feat(search): --json JSON Lines output (ripgrep-compatible)

crates/cli/src/main.rs

@@ -8,7 +8,7 @@ use sift_core::{
     CaseMode, ColorChoice, CompiledSearch, Error as SiftError, FilenameMode, GlobConfig,
     HiddenMode, IgnoreConfig, IgnoreSources, Index, IndexBuilder, OutputEmission, SearchFilter,
     SearchFilterConfig, SearchLineStyle, SearchMatchFlags, SearchMode, SearchOptions, SearchOutput,
-    SearchRecordStyle, SearchStats, VisibilityConfig,
+    SearchOutputFormat, SearchRecordStyle, SearchStats, VisibilityConfig,
 };
 
 #[derive(Parser)]
@@ -52,6 +52,17 @@ struct Cli {
     paths: PathArgs,
     #[command(flatten)]
     stats_decl: StatsDecl,
+    #[command(flatten)]
+    json_decl: JsonDecl,
+}
+
+/// Declares `--json` / `--no-json` for clap; effective value uses [`resolve_json_from_args`].
+#[derive(Args)]
+struct JsonDecl {
+    #[arg(long = "json", action = ArgAction::SetTrue)]
+    _json: bool,
+    #[arg(long = "no-json", action = ArgAction::SetTrue)]
+    _no_json: bool,
 }
 
 /// Declares `--stats` for clap; effective value uses [`resolve_stats_from_args`].
@@ -363,6 +374,23 @@ fn resolve_stats_from_args(args: &[String]) -> bool {
     result
 }
 
+/// `--json` / `--no-json`; later flag wins (ripgrep-style).
+fn resolve_json_from_args(args: &[String]) -> bool {
+    let mut last_idx = 0usize;
+    let mut result = false;
+    for (i, arg) in args.iter().enumerate() {
+        if arg == "--json" && i >= last_idx {
+            last_idx = i;
+            result = true;
+        }
+        if arg == "--no-json" && i >= last_idx {
+            last_idx = i;
+            result = false;
+        }
+    }
+    result
+}
+
 fn resolve_color_from_args(args: &[String]) -> ColorChoice {
     let mut result = ColorChoice::Auto;
     let mut i = 0usize;
@@ -897,12 +925,14 @@ fn excluded_search_paths(search_root: &Path, sift_dir: &Path) -> Vec<PathBuf> {
 }
 
 const fn search_output(
+    format: SearchOutputFormat,
     effective_mode: SearchMode,
     quiet: bool,
     lines: SearchLineStyle,
     records: SearchRecordStyle,
 ) -> SearchOutput {
     SearchOutput {
+        format,
         mode: effective_mode,
         emission: if quiet {
             OutputEmission::Quiet
@@ -954,6 +984,7 @@ struct SearchOutputCtx {
     mode: SearchModeCtx,
     lines: SearchLineResolveCtx,
     format: SearchFormatCtx,
+    output_format: SearchOutputFormat,
 }
 
 /// Resolved visibility, ignore sources, and glob case (from argv order + clap) for [`SearchFilterConfig`].
@@ -1028,12 +1059,14 @@ fn run_search_with_index(
         corpus_is_single_file,
     );
     let output = search_output(
+        out.output_format,
         out.mode.effective_mode,
         out.mode.quiet,
         SearchLineStyle {
             filename_mode,
             heading: out.lines.heading,
-            line_number: cli.out1.line_number,
+            line_number: cli.out1.line_number
+                || matches!(out.output_format, SearchOutputFormat::Json),
         },
         SearchRecordStyle {
             null_data: out.format.null_data,
@@ -1066,12 +1099,14 @@ fn run_search_walk(
     let filename_mode =
         effective_filename_mode(out.lines.with_filename, out.lines.is_path_mode, false);
     let output = search_output(
+        out.output_format,
         out.mode.effective_mode,
         out.mode.quiet,
         SearchLineStyle {
             filename_mode,
             heading: out.lines.heading,
-            line_number: cli.out1.line_number,
+            line_number: cli.out1.line_number
+                || matches!(out.output_format, SearchOutputFormat::Json),
         },
         SearchRecordStyle {
             null_data: out.format.null_data,
@@ -1099,7 +1134,6 @@ fn run_search(cli: &Cli) -> anyhow::Result<bool> {
     )?;
 
