From a58a2cba41916ad05df08199972763e0850df1d4 Mon Sep 17 00:00:00 2001
From: Emil Lerch <emil@lerch.org>
Date: Fri, 25 Aug 2023 15:34:52 -0700
Subject: [PATCH] adjustments in prep for hash optimization

---
 Package.zig            |  2 +-
 codegen/src/Hasher.zig | 32 ++++++++++++++++++++++++++++----
 codegen/src/main.zig   | 18 ++++++++++++------
 3 files changed, 41 insertions(+), 11 deletions(-)

diff --git a/Package.zig b/Package.zig
index f051a89..eea3212 100644
--- a/Package.zig
+++ b/Package.zig
@@ -202,7 +202,7 @@ pub fn fetchAndUnpack(
         // Of course, if the ignore rules above omit the file from the package, then everything
         // is fine and no error should be raised.
 
-        break :a try Hasher.computeDirectoryHash(thread_pool, .{ .dir = tmp_directory.handle });
+        break :a try Hasher.computeDirectoryHash(thread_pool, .{ .dir = tmp_directory.handle }, &.{});
     };
 
     const pkg_dir_sub_path = "p" ++ s ++ Hasher.hexDigest(actual_hash);
diff --git a/codegen/src/Hasher.zig b/codegen/src/Hasher.zig
index dcfd4ae..faed794 100644
--- a/codegen/src/Hasher.zig
+++ b/codegen/src/Hasher.zig
@@ -1,7 +1,8 @@
 const builtin = @import("builtin");
 const std = @import("std");
 const Hash = std.crypto.hash.sha2.Sha256;
-const HashedFile = struct {
+
+pub const HashedFile = struct {
     fs_path: []const u8,
     normalized_path: []const u8,
     hash: [Hash.digest_length]u8,
@@ -74,9 +75,28 @@ pub fn hex64(x: u64) [16]u8 {
     }
     return result;
 }
+
+pub const walkerFn = *const fn (std.fs.IterableDir.Walker.WalkerEntry) bool;
+
+fn included(entry: std.fs.IterableDir.Walker.WalkerEntry) bool {
+    _ = entry;
+    return true;
+}
+fn excluded(entry: std.fs.IterableDir.Walker.WalkerEntry) bool {
+    _ = entry;
+    return false;
+}
+pub const ComputeDirectoryOptions = struct {
+    isIncluded: walkerFn = included,
+    isExcluded: walkerFn = excluded,
+    fileHashes: []*HashedFile = undefined,
+    needFileHashes: bool = false,
+};
+
 pub fn computeDirectoryHash(
     thread_pool: *std.Thread.Pool,
     dir: std.fs.IterableDir,
+    options: *ComputeDirectoryOptions,
 ) ![Hash.digest_length]u8 {
     const gpa = thread_pool.allocator;
 
@@ -105,11 +125,14 @@ pub fn computeDirectoryHash(
                 .file => {},
                 else => return error.IllegalFileTypeInPackage,
             }
-            const hashed_file = try arena.create(HashedFile);
-            const fs_path = try arena.dupe(u8, entry.path);
+            if (options.isExcluded(entry) or !options.isIncluded(entry))
+                continue;
+            const alloc = if (options.needFileHashes) gpa else arena;
+            const hashed_file = try alloc.create(HashedFile);
+            const fs_path = try alloc.dupe(u8, entry.path);
             hashed_file.* = .{
                 .fs_path = fs_path,
-                .normalized_path = try normalizePath(arena, fs_path),
+                .normalized_path = try normalizePath(alloc, fs_path),
                 .hash = undefined, // to be populated by the worker
                 .failure = undefined, // to be populated by the worker
             };
@@ -132,6 +155,7 @@ pub fn computeDirectoryHash(
         hasher.update(&hashed_file.hash);
     }
     if (any_failures) return error.DirectoryHashUnavailable;
+    if (options.needFileHashes) options.fileHashes = try all_files.toOwnedSlice();
     return hasher.finalResult();
 }
 fn workerHashFile(dir: std.fs.Dir, hashed_file: *HashedFile, wg: *std.Thread.WaitGroup) void {
diff --git a/codegen/src/main.zig b/codegen/src/main.zig
index 583895e..dc832a3 100644
--- a/codegen/src/main.zig
+++ b/codegen/src/main.zig
@@ -62,24 +62,30 @@ pub fn main() anyerror!void {
     }
     if (files_processed == 0) {
         // no files specified, look for json files in models directory or cwd
+        // this is our normal mode of operation and where initial optimizations
+        // can be made
         if (models_dir) |m| {
             var cwd = try std.fs.cwd().openDir(".", .{});
             defer cwd.close();
             defer cwd.setAsCwd() catch unreachable;
 
             try m.dir.setAsCwd();
-            var mi = m.iterate();
-            while (try mi.next()) |e| {
-                if ((e.kind == .file or e.kind == .sym_link) and
-                    std.mem.endsWith(u8, e.name, ".json"))
-                    try processFile(e.name, stdout, output_dir, manifest);
-            }
+            try processDirectories(m, output_dir, stdout, manifest);
         }
     }
 
     if (args.len == 0)
         _ = try generateServices(allocator, ";", std.io.getStdIn(), stdout);
 }
+fn processDirectories(models_dir: std.fs.IterableDir, output_dir: std.fs.Dir, stdout: anytype, manifest: anytype) !void {
+    // Do this in a brain dead fashion, no optimization
+    var mi = models_dir.iterate();
+    while (try mi.next()) |e| {
+        if ((e.kind == .file or e.kind == .sym_link) and
+            std.mem.endsWith(u8, e.name, ".json"))
+            try processFile(e.name, stdout, output_dir, manifest);
+    }
+}
 
 fn processFile(file_name: []const u8, stdout: anytype, output_dir: std.fs.Dir, manifest: anytype) !void {
     // It's probably best to create our own allocator here so we can deint at the end and