Run indentation tests on a part of the Helix source code.

Add C++ indent test file.

4 files changed, 211 insertions, 23 deletions

diff --git a/helix-core/tests/data/indent/cpp.cpp b/helix-core/tests/data/indent/cpp.cpp
new file mode 100644
index 00000000..6e7f3a88
--- /dev/null
+++ b/helix-core/tests/data/indent/cpp.cpp
@@ -0,0 +1,48 @@
+std::vector<std::string>
+fn_with_many_parameters(int parm1, long parm2, float parm3, double parm4,
+                        char* parm5, bool parm6);
+
+std::vector<std::string>
+fn_with_many_parameters(int parm1, long parm2, float parm3, double parm4,
+                        char* parm5, bool parm6) {
+  auto lambda = []() {
+    return 0;
+  };
+  auto lambda_with_a_really_long_name_that_uses_a_whole_line
+    = [](int some_more_aligned_parameters,
+         std::string parm2) {
+      do_smth();
+    };
+  if (brace_on_same_line) {
+    do_smth();
+  } else if (brace_on_next_line)
+  {
+    do_smth();
+  } else if (another_condition) {
+    do_smth();
+  }
+  else {
+    do_smth();
+  }
+  if (inline_if_statement)
+    do_smth();
+  if (another_inline_if_statement)
+    return [](int parm1, char* parm2) {
+      this_is_a_really_pointless_lambda();
+    };
+
+  switch (var) {
+  case true:
+    return -1;
+  case false:
+    return 42;
+  }
+}
+
+class MyClass : public MyBaseClass {
+public:
+  MyClass();
+  void public_fn();
+private:
+  super_secret_private_fn();
+}
diff --git a/helix-core/tests/data/indent/indent.rs b/helix-core/tests/data/indent/indent.rs
deleted file mode 120000
index 2ac16cf9..00000000
--- a/helix-core/tests/data/indent/indent.rs
+++ /dev/null
@@ -1 +0,0 @@
-../../../src/indent.rs
\ No newline at end of file
diff --git a/helix-core/tests/data/indent/languages.toml b/helix-core/tests/data/indent/languages.toml
index 3206f124..fa02e451 100644
--- a/helix-core/tests/data/indent/languages.toml
+++ b/helix-core/tests/data/indent/languages.toml
@@ -11,3 +11,16 @@ indent = { tab-width = 4, unit = "    " }
 [[grammar]]
 name = "rust"
 source = { git = "https://github.com/tree-sitter/tree-sitter-rust", rev = "0431a2c60828731f27491ee9fdefe25e250ce9c9" }
+
+[[language]]
+name = "cpp"
+scope = "source.cpp"
+injection-regex = "cpp"
+file-types = ["cc", "hh", "c++", "cpp", "hpp", "h", "ipp", "tpp", "cxx", "hxx", "ixx", "txx", "ino", "C", "H"]
+roots = []
+comment-token = "//"
+indent = { tab-width = 2, unit = "  " }
+
+[[grammar]]
+name = "cpp"
+source = { git = "https://github.com/tree-sitter/tree-sitter-cpp", rev = "2d2c4aee8672af4c7c8edff68e7dd4c07e88d2b1" }
diff --git a/helix-core/tests/indent.rs b/helix-core/tests/indent.rs
index 1dec9989..26010c90 100644
--- a/helix-core/tests/indent.rs
+++ b/helix-core/tests/indent.rs
@@ -1,20 +1,122 @@
 use helix_core::{
     indent::{indent_level_for_line, treesitter_indent_for_pos, IndentStyle},
-    syntax::Loader,
+    syntax::{Configuration, Loader},
     Syntax,
 };
-use std::path::PathBuf;
+use ropey::Rope;
+use std::{ops::Range, path::PathBuf, process::Command};
 
