From 895bd71fa058c52a1236b22693ecef9199de9c76 Mon Sep 17 00:00:00 2001
From: Turiiya <34311583+ttytm@users.noreply.github.com>
Date: Sat, 29 Jun 2024 10:10:10 +0200
Subject: [PATCH] builtin: improve snake to camel case conversion (#21755)

---
 vlib/builtin/string.v      | 69 +++++++++++++++++---------------------
 vlib/builtin/string_test.v | 11 ++++--
 2 files changed, 40 insertions(+), 40 deletions(-)

diff --git a/vlib/builtin/string.v b/vlib/builtin/string.v
index 1ae24f61cf2a23..3ef2cd6e63b464 100644
--- a/vlib/builtin/string.v
+++ b/vlib/builtin/string.v
@@ -2651,56 +2651,49 @@ pub fn (s string) camel_to_snake() string {
 		return s.to_lower()
 	}
 	mut b := unsafe { malloc_noscan(2 * s.len + 1) }
+	// Rather than checking whether the iterator variable is > 1 inside the loop,
+	// handle the first two chars separately to reduce load.
+	mut pos := 2
 	mut prev_is_upper := false
-	first_char, second_char := if s[0].is_capital() {
-		lower_first_c := s[0] + 32
-		lower_second_c := if s[1].is_capital() {
-			prev_is_upper = true
-			s[1] + 32
-		} else {
-			s[1]
-		}
-		lower_first_c, lower_second_c
-	} else {
-		first_c := s[0]
-		second_c := if s[1].is_capital() {
-			if first_c == `_` { s[1] + 32 } else { u8(`_`) }
+	unsafe {
+		if s[0].is_capital() {
+			b[0] = s[0] + 32
+			b[1] = if s[1].is_capital() {
+				prev_is_upper = true
+				s[1] + 32
+			} else {
+				s[1]
+			}
 		} else {
-			s[1]
+			b[0] = s[0]
+			if s[1].is_capital() {
+				prev_is_upper = true
+				if s[0] != `_` && s.len > 2 && !s[2].is_capital() {
+					b[1] = `_`
+					b[2] = s[1] + 32
+					pos = 3
+				} else {
+					b[1] = s[1] + 32
+				}
+			} else {
+				b[1] = s[1]
+			}
 		}
-		first_c, second_c
 	}
-	unsafe {
-		b[0] = first_char
-		b[1] = second_char
-	}
-	mut pos := 2
-	mut prev_char := second_char
-	mut lower_c := `_`
-	mut c_is_upper := false
-	for i in pos .. s.len {
+	for i := 2; i < s.len; i++ {
 		c := s[i]
-		c_is_upper = c.is_capital()
-		lower_c = if c_is_upper { c + 32 } else { c }
-		if !prev_is_upper && c_is_upper {
-			// aB => a_b, if prev has `_`, then do not add `_`
+		c_is_upper := c.is_capital()
+		// Cases: `aBcd == a_bcd` || `ABcd == ab_cd`
+		if ((c_is_upper && !prev_is_upper)
+			|| (!c_is_upper && prev_is_upper && s[i - 2].is_capital())) && c != `_` {
 			unsafe {
 				if b[pos - 1] != `_` {
 					b[pos] = `_`
 					pos++
 				}
 			}
-		} else if prev_is_upper && !c_is_upper && c != `_` {
-			// Ba => _ba, if prev has `_`, then do not add `_`
-			unsafe {
-				if b[pos - 2] != `_` {
-					prev_char = b[pos - 1]
-					b[pos - 1] = `_`
-					b[pos] = prev_char
-					pos++
-				}
-			}
 		}
+		lower_c := if c_is_upper { c + 32 } else { c }
 		unsafe {
 			b[pos] = lower_c
 		}
diff --git a/vlib/builtin/string_test.v b/vlib/builtin/string_test.v
index b9cf2dcf912c77..95b7003392e555 100644
--- a/vlib/builtin/string_test.v
+++ b/vlib/builtin/string_test.v
@@ -1524,12 +1524,19 @@ fn test_contains_byte() {
 
 fn test_camel_to_snake() {
 	assert 'Abcd'.camel_to_snake() == 'abcd'
+	assert 'aBcd'.camel_to_snake() == 'a_bcd'
+	assert 'AAbb'.camel_to_snake() == 'aa_bb'
 	assert 'aaBB'.camel_to_snake() == 'aa_bb'
 	assert 'aaBbCcDD'.camel_to_snake() == 'aa_bb_cc_dd'
-	assert 'BBaa'.camel_to_snake() == 'b_baa'
+	assert 'AAbbCC'.camel_to_snake() == 'aa_bb_cc'
+	assert 'aaBBcc'.camel_to_snake() == 'aa_bb_cc'
 	assert 'aa_BB'.camel_to_snake() == 'aa_bb'
+	assert 'aa__BB'.camel_to_snake() == 'aa__bb'
 	assert 'JVM_PUBLIC_ACC'.camel_to_snake() == 'jvm_public_acc'
-	assert '_ISspace'.camel_to_snake() == '_i_sspace'
+	assert '_ISspace'.camel_to_snake() == '_is_space'
+	assert '_aBcd'.camel_to_snake() == '_a_bcd'
+	assert '_a_Bcd'.camel_to_snake() == '_a_bcd'
+	assert '_AbCDe_'.camel_to_snake() == '_ab_cd_e_'
 }
 
 fn test_snake_to_camel() {