) => Postcondition::Safe,
// Changing the heading level, enum numbering, or list bullet
// changes the next layer.
Self::EnumNumbering(_) => Postcondition::Unsafe,
// These are expressions that can be replaced by other expressions.
Self::Ident(_)
| Self::Bool(_)
| Self::Int(_)
| Self::Float(_)
| Self::Length(_, _)
| Self::Angle(_, _)
| Self::Percentage(_)
| Self::Str(_)
| Self::Fraction(_)
| Self::Array
| Self::Dict
| Self::Group => Postcondition::AtomicPrimary,
Self::Call
| Self::Unary
| Self::Binary
| Self::CallArgs
| Self::Named
| Self::Spread => Postcondition::UnsafeLayer,
// The closure is a bit magic with the let expression, and also it
// is not atomic.
Self::Closure | Self::ClosureParams => Postcondition::UnsafeLayer,
// These can appear as bodies and would trigger an error if they
// became something else.
Self::Template => Postcondition::SameKind(None),
Self::Block => Postcondition::SameKind(Some(TokenMode::Code)),
Self::ForExpr
| Self::WhileExpr
| Self::IfExpr
| Self::LetExpr
| Self::SetExpr
| Self::ImportExpr
| Self::IncludeExpr => Postcondition::AtomicPrimary,
Self::WithExpr | Self::ForPattern | Self::ImportItems => {
Postcondition::UnsafeLayer
}
// These can appear everywhere and must not change to other stuff
// because that could change the outer expression.
Self::LineComment | Self::BlockComment => Postcondition::SameKind(None),
Self::Error(_, _) | Self::Unknown(_) => Postcondition::Unsafe,
}
}
/// The appropriate precondition for the type.
pub fn pre(&self) -> Precondition {
match self {
Self::Heading | Self::Enum | Self::List => Precondition::AtStart,
Self::TextInLine(_) => Precondition::NotAtStart,
Self::Linebreak => Precondition::RightWhitespace,
_ => Precondition::None,
}
}
}
/// This enum describes what conditions a node has for being replaced by a new
/// parse result.
///
/// Safe nodes are replaced by the new parse result from the respective mode.
/// They can be replaced by multiple tokens. If a token is inserted in Markup
/// mode and the next token would not be `at_start` there needs to be a forward
/// check for a `EnsureAtStart` node. If this fails, the parent has to be
/// reparsed. if the direct whitespace sibling of a `EnsureRightWhitespace` is
/// `Unsafe`. Similarly, if a `EnsureRightWhitespace` token is one of the last
/// tokens to be inserted, the edit is invalidated if there is no following
/// whitespace. The atomic nodes may only be replaced by other atomic nodes. The
/// unsafe layers cannot be used but allow children access, the unsafe nodes do
/// neither.
///
/// *Procedure:*
/// 1. Check if the node is safe - if unsafe layer recurse, if unsafe, return
/// None.
/// 2. Reparse with appropriate node kind and `at_start`.
/// 3. Check whether the topmost group is terminated and the range was
/// completely consumed, otherwise return None.
/// 4. Check if the type criteria are met.
/// 5. If the node is not at the end of the tree, check if Strings etc. are
/// terminated.
/// 6. If this is markup, check the following things:
/// - The `at_start` conditions of the next non-comment and non-space(0) node
/// are met.
/// - The first node is whitespace or the previous siblings are not
/// `EnsureRightWhitespace`.
/// - If any of those fails, return None.
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
pub enum Postcondition {
/// Changing this node can never have an influence on the other nodes.
Safe,
/// This node has to be replaced with a single token of the same kind.
SameKind(Option<TokenMode>),
/// Changing this node into a single atomic expression is allowed if it
/// appears in code mode, otherwise it is safe.
AtomicPrimary,
/// Changing an unsafe layer node changes what the parents or the
/// surrounding nodes would be and is therefore disallowed. Change the
/// parents or children instead. If it appears in Markup, however, it is
/// safe to change.
UnsafeLayer,
/// Changing an unsafe node or any of its children will trigger undefined
/// behavior. Change the parents instead.
Unsafe,
}
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
pub enum Precondition {
/// These nodes depend on being at the start of a line. Reparsing of safe
/// left neighbors has to check this invariant. Otherwise, this node is
/// safe.
AtStart,
/// These nodes depend on not being at the start of a line. Reparsing of
/// safe left neighbors has to check this invariant. Otherwise, this node is
/// safe.
NotAtStart,
/// These nodes must be followed by whitespace.
RightWhitespace,
/// No additional requirements.
