From 8846b406ebc48e97bd7e653215960ad999531c1e Mon Sep 17 00:00:00 2001
From: Alex Sharov <AskAlexSharov@gmail.com>
Date: Sun, 27 Mar 2022 13:41:16 +0700
Subject: [PATCH] gsa: remove terminator (make it closer to SAIS) #400

---
 sais/gsa/gsa_test.go | 23 +++++++++-------
 sais/gsa/gsaca.go    | 62 ++++++++++++++++----------------------------
 2 files changed, 36 insertions(+), 49 deletions(-)

diff --git a/sais/gsa/gsa_test.go b/sais/gsa/gsa_test.go
index 9d76b616b..a72c2805d 100644
--- a/sais/gsa/gsa_test.go
+++ b/sais/gsa/gsa_test.go
@@ -8,11 +8,15 @@ import (
 )
 
 func TestExampleGSA(t *testing.T) {
-	R := [][]byte{[]byte("hihihi")}
+	R := [][]byte{
+		[]byte("hihihi"),
+		[]byte("hihihi"),
+		[]byte("hihihi"),
+	}
 	str, n := ConcatAll(R)
-	sa := make([]uint, SaSize(n))
-	lcp := make([]int, LcpSize(n))
-	da := make([]int32, DaSize(n))
+	sa := make([]uint, n)
+	lcp := make([]int, n)
+	da := make([]int32, n)
 	_ = GSA(str, sa, lcp, da)
 	PrintArrays(str, sa, lcp, da, n)
 }
@@ -20,9 +24,10 @@ func TestExampleGSA(t *testing.T) {
 func TestGSA(t *testing.T) {
 	R := [][]byte{{4, 5, 6, 4, 5, 6, 4, 5, 6}}
 	str, n := ConcatAll(R)
-	sa := make([]uint, SaSize(n))
+	sa := make([]uint, n)
 	lcp := make([]int, n)
-	_ = GSA(str, sa, lcp, nil)
+	da := make([]int32, n)
+	_ = GSA(str, sa, lcp, da)
 	assert.Equal(t, []uint{10, 9, 6, 3, 0, 7, 4, 1, 8, 5, 2}, sa[:n])
 }
 
@@ -57,9 +62,9 @@ func BenchmarkName2(b *testing.B) {
 		R = append(R, []byte("hihihi"))
 	}
 	str, n := ConcatAll(R)
-	sa := make([]uint, SaSize(n))
-	lcp := make([]int, LcpSize(n))
-	da := make([]int32, DaSize(n))
+	sa := make([]uint, n)
+	lcp := make([]int, n)
+	da := make([]int32, n)
 	b.ResetTimer()
 	for i := 0; i < b.N; i++ {
 		_ = GSA(str, sa, lcp, da)
diff --git a/sais/gsa/gsaca.go b/sais/gsa/gsaca.go
index 8b1e3f184..84908c413 100644
--- a/sais/gsa/gsaca.go
+++ b/sais/gsa/gsaca.go
@@ -2,6 +2,7 @@ package gsa
 
 /*
 #include "gsacak.h"
+#cgo CFLAGS: -DTERMINATOR=0 -DM64=1 -Dm64=1
 */
 import "C"
 import (
@@ -12,66 +13,47 @@ import (
 // Implementation from https://github.com/felipelouza/gsufsort
 // see also: https://almob.biomedcentral.com/track/pdf/10.1186/s13015-020-00177-y.pdf
 // see also: https://almob.biomedcentral.com/track/pdf/10.1186/s13015-017-0117-9.pdf
-
-func SaSize(l int) int {
-	var a uint
-	return l * int(unsafe.Sizeof(a))
-}
-func LcpSize(l int) int {
-	var a uint
-	return l * int(unsafe.Sizeof(a))
-}
-func DaSize(l int) int {
-	var a C.int_da
-	return l * int(unsafe.Sizeof(a))
-}
 func PrintArrays(str []byte, sa []uint, lcp []int, da []int32, n int) {
+	// remove terminator
+	sa = sa[1:]
+	lcp = lcp[1:]
+	terminatorDa := da[0]
+	da = da[1:]
+	n = n - 1
+
 	fmt.Printf("i\t")
-	if da != nil {
-		fmt.Printf("da\t")
-	}
+	fmt.Printf("sa\t")
 	if lcp != nil {
 		fmt.Printf("lcp\t")
 	}
-	fmt.Printf("gsa\t")
+	if da != nil {
+		fmt.Printf("gsa\t")
+	}
 	fmt.Printf("suffixes\t")
 	fmt.Printf("\n")
 	for i := 0; i < n; i++ {
 		fmt.Printf("%d\t", i)
+		fmt.Printf("%d\t", sa[i])
 		if lcp != nil {
 			fmt.Printf("%d\t", lcp[i])
 		}
 
-		if da != nil {
-			pos := sa[da[i]]
+		if da != nil { // gsa
+			pos := uint(terminatorDa)
+			if da[i] > 0 {
+				pos = sa[da[i]-1]
+			}
 			value := sa[i]
 			if da[i] != 0 {
 				value = sa[i] - pos - 1
 			}
 			fmt.Printf("(%d %d)\t", da[i], value)
-		} else {
-			fmt.Printf("%d\t", sa[i])
 		}
-		/*
-			if(gsa){
-			      da_value = (light)?rankbv_rank1(rbv,SA[i]):DA[i];
-			      printf("(%" PRIdA ", ", da_value);
+		//bwt
+		//	char c = (SA[i])? T[SA[i]-1]-1:terminal;
+		//	if(c==0) c = '$';
+		//	printf("%c\t",c);
 
-			      int_t pos;
-			      if(last_end) pos=SA[da_value];
-			      else pos=SA[da_value-1];
-
-			      int_t value = (da_value==0)?SA[i]:SA[i]-pos-1;
-			      printf("%" PRIdN ")   \t", value);
-			    }
-		*/
-		/*
-		  if(bwt){
-		      char c = (SA[i])? T[SA[i]-1]-1:terminal;
-		      if(c==0) c = '$';
-		      printf("%c\t",c);
-		    }
-		*/
 		for j := sa[i]; int(j) < n; j++ {
 			if str[j] == 1 {
 				fmt.Printf("$")