Migrate diagnostics list output to use icu list formatter.

2022-11-06 14:43:25 +08:00 · 2022-11-06 14:43:25 +08:00 · a775004322
commit a775004322
parent 42d3bda08c
7 changed files with 296 additions and 28 deletions
--- a/Cargo.lock
+++ b/Cargo.lock
@ -211,7 +211,7 @@ checksum = "ba3569f383e8f1598449f1a423e72e99569137b47740b1da11ef19af3d5c3223"
 dependencies = [
 "lazy_static",
 "memchr",
- "regex-automata",
+ "regex-automata 0.1.10",
 ]

 [[package]]
@ -1707,6 +1707,73 @@ version = "2.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3c1ad908cc71012b7bea4d0c53ba96a8cba9962f048fa68d143376143d863b7a"

+[[package]]
+name = "icu_list"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c40218275f081c4493f190357c5395647b06734c2dc3dcb41cc099a0f60168b1"
+dependencies = [
+ "displaydoc",
+ "icu_locid",
+ "icu_provider",
+ "regex-automata 0.2.0",
+ "writeable",
+ "zerovec",
+]
+
+[[package]]
+name = "icu_locid"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "34b3de5d99a0e275fe6193b9586dbf37364daebc0d39c89b5cf8376a53b789e8"
+dependencies = [
+ "displaydoc",
+ "litemap",
+ "tinystr",
+ "writeable",
+ "zerovec",
+]
+
+[[package]]
+name = "icu_provider"
+version = "1.0.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2f911086e3c521a8a824d4f8bfd87769645ced2f07ff913b521c0d793be07100"
+dependencies = [
+ "displaydoc",
+ "icu_locid",
+ "icu_provider_macros",
+ "stable_deref_trait",
+ "writeable",
+ "yoke",
+ "zerofrom",
+ "zerovec",
+]
+
+[[package]]
+name = "icu_provider_adapters"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "980c71d8a91b246ebbb97847178a4b816eea39d1d550c70ee566384555bb6545"
+dependencies = [
+ "icu_locid",
+ "icu_provider",
+ "tinystr",
+ "yoke",
+ "zerovec",
+]
+
+[[package]]
+name = "icu_provider_macros"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "38cf6f5b65cf81f0b4298da647101acbfe6ae0e25263f92bd7a22597e9d6d606"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
 [[package]]
 name = "idna"
 version = "0.2.0"
@ -2034,6 +2101,12 @@ dependencies = [
 "walkdir",
 ]

+[[package]]
+name = "litemap"
+version = "0.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f34a3f4798fac63fb48cf277eefa38f94d3443baff555bb98e4f56bc9092368e"
+
 [[package]]
 name = "lld-wrapper"
 version = "0.1.0"
@ -2100,7 +2173,7 @@ version = "0.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "8263075bb86c5a1b1427b5ae862e8889656f126e9f77c484496e8b47cf5c5558"
 dependencies = [
- "regex-automata",
+ "regex-automata 0.1.10",
 ]

 [[package]]
@ -2927,6 +3000,15 @@ dependencies = [
 "regex-syntax",
 ]

+[[package]]
+name = "regex-automata"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e9368763f5a9b804326f3af749e16f9abf378d227bcdee7634b13d8f17793782"
+dependencies = [
+ "memchr",
+]
+
 [[package]]
 name = "regex-syntax"
 version = "0.6.26"
@ -3202,6 +3284,16 @@ dependencies = [
 "rustc_span",
 ]

+[[package]]
+name = "rustc_baked_icu_data"
+version = "0.0.0"
+dependencies = [
+ "icu_list",
+ "icu_locid",
+ "icu_provider",
+ "litemap",
+]
+
 [[package]]
 name = "rustc_borrowck"
 version = "0.0.0"
@ -3422,13 +3514,18 @@ version = "0.0.0"
 dependencies = [
 "fluent-bundle",
 "fluent-syntax",
+ "icu_list",
+ "icu_locid",
+ "icu_provider_adapters",
 "intl-memoizer",
+ "rustc_baked_icu_data",
 "rustc_data_structures",
 "rustc_macros",
 "rustc_serialize",
 "rustc_span",
 "tracing",
 "unic-langid",
+ "writeable",
 ]