None,
}
impl Postcondition {
pub fn unsafe_interior(&self) -> bool {
match self {
Self::Unsafe => true,
_ => false,
}
}
pub fn markup_safe(&self) -> bool {
match self {
Self::Safe | Self::UnsafeLayer => true,
Self::SameKind(tm) => tm.map_or(false, |tm| tm != TokenMode::Markup),
_ => false,
}
}
}

View File

@ -2,6 +2,7 @@
pub mod ast;
mod highlight;
mod incremental;
mod pretty;
mod span;
@ -10,16 +11,14 @@ use std::ops::Range;
use std::rc::Rc;
pub use highlight::*;
pub use incremental::*;
pub use pretty::*;
pub use span::*;
use self::ast::{MathNode, RawNode, TypedNode};
use crate::diag::Error;
use crate::geom::{AngularUnit, LengthUnit};
use crate::parse::{
parse_atomic, parse_block, parse_comment, parse_markup, parse_markup_elements,
parse_template, TokenMode,
};
use crate::parse::TokenMode;
use crate::source::SourceId;
use crate::util::EcoString;
@ -87,29 +86,6 @@ impl Green {
Self::Token(data) => data.kind = kind,
}
}
/// Find the innermost child that is incremental safe.
fn incremental(
&mut self,
edit: &str,
replace: Span,
replacement_len: usize,
offset: usize,
parent_mode: TokenMode,
outermost: bool,
) -> Result<Range<usize>, ()> {
match self {
Green::Node(n) => Rc::make_mut(n).incremental_int(
edit,
replace,
replacement_len,
offset,
parent_mode,
outermost,
),
Green::Token(_) => Err(()),
}
}
}
impl Default for Green {
@ -194,8 +170,22 @@ impl GreenNode {
self.children[child_idx_range.clone()].iter().map(Green::len).sum();
let new_len: usize = replacement.iter().map(Green::len).sum();
if self.erroneous {
if self.children[child_idx_range.clone()].iter().any(Green::erroneous) {
// the old range was erroneous but we do not know if anywhere
// else was so we have to iterate over the whole thing.
self.erroneous = self.children[.. child_idx_range.start]
.iter()
.any(Green::erroneous)
|| self.children[child_idx_range.end ..].iter().any(Green::erroneous);
}
// in this case nothing changes so we do not have to bother.
}
// the or assignment operator is not lazy.
self.erroneous = self.erroneous || replacement.iter().any(Green::erroneous);
self.children.splice(child_idx_range, replacement);
self.erroneous = self.children.iter().any(|x| x.erroneous());
self.data.set_len(self.data.len + new_len - old_len);
}
@ -203,250 +193,6 @@ impl GreenNode {
self.data.len = self.data.len() + new_len - old_len;
self.erroneous = self.children.iter().any(|x| x.erroneous());
}
/// Find the innermost child that is incremental safe.
pub fn incremental(
&mut self,
src: &str,
replace: Span,
replacement_len: usize,
) -> Result<Range<usize>, ()> {
self.incremental_int(src, replace, replacement_len, 0, TokenMode::Markup, true)
}
fn incremental_int(
&mut self,
src: &str,
replace: Span,
replacement_len: usize,
mut offset: usize,
parent_mode: TokenMode,
outermost: bool,
) -> Result<Range<usize>, ()> {
let kind = self.kind().clone();
let mode = kind.mode().contextualize(parent_mode);
let mut loop_result = None;
let mut child_at_start = true;
let last = self.children.len() - 1;
let mut start = None;
for (i, child) in self.children.iter_mut().enumerate() {
let child_span = Span::new(replace.source, offset, offset + child.len());
if child_span.surrounds(replace)
&& start.is_none()
&& ((replace.start != child_span.end && replace.end != child_span.start)
|| mode == TokenMode::Code
|| i == last)
{
let old_len = child.len();
// First, we try if the child has another, more specific applicable child.
if !kind.incremental_safety().unsafe_interior() {
if let Ok(range) = child.incremental(
src,
replace,
replacement_len,
offset,
kind.mode().child_mode(),
i == last && outermost,
) {
let new_len = child.len();
self.update_child_len(new_len, old_len);
return Ok(range);
}
}
// This didn't work, so we try to replace the child at this
// level.
