feat: implement json type returner

crates/cli/Cargo.toml

@@ -36,6 +36,10 @@ harness = false
 name = "ui_github"
 harness = false
 
+[[test]]
+name = "ui_json"
+harness = false
+
 [features]
 python = ["sqruff-lib/python"]
 codegen-docs = ["clap-markdown", "minijinja", "serde", "python"]

crates/cli/src/commands.rs

@@ -51,6 +51,7 @@ pub(crate) struct FixArgs {
 pub(crate) enum Format {
     Human,
     GithubAnnotationNative,
+    Json,
 }
 
 impl Default for Format {

crates/cli/src/main.rs

@@ -1,6 +1,7 @@
 use clap::Parser as _;
 use commands::Format;
 use sqruff_lib::cli::formatters::Formatter;
+use sqruff_lib::cli::json::JsonFormatter;
 use sqruff_lib::cli::{
     formatters::OutputStreamFormatter,
     github_annotation_native_formatter::GithubAnnotationNativeFormatter,
@@ -103,6 +104,10 @@ pub(crate) fn linter(config: FluffConfig, format: Format) -> Linter {
             let formatter = GithubAnnotationNativeFormatter::new(output_stream);
             Arc::new(formatter)
         }
+        Format::Json => {
+            let formatter = JsonFormatter::default();
+            Arc::new(formatter)
+        }
     };
 
     Linter::new(config, Some(formatter), None)

crates/cli/tests/json/LT01_noqa.exitcode

@@ -0,0 +1 @@
+0

crates/cli/tests/json/LT01_noqa.sql

@@ -0,0 +1 @@
+SELECT   1; --noqa: LT01

crates/cli/tests/json/LT01_noqa.stdout

@@ -0,0 +1 @@
+{"tests/lint/LT01_noqa.sql":[]}

crates/cli/tests/json/hql_file.exitcode

@@ -0,0 +1 @@
+1

crates/cli/tests/json/hql_file.hql

@@ -0,0 +1 @@
+SELECT   1;

crates/cli/tests/json/hql_file.stdout

@@ -0,0 +1 @@
+{"tests/lint/hql_file.hql":[{"range":{"start":{"line":1,"character":7},"end":{"line":1,"character":7}},"message":"Expected only single space before \"1\". Found \"   \".","severity":"Error","source":"sqruff"},{"range":{"start":{"line":1,"character":11},"end":{"line":1,"character":11}},"message":"Files must end with a single trailing newline.","severity":"Error","source":"sqruff"}]}

crates/cli/tests/json/test_fail_whitespace_before_comma.exitcode

@@ -0,0 +1 @@
+1

crates/cli/tests/json/test_fail_whitespace_before_comma.sql

@@ -0,0 +1 @@
+SELECT 1 ,4

crates/cli/tests/json/test_fail_whitespace_before_comma.stdout

@@ -0,0 +1 @@
+{"tests/lint/test_fail_whitespace_before_comma.sql":[{"range":{"start":{"line":1,"character":8},"end":{"line":1,"character":8}},"message":"Column expression without alias. Use explicit `AS` clause.","severity":"Error","source":"sqruff"},{"range":{"start":{"line":1,"character":11},"end":{"line":1,"character":11}},"message":"Column expression without alias. Use explicit `AS` clause.","severity":"Error","source":"sqruff"},{"range":{"start":{"line":1,"character":9},"end":{"line":1,"character":9}},"message":"Unexpected whitespace before comma.","severity":"Error","source":"sqruff"},{"range":{"start":{"line":1,"character":11},"end":{"line":1,"character":11}},"message":"Expected single whitespace between \",\" and \"4\".","severity":"Error","source":"sqruff"},{"range":{"start":{"line":1,"character":12},"end":{"line":1,"character":12}},"message":"Files must end with a single trailing newline.","severity":"Error","source":"sqruff"}]}

