From 6c6d9d9c2c385976105dd977b8edd4be0e4587e7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Motiejus=20Jak=C5=A1tys?= <motiejus@jakstys.lt>
Date: Wed, 6 Jul 2022 12:32:49 +0300
Subject: [PATCH] replace users_arr and groups_arr with smaller arrays

RSS savings are trivial, it was not worth it.
---
 src/Corpus.zig | 117 ++++++++++++++++++++++++++-----------------------
 1 file changed, 62 insertions(+), 55 deletions(-)

diff --git a/src/Corpus.zig b/src/Corpus.zig
index d0bc4ff..1a31d6d 100644
--- a/src/Corpus.zig
+++ b/src/Corpus.zig
@@ -40,35 +40,76 @@ pub fn init(
     if (usersConst.len >= math.maxInt(u32)) return error.TooMany;
     if (groupsConst.len >= math.maxInt(u32)) return error.TooMany;
 
+    var users = MultiArrayList(User){};
+    var groups = MultiArrayList(Group){};
+    var getgr_bufsize: usize = 0;
+    var getpw_bufsize: usize = 0;
+
     var arena = ArenaAllocator.init(baseAllocator);
     var allocator = arena.allocator();
     errdefer arena.deinit();
 
-    var groups_arr = try allocator.alloc(Group, groupsConst.len);
-    var users_arr = try allocator.alloc(User, usersConst.len);
+    const NameIdx = struct {
+        name: []const u8,
+        idx: usize,
+    };
 
-    var getgr_bufsize: usize = 0;
-    for (groupsConst) |*group, i| {
-        groups_arr[i] = try group.clone(allocator);
-        getgr_bufsize = math.max(getgr_bufsize, group.strlenZ());
-    }
-    var getpw_bufsize: usize = 0;
-    for (usersConst) |*user, i| {
-        users_arr[i] = try user.clone(allocator);
-        getpw_bufsize = math.max(getpw_bufsize, user.strlenZ());
+    const GidIdx = struct {
+        gid: u32,
+        idx: usize,
+    };
+
+    const Compare = struct {
+        fn name(_: void, a: NameIdx, b: NameIdx) bool {
+            var utf8_a = (unicode.Utf8View.init(a.name) catch unreachable).iterator();
+            var utf8_b = (unicode.Utf8View.init(b.name) catch unreachable).iterator();
+            while (utf8_a.nextCodepoint()) |codepoint_a| {
+                if (utf8_b.nextCodepoint()) |codepoint_b| {
+                    if (codepoint_a == codepoint_b) {
+                        continue;
+                    } else return codepoint_a < codepoint_b;
+                }
+
+                // a is a prefix of b. It is thus shorter.
+                return true;
+            }
+            // b is a prefix of a
+            return false;
+        }
+        fn gid(_: void, a: GidIdx, b: GidIdx) bool {
+            return a.gid < b.gid;
+        }
+    };
+
+    {
+        var name_idx = try baseAllocator.alloc(NameIdx, usersConst.len);
+        defer baseAllocator.free(name_idx);
+        for (usersConst) |user, i|
+            name_idx[i] = NameIdx{ .name = user.name, .idx = i };
+        sort.sort(NameIdx, name_idx, {}, Compare.name);
+
+        try users.ensureTotalCapacity(allocator, usersConst.len);
+        for (name_idx) |entry| {
+            const user = try usersConst[entry.idx].clone(allocator);
+            users.appendAssumeCapacity(user);
+            getpw_bufsize = math.max(getpw_bufsize, user.strlenZ());
+        }
     }
 
-    sort.sort(User, users_arr, {}, cmpUser);
-    sort.sort(Group, groups_arr, {}, cmpGroup);
+    {
+        var gid_idx = try baseAllocator.alloc(GidIdx, groupsConst.len);
+        defer baseAllocator.free(gid_idx);
+        for (groupsConst) |group, i|
+            gid_idx[i] = GidIdx{ .gid = group.gid, .idx = i };
+        sort.sort(GidIdx, gid_idx, {}, Compare.gid);
 
-    var users = MultiArrayList(User){};
-    try users.ensureTotalCapacity(allocator, users_arr.len);
-    for (users_arr) |user|
-        users.appendAssumeCapacity(user);
-    var groups = MultiArrayList(Group){};
-    try groups.ensureTotalCapacity(allocator, groups_arr.len);
-    for (groups_arr) |group|
-        groups.appendAssumeCapacity(group);
+        try groups.ensureTotalCapacity(allocator, groupsConst.len);
+        for (gid_idx) |entry| {
+            const group = try groupsConst[entry.idx].clone(allocator);
+            groups.appendAssumeCapacity(group);
+            getgr_bufsize = math.max(getgr_bufsize, group.strlenZ());
+        }
+    }
 
     // verify whatever comes to cmph are unique: user names
     var name2user = StringHashMap(u32).init(allocator);
@@ -167,28 +208,6 @@ pub fn deinit(self: *Corpus) void {
     self.* = undefined;
 }
 
-// cmpUser compares two users for sorting. By username's utf8 codepoints, ascending.
-fn cmpUser(_: void, a: User, b: User) bool {
-    var utf8_a = (unicode.Utf8View.init(a.name) catch unreachable).iterator();
-    var utf8_b = (unicode.Utf8View.init(b.name) catch unreachable).iterator();
-    while (utf8_a.nextCodepoint()) |codepoint_a| {
-        if (utf8_b.nextCodepoint()) |codepoint_b| {
-            if (codepoint_a == codepoint_b) {
-                continue;
-            } else return codepoint_a < codepoint_b;
-        }
-
-        // a is a prefix of b. It is thus shorter.
-        return true;
-    }
-    // b is a prefix of a
-    return false;
-}
-
-fn cmpGroup(_: void, a: Group, b: Group) bool {
-    return a.gid < b.gid;
-}
-
 fn testUser(name: []const u8) User {
     var result = mem.zeroes(User);
     result.name = name;
@@ -198,18 +217,6 @@ fn testUser(name: []const u8) User {
 const testing = std.testing;
 const someMembers = @import("Group.zig").someMembers;
 
-test "users compare function" {
-    const a = testUser("a");
-    const b = testUser("b");
-    const bb = testUser("bb");
-    try testing.expect(cmpUser({}, a, b));
-    try testing.expect(!cmpUser({}, b, a));
-    try testing.expect(cmpUser({}, a, bb));
-    try testing.expect(!cmpUser({}, bb, a));
-    try testing.expect(cmpUser({}, bb, b));
-    try testing.expect(!cmpUser({}, b, bb));
-}
-
 pub fn testCorpus(allocator: Allocator) !Corpus {
     const users = [_]User{ User{
         .uid = 0,