let (function, policy) =
child.kind().reparsing_function(kind.mode().child_mode());
let function = function?;
loop_result = Some((
i .. i + 1,
child_span,
i == last && outermost,
function,
policy,
));
break;
} else if start.is_none()
&& child_span.contains(replace.start)
&& mode == TokenMode::Markup
&& child.kind().incremental_safety().markup_safe()
{
start = Some((i, offset));
} else if child_span.contains(replace.end)
&& (replace.end != child_span.end || i == last)
&& mode == TokenMode::Markup
&& child.kind().incremental_safety().markup_safe()
{
if let Some((start, start_offset)) = start {
let (function, policy) =
child.kind().reparsing_function(kind.mode().child_mode());
let function = function?;
loop_result = Some((
start .. i + 1,
Span::new(replace.source, start_offset, offset + child.len()),
i == last && outermost,
function,
policy,
));
}
break;
} else if start.is_some()
&& (mode != TokenMode::Markup
|| !child.kind().incremental_safety().markup_safe())
{
break;
}
offset += child.len();
child_at_start = child.kind().is_at_start(child_at_start);
}
// We now have a child that we can replace and a function to do so if
// the loop found any results at all.
let (child_idx_range, child_span, child_outermost, func, policy) =
loop_result.ok_or(())?;
let src_span = child_span.inserted(replace, replacement_len);
let recompile_range = if policy == IncrementalSafety::AtomicPrimary {
src_span.start .. src.len()
} else {
src_span.to_range()
};
let (mut new_children, unterminated) =
func(&src[recompile_range], child_at_start).ok_or(())?;
let insertion = match check_invariants(
&new_children,
self.children(),
unterminated,
child_idx_range.clone(),
child_outermost,
child_at_start,
mode,
src_span,
policy,
) {
InvariantResult::Ok => Ok(new_children),
InvariantResult::UseFirst => Ok(vec![std::mem::take(&mut new_children[0])]),
InvariantResult::Error => Err(()),
}?;
self.replace_child_range(child_idx_range, insertion);
Ok(src_span.to_range())
}
}
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
enum InvariantResult {
Ok,
UseFirst,
Error,
}
fn check_invariants(
use_children: &[Green],
old_children: &[Green],
unterminated: bool,
child_idx_range: Range<usize>,
outermost: bool,
child_at_start: bool,
mode: TokenMode,
src_span: Span,
policy: IncrementalSafety,
) -> InvariantResult {
let (new_children, ok) = if policy == IncrementalSafety::AtomicPrimary {
if use_children.iter().map(Green::len).sum::<usize>() == src_span.len() {
(use_children, InvariantResult::Ok)
} else if use_children[0].len() == src_span.len() {
(&use_children[0 .. 1], InvariantResult::UseFirst)
} else {
return InvariantResult::Error;
}
} else {
(use_children, InvariantResult::Ok)
};
let child_mode = old_children[child_idx_range.start].kind().mode().child_mode();
// Check if the children / child has the right type.
let require_single = match policy {
IncrementalSafety::AtomicPrimary | IncrementalSafety::SameKind => true,
IncrementalSafety::SameKindInCode if child_mode == TokenMode::Code => true,
_ => false,
};
if require_single {
if new_children.len() != 1 {
return InvariantResult::Error;
}
if match policy {
IncrementalSafety::SameKind => true,
IncrementalSafety::SameKindInCode => child_mode == TokenMode::Code,
_ => false,
} {
if old_children[child_idx_range.start].kind() != new_children[0].kind() {
return InvariantResult::Error;
}
}
}
// Do not accept unclosed nodes if the old node did not use to be at the
// right edge of the tree.
if !outermost && unterminated {
return InvariantResult::Error;
}
// Check if the neighbor invariants are still true.