 #[test]
 fn test_treesitter_indent_rust() {
-    test_treesitter_indent("rust.rs", "source.rust");
+    standard_treesitter_test("rust.rs", "source.rust");
 }
+
+#[test]
+fn test_treesitter_indent_cpp() {
+    standard_treesitter_test("cpp.cpp", "source.cpp");
+}
+
 #[test]
-fn test_treesitter_indent_rust_2() {
-    test_treesitter_indent("indent.rs", "source.rust");
-    // TODO Use commands.rs as indentation test.
-    // Currently this fails because we can't align the parameters of a closure yet
-    // test_treesitter_indent("commands.rs", "source.rust");
+fn test_treesitter_indent_rust_helix() {
+    // We pin a specific git revision to prevent unrelated changes from causing the indent tests to fail.
+    // Ideally, someone updates this once in a while and fixes any errors that occur.
+    let rev = "af382768cdaf89ff547dbd8f644a1bddd90e7c8f";
+    let files = Command::new("git")
+        .args([
+            "ls-tree",
+            "-r",
+            "--name-only",
+            "--full-tree",
+            rev,
+            "helix-term/src",
+        ])
+        .output()
+        .unwrap();
+    let files = String::from_utf8(files.stdout).unwrap();
+
+    let ignored_files = vec![
+        // Contains many macros that tree-sitter does not parse in a meaningful way and is otherwise not very interesting
+        "helix-term/src/health.rs",
+    ];
+
+    for file in files.split_whitespace() {
+        if ignored_files.contains(&file) {
+            continue;
+        }
+        let ignored_lines: Vec<Range<usize>> = match file {
+            "helix-term/src/application.rs" => vec![
+                // We can't handle complicated indent rules inside macros (`json!` in this case) since
+                // the tree-sitter grammar only parses them as `token_tree` and `identifier` nodes.
+                1045..1051,
+            ],
+            "helix-term/src/commands.rs" => vec![
+                // This is broken because of the current handling of `call_expression`
+                // (i.e. having an indent query for it but outdenting again in specific cases).
+                // The indent query is needed to correctly handle multi-line arguments in function calls
+                // inside indented `field_expression` nodes (which occurs fairly often).
+                //
+                // Once we have the `@indent.always` capture type, it might be possible to just have an indent
+                // capture for the `arguments` field of a call expression. That could enable us to correctly
+                // handle this.
+                2226..2230,
+            ],
+            "helix-term/src/commands/dap.rs" => vec![
+                // Complex `format!` macro
+                46..52,
+            ],
+            "helix-term/src/commands/lsp.rs" => vec![
+                // Macro
+                624..627,
+                // Return type declaration of a closure. `cargo fmt` adds an additional space here,
+                // which we cannot (yet) model with our indent queries.
+                878..879,
+                // Same as in `helix-term/src/commands.rs`
+                1335..1343,
+            ],
+            "helix-term/src/config.rs" => vec![
+                // Multiline string
+                146..152,
+            ],
+            "helix-term/src/keymap.rs" => vec![
+                // Complex macro (see above)
+                456..470,
+                // Multiline string without indent
+                563..567,
+            ],
+            "helix-term/src/main.rs" => vec![
+                // Multiline string
+                44..70,
+            ],
+            "helix-term/src/ui/completion.rs" => vec![
+                // Macro
+                218..232,
+            ],
+            "helix-term/src/ui/editor.rs" => vec![
+                // The chained function calls here are not indented, probably because of the comment
+                // in between. Since `cargo fmt` doesn't even attempt to format it, there's probably
+                // no point in trying to indent this correctly.
+                342..350,
+            ],
+            "helix-term/src/ui/lsp.rs" => vec![
+                // Macro
+                56..61,
+            ],
+            "helix-term/src/ui/statusline.rs" => vec![
+                // Same as in `helix-term/src/commands.rs`
+                436..442,
+                450..456,
+            ],
+            _ => Vec::new(),
+        };
+
+        let git_object = rev.to_string() + ":" + file;
+        let content = Command::new("git")
+            .args(["cat-file", "blob", &git_object])
+            .output()
+            .unwrap();
+        let doc = Rope::from_reader(&mut content.stdout.as_slice()).unwrap();
+        test_treesitter_indent(file, doc, "source.rust", ignored_lines);
+    }
 }
 
