Move lint definition generation into xtask/codegen

Author: Veykril

Cargo.lock

@@ -44,13 +44,10 @@ line-index.workspace = true
 
 [dev-dependencies]
 expect-test = "1.4.0"
-oorandom = "11.1.3"
-xshell.workspace = true
 
 # local deps
 test-utils.workspace = true
 test-fixture.workspace = true
-sourcegen.workspace = true
 
 [lints]
 workspace = true

crates/ide-db/Cargo.toml

@@ -412,9 +412,3 @@ impl SnippetCap {
         }
     }
 }
-
-#[cfg(test)]
-mod tests {
-    mod line_index;
-    mod sourcegen_lints;
-}

crates/ide-db/src/lib.rs

@@ -1,6 +1,4 @@
 #![allow(clippy::print_stderr)]
-#[cfg(not(feature = "in-rust-tree"))]
-mod sourcegen;
 
 use ide_db::{
     assists::AssistResolveStrategy, base_db::SourceDatabaseExt, LineIndexDatabase, RootDatabase,

crates/ide-db/src/tests/line_index.rs

@@ -10,5 +10,8 @@ edition = "2021"
 text-size = "1.1.1"
 nohash-hasher = "0.2.0"
 
+[dev-dependencies]
+oorandom = "11.1.3"
+
 [lints]
-workspace = true
+workspace = true

crates/ide-diagnostics/src/tests.rs

@@ -142,3 +142,56 @@ fn test_to_wide() {
     let wide_line_col = line_index.to_wide(WideEncoding::Utf16, line_col.unwrap());
     assert_eq!(wide_line_col, Some(WideLineCol { line: 5, col: 4 }));
 }
+
+#[test]
+fn test_every_chars() {
+    let text: String = {
+        let mut chars: Vec<char> = ((0 as char)..char::MAX).collect(); // Neat!
+        chars.extend("\n".repeat(chars.len() / 16).chars());
+        let seed = std::hash::Hasher::finish(&std::hash::BuildHasher::build_hasher(
+            #[allow(clippy::disallowed_types)]
+            &std::collections::hash_map::RandomState::new(),
+        ));
+        let mut rng = oorandom::Rand32::new(seed);
+        let mut rand_index = |i| rng.rand_range(0..i as u32) as usize;
+        let mut remaining = chars.len() - 1;
+        while remaining > 0 {
+            let index = rand_index(remaining);
+            chars.swap(remaining, index);
+            remaining -= 1;
+        }
+        chars.into_iter().collect()
+    };
+    assert!(text.contains('💩')); // Sanity check.
+
+    let line_index = LineIndex::new(&text);
+
+    let mut lin_col = LineCol { line: 0, col: 0 };
+    let mut col_utf16 = 0;
+    let mut col_utf32 = 0;
+    for (offset, c) in text.char_indices() {
+        let got_offset = line_index.offset(lin_col).unwrap();
+        assert_eq!(usize::from(got_offset), offset);
+
+        let got_lin_col = line_index.line_col(got_offset);
+        assert_eq!(got_lin_col, lin_col);
+
+        for (enc, col) in [(WideEncoding::Utf16, col_utf16), (WideEncoding::Utf32, col_utf32)] {
+            let wide_lin_col = line_index.to_wide(enc, lin_col).unwrap();
+            let got_lin_col = line_index.to_utf8(enc, wide_lin_col).unwrap();
+            assert_eq!(got_lin_col, lin_col);
+            assert_eq!(wide_lin_col.col, col)
+        }
+
+        if c == '\n' {
+            lin_col.line += 1;
+            lin_col.col = 0;
+            col_utf16 = 0;
+            col_utf32 = 0;
+        } else {
+            lin_col.col += c.len_utf8() as u32;
+            col_utf16 += c.len_utf16() as u32;
+            col_utf32 += 1;
+        }
+    }
+}

lib/line-index/Cargo.toml

@@ -9,14 +9,20 @@ use crate::{flags, project_root};
 
 pub(crate) mod assists_doc_tests;
 pub(crate) mod diagnostics_docs;
+mod lints;
 
 impl flags::Codegen {
     pub(crate) fn run(self, _sh: &Shell) -> anyhow::Result<()> {
         match self.codegen_type.unwrap_or_default() {
             flags::CodegenType::All => {
+                diagnostics_docs::generate(self.check);
                 assists_doc_tests::generate(self.check);
+                // lints::generate(self.check) Updating clones the rust repo, so don't run it unless
+                // explicitly asked for
             }
             flags::CodegenType::AssistsDocTests => assists_doc_tests::generate(self.check),
+            flags::CodegenType::DiagnosticsDocs => diagnostics_docs::generate(self.check),
+            flags::CodegenType::LintDefinitions => lints::generate(self.check),
         }
         Ok(())
     }

lib/line-index/src/tests.rs

@@ -7,7 +7,7 @@ use crate::{
     project_root,
 };
 
