feat: add ruff --statistics (#2284)

Closes #2284.

ruff_cli/src/args.rs

@@ -211,6 +211,15 @@ pub struct CheckArgs {
     update_check: bool,
     #[clap(long, overrides_with("update_check"), hide = true)]
     no_update_check: bool,
+    /// Show counts for every rule with at least one violation.
+    #[arg(
+        long,
+        // Unsupported default-command arguments.
+        conflicts_with = "diff",
+        conflicts_with = "show_source",
+        conflicts_with = "watch",
+    )]
+    pub statistics: bool,
     /// Enable automatic additions of `noqa` directives to failing lines.
     #[arg(
         long,
@@ -218,6 +227,7 @@ pub struct CheckArgs {
         conflicts_with = "show_files",
         conflicts_with = "show_settings",
         // Unsupported default-command arguments.
+        conflicts_with = "statistics",
         conflicts_with = "stdin_filename",
         conflicts_with = "watch",
     )]
@@ -230,6 +240,7 @@ pub struct CheckArgs {
         // conflicts_with = "show_files",
         conflicts_with = "show_settings",
         // Unsupported default-command arguments.
+        conflicts_with = "statistics",
         conflicts_with = "stdin_filename",
         conflicts_with = "watch",
     )]
@@ -242,6 +253,7 @@ pub struct CheckArgs {
         conflicts_with = "show_files",
         // conflicts_with = "show_settings",
         // Unsupported default-command arguments.
+        conflicts_with = "statistics",
         conflicts_with = "stdin_filename",
         conflicts_with = "watch",
     )]
@@ -316,6 +328,7 @@ impl CheckArgs {
                 no_cache: self.no_cache,
                 show_files: self.show_files,
                 show_settings: self.show_settings,
+                statistics: self.statistics,
                 stdin_filename: self.stdin_filename,
                 watch: self.watch,
             },
@@ -371,6 +384,7 @@ pub struct Arguments {
     pub no_cache: bool,
     pub show_files: bool,
     pub show_settings: bool,
+    pub statistics: bool,
     pub stdin_filename: Option<PathBuf>,
     pub watch: bool,
 }

ruff_cli/src/main.rs

@@ -244,7 +244,11 @@ fn check(args: CheckArgs, log_level: LogLevel) -> Result<ExitCode> {
         // unless we're writing fixes via stdin (in which case, the transformed
         // source code goes to stdout).
         if !(is_stdin && matches!(autofix, fix::FixMode::Apply | fix::FixMode::Diff)) {
-            printer.write_once(&diagnostics)?;
+            if cli.statistics {
+                printer.write_statistics(&diagnostics)?;
+            } else {
+                printer.write_once(&diagnostics)?;
+            }
         }
 
         // Check for updates if we're in a non-silent log level.

ruff_cli/src/printer.rs

@@ -43,6 +43,13 @@ struct ExpandedMessage<'a> {
     filename: &'a str,
 }
 
+#[derive(Serialize)]
+struct ExpandedStatistics<'a> {
+    count: usize,
+    code: &'a str,
+    message: String,
+}
+
 struct SerializeRuleAsCode<'a>(&'a Rule);
 
 impl Serialize for SerializeRuleAsCode<'_> {
@@ -336,6 +343,77 @@ impl<'a> Printer<'a> {
         Ok(())
     }
 
+    pub fn write_statistics(&self, diagnostics: &Diagnostics) -> Result<()> {
+        let mut violations = diagnostics
+            .messages
+            .iter()
+            .map(|message| message.kind.rule())
+            .collect::<Vec<_>>();
+        violations.sort();
+        violations.dedup();
+
+        let statistics = violations
+            .iter()
+            .map(|rule| ExpandedStatistics {
+                code: rule.code(),
+                count: diagnostics
+                    .messages
+                    .iter()
+                    .filter(|message| message.kind.rule() == *rule)
+                    .count(),
+                message: diagnostics
+                    .messages
+                    .iter()
+                    .find(|message| message.kind.rule() == *rule)
+                    .map(|message| message.kind.body())
+                    .unwrap(),
+            })
+            .collect::<Vec<_>>();
+
+        let mut stdout = BufWriter::new(io::stdout().lock());
+        match self.format {
+            SerializationFormat::Text => {
+                // Compute the maximum number of digits in the count and code, for all messages, to enable
+                // pretty-printing.
+                let count_width = num_digits(
+                    statistics
+                        .iter()
+                        .map(|statistic| statistic.count)
+                        .max()
+                        .unwrap(),
+                );
+                let code_width = statistics
+                    .iter()
+                    .map(|statistic| statistic.code.len())
+                    .max()
+                    .unwrap();
+
+                // By default, we mimic Flake8's `--statistics` format.
+                for msg in statistics {
+                    writeln!(
+                        stdout,
+                        "{:>count_width$}\t{:<code_width$}\t{}",
+                        msg.count, msg.code, msg.message
+                    )?;
+                }
+                return Ok(());
+            }
+            SerializationFormat::Json => {
+                writeln!(stdout, "{}", serde_json::to_string_pretty(&statistics)?)?;
+            }
+            _ => {
+                anyhow::bail!(
+                    "Unsupported serialization format for statistics: {:?}",
+                    self.format
+                )
+            }
+        }
+
+        stdout.flush()?;
+
+        Ok(())
+    }
+
     pub fn write_continuously(&self, diagnostics: &Diagnostics) -> Result<()> {
         if matches!(self.log_level, LogLevel::Silent) {
             return Ok(());

ruff_cli/tests/integration_test.rs

@@ -182,3 +182,21 @@ fn explain_status_codes() -> Result<()> {
     cmd.args(["--explain", "RUF404"]).assert().failure();
     Ok(())
 }
+
+#[test]
+fn show_statistics() -> Result<()> {
+    let mut cmd = Command::cargo_bin(BIN_NAME)?;
+    let output = cmd
+        .args(["-", "--format", "text", "--select", "F401", "--statistics"])
+        .write_stdin("import sys\nimport os\n\nprint(os.getuid())\n")
+        .assert()
+        .failure();
+    assert_eq!(
+        str::from_utf8(&output.get_output().stdout)?
+            .lines()
+            .last()
+            .unwrap(),
+        "1\tF401\t`sys` imported but unused"
+    );
+    Ok(())
+}