     let args: Vec<String> = std::env::args().collect();
-    let print_stats = resolve_stats_from_args(&args);
     let invert_match = resolve_invert_match_from_args(&args);
     let glob_case_insensitive = resolve_glob_case_insensitive_from_args(&args);
     let (hidden, ignore_sources, require_git) = resolve_visibility_and_ignore(&args);
@@ -1138,6 +1172,22 @@ fn run_search(cli: &Cli) -> anyhow::Result<bool> {
         mode
     };
 
+    let use_json = resolve_json_from_args(&args);
+    if use_json {
+        match effective_mode {
+            SearchMode::Count
+            | SearchMode::CountMatches
+            | SearchMode::FilesWithMatches
+            | SearchMode::FilesWithoutMatch => {
+                anyhow::bail!(
+                    "sift: --json cannot be used with --count, --count-matches, --files-with-matches, or --files-without-match"
+                );
+            }
+            SearchMode::Standard | SearchMode::OnlyMatching => {}
+        }
+    }
+    let print_stats = resolve_stats_from_args(&args) || use_json;
+
     let query = CompiledSearch::new(&patterns, opts).map_err(|e| anyhow::anyhow!("{e}"))?;
     let cwd = std::env::current_dir()?;
 
@@ -1157,6 +1207,11 @@ fn run_search(cli: &Cli) -> anyhow::Result<bool> {
             is_path_mode,
         },
         format: SearchFormatCtx { null_data, color },
+        output_format: if use_json {
+            SearchOutputFormat::Json
+        } else {
+            SearchOutputFormat::Text
+        },
     };
     let filter = SearchFilterCtx {
         hidden,

crates/cli/tests/integration_json.rs

@@ -0,0 +1,87 @@
+//! `--json` JSON Lines output (ripgrep-compatible wire format via `grep-printer`).
+
+mod common;
+
+use std::ffi::OsString;
+use std::fs;
+
+use common::{BuildIndexOptions, assert_success, fresh_dir};
+
+#[test]
+fn json_emits_begin_match_end_and_summary() {
+    let root = fresh_dir("json-basic");
+    fs::write(root.join("a.txt"), "hello world\n").unwrap();
+    let idx = root.join(".sift");
+    BuildIndexOptions::default().run(Some(&root), &idx, std::path::Path::new("."));
+
+    let mut cmd = common::command(Some(&root));
+    cmd.arg("--sift-dir").arg(&idx);
+    cmd.args([OsString::from("hello"), OsString::from("--json")]);
+    let output = cmd.output().unwrap();
+    assert_success(&output);
+    let stdout = String::from_utf8_lossy(&output.stdout);
+    let lines: Vec<&str> = stdout.lines().collect();
+    assert!(
+        lines.len() >= 4,
+        "expected begin, match, end, summary lines; got {}: {stdout:?}",
+        lines.len()
+    );
+    assert!(
+        lines[0].contains("\"type\":\"begin\""),
+        "begin: {}",
+        lines[0]
+    );
+    assert!(
+        lines[1].contains("\"type\":\"match\""),
+        "match: {}",
+        lines[1]
+    );
+    assert!(lines[2].contains("\"type\":\"end\""), "end: {}", lines[2]);
+    assert!(
+        lines[lines.len() - 1].contains("\"type\":\"summary\""),
+        "summary: {}",
+        lines[lines.len() - 1]
+    );
+}
+
+#[test]
+fn json_implies_stats_on_stderr() {
+    let root = fresh_dir("json-stats-stderr");
+    fs::write(root.join("a.txt"), "x\n").unwrap();
+    let idx = root.join(".sift");
+    BuildIndexOptions::default().run(Some(&root), &idx, std::path::Path::new("."));
+
+    let mut cmd = common::command(Some(&root));
+    cmd.arg("--sift-dir").arg(&idx);
+    cmd.args([OsString::from("x"), OsString::from("--json")]);
+    let output = cmd.output().unwrap();
+    assert_success(&output);
+    let stderr = String::from_utf8_lossy(&output.stderr);
+    assert!(
+        stderr.contains("matches") && stderr.contains("bytes searched"),
+        "expected --stats-style lines on stderr: {stderr:?}"
+    );
+}
+
+#[test]
+fn json_with_count_exits_error() {
+    let root = fresh_dir("json-bad-count");
+    fs::write(root.join("a.txt"), "a\n").unwrap();
+    let idx = root.join(".sift");
+    BuildIndexOptions::default().run(Some(&root), &idx, std::path::Path::new("."));
+
+    let mut cmd = common::command(Some(&root));
+    cmd.arg("--sift-dir").arg(&idx);
+    cmd.args([
+        OsString::from("a"),
+        OsString::from("--json"),
+        OsString::from("--count"),
+    ]);
+    let output = cmd.output().unwrap();
+    assert!(!output.status.success());
+    let stderr = String::from_utf8_lossy(&output.stderr);
+    assert!(
+        stderr.contains("--json") && stderr.contains("count"),
+        "expected conflict message: {stderr:?}"
+    );
+}

