From 2000f47ae539ff31c248626e4d142a3221522708 Mon Sep 17 00:00:00 2001
From: Kurt Kartaltepe <kkartaltepe@gmail.com>
Date: Wed, 19 May 2021 00:34:09 -0700
Subject: [PATCH] [set] Compute is_subset by comparing pages.

Test subsets one page at a time instead of by codepoints. On my machine
this is about 250x faster than the previous implementation.
---
 src/hb-set.hh       | 44 +++++++++++++++++++++++++++++++------
 test/api/test-set.c | 53 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 90 insertions(+), 7 deletions(-)

diff --git a/src/hb-set.hh b/src/hb-set.hh
index 9404ba261..28ece87ff 100644
--- a/src/hb-set.hh
+++ b/src/hb-set.hh
@@ -506,14 +506,44 @@ struct hb_set_t
 
   bool is_subset (const hb_set_t *larger_set) const
   {
-    if (get_population () > larger_set->get_population ())
-      return false;
+    if (unlikely(larger_set->is_empty ()))
+    {
+      return is_empty ();
+    }
 
-    /* TODO Optimize to use pages. */
-    hb_codepoint_t c = INVALID;
-    while (next (&c))
-      if (!larger_set->has (c))
-	return false;
+    uint32_t spi = 0;
+    for (uint32_t lpi = 0; spi < page_map.length && lpi < larger_set->page_map.length; lpi++)
+    {
+      uint32_t spm = page_map[spi].major;
+      uint32_t lpm = larger_set->page_map[lpi].major;
+      auto sp = page_at (spi);
+      auto lp = larger_set->page_at (lpi);
+
+      if (spm < lpm && !sp.is_empty ())
+      {
+        return false;
+      }
+
+      if (lpm < spm)
+      {
+        continue;
+      }
+
+      for (int j = 0; j < ARRAY_LENGTH_CONST (sp.v); j++)
+      {
+        if ((~lp.v[j] & sp.v[j]) != 0) { return false; }
+      }
+
+      spi++;
+    }
+
+    while (spi < page_map.length)
+    {
+      if (!page_at (spi++).is_empty ())
+      {
+        return false;
+      }
+    }
 
     return true;
   }
diff --git a/test/api/test-set.c b/test/api/test-set.c
index 30a47674f..21a48ffa6 100644
--- a/test/api/test-set.c
+++ b/test/api/test-set.c
@@ -215,6 +215,58 @@ static void test_set_union (void)
   hb_set_destroy (u);
 }
 
+static void
+test_set_subsets (void)
+{
+  hb_set_t *s = hb_set_create ();
+  hb_set_t *l = hb_set_create ();
+
+  hb_set_add (l, 0x0FFFF);
+  hb_set_add (s, 0x1FFFF);
+  g_assert (!hb_set_is_subset (s, l));
+  hb_set_clear (s);
+
+  hb_set_add (s, 0x0FFF0);
+  g_assert (!hb_set_is_subset (s, l));
+  hb_set_clear (s);
+
+  hb_set_add (s, 0x0AFFF);
+  g_assert (!hb_set_is_subset (s, l));
+
+  hb_set_clear (s);
+  g_assert (hb_set_is_subset (s, l));
+
+  hb_set_clear (l);
+  g_assert (hb_set_is_subset (s, l));
+
+  hb_set_add (s, 0x1FFFF);
+  g_assert (!hb_set_is_subset (s, l));
+  hb_set_clear (s);
+
+  hb_set_add (s, 0xFF);
+  hb_set_add (s, 0x1FFFF);
+  hb_set_add (s, 0x2FFFF);
+
+  hb_set_add (l, 0xFF);
+  hb_set_add (l, 0x1FFFF);
+  hb_set_add (l, 0x2FFFF);
+
+  g_assert (hb_set_is_subset (s, l));
+  hb_set_del (l, 0xFF);
+  g_assert (!hb_set_is_subset (s, l));
+  hb_set_add (l, 0xFF);
+
+  hb_set_del (l, 0x2FFFF);
+  g_assert (!hb_set_is_subset (s, l));
+  hb_set_add (l, 0x2FFFF);
+
+  hb_set_del (l, 0x1FFFF);
+  g_assert (!hb_set_is_subset (s, l));
+
+  hb_set_destroy (s);
+  hb_set_destroy (l);
+}
+
 static void
 test_set_algebra (void)
 {
@@ -528,6 +580,7 @@ main (int argc, char **argv)
   hb_test_init (&argc, &argv);
 
   hb_test_add (test_set_basic);
+  hb_test_add (test_set_subsets);
   hb_test_add (test_set_algebra);
   hb_test_add (test_set_iter);
   hb_test_add (test_set_empty);