refactor(build): bzip2 using translate-c

- Replace @cImport with bzip2 mod via translate-c
- Simplify genesis_download to use new bzip2 module import
- Fix path handling in cmd.zig and config.zig for validator-dir

Author: Ubuntu

build.zig

@@ -242,12 +242,6 @@ pub fn build(b: *Build) !void {
         .optimize = config.optimize,
     }).module("zstd");
 
-    const bzip2_dep = b.dependency("bzip2", .{
-        .target = config.target,
-        .optimize = config.optimize,
-    });
-    const bzip2_lib = bzip2_dep.artifact("bz");
-
     const ssl_mod = lsquic_dep.builder.dependency("boringssl", .{
         .target = config.target,
         .optimize = config.optimize,
@@ -288,11 +282,13 @@ pub fn build(b: *Build) !void {
     const gh_table = b.createModule(.{ .root_source_file = generateTable(b) });
 
     const sqlite_mod = genSqlite(b, config.target, config.optimize);
+    const bzip2_mod = genBzip2(b, config.target, config.optimize);
 
     // zig fmt: off
     const imports: []const Build.Module.Import = &.{
         .{ .name = "base58",        .module = base58_mod },
         .{ .name = "build-options", .module = build_options.createModule() },
+        .{ .name = "bzip2",         .module = bzip2_mod },
         .{ .name = "httpz",         .module = httpz_mod },
         .{ .name = "lsquic",        .module = lsquic_mod },
         .{ .name = "poseidon",      .module = poseidon_mod },
@@ -323,7 +319,6 @@ pub fn build(b: *Build) !void {
         .optimize = config.optimize,
         .imports = imports,
     });
-    sig_mod.linkLibrary(bzip2_lib);
 
     switch (config.ledger_db) {
         .rocksdb => sig_mod.addImport("rocksdb", rocksdb_mod),
@@ -345,7 +340,6 @@ pub fn build(b: *Build) !void {
     });
     sig_exe.root_module.addObject(memcpy);
     sig_exe.root_module.addImport("cli", cli_mod);
-    sig_exe.linkLibrary(bzip2_lib);
 
     // make sure pyroscope's got enough info to profile
     sig_exe.build_id = .fast;
@@ -370,7 +364,6 @@ pub fn build(b: *Build) !void {
     });
     unit_tests_exe.root_module.addObject(memcpy);
     unit_tests_exe.root_module.addImport("cli", cli_mod);
-    unit_tests_exe.linkLibrary(bzip2_lib);
     switch (config.ledger_db) {
         .rocksdb => unit_tests_exe.root_module.addImport("rocksdb", rocksdb_mod),
         .hashmap => {},
@@ -550,6 +543,47 @@ fn genSqlite(
     return mod;
 }
 
+fn genBzip2(
+    b: *Build,
+    target: std.Build.ResolvedTarget,
+    optimize: std.builtin.OptimizeMode,
+) *Build.Module {
+    const dep = b.dependency("bzip2", .{});
+
+    const lib = b.addLibrary(.{
+        .name = "bz",
+        .linkage = .static,
+        .root_module = b.createModule(.{
+            .target = target,
+            .optimize = optimize,
+        }),
+    });
+    lib.addCSourceFiles(.{
+        .root = dep.path("."),
+        .files = &.{
+            "bzlib.c",
+            "blocksort.c",
+            "compress.c",
+            "crctable.c",
+            "decompress.c",
+            "huffman.c",
+            "randtable.c",
+        },
+    });
+    lib.linkLibC();
+
+    const translate_c = b.addTranslateC(.{
+        .root_source_file = dep.path("bzlib.h"),
+        .target = target,
+        .optimize = optimize,
+        .link_libc = true,
+    });
+    const mod = translate_c.createModule();
+    mod.linkLibrary(lib);
+
+    return mod;
+}
+
 /// Reference/inspiration: https://kristoff.it/blog/improving-your-zls-experience/
 fn disableEmitBin(b: *Build) void {
     for (b.install_tls.step.dependencies.items) |*install_step_dep| {

