cmph may return a larger hash than there are items.

lib/DB.zig

@@ -275,6 +275,7 @@ fn getGroup(self: *const DB, group: PackedGroup, buf: *[]u8) error{OutOfMemory}!
 // get a CGroup entry by name.
 fn getgrnam(self: *const DB, name: []const u8, buf: *[]u8) error{OutOfMemory}!?CGroup {
     const idx = bdz.search(self.bdz_groupname, name);
+    if (idx >= self.header.num_groups) return null;
     const offset = self.idx_groupname2group[idx];
     const nbits = PackedGroup.alignment_bits;
     const group = PackedGroup.fromBytes(self.groups[offset << nbits ..]).group;
@@ -285,6 +286,7 @@ fn getgrnam(self: *const DB, name: []const u8, buf: *[]u8) error{OutOfMemory}!?C
 // get a CGroup entry by it's gid.
 fn getgrgid(self: *const DB, gid: u32, buf: *[]u8) error{OutOfMemory}!?CGroup {
     const idx = bdz.search_u32(self.bdz_gid, gid);
+    if (idx >= self.header.num_groups) return null;
     const offset = self.idx_gid2group[idx];
     const nbits = PackedGroup.alignment_bits;
     const group = PackedGroup.fromBytes(self.groups[offset << nbits ..]).group;
@@ -336,6 +338,8 @@ fn getUser(self: *const DB, user: PackedUser, buf: *[]u8) error{OutOfMemory}!CUs
 // get a CUser entry by name.
 fn getpwnam(self: *const DB, name: []const u8, buf: *[]u8) error{OutOfMemory}!?CUser {
     const idx = bdz.search(self.bdz_username, name);
+    // bdz may return a hash that's bigger than the number of users
+    if (idx >= self.header.num_users) return null;
     const offset = self.idx_name2user[idx];
     const nbits = PackedUser.alignment_bits;
     const user = PackedUser.fromBytes(self.users[offset << nbits ..]).user;
@@ -346,6 +350,7 @@ fn getpwnam(self: *const DB, name: []const u8, buf: *[]u8) error{OutOfMemory}!?C
 // get a CUser entry by uid.
 fn getpwuid(self: *const DB, uid: u32, buf: *[]u8) error{OutOfMemory}!?CUser {
     const idx = bdz.search_u32(self.bdz_uid, uid);
+    if (idx >= self.header.num_users) return null;
     const offset = self.idx_uid2user[idx];
     const nbits = PackedUser.alignment_bits;
     const user = PackedUser.fromBytes(self.users[offset << nbits ..]).user;

lib/cmph.zig

@@ -6,19 +6,9 @@ const assert = std.debug.assert;
 
 const bdz = @import("bdz.zig");
 
-// must be kept in sync with the definition in cmph_types.h
+const CMPH_BDZ = @cImport({
-const CMPH_ALGO = enum(c_int) {
+    @cInclude("cmph_types.h");
-    CMPH_BMZ,
+}).CMPH_BDZ;
-    CMPH_BMZ8,
-    CMPH_CHM,
-    CMPH_BRZ,
-    CMPH_FCH,
-    CMPH_BDZ,
-    CMPH_BDZ_PH,
-    CMPH_CHD_PH,
-    CMPH_CHD,
-    CMPH_COUNT,
-};
 
 extern fn cmph_io_vector_adapter(vector: [*]const [*:0]const u8, len: c_uint) [*]u8;
 extern fn cmph_io_vector_adapter_destroy(key_source: [*]u8) void;
@@ -29,18 +19,17 @@ extern fn cmph_new(config: [*]const u8) ?[*]u8;
 extern fn cmph_config_destroy(mph: [*]u8) void;
 extern fn cmph_packed_size(mphf: [*]const u8) u32;
 extern fn cmph_pack(mphf: [*]const u8, packed_mphf: [*]u8) void;
-extern fn cmph_destroy(mphf: [*]const u8) void;
+extern fn cmph_destroy(mphf: [*]u8) void;
 
 // pack packs cmph hashes for the given input and returns a slice ("cmph pack
 // minus first 4 bytes") for further storage. The slice must be freed by the
 // caller.
 pub fn pack(allocator: Allocator, input: [][*:0]const u8) error{OutOfMemory}![]const u8 {
-    assert(input.len <= math.maxInt(c_uint));
     const input_len = @intCast(c_uint, input.len);
     var source = cmph_io_vector_adapter(input.ptr, input_len);
     defer cmph_io_vector_adapter_destroy(source);
     var config = cmph_config_new(source) orelse return error.OutOfMemory;
-    cmph_config_set_algo(config, @enumToInt(CMPH_ALGO.CMPH_BDZ));
+    cmph_config_set_algo(config, CMPH_BDZ);
     cmph_config_set_b(config, 7);
     var mph = cmph_new(config) orelse return error.OutOfMemory;
     cmph_config_destroy(config);
@@ -104,24 +93,20 @@ test "basic pack/unpack" {
     const buf = try samplePack(testing.allocator);
     defer testing.allocator.free(buf);
     try testing.expect(buf.len < 100);
-
     var used: [items_len]bool = undefined;
-
     inline for (items) |elem| {
         const hashed = bdz.search(buf, elem);
         used[hashed] = true;
     }
 
-    for (used) |item| {
+    for (used) |item| try testing.expect(item);
-        try testing.expect(item);
-    }
 }
 
 // encodes a u32 to 6 bytes so no bytes except the last one is a '\0'.
 // This is useful for cmph-packing, where it accepts 0-terminated char*s.
 pub fn unzeroZ(x: u32) [6]u8 {
     var buf: [6]u8 = undefined;
-    std.mem.copy(u8, buf[0..], bdz.unzero(x)[0..]);
+    std.mem.copy(u8, &buf, &bdz.unzero(x));
     buf[5] = 0;
     return buf;
 }
@@ -141,9 +126,8 @@ test "pack u32" {
     const packed_mphf = try packU32(testing.allocator, keys);
     defer testing.allocator.free(packed_mphf);
     var hashes: [keys.len]u32 = undefined;
-    for (keys) |key, i| {
+    for (keys) |key, i|
         hashes[i] = bdz.search_u32(packed_mphf, key);
-    }
     sort.sort(u32, hashes[0..], {}, comptime sort.asc(u32));
     for (hashes) |hash, i|
         try testing.expectEqual(i, hash);
@@ -154,17 +138,9 @@ test "pack str" {
     const packed_mphf = try packStr(testing.allocator, keys[0..]);
     defer testing.allocator.free(packed_mphf);
     var hashes: [keys.len]u32 = undefined;
-    for (keys) |key, i| {
+    for (keys) |key, i|
         hashes[i] = bdz.search(packed_mphf, key);
-    }
     sort.sort(u32, hashes[0..], {}, comptime sort.asc(u32));
     for (hashes) |hash, i|
         try testing.expectEqual(i, hash);
 }
-
-test "CMPH_ALGO.CMPH_BDZ is in sync with our definition" {
-    const c = @cImport({
-        @cInclude("cmph_types.h");
-    });
-    try testing.expectEqual(c.CMPH_BDZ, @enumToInt(CMPH_ALGO.CMPH_BDZ));
-}