crates/cli/tests/ui_json.rs

@@ -0,0 +1,67 @@
+use std::fs;
+use std::path::PathBuf;
+
+use assert_cmd::Command;
+use expect_test::expect_file;
+
+fn main() {
+    let profile = if cfg!(debug_assertions) {
+        "debug"
+    } else {
+        "release"
+    };
+    let mut lint_dir = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
+    lint_dir.push("tests/json");
+
+    // Iterate over each test file in the directory
+    for entry in fs::read_dir(&lint_dir).unwrap() {
+        let entry = entry.unwrap();
+        let path = entry.path();
+
+        // Check if the file has a .sql or .hql extension
+        if path
+            .extension()
+            .and_then(|e| e.to_str())
+            .map_or(false, |ext| ext == "sql" || ext == "hql")
+        {
+            // Construct the path to the sqruff binary
+            let mut sqruff_path = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
+            sqruff_path.push(format!("../../target/{}/sqruff", profile));
+
+            // Set up the command with arguments
+            let mut cmd = Command::new(sqruff_path);
+
+            cmd.arg("lint");
+            cmd.arg(path.to_str().unwrap());
+            cmd.arg("-f");
+            cmd.arg("json");
+            // Set the HOME environment variable to the fake home directory
+            cmd.env("HOME", PathBuf::from(env!("CARGO_MANIFEST_DIR")));
+
+            // Run the command and capture the output
+            let assert = cmd.assert();
+
+            // Construct the expected output file paths
+            let mut expected_output_path_stderr = path.clone();
+            expected_output_path_stderr.set_extension("stderr");
+            let mut expected_output_path_stdout = path.clone();
+            expected_output_path_stdout.set_extension("stdout");
+            let mut expected_output_path_exitcode = path.clone();
+            expected_output_path_exitcode.set_extension("exitcode");
+
+            // Read the expected output
+            let output = assert.get_output();
+            let stderr_str = std::str::from_utf8(&output.stderr).unwrap();
+            let stdout_str = std::str::from_utf8(&output.stdout).unwrap();
+            let exit_code_str = output.status.code().unwrap().to_string();
+
+            let test_dir_str = lint_dir.to_string_lossy().to_string();
+            let stderr_normalized: String = stderr_str.replace(&test_dir_str, "tests/lint");
+            let stdout_normalized: String = stdout_str.replace(&test_dir_str, "tests/lint");
+
+            expect_file![expected_output_path_stderr].assert_eq(&stderr_normalized);
+            expect_file![expected_output_path_stdout].assert_eq(&stdout_normalized);
+            expect_file![expected_output_path_exitcode].assert_eq(&exit_code_str);
+        }
+    }
+}

crates/lib/Cargo.toml

@@ -34,8 +34,6 @@ name = "depth_map"
 harness = false
 
 [features]
-serde = ["dep:serde"]
-stringify = ["dep:serde_yaml", "serde"]
 python = ["pyo3"]
 
 [dependencies]
@@ -64,8 +62,9 @@ nohash-hasher = "0.2.0"
 rustc-hash = "2.1.0"
 strum_macros = "0.26.4"
 strum = "0.26.3"
-serde = { version = "1.0", features = ["derive"], optional = true }
+serde = { version = "1.0", features = ["derive"] }
 serde_yaml = { version = "0.9.34", optional = true }
+serde_json = "1"
 append-only-vec = "0.1.5"
 
 # Only activated on python
@@ -81,7 +80,7 @@ jemallocator = { version = "0.6.0", package = "tikv-jemallocator" }
 pprof = { version = "0.14", features = ["flamegraph", "criterion"] }
 
 [dev-dependencies]
-sqruff-lib = { path = ".", features = ["serde"] }
+sqruff-lib = { path = "." }
 serde_yaml = "0.9.34"
 criterion = "0.5"
 expect-test = "1.5"

crates/lib/src/cli.rs

@@ -1,2 +1,4 @@
 pub mod formatters;
 pub mod github_annotation_native_formatter;
+pub mod json;
+pub mod json_types;

crates/lib/src/cli/json.rs

@@ -0,0 +1,50 @@
+use std::sync::Mutex;
+
+use crate::core::{config::FluffConfig, linter::linted_file::LintedFile};
+
+use super::{
+    formatters::Formatter,
+    json_types::{Diagnostic, DiagnosticCollection, DiagnosticSeverity},
+};
+
+#[derive(Default)]
+pub struct JsonFormatter {
+    violations: Mutex<DiagnosticCollection>,
+}
+
+impl Formatter for JsonFormatter {
+    fn dispatch_file_violations(&self, linted_file: &LintedFile, only_fixable: bool) {
+        let violations = linted_file.get_violations(only_fixable.then_some(true));
+        let mut lock = self.violations.lock().unwrap();
+        lock.entry(linted_file.path.clone()).or_default().extend(
+            violations
+                .iter()
+                .map(|err| Diagnostic::from(err.clone()))
+                .collect::<Vec<_>>(),
+        );
+    }
+
+    fn has_fail(&self) -> bool {
+        let lock = self.violations.lock().unwrap();
+        lock.values().any(|v| {
+            v.iter()
+                .any(|d| matches!(&d.severity, DiagnosticSeverity::Error))
+        })
+    }
+
+    fn completion_message(&self) {
+        let lock = self.violations.lock().unwrap();
+        let json = serde_json::to_string(&*lock).unwrap();
+        println!("{}", json);
+    }
+
+    fn dispatch_template_header(
+        &self,
+        _f_name: String,
+        _linter_config: FluffConfig,
+        _file_config: FluffConfig,
+    ) {
+    }
+
+    fn dispatch_parse_header(&self, _f_name: String) {}
+}

