rufio 1.1.0 → 1.3.0

data/lib/rufio.rb

@@ -4,6 +4,8 @@ require_relative 'rufio/version'
 require_relative 'rufio/config'
 require_relative 'rufio/config_loader'
 require_relative 'rufio/color_helper'
+require_relative 'rufio/key_input'
+require_relative 'rufio/shell_init'
 require_relative 'rufio/directory_listing'
 require_relative 'rufio/filter_manager'
 require_relative 'rufio/selection_manager'
@@ -22,6 +24,9 @@ require_relative 'rufio/search_controller'
 require_relative 'rufio/keybind_handler'
 require_relative 'rufio/ansi_line_parser'
 require_relative 'rufio/syntax_highlighter'
+require_relative 'rufio/zip_archive'
+require_relative 'rufio/zip_extractor'
+require_relative 'rufio/zip_writer'
 require_relative 'rufio/file_preview'
 require_relative 'rufio/terminal_ui'
 require_relative 'rufio/application'

data/lib_zig/rufio_native/Makefile

@@ -1,34 +1,71 @@
-# Makefile for Zig Ruby extension
+# Makefile for Zig native library (Ruby ABI 非依存・Fiddle 経由でロード)
+#
+#   make          現在の環境向けにビルド
+#   make install  現在の環境向けだけを Gem 同梱と同じ設定でビルドして lib/rufio/native/<platform>/ に配置
+#   make dist     Gem 同梱用に全プラットフォームをクロスコンパイル
+#   make test     検索エンジンの単体テスト
 
-RUBY := ruby
 ZIG := zig
-
-# Rubyの設定を取得
-RUBY_INCLUDE := $(shell $(RUBY) -e "puts RbConfig::CONFIG['rubyhdrdir']")
-RUBY_ARCH_INCLUDE := $(shell $(RUBY) -e "puts RbConfig::CONFIG['rubyarchhdrdir']")
-RUBY_LIB_DIR := $(shell $(RUBY) -e "puts RbConfig::CONFIG['libdir']")
-RUBY_VERSION := $(shell $(RUBY) -e "puts RbConfig::CONFIG['MAJOR'] + '.' + RbConfig::CONFIG['MINOR']")
+RUBY := ruby
 
 # コンパイルフラグ
 ZIG_FLAGS := -O ReleaseFast -dynamic -lc
-ZIG_FLAGS += -I$(RUBY_INCLUDE)
-ZIG_FLAGS += -I$(RUBY_ARCH_INCLUDE)
-ZIG_FLAGS += -L$(RUBY_LIB_DIR)
-ZIG_FLAGS += -lruby.$(RUBY_VERSION)
 
 # ターゲット
 TARGET := librufio_zig.dylib
+BENCH := rufio_search
 SRC := src/main.zig
+SOURCES := src/main.zig src/search.zig src/os.zig src/zip.zig
+NATIVE_DIR := ../../lib/rufio/native
+BUILD_DIR := zig-out/dist
+
+# Gem 同梱プラットフォーム（ディレクトリ名:Zig ターゲット:ファイル名）
+# Linux は glibc 2.17（CentOS 7 世代）以降で動くようにリンクする
+PLATFORMS := \
+	aarch64-macos:aarch64-macos.11.0:librufio_zig.dylib \
+	x86_64-macos:x86_64-macos.10.15:librufio_zig.dylib \
+	x86_64-linux-gnu:x86_64-linux-gnu.2.17:librufio_zig.so \
+	aarch64-linux-gnu:aarch64-linux-gnu.2.17:librufio_zig.so \
+	x86_64-linux-musl:x86_64-linux-musl:librufio_zig.so \
+	aarch64-linux-musl:aarch64-linux-musl:librufio_zig.so \
+	x86_64-windows:x86_64-windows-gnu:rufio_zig.dll \
+	aarch64-windows:aarch64-windows-gnu:rufio_zig.dll
 
 all: $(TARGET)
 