if mode == TokenMode::Markup {
if child_idx_range.start > 0 {
if old_children[child_idx_range.start - 1].kind().incremental_safety()
== IncrementalSafety::EnsureRightWhitespace
&& !new_children[0].kind().is_whitespace()
{
return InvariantResult::Error;
}
}
let mut new_at_start = child_at_start;
for child in new_children {
new_at_start = child.kind().is_at_start(new_at_start);
}
for child in &old_children[child_idx_range.end ..] {
if child.kind().is_trivia() {
new_at_start = child.kind().is_at_start(new_at_start);
continue;
}
match child.kind().incremental_safety() {
IncrementalSafety::EnsureAtStart if !new_at_start => {
return InvariantResult::Error;
}
IncrementalSafety::EnsureNotAtStart if new_at_start => {
return InvariantResult::Error;
}
_ => {}
}
break;
}
if new_children.last().map(|x| x.kind().incremental_safety())
== Some(IncrementalSafety::EnsureRightWhitespace)
&& old_children.len() > child_idx_range.end
{
if !old_children[child_idx_range.end].kind().is_whitespace() {
return InvariantResult::Error;
}
}
}
ok
}
impl From<GreenNode> for Green {
@ -1053,190 +799,6 @@ impl NodeKind {
}
}
pub fn reparsing_function(
&self,
parent_mode: TokenMode,
) -> (
Result<fn(&str, bool) -> Option<(Vec<Green>, bool)>, ()>,
IncrementalSafety,
) {
let policy = self.incremental_safety();
if policy.is_unsafe() {
return (Err(()), policy);
}
let contextualized = self.mode().contextualize(parent_mode);
let is_code = contextualized == TokenMode::Code;
if is_code && policy == IncrementalSafety::UnsafeLayer {
return (Err(()), policy);
}
if is_code && policy == IncrementalSafety::AtomicPrimary {
return (Ok(parse_atomic), policy);
}
if policy == IncrementalSafety::SameKind
|| (policy == IncrementalSafety::SameKindInCode && is_code)
{
let parser: fn(&str, bool) -> _ = match self {
NodeKind::Template => parse_template,
NodeKind::Block => parse_block,
NodeKind::LineComment | NodeKind::BlockComment => parse_comment,
_ => return (Err(()), policy),
};
return (Ok(parser), policy);
}
let parser: fn(&str, bool) -> _ = match contextualized {
TokenMode::Markup if self == &Self::Markup => parse_markup,
TokenMode::Markup => parse_markup_elements,
_ => return (Err(()), policy),
};
(Ok(parser), policy)
}
/// Whether it is safe to do incremental parsing on this node. Never allow
/// non-termination errors if this is not already the last leaf node.
pub fn incremental_safety(&self) -> IncrementalSafety {
match self {
// Replacing parenthesis changes if the expression is balanced and
// is therefore not safe.
Self::LeftBracket
| Self::RightBracket
| Self::LeftBrace
| Self::RightBrace
| Self::LeftParen
| Self::RightParen => IncrementalSafety::Unsafe,
// Replacing an operator can change whether the parent is an
// operation which makes it unsafe. The star can appear in markup.
Self::Star
| Self::Comma
| Self::Semicolon
| Self::Colon
| Self::Plus
| Self::Minus
| Self::Slash
| Self::Eq
| Self::EqEq
| Self::ExclEq
| Self::Lt
| Self::LtEq
| Self::Gt
| Self::GtEq
| Self::PlusEq
| Self::HyphEq
| Self::StarEq
| Self::SlashEq
| Self::Not
| Self::And
| Self::Or
| Self::With
| Self::Dots
| Self::Arrow => IncrementalSafety::Unsafe,
// These keywords are literals and can be safely be substituted with
// other expressions.
Self::None | Self::Auto => IncrementalSafety::AtomicPrimary,
// These keywords change what kind of expression the parent is and
// how far the expression would go.
Self::Let
| Self::If
| Self::Else
| Self::For
| Self::In
| Self::While
| Self::Break
| Self::Continue
| Self::Return
| Self::Set
| Self::Import
| Self::Include
| Self::From => IncrementalSafety::Unsafe,
// This is a backslash followed by a space. But changing it to
// anything else is fair game.
Self::Linebreak => IncrementalSafety::EnsureRightWhitespace,
Self::Markup => IncrementalSafety::SameKind,
Self::Space(_) => IncrementalSafety::SameKindInCode,
// These are all replaceable by other tokens.
Self::Parbreak
| Self::Text(_)
| Self::NonBreakingSpace
| Self::EnDash
| Self::EmDash
| Self::Escape(_)
| Self::Strong
| Self::Emph => IncrementalSafety::Safe,
// This is text that needs to be not `at_start`, otherwise it would
// start one of the below items.
Self::TextInLine(_) => IncrementalSafety::EnsureNotAtStart,
// These have to be `at_start` so they must be preceeded with a
// Space(n) with n > 0 or a Parbreak.
Self::Heading | Self::Enum | Self::List => IncrementalSafety::EnsureAtStart,
// Changing the heading level, enum numbering, or list bullet
// changes the next layer.
Self::EnumNumbering(_) => IncrementalSafety::Unsafe,
Self::Raw(_) | Self::Math(_) => IncrementalSafety::Safe,
// These are expressions that can be replaced by other expressions.
Self::Ident(_)
| Self::Bool(_)
| Self::Int(_)
| Self::Float(_)
| Self::Length(_, _)
| Self::Angle(_, _)
| Self::Percentage(_)
| Self::Str(_)
| Self::Fraction(_)
| Self::Array
| Self::Dict
| Self::Group => IncrementalSafety::AtomicPrimary,
Self::Call | Self::Unary | Self::Binary | Self::SetExpr => {
IncrementalSafety::UnsafeLayer
}
Self::CallArgs | Self::Named | Self::Spread => IncrementalSafety::UnsafeLayer,
// The closure is a bit magic with the let expression, and also it
// is not atomic.