 [[package]]
@ -4934,6 +5031,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "f8aeafdfd935e4a7fe16a91ab711fa52d54df84f9c8f7ca5837a9d1d902ef4c2"
 dependencies = [
 "displaydoc",
+ "zerovec",
 ]

 [[package]]
@ -5546,6 +5644,12 @@ version = "0.42.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "f40009d85759725a34da6d89a94e63d7bdc50a862acf0dbc7c8e488f1edcb6f5"

+[[package]]
+name = "writeable"
+version = "0.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f8e6ab4f5da1b24daf2c590cfac801bacb27b15b4f050e84eb60149ea726f06b"
+
 [[package]]
 name = "xattr"
 version = "0.2.2"
@ -5598,3 +5702,71 @@ checksum = "fe5c30ade05e61656247b2e334a031dfd0cc466fadef865bdcdea8d537951bf1"
 dependencies = [
 "winapi",
 ]
+
+[[package]]
+name = "yoke"
+version = "0.6.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1fe1d55ca72c32d573bfbd5cb2f0ca65a497854c44762957a6d3da96041a5184"
+dependencies = [
+ "serde",
+ "stable_deref_trait",
+ "yoke-derive",
+ "zerofrom",
+]
+
+[[package]]
+name = "yoke-derive"
+version = "0.6.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1346e4cd025ae818b88566eac7eb65ab33a994ea55f355c86889af2e7e56b14e"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+ "synstructure",
+]
+
+[[package]]
+name = "zerofrom"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "79e9355fccf72b04b7deaa99ce7a0f6630530acf34045391b74460fcd714de54"
+dependencies = [
+ "zerofrom-derive",
+]
+
+[[package]]
+name = "zerofrom-derive"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2e8aa86add9ddbd2409c1ed01e033cd457d79b1b1229b64922c25095c595e829"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+ "synstructure",
+]
+
+[[package]]
+name = "zerovec"
+version = "0.9.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b9d919a74c17749ccb17beaf6405562e413cd94e98ba52ca1e64bbe7eefbd8b8"
+dependencies = [
+ "yoke",
+ "zerofrom",
+ "zerovec-derive",
+]
+
+[[package]]
+name = "zerovec-derive"
+version = "0.9.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "490e5f878c2856225e884c35927e7ea6db3c24cdb7229b72542c7526ad7ed49e"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+ "synstructure",
+]
--- a/compiler/rustc_baked_icu_data/Cargo.toml
+++ b/compiler/rustc_baked_icu_data/Cargo.toml
@ -5,5 +5,6 @@ edition = "2021"

 [dependencies]
 icu_list = "1.0.0"
+icu_locid = "1.0.0"
 icu_provider = "1.0.1"
 litemap = "0.6.0"
--- a/compiler/rustc_baked_icu_data/src/lib.rs
+++ b/compiler/rustc_baked_icu_data/src/lib.rs
@ -2,16 +2,30 @@
 //!
 #![allow(elided_lifetimes_in_paths)]

-// generated with:
-// ```text
-// icu4x-datagen -W --pretty --fingerprint --use-separate-crates --cldr-tag latest --icuexport-tag latest \
-// --format mod -l en es fr it ja pt ru tr zh-Hans zh-Hant -k list/and@1 -o src/data
-// ```
+/* generated with:
+```text
+icu4x-datagen -W --pretty --fingerprint --use-separate-crates --cldr-tag latest --icuexport-tag latest \
+--format mod -l en es fr it ja pt ru tr zh-Hans zh-Hant -k list/and@1 -o src/data
+```
+*/

 mod data;

 pub use data::BakedDataProvider;