crates/lib/src/cli/json_types.rs

@@ -0,0 +1,100 @@
+use std::collections::BTreeMap;
+
+use serde::Serialize;
+use sqruff_lib_core::errors::SQLBaseError;
+
+impl From<SQLBaseError> for Diagnostic {
+    fn from(value: SQLBaseError) -> Self {
+        Diagnostic {
+            range: Range {
+                start: Position::new(value.line_no as u32, value.line_pos as u32),
+                end: Position::new(value.line_no as u32, value.line_pos as u32),
+            },
+            message: value.description,
+            severity: if value.warning {
+                DiagnosticSeverity::Warning
+            } else {
+                DiagnosticSeverity::Error
+            },
+            source: Some("sqruff".to_string()),
+            // code: todo!(),
+            // source: Some(value.get_source().to_string()),
+            // code: Some(DiagnosticCode {
+            //     value: value.rule_code().to_string(),
+            //     target: Uri::new("".to_string()),
+            // }),
+            // related_information: Vec::new(),
+            // tags: Vec::new(),
+        }
+    }
+}
+
+/// Represents a line and character position, such as the position of the cursor.
+#[derive(Serialize)]
+struct Position {
+    /// The zero-based line value.
+    line: u32,
+    /// The zero-based character value.
+    character: u32,
+}
+
+impl Position {
+    /// Creates a new `Position` instance.
+    fn new(line: u32, character: u32) -> Self {
+        Self { line, character }
+    }
+}
+
+/// A range represents an ordered pair of two positions. It is guaranteed that `start` is before or equal to `end`.
+#[derive(Serialize)]
+struct Range {
+    /// The start position. It is before or equal to `end`.
+    start: Position,
+    /// The end position. It is after or equal to `start`.
+    end: Position,
+}
+
+/// Represents a diagnostic, such as a compiler error or warning. Diagnostic objects are only valid in the scope of a file.
+#[derive(Serialize)]
+pub struct Diagnostic {
+    /// The range to which this diagnostic applies.
+    range: Range,
+    /// The human-readable message.
+    message: String,
+    /// The severity, default is {@link DiagnosticSeverity::Error error}.
+    pub severity: DiagnosticSeverity,
+    /// A human-readable string describing the source of this diagnostic, e.g. 'typescript' or 'super lint'.
+    source: Option<String>,
+    // A code or identifier for this diagnostic. Should be used for later processing, e.g. when providing {@link CodeActionContext code actions}.
+    // code: Option<DiagnosticCode>,
+    // TODO Maybe implement
+    // An array of related diagnostic information, e.g. when symbol-names within a scope collide all definitions can be marked via this property.
+    // related_information: Vec<DiagnosticRelatedInformation>,
+    // Additional metadata about the diagnostic.
+    // tags: Vec<DiagnosticTag>,
+}
+
+// /// Represents a related message and source code location for a diagnostic. This should be used to point to code locations that cause or are related to a diagnostics, e.g when duplicating a symbol in a scope.
+// #[derive(Serialize)]
+// struct DiagnosticCode {
+//     /// A code or identifier for this diagnostic.
+//     value: String,
+//     // TODO Maybe implement
+//     // A target URI to open with more information about the diagnostic error.
+//     // target: Uri,
+// }
+
+/// Represents the severity of diagnostics.
+#[derive(Serialize)]
+pub enum DiagnosticSeverity {
+    /// Something not allowed by the rules of a language or other means.
+    Error = 0,
+    /// Something suspicious but allowed.
+    Warning = 1,
+    /// Something to inform about but not a problem.
+    Information = 2,
+    /// Something to hint to a better way of doing it, like proposing a refactoring.
+    Hint = 3,
+}
+
+pub type DiagnosticCollection = BTreeMap<String, Vec<Diagnostic>>;

crates/lib/src/core/config.rs

@@ -458,9 +458,8 @@ impl ConfigLoader {
     }
 }
 
-#[derive(Debug, Clone, PartialEq, Default)]
-#[cfg_attr(any(test, feature = "serde"), derive(serde::Deserialize))]
-#[cfg_attr(any(test, feature = "serde"), serde(untagged))]
+#[derive(Debug, Clone, PartialEq, Default, serde::Deserialize)]
+#[serde(untagged)]
 pub enum Value {
     Int(i32),
     Bool(bool),

docs/cli.md

@@ -43,7 +43,7 @@ lint files
 
   Default value: `human`
 
-  Possible values: `human`, `github-annotation-native`
+  Possible values: `human`, `github-annotation-native`, `json`
 
 
 
@@ -65,7 +65,7 @@ fix files
 
   Default value: `human`
 
-  Possible values: `human`, `github-annotation-native`
+  Possible values: `human`, `github-annotation-native`, `json`