feat: add SQL pretty printer with keyword casing support

- Add pgls_pretty_print and pgls_pretty_print_codegen crates
- Implement SQL formatting with configurable keyword casing
- Support keyword_case and constant_case settings (default: lowercase)
- Integrate format command into CLI
- Add LSP formatting handler

Author: psteinroe

Cargo.lock

@@ -15,13 +15,20 @@ rust-version = "1.86.0"
 [workspace.dependencies]
 # supporting crates unrelated to postgres
 anyhow                   = "1.0.92"
+biome_console            = "=0.5.7"
 biome_deserialize        = "0.6.0"
 biome_deserialize_macros = "0.6.0"
+biome_diagnostics        = "=0.5.7"
 biome_js_factory         = "0.5.7"
 biome_js_formatter       = "0.5.7"
 biome_js_syntax          = "0.5.7"
-biome_rowan              = "0.5.7"
-biome_string_case        = "0.5.8"
+biome_json_parser        = "=0.5.7"
+biome_parser             = "=0.5.7"
+biome_rowan              = "=0.5.7"
+biome_string_case        = "=0.5.8"
+biome_text_edit          = "=0.5.7"
+biome_text_size          = "=0.5.7"
+biome_unicode_table      = "=0.5.7"
 bpaf                     = { version = "0.9.15", features = ["derive"] }
 criterion                = "0.5"
 crossbeam                = "0.8.4"
@@ -44,7 +51,9 @@ slotmap                  = "1.0.7"
 smallvec                 = { version = "1.13.2", features = ["union", "const_new", "serde"] }
 strum                    = { version = "0.27.1", features = ["derive"] }
 # this will use tokio if available, otherwise async-std
+camino                   = "1.1.9"
 convert_case             = "0.6.0"
+dir-test                 = "0.4.1"
 prost                    = "0.13.5"
 prost-reflect            = "0.15.3"
 protox                   = "0.8.0"
@@ -77,8 +86,9 @@ pgls_lexer                  = { path = "./crates/pgls_lexer", version = "0.0.0"
 pgls_lexer_codegen          = { path = "./crates/pgls_lexer_codegen", version = "0.0.0" }
 pgls_lsp                    = { path = "./crates/pgls_lsp", version = "0.0.0" }
 pgls_markup                 = { path = "./crates/pgls_markup", version = "0.0.0" }
-pgls_matcher                = { path = "./crates/pgls_matcher", version = "0.0.0" }
 pgls_plpgsql_check          = { path = "./crates/pgls_plpgsql_check", version = "0.0.0" }
+pgls_pretty_print           = { path = "./crates/pgls_pretty_print", version = "0.0.0" }
+pgls_pretty_print_codegen   = { path = "./crates/pgls_pretty_print_codegen", version = "0.0.0" }
 pgls_query                  = { path = "./crates/pgls_query", version = "0.0.0" }
 pgls_query_ext              = { path = "./crates/pgls_query_ext", version = "0.0.0" }
 pgls_query_macros           = { path = "./crates/pgls_query_macros", version = "0.0.0" }

Cargo.toml

@@ -0,0 +1,109 @@
+use crate::cli_options::CliOptions;
+use crate::commands::get_files_to_process_with_cli_options;
+use crate::execute::run_files;
+use crate::reporter::Report;
+use crate::{CliDiagnostic, CliSession, VcsIntegration};
+use crate::{ExecutionConfig, ExecutionMode, VcsTargeting};
+use pgls_configuration::PartialConfiguration;
+use pgls_diagnostics::category;
+use pgls_fs::FileSystem;
+use pgls_workspace::DynRef;
+use std::ffi::OsString;
+
+pub struct FormatArgs {
+    pub configuration: Option<PartialConfiguration>,
+    pub paths: Vec<OsString>,
+    pub write: bool,
+    pub staged: bool,
+    pub changed: bool,
+    pub since: Option<String>,
+}
+
+pub fn format(
+    mut session: CliSession,
+    cli_options: &CliOptions,
+    args: FormatArgs,
+) -> Result<(), CliDiagnostic> {
+    validate_args(&args)?;
+
+    let configuration = session.prepare_with_config(cli_options, args.configuration.clone())?;
+    session.setup_workspace(configuration.clone(), VcsIntegration::Enabled)?;
+
+    let paths = resolve_paths(session.fs(), &configuration, &args)?;
+
+    let vcs = VcsTargeting {
+        staged: args.staged,
+        changed: args.changed,
+    };
+
+    let mode = ExecutionMode::Format {
+        write: args.write,
+        vcs,
+    };
+    let execution = ExecutionConfig::new(mode, u32::MAX);
+
+    let report: Report = run_files(&mut session, &execution, paths)?;
+
+    let exit_result = enforce_exit_codes(cli_options, &report);
+    session.report("format", cli_options, &report)?;
+    exit_result
+}
+
+fn resolve_paths(
+    fs: &DynRef<'_, dyn FileSystem>,
+    configuration: &PartialConfiguration,
+    args: &FormatArgs,
+) -> Result<Vec<OsString>, CliDiagnostic> {
+    let mut paths = get_files_to_process_with_cli_options(
+        args.since.as_deref(),
+        args.changed,
+        args.staged,
+        fs,
+        configuration,
+    )?
+    .unwrap_or_else(|| args.paths.clone());
+
+    if paths.is_empty() {
+        if let Some(current_dir) = fs.working_directory() {
+            paths.push(current_dir.into_os_string());
+        }
+    }
+
+    Ok(paths)
+}
+
+fn enforce_exit_codes(cli_options: &CliOptions, payload: &Report) -> Result<(), CliDiagnostic> {
+    let traversal = payload.traversal.as_ref();
+    let processed = traversal.map_or(0, |t| t.changed + t.unchanged);
+    let skipped = traversal.map_or(0, |t| t.skipped);
+
+    if processed.saturating_sub(skipped) == 0 && !cli_options.no_errors_on_unmatched {
+        return Err(CliDiagnostic::no_files_processed());
+    }
+
+    let errors = payload.errors;
+    let category = category!("format");
+
+    if errors > 0 {
+        return Err(CliDiagnostic::check_error(category));
+    }
+
+    Ok(())
+}
+
+fn validate_args(args: &FormatArgs) -> Result<(), CliDiagnostic> {
+    if args.since.is_some() {
+        if !args.changed {
+            return Err(CliDiagnostic::incompatible_arguments("since", "changed"));
+        }
+        if args.staged {
+            return Err(CliDiagnostic::incompatible_arguments("since", "staged"));
+        }
+    }
+
+    if args.changed && args.staged {
+        return Err(CliDiagnostic::incompatible_arguments("changed", "staged"));
+    }
+
+    Ok(())
+}

