From 34fc2bfde5108956369769352cf99eb75bd54505 Mon Sep 17 00:00:00 2001
From: Daniel Lemire <daniel@lemire.me>
Date: Wed, 20 Mar 2024 18:10:06 -0400
Subject: [PATCH] faster faster iandNotRun16

Co-authored-by: Neena Dugar <neena@lightstep.com>
---
 arraycontainer.go | 57 +++++++++++++++++++++++++++++++++++------------
 1 file changed, 43 insertions(+), 14 deletions(-)

diff --git a/arraycontainer.go b/arraycontainer.go
index 1fa626a6..80fa676e 100644
--- a/arraycontainer.go
+++ b/arraycontainer.go
@@ -664,25 +664,54 @@ func (ac *arrayContainer) iandNot(a container) container {
 }
 
 func (ac *arrayContainer) iandNotRun16(rc *runContainer16) container {
-	if len(ac.content) == 0 {
+	// Fast path: if either the array container or the run container is empty, the result is the array.
+	if ac.isEmpty() || rc.isEmpty() {
 		// Empty
 		return ac
 	}
-
-	for _, run := range rc.iv {
-		if run.start > ac.maximum() {
-			// Since the runs are sorted, we can stop here. (No subsequent runs will
-			// overlap with the array container.)
-			break
-		}
-		if run.last() < ac.minimum() {
-			// This run is entirely before the array container. We can skip it.
-			continue
+	// Fast path: if the run container is full, the result is empty.
+	if rc.isFull() {
+		ac.content = ac.content[:0]
+		return ac
+	}
+	current_run := 0
+	// All values in [start_run, end_end] are part of the run
+	start_run := rc.iv[current_run].start
+	end_end := start_run + rc.iv[current_run].length
+	// We are going to read values in the array at index i, and we are
+	// going to write them at index pos. So we do in-place processing.
+	// We always have that pos <= i by construction. So we can either
+	// overwrite a value just read, or a value that was previous read.
+	pos := 0
+	i := 0
+	for ; i < len(ac.content); i++ {
+		if ac.content[i] < start_run {
+			// the value in the array appears before the run [start_run, end_end]
+			ac.content[pos] = ac.content[i]
+			pos++
+		} else if ac.content[i] <= end_end {
+			// nothing to do, the value is in the array but also in the run.
+		} else {
+			// We have the value in the array after the run. We cannot tell
+			// whether we need to keep it or not. So let us move to another run.
+			if current_run+1 < len(rc.iv) {
+				current_run++
+				start_run = rc.iv[current_run].start
+				end_end = start_run + rc.iv[current_run].length
+				i-- // retry with the same i
+			} else {
+				// We have exhausted the number of runs. We can keep the rest of the values
+				// from i to len(ac.content) - 1 inclusively.
+				break // We are done, the rest of the array will be kept
+			}
 		}
-
-		out := ac.iremoveRange(int(run.start), int(run.start)+int(run.length)+1).(*arrayContainer)
-		*ac = *out
 	}
+	for ; i < len(ac.content); i++ {
+		ac.content[pos] = ac.content[i]
+		pos++
+	}
+	// We 'shink' the slice.
+	ac.content = ac.content[:pos]
 	return ac
 }