-fn generate(check: bool) {
+pub(crate) fn generate(check: bool) {
     let diagnostics = Diagnostic::collect().unwrap();
     if !check {
         let contents =

xtask/src/codegen.rs

@@ -2,18 +2,18 @@
 //! and lints from rustc, rustdoc, and clippy.
 use std::{borrow::Cow, fs, path::Path};
 
-use itertools::Itertools;
 use stdx::format_to;
-use test_utils::project_root;
 use xshell::{cmd, Shell};
 
+use crate::{
+    codegen::{add_preamble, ensure_file_contents, list_files, reformat},
+    project_root,
+};
+
 const DESTINATION: &str = "crates/ide-db/src/generated/lints.rs";
 
-/// This clones rustc repo, and so is not worth to keep up-to-date. We update
-/// manually by un-ignoring the test from time to time.
-#[test]
-#[ignore]
-fn sourcegen_lint_completions() {
+/// This clones rustc repo, and so is not worth to keep up-to-date on a constant basis.
+pub(crate) fn generate(check: bool) {
     let sh = &Shell::new().unwrap();
 
     let rust_repo = project_root().join("./target/rust");
@@ -73,10 +73,10 @@ pub struct LintGroup {
     .unwrap();
     generate_descriptor_clippy(&mut contents, &lints_json);
 
-    let contents = sourcegen::add_preamble("sourcegen_lints", sourcegen::reformat(contents));
+    let contents = add_preamble("sourcegen_lints", reformat(contents));
 
     let destination = project_root().join(DESTINATION);
-    sourcegen::ensure_file_contents(destination.as_path(), &contents);
+    ensure_file_contents(destination.as_path(), &contents, check);
 }
 
 /// Parses the output of `rustdoc -Whelp` and prints `Lint` and `LintGroup` constants into `buf`.
@@ -130,10 +130,9 @@ fn generate_lint_descriptor(sh: &Shell, buf: &mut String) {
         )
     });
 
-    let lints = lints
-        .chain(lint_groups)
-        .sorted_by(|(ident, ..), (ident2, ..)| ident.cmp(ident2))
-        .collect::<Vec<_>>();
+    let mut lints = lints.chain(lint_groups).collect::<Vec<_>>();
+    lints.sort_by(|(ident, ..), (ident2, ..)| ident.cmp(ident2));
+
     for (name, description, ..) in &lints {
         push_lint_completion(buf, &name.replace('-', "_"), description);
     }
@@ -177,10 +176,8 @@ fn generate_lint_descriptor(sh: &Shell, buf: &mut String) {
             )
         });
 
-    let lints_rustdoc = lints_rustdoc
-        .chain(lint_groups_rustdoc)
-        .sorted_by(|(ident, ..), (ident2, ..)| ident.cmp(ident2))
-        .collect::<Vec<_>>();
+    let mut lints_rustdoc = lints_rustdoc.chain(lint_groups_rustdoc).collect::<Vec<_>>();
+    lints_rustdoc.sort_by(|(ident, ..), (ident2, ..)| ident.cmp(ident2));
 
     for (name, description, ..) in &lints_rustdoc {
         push_lint_completion(buf, &name.replace('-', "_"), description)
@@ -212,7 +209,7 @@ fn find_and_slice<'a>(i: &'a str, p: &str) -> &'a str {
 fn generate_feature_descriptor(buf: &mut String, src_dir: &Path) {
     let mut features = ["language-features", "library-features"]
         .into_iter()
-        .flat_map(|it| sourcegen::list_files(&src_dir.join(it)))
+        .flat_map(|it| list_files(&src_dir.join(it)))
         // Get all `.md` files
         .filter(|path| path.extension() == Some("md".as_ref()))
         .map(|path| {
@@ -302,7 +299,7 @@ fn generate_descriptor_clippy(buf: &mut String, path: &Path) {
         let children = children.iter().map(|id| format!("clippy::{id}")).collect::<Vec<_>>();
         if !children.is_empty() {
             let lint_ident = format!("clippy::{id}");
-            let description = format!("lint group for: {}", children.iter().join(", "));
+            let description = format!("lint group for: {}", children.join(", "));
             push_lint_group(buf, &lint_ident, &description, &children);
         }
     }
@@ -331,7 +328,10 @@ fn push_lint_group(buf: &mut String, label: &str, description: &str, children: &
 
     push_lint_completion(buf, label, description);
 
-    let children = format!("&[{}]", children.iter().map(|it| format!("\"{it}\"")).join(", "));
+    let children = format!(
+        "&[{}]",
+        children.iter().map(|it| format!("\"{it}\"")).collect::<Vec<_>>().join(", ")
+    );
     format_to!(
         buf,
         r###"