crates/pgls_cli/src/commands/format.rs

@@ -12,6 +12,7 @@ pub(crate) mod check;
 pub(crate) mod clean;
 pub(crate) mod daemon;
 pub(crate) mod dblint;
+pub(crate) mod format;
 pub(crate) mod init;
 pub(crate) mod version;
 
@@ -71,6 +72,39 @@ pub enum PgLSCommand {
         paths: Vec<OsString>,
     },
 
+    /// Formats the requested files.
+    #[bpaf(command)]
+    Format {
+        #[bpaf(external(partial_configuration), hide_usage, optional)]
+        configuration: Option<PartialConfiguration>,
+
+        #[bpaf(external, hide_usage)]
+        cli_options: CliOptions,
+
+        /// Write formatted output back to files
+        #[bpaf(long("write"), switch)]
+        write: bool,
+
+        /// When set to true, only the files that have been staged (the ones prepared to be committed)
+        /// will be formatted. This option should be used when working locally.
+        #[bpaf(long("staged"), switch)]
+        staged: bool,
+
+        /// When set to true, only the files that have been changed compared to your `defaultBranch`
+        /// configuration will be formatted. This option should be used in CI environments.
+        #[bpaf(long("changed"), switch)]
+        changed: bool,
+
+        /// Use this to specify the base branch to compare against when you're using the --changed
+        /// flag and the `defaultBranch` is not set in your `postgres-language-server.jsonc`
+        #[bpaf(long("since"), argument("REF"))]
+        since: Option<String>,
+
+        /// Single file, single path or list of paths
+        #[bpaf(positional("PATH"), many)]
+        paths: Vec<OsString>,
+    },
+
     /// Starts the daemon server process.
     #[bpaf(command)]
     Start {
@@ -223,7 +257,8 @@ impl PgLSCommand {
         match self {
             PgLSCommand::Version(cli_options)
             | PgLSCommand::Check { cli_options, .. }
-            | PgLSCommand::Dblint { cli_options, .. } => Some(cli_options),
+            | PgLSCommand::Dblint { cli_options, .. }
+            | PgLSCommand::Format { cli_options, .. } => Some(cli_options),
             PgLSCommand::LspProxy { .. }
             | PgLSCommand::Start { .. }
             | PgLSCommand::Stop

crates/pgls_cli/src/commands/mod.rs

@@ -2,8 +2,11 @@ use pgls_console::markup;
 use pgls_diagnostics::adapters::{BpafError, IoError, SerdeJsonError};
 use pgls_diagnostics::{
     Advices, Category, Diagnostic, Error, LogCategory, MessageAndDescription, Severity, Visit,
+    category,
 };
+use pgls_text_edit::TextEdit;
 use pgls_workspace::WorkspaceError;
+use std::fmt::{self, Formatter};
 use std::process::{ExitCode, Termination};
 use std::{env::current_exe, fmt::Debug};
 
@@ -447,6 +450,41 @@ impl Termination for CliDiagnostic {
 )]
 pub struct StdinDiagnostic {}
 
+#[derive(Debug)]
+pub struct FormatDiffDiagnostic {
+    pub file_name: String,
+    pub diff: TextEdit,
+    pub start_line: u32,
+}
+
+impl Diagnostic for FormatDiffDiagnostic {
+    fn category(&self) -> Option<&'static Category> {
+        Some(category!("format"))
+    }
+
+    fn severity(&self) -> Severity {
+        Severity::Error
+    }
+
+    fn location(&self) -> pgls_diagnostics::Location<'_> {
+        pgls_diagnostics::Location::builder()
+            .resource(&self.file_name)
+            .build()
+    }
+
+    fn description(&self, fmt: &mut Formatter<'_>) -> fmt::Result {
+        write!(fmt, "File could be formatted using `--write`")
+    }
+
+    fn message(&self, fmt: &mut pgls_console::fmt::Formatter<'_>) -> std::io::Result<()> {
+        fmt.write_str("File could be formatted using `--write`")
+    }
+
+    fn advices(&self, visitor: &mut dyn Visit) -> std::io::Result<()> {
+        visitor.record_diff_with_offset(&self.diff, self.start_line)
+    }
+}
+
 #[cfg(test)]
 mod test {
     use crate::CliDiagnostic;