Merge branch 'main' into jarred/unicode-table

cmake/targets/BuildBun.cmake

@@ -860,7 +860,7 @@ endif()
 
 if(WIN32)
   target_link_options(${bun} PUBLIC
-    /STACK:0x1200000,0x100000
+    /STACK:0x1200000,0x200000
     /errorlimit:0
   )
   if(RELEASE)

misctools/generate-add-completions.ts

@@ -0,0 +1,231 @@
+import * as fs from "fs";
+import path from "path";
+import { execSync } from "child_process";
+
+interface LetterGroup {
+  offset: number;
+  length: number;
+  packages: string[];
+}
+
+// Read and parse input file
+const content = fs.readFileSync(path.join(__dirname, "..", "src", "cli", "add_completions.txt"), "utf8");
+const packages = content
+  .split("\n")
+  .map(line => line.trim())
+  .filter(line => line.length > 0)
+  .sort();
+
+// Group packages by first letter
+const letterGroups = new Map<string, LetterGroup>();
+let currentOffset = 0;
+let maxListSize = 0;
+
+for (const pkg of packages) {
+  if (pkg.length === 0) continue;
+  const firstLetter = pkg[0].toLowerCase();
+  if (!letterGroups.has(firstLetter)) {
+    letterGroups.set(firstLetter, {
+      offset: currentOffset,
+      length: 0,
+      packages: [],
+    });
+  }
+  const group = letterGroups.get(firstLetter)!;
+  group.packages.push(pkg);
+  group.length++;
+  maxListSize = Math.max(maxListSize, group.length);
+}
+
+// Helper to ensure temp dir exists
+const tmpDir = path.join(__dirname, "tmp");
+if (!fs.existsSync(tmpDir)) {
+  fs.mkdirSync(tmpDir);
+}
+
+// Create a single buffer with all package data
+const dataChunks: Buffer[] = [];
+let totalUncompressed = 0;
+
+// Store total package count first
+const totalCountBuf = Buffer.alloc(4);
+totalCountBuf.writeUInt32LE(packages.length, 0);
+dataChunks.push(totalCountBuf);
+totalUncompressed += 4;
+
+// Then all packages with length prefixes
+for (const pkg of packages) {
+  const lenBuf = Buffer.alloc(2);
+  lenBuf.writeUInt16LE(pkg.length, 0);
+  dataChunks.push(lenBuf);
+  dataChunks.push(Buffer.from(pkg, "utf8"));
+  totalUncompressed += 2 + pkg.length;
+}
+
+const uncompressedData = Buffer.concat(dataChunks);
+
+// Write to temp file and compress with zstd
+const uncompressedPath = path.join(tmpDir, "packages.bin");
+const compressedPath = path.join(tmpDir, "packages.bin.zst");
+
+fs.writeFileSync(uncompressedPath, uncompressedData);
+execSync(`zstd -1 --rm -f "${uncompressedPath}" -o "${compressedPath}"`);
+
+// Read back compressed data
+const compressedData = fs.readFileSync(compressedPath);
+fs.unlinkSync(compressedPath);
+
+// Calculate compression ratio
+const totalCompressed = compressedData.length;
+const ratio = ((totalCompressed / totalUncompressed) * 100).toFixed(1);
+
+console.log("\nCompression statistics:");
+console.log(`Uncompressed size: ${totalUncompressed} bytes`);
+console.log(`Compressed size: ${totalCompressed} bytes`);
+console.log(`Compression ratio: ${ratio}%`);
+
+// Generate Zig code
+const chunks: string[] = [];
+
+// Header with comments and imports
+chunks.push(`// Auto-generated file. Do not edit.
+// To regenerate this file, run:
+// 
+//   bun misctools/generate-add-completions.ts
+//
+// If you update add_completions.txt, then you should run this script again.
+//
+// This used to be a comptime block, but it made the build too slow.
+// Compressing the completions list saves about 100 KB of binary size.
+const std = @import("std");
+const bun = @import("root").bun;
+const zstd = bun.zstd;
+const Environment = bun.Environment;
+
+pub const FirstLetter = enum(u8) {
+    a = 'a',
+    b = 'b',
+    c = 'c',
+    d = 'd',
+    e = 'e',
+    f = 'f',
+    g = 'g',
+    h = 'h',
+    i = 'i',
+    j = 'j',
+    k = 'k',
+    l = 'l',
+    m = 'm',
+    n = 'n',
+    o = 'o',
+    p = 'p',
+    q = 'q',
+    r = 'r',
+    s = 's',
+    t = 't',
+    u = 'u',
+    v = 'v',
+    w = 'w',
+    x = 'x',
+    y = 'y',
+    z = 'z',
+};`);
+