-$(TARGET): $(SRC)
+$(TARGET): $(SOURCES)
 	$(ZIG) build-lib $(SRC) $(ZIG_FLAGS) -femit-bin=$@
 
+# ripgrep との比較用 CLI
+$(BENCH): src/cli.zig src/search.zig src/os.zig
+	$(ZIG) build-exe src/cli.zig -O ReleaseFast -lc -femit-bin=$@
+
+test:
+	$(ZIG) test src/search_test.zig
+
+bench: $(BENCH)
+	./bench.sh
+
+# ONLY=<platform> で 1 つだけビルドする
+dist: $(SOURCES)
+	@for entry in $(PLATFORMS); do \
+		dir=$${entry%%:*}; rest=$${entry#*:}; triple=$${rest%%:*}; file=$${rest#*:}; \
+		if [ -n "$(ONLY)" ] && [ "$(ONLY)" != "$$dir" ]; then continue; fi; \
+		echo "==> $$dir ($$triple)"; \
+		mkdir -p $(BUILD_DIR)/$$dir $(NATIVE_DIR)/$$dir || exit 1; \
+		$(ZIG) build-lib $(SRC) -target $$triple $(ZIG_FLAGS) -fstrip \
+			-femit-bin=$(BUILD_DIR)/$$dir/$$file || exit 1; \
+		cp $(BUILD_DIR)/$$dir/$$file $(NATIVE_DIR)/$$dir/$$file || exit 1; \
+	done
+
 clean:
-	rm -f $(TARGET) *.o
+	rm -rf $(TARGET) $(BENCH) *.o $(BUILD_DIR)
 
-install: $(TARGET)
-	cp $(TARGET) ../../lib/rufio/native/rufio_zig.bundle
+# 現在の環境のプラットフォーム名は Ruby 側の判定（NativeScannerZigFFI.platform_dir）に合わせ、
+# Gem 同梱と同じ設定でビルドする
+install:
+	@platform=$$($(RUBY) -I../../lib -rrufio/native_scanner -rrufio/native_scanner_zig -e 'print Rufio::NativeScannerZigFFI.platform_dir.to_s' 2>/dev/null); \
+	if [ -z "$$platform" ]; then echo "unsupported platform"; exit 1; fi; \
+	$(MAKE) --no-print-directory dist ONLY=$$platform
 
-.PHONY: all clean install
+.PHONY: all clean install test bench dist

data/lib_zig/rufio_native/bench.sh

@@ -0,0 +1,36 @@
+#!/usr/bin/env bash
+# rufio_search と ripgrep の速度比較（ウォームキャッシュ、各 10 回の中央値）
+# 使い方: ./bench.sh [DIR] [QUERY...]
+set -euo pipefail
+cd "$(dirname "$0")"
+
+DIR=${1:-$(cd ../.. && pwd)}
+shift || true
+QUERIES=("$@")
+[ ${#QUERIES[@]} -eq 0 ] && QUERIES=(rb_define_method zzqxj_not_found "the" "日本語")
+RUNS=${RUNS:-10}
+
+[ -x ./rufio_search ] || make rufio_search >/dev/null
+
+measure() {
+  python3 - "$RUNS" "$@" <<'EOF'
+import subprocess, sys, time, statistics
+runs = int(sys.argv[1]); cmd = sys.argv[2:]
+for _ in range(2):
+    subprocess.run(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+times = []
+for _ in range(runs):
+    t = time.perf_counter()
+    subprocess.run(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+    times.append((time.perf_counter() - t) * 1000)
+print(f"median {statistics.median(times):8.1f} ms   min {min(times):8.1f} ms")
+EOF
+}
+
+echo "corpus: $DIR ($(find "$DIR" -type f | wc -l | tr -d ' ') files, $(du -sh "$DIR" | cut -f1))"
+for q in "${QUERIES[@]}"; do
+  hits=$(./rufio_search "$q" "$DIR" | wc -l | tr -d ' ')
+  echo "== query: $q ($hits lines)"
+  printf '  rufio_search : '; measure ./rufio_search "$q" "$DIR"
+  printf '  rg -uu -i -F : '; measure rg -uu -i -F -n "$q" "$DIR"
+done

data/lib_zig/rufio_native/build.zig

@@ -4,42 +4,39 @@ pub fn build(b: *std.Build) void {
     const target = b.standardTargetOptions(.{});
     const optimize = b.standardOptimizeOption(.{});
 
