Using NormPropsData in NormData; No Ziglyph deps in NOrmalizer

4 files changed, 67 insertions, 9 deletions

diff --git a/src/NormData.zig b/src/NormData.zig
index 8923382..7c2a09b 100644
--- a/src/NormData.zig
+++ b/src/NormData.zig
@@ -5,11 +5,13 @@ const CanonData = @import("CanonData");
 const CccData = @import("CombiningData");
 const CompatData = @import("CompatData");
 const HangulData = @import("HangulData");
+const NormPropsData = @import("NormPropsData");
 
 canon_data: CanonData,
 ccc_data: CccData,
 compat_data: CompatData,
 hangul_data: HangulData,
+normp_data: NormPropsData,
 
 const Self = @This();
 
@@ -19,6 +21,7 @@ pub fn init(allocator: std.mem.Allocator) !Self {
         .ccc_data = try CccData.init(allocator),
         .compat_data = try CompatData.init(allocator),
         .hangul_data = try HangulData.init(allocator),
+        .normp_data = try NormPropsData.init(allocator),
     };
 }
 
@@ -27,4 +30,5 @@ pub fn deinit(self: *Self) void {
     self.ccc_data.deinit();
     self.compat_data.deinit();
     self.hangul_data.deinit();
+    self.normp_data.deinit();
 }
diff --git a/src/NormPropsData.zig b/src/NormPropsData.zig
new file mode 100644
index 0000000..3c49712
--- /dev/null
+++ b/src/NormPropsData.zig
@@ -0,0 +1,53 @@
+const std = @import("std");
+const builtin = @import("builtin");
+const compress = std.compress;
+const mem = std.mem;
+const testing = std.testing;
+
+allocator: mem.Allocator,
+s1: []u16 = undefined,
+s2: []u4 = undefined,
+
+const Self = @This();
+
+pub fn init(allocator: mem.Allocator) !Self {
+    const decompressor = compress.deflate.decompressor;
+    const in_bytes = @embedFile("normp");
+    var in_fbs = std.io.fixedBufferStream(in_bytes);
+    var in_decomp = try decompressor(allocator, in_fbs.reader(), null);
+    defer in_decomp.deinit();
+    var reader = in_decomp.reader();
+
+    const endian = builtin.cpu.arch.endian();
+    var self = Self{ .allocator = allocator };
+
+    const stage_1_len: u16 = try reader.readInt(u16, endian);
+    self.s1 = try allocator.alloc(u16, stage_1_len);
+    for (0..stage_1_len) |i| self.s1[i] = try reader.readInt(u16, endian);
+
+    const stage_2_len: u16 = try reader.readInt(u16, endian);
+    self.s2 = try allocator.alloc(u4, stage_2_len);
+    for (0..stage_2_len) |i| self.s2[i] = @intCast(try reader.readInt(u8, endian));
+
+    return self;
+}
+
+pub fn deinit(self: *Self) void {
+    self.allocator.free(self.s1);
+    self.allocator.free(self.s2);
+}
+
+/// Returns true if `cp` is already in NFD form.
+pub inline fn isNfd(self: Self, cp: u21) bool {
+    return self.s2[self.s1[cp >> 8] + (cp & 0xff)] & 1 == 0;
+}
+
+/// Returns true if `cp` is already in NFKD form.
+pub inline fn isNfkd(self: Self, cp: u21) bool {
+    return self.s2[self.s1[cp >> 8] + (cp & 0xff)] & 2 == 0;
+}
+
+/// Returns true if `cp` is not allowed in any normalized form.
+pub inline fn isFcx(self: Self, cp: u21) bool {
+    return self.s2[self.s1[cp >> 8] + (cp & 0xff)] & 4 == 4;
+}
diff --git a/src/Normalizer.zig b/src/Normalizer.zig
index d1d7cee..26177ac 100644
--- a/src/Normalizer.zig
+++ b/src/Normalizer.zig
@@ -6,8 +6,6 @@ const std = @import("std");
 const testing = std.testing;
 
 const CodePointIterator = @import("code_point").Iterator;
-const norm_props = @import("ziglyph").normalization_props;
-
 pub const NormData = @import("NormData");
 
 norm_data: *NormData,
@@ -109,7 +107,10 @@ pub fn decompose(self: Self, cp: u21, form: Form) Decomp {
     var dc = Decomp{ .form = form };
 
     // ASCII or NFD / NFKD quick checks.
-    if (cp <= 127 or (form == .nfd and norm_props.isNfd(cp)) or (form == .nfkd and norm_props.isNfkd(cp))) {
+    if (cp <= 127 or
+        (form == .nfd and self.norm_data.normp_data.isNfd(cp)) or
+        (form == .nfkd and self.norm_data.normp_data.isNfkd(cp)))
+    {
         dc.cps[0] = cp;
         return dc;
     }
@@ -436,7 +437,7 @@ fn nfxc(self: Self, allocator: std.mem.Allocator, str: []const u8, form: Form) !
                 if (!processed_hangul) {
                     // L -> C not Hangul.
                     if (self.norm_data.canon_data.toNfc(.{ L, C })) |P| {
-                        if (!norm_props.isFcx(P)) {
+                        if (!self.norm_data.normp_data.isFcx(P)) {
                             d_list.items[sidx] = P;
                             d_list.items[i] = tombstone; // Mark for deletion.
                             deleted += 1;
diff --git a/src/main.zig b/src/main.zig
index 2c2cf8c..15dca16 100644
--- a/src/main.zig
+++ b/src/main.zig
@@ -25,9 +25,9 @@ pub fn main() !void {
     _ = args_iter.skip();
     const in_path = args_iter.next() orelse return error.MissingArg;
 
-    var gpa = std.heap.GeneralPurposeAllocator(.{}){};
-    defer _ = gpa.deinit();
-    const allocator = gpa.allocator();
+    var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator);
+    defer arena.deinit();
+    const allocator = arena.allocator();
 
     const input = try std.fs.cwd().readFileAlloc(allocator, in_path, std.math.maxInt(u32));
     defer allocator.free(input);
@@ -51,9 +51,9 @@ pub fn main() !void {
     // while (iter.next()) |_| result += 1;
     // while (iter.next()) |line| result += strWidth(line, &data);
     while (iter.next()) |line| {
-        var nfc = try n.nfc(allocator, line);
+        const nfc = try n.nfc(allocator, line);
         result += nfc.slice.len;
-        nfc.deinit();
+        // nfc.deinit();
     }
 
     std.debug.print("result: {}, took: {}\n", .{ result, timer.lap() / std.time.ns_per_ms });