From 9b435e69cb9f1572728b38457fabc9636bc47143 Mon Sep 17 00:00:00 2001
From: Jose Colon Rodriguez
Date: Fri, 1 Mar 2024 19:28:41 -0400
Subject: Changes when case folded check; 20ms faster

---
 src/FoldData.zig   | 14 ++++++++++++--
 src/Normalizer.zig | 30 ++++++++++++++++++++++++++----
 2 files changed, 38 insertions(+), 6 deletions(-)

(limited to 'src')

diff --git a/src/FoldData.zig b/src/FoldData.zig
index 139c677..2a9a1f5 100644
--- a/src/FoldData.zig
+++ b/src/FoldData.zig
@@ -5,6 +5,7 @@ const mem = std.mem;
 
 allocator: mem.Allocator,
 fold: [][]u21 = undefined,
+cwcf: []bool = undefined,
 
 const Self = @This();
 
@@ -20,18 +21,21 @@ pub fn init(allocator: mem.Allocator) !Self {
     var self = Self{
         .allocator = allocator,
         .fold = try allocator.alloc([]u21, 0x110000),
+        .cwcf = try allocator.alloc(bool, 0x110000),
     };
 
     @memset(self.fold, &.{});
+    @memset(self.cwcf, false);
 
     while (true) {
         const len: u8 = try reader.readInt(u8, endian);
         if (len == 0) break;
         const cp = try reader.readInt(u24, endian);
-        self.fold[cp] = try allocator.alloc(u21, len - 1);
+        self.fold[cp >> 1] = try allocator.alloc(u21, len - 1);
         for (0..len - 1) |i| {
-            self.fold[cp][i] = @intCast(try reader.readInt(u24, endian));
+            self.fold[cp >> 1][i] = @intCast(try reader.readInt(u24, endian));
         }
+        self.cwcf[cp >> 1] = cp & 1 == 1;
     }
 
     return self;
@@ -40,9 +44,15 @@ pub fn init(allocator: mem.Allocator) !Self {
 pub fn deinit(self: *Self) void {
     for (self.fold) |slice| self.allocator.free(slice);
     self.allocator.free(self.fold);
+    self.allocator.free(self.cwcf);
 }
 
 /// Returns the case fold for `cp`.
 pub inline fn caseFold(self: Self, cp: u21) []const u21 {
     return self.fold[cp];
 }
+
+/// Returns true when caseFold(NFD(`cp`)) != NFD(`cp`).
+pub inline fn changesWhenCaseFolded(self: Self, cp: u21) bool {
+    return self.cwcf[cp];
+}
diff --git a/src/Normalizer.zig b/src/Normalizer.zig
index c68b2ec..5a26dfa 100644
--- a/src/Normalizer.zig
+++ b/src/Normalizer.zig
@@ -389,6 +389,12 @@ fn nfkdCodePoints(
     return try dcp_list.toOwnedSlice();
 }
 
+fn changesWhenCaseFolded(self: Self, cps: []const u21) bool {
+    return for (cps) |cp| {
+        if (self.norm_data.fold_data.changesWhenCaseFolded(cp)) break true;
+    } else false;
+}
+
 pub fn eqlIgnoreCase(
     self: Self,
     allocator: mem.Allocator,
@@ -397,10 +403,18 @@ pub fn eqlIgnoreCase(
 ) !bool {
     if (ascii.isAsciiOnly(a) and ascii.isAsciiOnly(b)) return std.ascii.eqlIgnoreCase(a, b);
 
+    // Process a
     const nfd_a = try self.nfxdCodePoints(allocator, a, .nfd);
     defer allocator.free(nfd_a);
-    const cf_nfd_a = try self.caseFold(allocator, nfd_a);
-    defer allocator.free(cf_nfd_a);
+
+    var need_frr_cf_nfd_a = false;
+    var cf_nfd_a: []const u21 = nfd_a;
+    if (self.changesWhenCaseFolded(nfd_a)) {
+        cf_nfd_a = try self.caseFold(allocator, nfd_a);
+        need_frr_cf_nfd_a = true;
+    }
+    defer if (need_frr_cf_nfd_a) allocator.free(cf_nfd_a);
+
     const nfkd_cf_nfd_a = try self.nfkdCodePoints(allocator, cf_nfd_a);
     defer allocator.free(nfkd_cf_nfd_a);
     const cf_nfkd_cf_nfd_a = try self.caseFold(allocator, nfkd_cf_nfd_a);
@@ -408,10 +422,18 @@ pub fn eqlIgnoreCase(
     const nfkd_cf_nfkd_cf_nfd_a = try self.nfkdCodePoints(allocator, cf_nfkd_cf_nfd_a);
     defer allocator.free(nfkd_cf_nfkd_cf_nfd_a);
 
+    // Process b
     const nfd_b = try self.nfxdCodePoints(allocator, b, .nfd);
     defer allocator.free(nfd_b);
-    const cf_nfd_b = try self.caseFold(allocator, nfd_b);
-    defer allocator.free(cf_nfd_b);
+
+    var need_frr_cf_nfd_b = false;
+    var cf_nfd_b: []const u21 = nfd_b;
+    if (self.changesWhenCaseFolded(nfd_b)) {
+        cf_nfd_b = try self.caseFold(allocator, nfd_b);
+        need_frr_cf_nfd_b = true;
+    }
+    defer if (need_frr_cf_nfd_b) allocator.free(cf_nfd_b);
+
     const nfkd_cf_nfd_b = try self.nfkdCodePoints(allocator, cf_nfd_b);
     defer allocator.free(nfkd_cf_nfd_b);
     const cf_nfkd_cf_nfd_b = try self.caseFold(allocator, nfkd_cf_nfd_b);
-- 
cgit v1.2.3