 #[test]
@@ -50,20 +152,41 @@ fn test_indent_level_for_line_with_spaces_and_tabs() {
     assert_eq!(indent_level, 2)
 }
 
-fn test_treesitter_indent(file_name: &str, lang_scope: &str) {
+fn indent_tests_dir() -> PathBuf {
     let mut test_dir = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
     test_dir.push("tests/data/indent");
+    test_dir
+}
+
+fn indent_test_path(name: &str) -> PathBuf {
+    let mut path = indent_tests_dir();
+    path.push(name);
+    path
+}
 
-    let mut test_file = test_dir.clone();
-    test_file.push(file_name);
-    let test_file = std::fs::File::open(test_file).unwrap();
+fn indent_tests_config() -> Configuration {
+    let mut config_path = indent_tests_dir();
+    config_path.push("languages.toml");
+    let config = std::fs::read_to_string(config_path).unwrap();
+    toml::from_str(&config).unwrap()
+}
+
+fn standard_treesitter_test(file_name: &str, lang_scope: &str) {
+    let test_path = indent_test_path(file_name);
+    let test_file = std::fs::File::open(test_path).unwrap();
     let doc = ropey::Rope::from_reader(test_file).unwrap();
+    test_treesitter_indent(file_name, doc, lang_scope, Vec::new())
+}
 
-    let mut config_file = test_dir;
-    config_file.push("languages.toml");
-    let config = std::fs::read_to_string(config_file).unwrap();
-    let config = toml::from_str(&config).unwrap();
-    let loader = Loader::new(config);
+/// Test that all the lines in the given file are indented as expected.
+/// ignored_lines is a list of (1-indexed) line ranges that are excluded from this test.
+fn test_treesitter_indent(
+    test_name: &str,
+    doc: Rope,
+    lang_scope: &str,
+    ignored_lines: Vec<std::ops::Range<usize>>,
+) {
+    let loader = Loader::new(indent_tests_config());
 
     // set runtime path so we can find the queries
     let mut runtime = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR"));
@@ -71,6 +194,7 @@ fn test_treesitter_indent(file_name: &str, lang_scope: &str) {
     std::env::set_var("HELIX_RUNTIME", runtime.to_str().unwrap());
 
     let language_config = loader.language_config_for_scope(lang_scope).unwrap();
+    let indent_style = IndentStyle::from_str(&language_config.indent.as_ref().unwrap().unit);
     let highlight_config = language_config.highlight_config(&[]).unwrap();
     let text = doc.slice(..);
     let syntax = Syntax::new(text, highlight_config, std::sync::Arc::new(loader)).unwrap();
@@ -78,14 +202,17 @@ fn test_treesitter_indent(file_name: &str, lang_scope: &str) {
 
     for i in 0..doc.len_lines() {
         let line = text.line(i);
+        if ignored_lines.iter().any(|range| range.contains(&(i + 1))) {
+            continue;
+        }
         if let Some(pos) = helix_core::find_first_non_whitespace_char(line) {
-            let tab_and_indent_width: usize = 4;
+            let tab_width: usize = 4;
             let suggested_indent = treesitter_indent_for_pos(
                 indent_query,
                 &syntax,
-                &IndentStyle::Spaces(tab_and_indent_width as u8),
-                tab_and_indent_width,
-                tab_and_indent_width,
+                &indent_style,
+                tab_width,
+                indent_style.indent_width(tab_width),
                 text,
                 i,
                 text.line_to_char(i) + pos,
@@ -94,7 +221,8 @@ fn test_treesitter_indent(file_name: &str, lang_scope: &str) {
             .unwrap();
             assert!(
                 line.get_slice(..pos).map_or(false, |s| s == suggested_indent),
-                "Wrong indentation on line {}:\n\"{}\" (original line)\n\"{}\" (suggested indentation)\n",
+                "Wrong indentation for file {:?} on line {}:\n\"{}\" (original line)\n\"{}\" (suggested indentation)\n",
+                test_name,
                 i+1,
                 line.slice(..line.len_chars()-1),
                 suggested_indent,