-pub fn baked_data_provider() -> BakedDataProvider {
+pub const fn baked_data_provider() -> BakedDataProvider {
    data::BakedDataProvider
 }
+
+pub mod supported_locales {
+    pub const EN: icu_locid::Locale = icu_locid::locale!("en");
+    pub const ES: icu_locid::Locale = icu_locid::locale!("es");
+    pub const FR: icu_locid::Locale = icu_locid::locale!("fr");
+    pub const IT: icu_locid::Locale = icu_locid::locale!("it");
+    pub const JA: icu_locid::Locale = icu_locid::locale!("ja");
+    pub const PT: icu_locid::Locale = icu_locid::locale!("pt");
+    pub const RU: icu_locid::Locale = icu_locid::locale!("ru");
+    pub const TR: icu_locid::Locale = icu_locid::locale!("tr");
+    pub const ZH_HANS: icu_locid::Locale = icu_locid::locale!("zh-Hans");
+    pub const ZH_HANT: icu_locid::Locale = icu_locid::locale!("zh-Hant");
+}
--- a/compiler/rustc_error_messages/Cargo.toml
+++ b/compiler/rustc_error_messages/Cargo.toml
@ -9,9 +9,14 @@ edition = "2021"
 fluent-bundle = "0.15.2"
 fluent-syntax = "0.11"
 intl-memoizer = "0.5.1"
+rustc_baked_icu_data = { path = "../rustc_baked_icu_data" }
 rustc_data_structures = { path = "../rustc_data_structures" }
 rustc_serialize = { path = "../rustc_serialize" }
 rustc_span = { path = "../rustc_span" }
 rustc_macros = { path = "../rustc_macros" }
 tracing = "0.1"
 unic-langid = { version = "0.9.0", features = ["macros"] }
+icu_list = "1.0.0"
+writeable = "0.5.0"
+icu_locid = "1.0.0"
+icu_provider_adapters = "1.0.0"
--- a/compiler/rustc_error_messages/src/lib.rs
+++ b/compiler/rustc_error_messages/src/lib.rs
@ -10,6 +10,7 @@

 use fluent_bundle::FluentResource;
 use fluent_syntax::parser::ParserError;
+use icu_provider_adapters::fallback::{LocaleFallbackProvider, LocaleFallbacker};
 use rustc_data_structures::sync::Lrc;
 use rustc_macros::{fluent_messages, Decodable, Encodable};
 use rustc_span::Span;
@ -30,8 +31,7 @@
 #[cfg(not(parallel_compiler))]
 use intl_memoizer::IntlLangMemoizer;

-pub use fluent_bundle::{self, FluentArgs, FluentError, FluentValue};
-
+pub use fluent_bundle::{self, types::FluentType, FluentArgs, FluentError, FluentValue};
 pub use unic_langid::{langid, LanguageIdentifier};