-    // 動的ライブラリを作成
-    const lib = b.addExecutable(.{
+    // 動的ライブラリ（Ruby ABI 非依存・Fiddle 経由でロード）
+    const lib = b.addLibrary(.{
         .name = "rufio_zig",
-        .root_source_file = b.path("src/main.zig"),
-        .target = target,
-        .optimize = optimize,
+        .linkage = .dynamic,
+        .root_module = b.createModule(.{
+            .root_source_file = b.path("src/main.zig"),
+            .target = target,
+            .optimize = optimize,
+            .link_libc = true,
+        }),
     });
-
-    // 共有ライブラリとしてリンク
-    lib.linkage = .dynamic;
-    lib.linkLibC();
-
-    // RubyのヘッダーとライブラリへのパスをRubyから取得
-    var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator);
-    defer arena.deinit();
-    const allocator = arena.allocator();
-
-    // Rubyのインクルードパスを取得
-    const ruby_include_result = b.run(&.{ "ruby", "-e", "puts RbConfig::CONFIG['rubyhdrdir']" });
-    const ruby_include = std.mem.trim(u8, ruby_include_result, " \n\r\t");
-
-    const ruby_arch_include_result = b.run(&.{ "ruby", "-e", "puts RbConfig::CONFIG['rubyarchhdrdir']" });
-    const ruby_arch_include = std.mem.trim(u8, ruby_arch_include_result, " \n\r\t");
-
-    const ruby_lib_dir_result = b.run(&.{ "ruby", "-e", "puts RbConfig::CONFIG['libdir']" });
-    const ruby_lib_dir = std.mem.trim(u8, ruby_lib_dir_result, " \n\r\t");
-
-    // インクルードパスを追加
-    lib.addIncludePath(.{ .cwd_relative = ruby_include });
-    lib.addIncludePath(.{ .cwd_relative = ruby_arch_include });
-
-    // ライブラリパスを追加
-    lib.addLibraryPath(.{ .cwd_relative = ruby_lib_dir });
-
-    // Rubyライブラリとリンク
-    lib.linkSystemLibrary("ruby.3.4");
-
     b.installArtifact(lib);
+
+    // ripgrep との比較用 CLI
+    const cli = b.addExecutable(.{
+        .name = "rufio_search",
+        .root_module = b.createModule(.{
+            .root_source_file = b.path("src/cli.zig"),
+            .target = target,
+            .optimize = optimize,
+            .link_libc = true,
+        }),
+    });
+    b.installArtifact(cli);
+
+    const tests = b.addTest(.{
+        .root_module = b.createModule(.{
+            .root_source_file = b.path("src/search_test.zig"),
+            .target = target,
+            .optimize = optimize,
+            .link_libc = true,
+        }),
+    });
+    const test_step = b.step("test", "検索エンジンの単体テストを実行");
+    test_step.dependOn(&b.addRunArtifact(tests).step);
 }