Self::Closure | Self::ClosureParams => IncrementalSafety::UnsafeLayer,
// These can appear as bodies and would trigger an error if they
// became something else.
Self::Template | Self::Block => IncrementalSafety::SameKindInCode,
Self::ForExpr
| Self::WhileExpr
| Self::IfExpr
| Self::LetExpr
| Self::ImportExpr
| Self::IncludeExpr => IncrementalSafety::AtomicPrimary,
Self::WithExpr | Self::ForPattern | Self::ImportItems => {
IncrementalSafety::UnsafeLayer
}
// These can appear everywhere and must not change to other stuff
// because that could change the outer expression.
Self::LineComment | Self::BlockComment => IncrementalSafety::SameKind,
Self::Error(_, _) | Self::Unknown(_) => IncrementalSafety::Unsafe,
}
}
/// A human-readable name for the kind.
pub fn as_str(&self) -> &'static str {
match self {
@ -1351,95 +913,6 @@ impl Display for NodeKind {
}
}
/// This enum describes what conditions a node has for being replaced by a new
/// parse result.
///
/// Safe nodes are replaced by the new parse result from the respective mode.
/// They can be replaced by multiple tokens. If a token is inserted in Markup
/// mode and the next token would not be `at_start` there needs to be a forward
/// check for a `EnsureAtStart` node. If this fails, the parent has to be
/// reparsed. if the direct whitespace sibling of a `EnsureRightWhitespace` is
/// `Unsafe`. Similarly, if a `EnsureRightWhitespace` token is one of the last
/// tokens to be inserted, the edit is invalidated if there is no following
/// whitespace. The atomic nodes may only be replaced by other atomic nodes. The
/// unsafe layers cannot be used but allow children access, the unsafe nodes do
/// neither.
///
/// *Procedure:*
/// 1. Check if the node is safe - if unsafe layer recurse, if unsafe, return
/// None.
/// 2. Reparse with appropriate node kind and `at_start`.
/// 3. Check whether the topmost group is terminated and the range was
/// completely consumed, otherwise return None.
/// 4. Check if the type criteria are met.
/// 5. If the node is not at the end of the tree, check if Strings etc. are
/// terminated.
/// 6. If this is markup, check the following things:
/// - The `at_start` conditions of the next non-comment and non-space(0) node
/// are met.
/// - The first node is whitespace or the previous siblings are not
/// `EnsureRightWhitespace`.
/// - If any of those fails, return None.
#[derive(Debug, Copy, Clone, Eq, PartialEq, Hash)]
pub enum IncrementalSafety {
/// Changing this node can never have an influence on the other nodes.
Safe,
/// This node has to be replaced with a single token of the same kind.
SameKind,
/// This node has to be replaced with a single token of the same kind if in
/// code mode.
SameKindInCode,
/// These nodes depend on being at the start of a line. Reparsing of safe
/// left neighbors has to check this invariant. Otherwise, this node is
/// safe.
EnsureAtStart,
/// These nodes depend on not being at the start of a line. Reparsing of
/// safe left neighbors has to check this invariant. Otherwise, this node is
/// safe.
EnsureNotAtStart,
/// These nodes must be followed by whitespace.
EnsureRightWhitespace,
/// Changing this node into a single atomic expression is allowed if it
/// appears in code mode, otherwise it is safe.
AtomicPrimary,
/// Changing an unsafe layer node changes what the parents or the
/// surrounding nodes would be and is therefore disallowed. Change the
/// parents or children instead. If it appears in Markup, however, it is
/// safe to change.
UnsafeLayer,
/// Changing an unsafe node or any of its children will trigger undefined
/// behavior. Change the parents instead.
Unsafe,
}
impl IncrementalSafety {
pub fn unsafe_interior(&self) -> bool {
match self {
Self::Unsafe => true,
_ => false,
}
}
pub fn is_unsafe(&self) -> bool {
match self {
Self::UnsafeLayer | Self::Unsafe => true,
_ => false,
}
}
pub fn markup_safe(&self) -> bool {
match self {
Self::Safe
| Self::SameKindInCode
| Self::EnsureAtStart
| Self::EnsureNotAtStart
| Self::EnsureRightWhitespace
| Self::UnsafeLayer => true,
_ => false,
}
}
}
/// This enum describes which mode a token of [`NodeKind`] can appear in.
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
pub enum NodeMode {