 // Generates `DEFAULT_LOCALE_RESOURCES` static and `fluent_generated` module.
@ -541,3 +541,91 @@ fn from(spans: Vec<Span>) -> MultiSpan {
        MultiSpan::from_spans(spans)
    }
 }
+
+fn icu_locale_from_unic_langid(lang: LanguageIdentifier) -> Option<icu_locid::Locale> {
+    icu_locid::Locale::try_from_bytes(lang.to_string().as_bytes()).ok()
+}
+
+pub fn fluent_value_from_str_list_sep_by_and<'source>(
+    l: Vec<Cow<'source, str>>,
+) -> FluentValue<'source> {
+    // Fluent requires 'static value here for its AnyEq usages.
+    #[derive(Clone, PartialEq, Debug)]
+    struct FluentStrListSepByAnd(Vec<String>);
+
+    impl FluentType for FluentStrListSepByAnd {
+        fn duplicate(&self) -> Box<dyn FluentType + Send> {
+            Box::new(self.clone())
+        }
+
+        fn as_string(&self, intls: &intl_memoizer::IntlLangMemoizer) -> Cow<'static, str> {
+            let result = intls
+                .with_try_get::<MemoizableListFormatter, _, _>((), |list_formatter| {
+                    list_formatter.format_to_string(self.0.iter())
+                })
+                .unwrap();
+            Cow::Owned(result)
+        }
+
+        #[cfg(not(parallel_compiler))]
+        fn as_string_threadsafe(
+            &self,
+            _intls: &intl_memoizer::concurrent::IntlLangMemoizer,
+        ) -> Cow<'static, str> {
+            unreachable!("`as_string_threadsafe` is not used in non-parallel rustc")
+        }
+
+        #[cfg(parallel_compiler)]
+        fn as_string_threadsafe(
+            &self,
+            intls: &intl_memoizer::concurrent::IntlLangMemoizer,
+        ) -> Cow<'static, str> {
+            let result = intls
+                .with_try_get::<MemoizableListFormatter, _, _>((), |list_formatter| {
+                    list_formatter.format_to_string(self.0.iter())
+                })
+                .unwrap();
+            Cow::Owned(result)
+        }
+    }
+
+    struct MemoizableListFormatter(icu_list::ListFormatter);
+
+    impl std::ops::Deref for MemoizableListFormatter {
+        type Target = icu_list::ListFormatter;
+        fn deref(&self) -> &Self::Target {
+            &self.0
+        }
+    }
+
+    impl intl_memoizer::Memoizable for MemoizableListFormatter {
+        type Args = ();
+        type Error = ();
+
+        fn construct(lang: LanguageIdentifier, _args: Self::Args) -> Result<Self, Self::Error>
+        where
+            Self: Sized,
+        {
+            let baked_data_provider = rustc_baked_icu_data::baked_data_provider();
+            let locale_fallbacker = LocaleFallbacker::try_new_unstable(&baked_data_provider);
+            let data_provider = LocaleFallbackProvider::new_with_fallbacker(
+                &baked_data_provider,
+                locale_fallbacker,
+            );
+            let locale = icu_locale_from_unic_langid(lang)
+                .unwrap_or_else(|| rustc_baked_icu_data::supported_locales::EN);
+            let list_formatter = icu_list::ListFormatter::try_new_and_with_length_unstable(
+                &data_provider,
+                &locale.into(),
+                icu_list::ListLength::Wide,
+            )
+            .expect("Failed to create list formatter");
+
+            Ok(MemoizableListFormatter(list_formatter))
+        }
+    }
+
+    let l = l.into_iter().map(|x| x.into_owned()).collect();
+
+    FluentValue::Custom(Box::new(FluentStrListSepByAnd(l)))
+}
--- a/compiler/rustc_errors/src/diagnostic.rs
+++ b/compiler/rustc_errors/src/diagnostic.rs
@ -4,6 +4,7 @@
    SubdiagnosticMessage, Substitution, SubstitutionPart, SuggestionStyle,
 };
 use rustc_data_structures::fx::FxHashMap;
+use rustc_error_messages::fluent_value_from_str_list_sep_by_and;
 use rustc_error_messages::FluentValue;
 use rustc_lint_defs::{Applicability, LintExpectationId};
 use rustc_span::edition::LATEST_STABLE_EDITION;
@ -34,6 +35,7 @@
 pub enum DiagnosticArgValue<'source> {
    Str(Cow<'source, str>),
    Number(usize),
+    StrListSepByAnd(Vec<Cow<'source, str>>),
 }

 /// Converts a value of a type into a `DiagnosticArg` (typically a field of an `IntoDiagnostic`
@ -58,6 +60,7 @@ fn into(self) -> FluentValue<'source> {
        match self {
            DiagnosticArgValue::Str(s) => From::from(s),
            DiagnosticArgValue::Number(n) => From::from(n),
+            DiagnosticArgValue::StrListSepByAnd(l) => fluent_value_from_str_list_sep_by_and(l),
        }
    }
 }
--- a/compiler/rustc_errors/src/diagnostic_impls.rs
+++ b/compiler/rustc_errors/src/diagnostic_impls.rs
@ -11,7 +11,6 @@
 use rustc_target::spec::{PanicStrategy, SplitDebuginfo, StackProtector, TargetTriple};
 use std::borrow::Cow;
 use std::fmt;
-use std::fmt::Write;
 use std::num::ParseIntError;
 use std::path::{Path, PathBuf};
 use std::process::ExitStatus;
@ -191,23 +190,9 @@ fn from(v: Vec<Symbol>) -> Self {

 impl IntoDiagnosticArg for DiagnosticSymbolList {
    fn into_diagnostic_arg(self) -> DiagnosticArgValue<'static> {
-        // FIXME: replace the logic here with a real list formatter
-        let symbols = match &self.0[..] {
-            [symbol] => format!("`{symbol}`"),
-            [symbol, last] => {
-                format!("`{symbol}` and `{last}`",)
-            }
-            [symbols @ .., last] => {
-                let mut result = String::new();
-                for symbol in symbols {
-                    write!(result, "`{symbol}`, ").unwrap();
-                }
-                write!(result, "and `{last}`").unwrap();
-                result
-            }
-            [] => unreachable!(),
-        };
-        DiagnosticArgValue::Str(Cow::Owned(symbols))
+        DiagnosticArgValue::StrListSepByAnd(
+            self.0.into_iter().map(|sym| Cow::Owned(format!("`{sym}`"))).collect(),
+        )
    }
 }