data/lib_zig/rufio_native/src/cli.zig

@@ -0,0 +1,99 @@
+//! ripgrep と同条件で比較するためのベンチマーク用 CLI。
+//! 使い方: rufio_search [-j N] [-r N] QUERY DIR...
+//! -r N: キャッシュを使って同じ検索を N 回繰り返し、各回の所要時間を標準エラーに出す（G mode の再検索を再現）
+//! 出力は rg -n と同じ `path:line:text` 形式（text はマッチ周辺の最大 1KB）。
+const std = @import("std");
+const c = std.c;
+const search = @import("search.zig");
+const os = @import("os.zig");
+
+fn writeAll(bytes: []const u8) void {
+    os.writeAll(1, bytes);
+}
+
+pub fn main(init: std.process.Init.Minimal) !void {
+    var args = try init.args.iterateAllocator(std.heap.c_allocator);
+    defer args.deinit();
+    _ = args.skip();
+    var threads: usize = 0;
+    var repeat: usize = 0;
+    var query: ?[]const u8 = null;
+    var roots: std.ArrayList(u8) = .empty;
+    const alloc = std.heap.c_allocator;
+    defer roots.deinit(alloc);
+
+    while (args.next()) |arg| {
+        if (std.mem.eql(u8, arg, "-j")) {
+            threads = try std.fmt.parseInt(usize, args.next() orelse return error.MissingValue, 10);
+        } else if (std.mem.eql(u8, arg, "-r")) {
+            repeat = try std.fmt.parseInt(usize, args.next() orelse return error.MissingValue, 10);
+        } else if (query == null) {
+            query = arg;
+        } else {
+            if (roots.items.len > 0) try roots.append(alloc, 0);
+            try roots.appendSlice(alloc, std.mem.trimEnd(u8, arg, "/"));
+        }
+    }
+    const q = query orelse {
+        writeAll("usage: rufio_search [-j N] [-r N] QUERY DIR...\n");
+        return;
+    };
+    if (roots.items.len == 0) try roots.append(alloc, '.');
+
+    if (repeat > 0) {
+        const cache = try search.Cache.create(256 * 1024 * 1024);
+        defer cache.destroy();
+        var i: usize = 0;
+        while (i < repeat) : (i += 1) {
+            const t0 = os.nowNs();
+            const lines = try runSearch(roots.items, q, threads, cache, i + 1 == repeat);
+            const ms = @as(f64, @floatFromInt(os.nowNs() - t0)) / 1e6;
+            var msg: [128]u8 = undefined;
+            os.writeAll(2, try std.fmt.bufPrint(&msg, "run {d}: {d:.1} ms, {d} lines, cache {d} MB\n", .{ i + 1, ms, lines, cache.cachedBytes() >> 20 }));
+        }
+        return;
+    }
+    _ = try runSearch(roots.items, q, threads, null, true);
+}
+
+fn runSearch(roots: []const u8, q: []const u8, threads: usize, cache: ?*search.Cache, print: bool) !usize {
+    const alloc = std.heap.c_allocator;
+    const s = try search.Search.create(roots, q, "", "", threads, cache);
+    defer s.destroy();
+
+    var lines: usize = 0;
+    var out: std.ArrayList(u8) = .empty;
+    defer out.deinit(alloc);
+    var line_buf: [24]u8 = undefined;
+    while (true) {
+        const state = s.getState();
+        const data = s.drain();
+        var pos: usize = 0;
+        while (pos + search.RECORD_HEADER <= data.len) {
+            const kind = data[pos];
+            const line = std.mem.readInt(u64, data[pos + 5 ..][0..8], .little);
+            const a_len = std.mem.readInt(u32, data[pos + 17 ..][0..4], .little);
+            const b_len = std.mem.readInt(u32, data[pos + 21 ..][0..4], .little);
+            const a = data[pos + search.RECORD_HEADER ..][0..a_len];
+            const b = data[pos + search.RECORD_HEADER + a_len ..][0..b_len];
+            pos += search.RECORD_HEADER + a_len + b_len;
+            if (kind != @intFromEnum(search.Kind.match)) continue;
+            lines += 1;
+            if (!print) continue;
+            try out.appendSlice(alloc, a);
+            try out.appendSlice(alloc, try std.fmt.bufPrint(&line_buf, ":{d}:", .{line}));
+            try out.appendSlice(alloc, std.mem.trimEnd(u8, b, "\r\n"));
+            try out.append(alloc, '\n');
+        }
+        if (out.items.len >= 1 << 16) {
+            writeAll(out.items);
+            out.clearRetainingCapacity();
+        }
+        if (data.len == 0) {
+            if (state != .scanning) break;
+            os.sleepMicros(50);
+        }
+    }
+    writeAll(out.items);
+    return lines;
+}