build.zig.zon

@@ -54,8 +54,8 @@
             .hash = "N-V-__8AAH-mpwB7g3MnqYU-ooUBF1t99RP27dZ9addtMVXD",
         },
         .bzip2 = .{
-            .url = "git+https://github.com/allyourcodebase/bzip2?ref=1.0.8#d7668d5ee5eb55489c35632ecf0b0107c7a77569",
-            .hash = "bzip2-1.0.8-m7CPz2QGAADdAt97rbHaz6ekIt864HgHs9rZ3fsYF5Fy",
+            .url = "https://github.com/libarchive/bzip2/archive/refs/tags/bzip2-1.0.8.zip",
+            .hash = "N-V-__8AANChDwAITDmkzIqB-9m_iaOXSquhghM2-AqDFcX8",
         },
     },
 }

src/cmd.zig

@@ -104,7 +104,7 @@ pub fn main() !void {
     current_config.metrics_port = cmd.metrics_port;
     current_config.log_file = cmd.log_file;
     current_config.tee_logs = cmd.tee_logs;
-    current_config.validator_dir = cmd.validator_dir;
+    current_config.validator_dir = try std.fs.realpathAlloc(gpa, cmd.validator_dir);
 
     // If no subcommand was provided, print a friendly header and help information.
     const subcmd = cmd.subcmd orelse {
@@ -1328,7 +1328,7 @@ fn ensureGenesis(
         &.{ cfg.validator_dir, "genesis.bin" },
     );
     errdefer allocator.free(existing_path);
-    if (!std.meta.isError(std.fs.accessAbsolute(existing_path, .{}))) {
+    if (!std.meta.isError(std.fs.cwd().access(existing_path, .{}))) {
         logger.info().logf("Using existing genesis file: {s}", .{existing_path});
         return existing_path;
     }

src/config.zig

@@ -61,7 +61,7 @@ pub const Cmd = struct {
         if (std.mem.eql(u8, param_value, sig.VALIDATOR_DIR ++ default_suffix)) {
             return try std.fs.path.join(allocator, &.{ self.validator_dir, default_suffix });
         }
-        return param_value;
+        return try allocator.dupe(u8, param_value);
     }
 };
 

src/core/genesis_download.zig

@@ -6,9 +6,7 @@
 
 const std = @import("std");
 const sig = @import("../sig.zig");
-const bzlib_c = @cImport({
-    @cInclude("bzlib.h");
-});
+const bzip2 = @import("bzip2");
 
 const Allocator = std.mem.Allocator;
 
@@ -93,12 +91,12 @@ fn downloadGenesisArchive(
 /// Decompresses bz2 data using libbz2
 fn decompressBz2(allocator: Allocator, compressed_data: []const u8) DownloadError![]u8 {
     // Start with an estimate of 10x compression ratio
-    var dest_len: c_uint = @intCast(@min(compressed_data.len * 10, 500 * 1024 * 1024));
+    var dest_len: u32 = @intCast(@min(compressed_data.len * 10, 500 * 1024 * 1024));
     var decompressed = allocator.alloc(u8, dest_len) catch return error.OutOfMemory;
     errdefer allocator.free(decompressed);
 
     while (true) {
-        const result = bzlib_c.BZ2_bzBuffToBuffDecompress(
+        const result = bzip2.BZ2_bzBuffToBuffDecompress(
             decompressed.ptr,
             &dest_len,
             @constCast(@ptrCast(compressed_data.ptr)),
@@ -108,21 +106,22 @@ fn decompressBz2(allocator: Allocator, compressed_data: []const u8) DownloadErro
         );
 
         switch (result) {
-            bzlib_c.BZ_OK => {
+            bzip2.BZ_OK => {
                 // Shrink to actual size
                 if (dest_len < decompressed.len) {
-                    decompressed = allocator.realloc(decompressed, dest_len) catch decompressed;
+                    decompressed = allocator.realloc(decompressed, dest_len) catch
+                        return error.OutOfMemory;
                 }
-                return decompressed[0..dest_len];
+                return decompressed;
             },
-            bzlib_c.BZ_OUTBUFF_FULL => {
-                // Need more space, double the buffer
-                allocator.free(decompressed);
+            bzip2.BZ_OUTBUFF_FULL => {
                 dest_len = dest_len * 2;
                 if (dest_len > 1024 * 1024 * 1024) { // 1GB limit
                     return error.Bz2DecompressError;
                 }
-                decompressed = allocator.alloc(u8, dest_len) catch return error.OutOfMemory;
+                decompressed = allocator.realloc(decompressed, dest_len) catch {
+                    return error.OutOfMemory;
+                };
             },
             else => return error.Bz2DecompressError,
         }