Keep attributes at u64->u64

Author: lalinsky

src/FileSegment.zig

@@ -8,7 +8,6 @@ const KeepOrDelete = common.KeepOrDelete;
 
 const Item = @import("segment.zig").Item;
 const SegmentInfo = @import("segment.zig").SegmentInfo;
-const SegmentBase = @import("segment.zig").SegmentBase;
 
 const filefmt = @import("filefmt.zig");
 
@@ -21,7 +20,7 @@ pub const Options = struct {
 allocator: std.mem.Allocator,
 dir: std.fs.Dir,
 info: SegmentInfo = .{},
-attributes: std.StringHashMapUnmanaged([]const u8) = .{},
+attributes: std.AutoHashMapUnmanaged(u64, u64) = .{},
 docs: std.AutoHashMap(u32, bool),
 index: std.ArrayList(u32),
 block_size: usize = 0,
@@ -32,8 +31,6 @@ delete_in_deinit: bool = false,
 
 raw_data: ?[]align(std.mem.page_size) u8 = null,
 
-usingnamespace SegmentBase(Self);
-
 pub fn init(allocator: std.mem.Allocator, options: Options) Self {
     return Self{
         .allocator = allocator,
@@ -45,7 +42,7 @@ pub fn init(allocator: std.mem.Allocator, options: Options) Self {
 }
 
 pub fn deinit(self: *Self, delete_file: KeepOrDelete) void {
-    self.deinitAttributes(self.allocator);
+    self.attributes.deinit(self.allocator);
     self.docs.deinit();
     self.index.deinit();
 

src/MemorySegment.zig

@@ -6,7 +6,6 @@ const SearchResults = common.SearchResults;
 const KeepOrDelete = common.KeepOrDelete;
 const SegmentInfo = @import("segment.zig").SegmentInfo;
 const Item = @import("segment.zig").Item;
-const SegmentBase = @import("segment.zig").SegmentBase;
 
 const Change = @import("change.zig").Change;
 
@@ -20,13 +19,11 @@ pub const Options = struct {};
 
 allocator: std.mem.Allocator,
 info: SegmentInfo = .{},
-attributes: std.StringHashMapUnmanaged([]const u8) = .{},
+attributes: std.AutoHashMapUnmanaged(u64, u64) = .{},
 docs: std.AutoHashMap(u32, bool),
 items: std.ArrayList(Item),
 frozen: bool = false,
 
-usingnamespace SegmentBase(Self);
-
 pub fn init(allocator: std.mem.Allocator, opts: Options) Self {
     _ = opts;
     return .{
@@ -39,7 +36,7 @@ pub fn init(allocator: std.mem.Allocator, opts: Options) Self {
 pub fn deinit(self: *Self, delete_file: KeepOrDelete) void {
     _ = delete_file;
 
-    self.deinitAttributes(self.allocator);
+    self.attributes.deinit(self.allocator);
     self.docs.deinit();
     self.items.deinit();
 }
@@ -104,13 +101,9 @@ pub fn build(self: *Self, changes: []const Change) !void {
                 }
             },
             .set_attribute => |op| {
-                const result = self.attributes.getOrPutAssumeCapacity(op.name);
+                const result = self.attributes.getOrPutAssumeCapacity(op.key);
                 if (!result.found_existing) {
-                    errdefer self.attributes.removeByPtr(result.key_ptr);
-                    result.key_ptr.* = try self.allocator.dupe(u8, op.name);
-                    errdefer self.allocator.free(result.key_ptr.*);
-                    result.value_ptr.* = try self.allocator.dupe(u8, op.value);
-                    errdefer self.allocator.free(result.value_ptr.*);
+                    result.value_ptr.* = op.value;
                 }
             },
         }
@@ -124,8 +117,13 @@ pub fn cleanup(self: *Self) void {
 }
 
 pub fn merge(self: *Self, merger: *SegmentMerger(Self)) !void {
+    std.debug.assert(self.allocator.ptr == merger.allocator.ptr);
+
     self.info = merger.segment.info;
 
+    self.attributes.deinit(self.allocator);
+    self.attributes = merger.segment.attributes.move();
+
     self.docs.deinit();
     self.docs = merger.segment.docs.move();
 

src/change.zig

@@ -18,8 +18,8 @@ pub const Delete = struct {
 };
 
 pub const SetAttribute = struct {
-    name: []const u8,
-    value: []const u8,
+    key: u64,
+    value: u64,
 
     pub fn msgpackFormat() msgpack.StructFormat {
         return .{ .as_map = .{ .key = .{ .field_name_prefix = 1 } } };

src/segment.zig

@@ -97,16 +97,3 @@ test "Item array sort" {
         Item{ .hash = 2, .id = 200 },
     }, items);
 }
-
-pub fn SegmentBase(comptime T: type) type {
-    return struct {
-        pub fn deinitAttributes(self: *T, allocator: std.mem.Allocator) void {
-            var iter = self.attributes.iterator();
-            while (iter.next()) |entry| {
-                allocator.free(entry.key_ptr.*);
-                allocator.free(entry.value_ptr.*);
-            }
-            self.attributes.deinit(allocator);
-        }
-    };
-}

src/segment_merger.zig

@@ -7,6 +7,7 @@ const SharedPtr = @import("utils/shared_ptr.zig").SharedPtr;
 
 pub const MergedSegmentInfo = struct {
     info: SegmentInfo = .{},
+    attributes: std.AutoHashMapUnmanaged(u64, u64) = .{},
     docs: std.AutoHashMap(u32, bool),
 };
 
@@ -76,16 +77,29 @@ pub fn SegmentMerger(comptime Segment: type) type {
                 return error.NoSources;
             }
 
+            var total_attributes: u32 = 0;
             var total_docs: u32 = 0;
             for (sources, 0..) |source, i| {
                 if (i == 0) {
                     self.segment.info = source.reader.segment.info;
                 } else {
                     self.segment.info = SegmentInfo.merge(self.segment.info, source.reader.segment.info);
                 }
+                total_attributes += source.reader.segment.attributes.count();
                 total_docs += source.reader.segment.docs.count();
             }
 
+            try self.segment.attributes.ensureTotalCapacity(self.allocator, total_attributes);
+            for (sources) |*source| {
+                const segment = source.reader.segment;
+                var iter = segment.attributes.iterator();
+                while (iter.next()) |entry| {
+                    const key = entry.key_ptr.*;
+                    const value = entry.value_ptr.*;
+                    self.segment.attributes.putAssumeCapacity(key, value);
+                }
+            }
+
             try self.segment.docs.ensureTotalCapacity(total_docs);
             for (sources) |*source| {
                 const segment = source.reader.segment;