crates/core/benches/search.rs

@@ -41,8 +41,8 @@ use std::hint::black_box;
 use sift_core::{
     CaseMode, ColorChoice, CompiledSearch, FilenameMode, GlobConfig, HiddenMode, IgnoreConfig,
     IgnoreSources, Index, IndexBuilder, OutputEmission, SearchFilter, SearchFilterConfig,
-    SearchLineStyle, SearchMatchFlags, SearchMode, SearchOptions, SearchOutput, SearchRecordStyle,
-    VisibilityConfig,
+    SearchLineStyle, SearchMatchFlags, SearchMode, SearchOptions, SearchOutput, SearchOutputFormat,
+    SearchRecordStyle, VisibilityConfig,
 };
 
 fn make_parity_corpus(root: &Path) {
@@ -176,6 +176,7 @@ fn default_filter() -> SearchFilterConfig {
 
 const fn output_std() -> SearchOutput {
     SearchOutput {
+        format: SearchOutputFormat::Text,
         mode: SearchMode::Standard,
         emission: OutputEmission::Normal,
         lines: SearchLineStyle {
@@ -192,6 +193,7 @@ const fn output_std() -> SearchOutput {
 
 const fn output_quiet(mode: SearchMode) -> SearchOutput {
     SearchOutput {
+        format: SearchOutputFormat::Text,
         mode,
         emission: OutputEmission::Quiet,
         lines: SearchLineStyle {

crates/core/src/bin/sift_profile/scenarios.rs

@@ -54,6 +54,7 @@ impl Scenario {
 
 const fn make_output(mode: SearchMode, emission: sift_core::OutputEmission) -> SearchOutput {
     SearchOutput {
+        format: sift_core::SearchOutputFormat::Text,
         mode,
         emission,
         lines: SearchLineStyle {

crates/core/src/lib.rs

@@ -16,8 +16,8 @@ pub use planner::TrigramPlan;
 pub use search::{
     CandidateInfo, CaseMode, ColorChoice, CompiledSearch, FilenameMode, GlobConfig, HiddenMode,
     IgnoreConfig, IgnoreSources, Match, OutputEmission, SearchFilter, SearchFilterConfig,
-    SearchLineStyle, SearchMatchFlags, SearchMode, SearchOptions, SearchOutput, SearchRecordStyle,
-    SearchStats, VisibilityConfig, walk_file_paths,
+    SearchLineStyle, SearchMatchFlags, SearchMode, SearchOptions, SearchOutput, SearchOutputFormat,
+    SearchRecordStyle, SearchStats, VisibilityConfig, walk_file_paths,
 };
 
 pub use ignore::{Walk, WalkBuilder};
@@ -55,6 +55,12 @@ pub enum Error {
     #[error("invalid max-count: 0 matches requested")]
     InvalidMaxCount,
 
+    #[error("JSON output is only supported for standard search (not count or file-list modes)")]
+    JsonOutputIncompatibleMode,
+
+    #[error("JSON serialization error: {0}")]
+    JsonSerialize(#[from] serde_json::Error),
+
     #[error("invalid index metadata: {0}")]
     InvalidMeta(PathBuf),