+// Add the compressed data
+chunks.push(`const compressed_data = [_]u8{${[...compressedData].join(",")}};`);
+
+// Add uncompressed size constant
+chunks.push(`const uncompressed_size: usize = ${totalUncompressed};`);
+
+// Generate index entries
+const indexEntries: string[] = [];
+let offset = 0;
+for (const letter of "abcdefghijklmnopqrstuvwxyz") {
+  const group = letterGroups.get(letter);
+  if (group) {
+    indexEntries.push(`        .${letter} = .{ .offset = ${offset}, .length = ${group.length} }`);
+    offset += group.length;
+  } else {
+    indexEntries.push(`        .${letter} = .{ .offset = ${offset}, .length = 0 }`);
+  }
+}
+
+// Generate index type and instance
+chunks.push(`pub const IndexEntry = struct {
+    offset: usize,
+    length: usize,
+};
+
+pub const Index = std.EnumArray(FirstLetter, IndexEntry);
+
+pub const index = Index.init(.{
+${indexEntries.join(",\n")}
+    });`);
+
+// Generate the decompression and access function
+chunks.push(`var decompressed_data: ?[]u8 = null;
+var packages_list: ?[][]const u8 = null;
+
+pub fn init(allocator: std.mem.Allocator) !void {
+    // Decompress data
+    var data = try allocator.alloc(u8, uncompressed_size);
+    errdefer allocator.free(data);
+
+    const result = zstd.decompress(data, &compressed_data);
+    decompressed_data = data[0..result.success];
+
+    // Parse package list
+    const total_count = std.mem.readInt(u32, data[0..4], .little);
+    var packages = try allocator.alloc([]const u8, total_count);
+    errdefer allocator.free(packages);
+
+    var pos: usize = 4;
+    var i: usize = 0;
+    while (i < total_count) : (i += 1) {
+        const len = std.mem.readInt(u16, data[pos..][0..2], .little);
+        pos += 2;
+        packages[i] = data[pos..pos + len];
+        pos += len;
+    }
+
+    packages_list = packages;
+}
+
+pub fn deinit(allocator: std.mem.Allocator) void {
+    if (packages_list) |pkgs| {
+        allocator.free(pkgs);
+        packages_list = null;
+    }
+
+    if (decompressed_data) |data| {
+        allocator.free(data);
+        decompressed_data = null;
+    }
+}
+
+pub fn getPackages(letter: FirstLetter) []const []const u8 {
+    const entry = index.get(letter);
+    if (entry.length == 0) return &[_][]const u8{};
+    
+    return packages_list.?[entry.offset..entry.offset + entry.length];
+}`);
+
+// Add biggest_list constant
+chunks.push(`pub const biggest_list: usize = ${maxListSize};`);
+
+// Write the output
+let zigCode = chunks.join("\n\n");
+
+zigCode = execSync("zig fmt --stdin", {
+  input: zigCode,
+  encoding: "utf8",
+}).toString();
+
+fs.writeFileSync(path.join(__dirname, "..", "src", "cli", "add_completions.zig"), zigCode);
+
+// Clean up temp dir
+try {
+  fs.rmdirSync(tmpDir);
+} catch {}
+
+console.log(`\nGenerated Zig completions for ${packages.length} packages`);

src/bit_set.zig

@@ -1,3 +1,8 @@
+//! This is a fork of Zig standard library bit_set.zig
+//! - https://github.com/ziglang/zig/pull/14129
+//! - AutoBitset which optimally chooses between a dynamic or static bitset.
+//! Prefer our fork over std.bit_set.zig.
+//!
 //! This file defines several variants of bit sets.  A bit set
 //! is a densely stored set of integers with a known maximum,
 //! in which each integer gets a single bit.  Bit sets have very
@@ -402,7 +407,7 @@ pub fn ArrayBitSet(comptime MaskIntType: type, comptime size: usize) type {
 
         /// Returns true if the bit at the specified index
         /// is present in the set, false otherwise.
-        pub inline fn isSet(self: *const Self, index: usize) bool {
+        pub fn isSet(self: *const Self, index: usize) bool {
             if (comptime Environment.allow_assert) bun.assert(index < bit_length);
             if (num_masks == 0) return false; // doesn't compile in this case
             return (self.masks[maskIndex(index)] & maskBit(index)) != 0;