From bfe0a7887fef67a91f46c35d41a6eccd38c3e9fe Mon Sep 17 00:00:00 2001
From: penguindark <57967770+penguindark@users.noreply.github.com>
Date: Sun, 25 Apr 2021 16:57:21 +0200
Subject: [PATCH] builtin,strconv: fix and optimize utf8 and formatting
 functions (#9874)

---
 vlib/builtin/utf8.v    | 23 +++++++++--------------
 vlib/strconv/f32_str.v |  8 ++++++++
 vlib/strconv/f64_str.v |  8 ++++++++
 vlib/strconv/format.v  | 14 ++++++++++++--
 4 files changed, 37 insertions(+), 16 deletions(-)

diff --git a/vlib/builtin/utf8.v b/vlib/builtin/utf8.v
index d2eca88cf2..f55d1b087a 100644
--- a/vlib/builtin/utf8.v
+++ b/vlib/builtin/utf8.v
@@ -103,16 +103,12 @@ fn utf8_len(c byte) int {
 }
 
 // Calculate string length for in number of codepoints
-fn utf8_str_len(s string) int {
+pub fn utf8_str_len(s string) int {
 	mut l := 0
-	for i := 0; i < s.len; i++ {
+	mut i := 0
+	for i < s.len {
 		l++
-		c := unsafe { s.str[i] }
-		if (c & (1 << 7)) != 0 {
-			for t := byte(1 << 6); (c & t) != 0; t >>= 1 {
-				i++
-			}
-		}
+		i += ((0xe5000000 >> ((unsafe { s.str[i] } >> 3) & 0x1e)) & 3) + 1
 	}
 	return l
 }
@@ -124,17 +120,16 @@ pub fn utf8_str_visible_length(s string) int {
 	mut l := 0
 	mut ul := 1
 	for i := 0; i < s.len; i += ul {
-		ul = 1
 		c := unsafe { s.str[i] }
-		if (c & (1 << 7)) != 0 {
-			for t := byte(1 << 6); (c & t) != 0; t >>= 1 {
-				ul++
-			}
-		}
+		ul = ((0xe5000000 >> ((unsafe { s.str[i] } >> 3) & 0x1e)) & 3) + 1
 		if i + ul > s.len { // incomplete UTF-8 sequence
 			return l
 		}
 		l++
+		// avoid the match if not needed
+		if ul == 1 {
+			continue
+		}
 		// recognize combining characters and wide characters
 		match ul {
 			2 {
diff --git a/vlib/strconv/f32_str.v b/vlib/strconv/f32_str.v
index 270c4c2458..a1e231c998 100644
--- a/vlib/strconv/f32_str.v
+++ b/vlib/strconv/f32_str.v
@@ -94,6 +94,14 @@ pub fn (d Dec32) get_string_32(neg bool, i_n_digit int, i_pad_digit int) string
 		x++
 	}
 
+	// no decimal digits needed, end here
+	if i_n_digit == 0 {
+		unsafe {
+			buf[i]=0
+			return 	tos(byteptr(&buf[0]), i)
+		}
+	}
+
 	if out_len >= 1 {
 		buf[y - x] = `.`
 		x++
diff --git a/vlib/strconv/f64_str.v b/vlib/strconv/f64_str.v
index f33d54797a..99c73bd082 100644
--- a/vlib/strconv/f64_str.v
+++ b/vlib/strconv/f64_str.v
@@ -110,6 +110,14 @@ fn (d Dec64) get_string_64(neg bool, i_n_digit int, i_pad_digit int) string {
 		x++
 	}
 
+	// no decimal digits needed, end here
+	if i_n_digit == 0 {
+		unsafe {
+			buf[i]=0
+			return 	tos(byteptr(&buf[0]), i)
+		}
+	}
+
 	if out_len >= 1 {
 		buf[y - x] = `.`
 		x++
diff --git a/vlib/strconv/format.v b/vlib/strconv/format.v
index bba96f2b36..f9c06efc32 100644
--- a/vlib/strconv/format.v
+++ b/vlib/strconv/format.v
@@ -29,7 +29,7 @@ enum Char_parse_state {
 	reset_params
 }
 
-enum Align_text {
+pub enum Align_text {
 	right = 0
 	left
 	center
@@ -176,6 +176,12 @@ pub fn f64_to_str_lnd(f f64, dec_digit int) string {
 			i++
 		}
 	}
+	
+	// no more digits needed, stop here
+	if dec_digit <= 0 {	
+		return unsafe { tos(res.data, dot_res_sp) }
+	} 
+
 	//println("r_i-d_pos: ${r_i - d_pos}")
 	if dot_res_sp >= 0 {
 		if (r_i - dot_res_sp) > dec_digit {
@@ -204,6 +210,7 @@ pub fn f64_to_str_lnd(f f64, dec_digit int) string {
 
 */
 pub struct BF_param {
+pub mut:
 	pad_ch       byte       = byte(` `)     // padding char
 	len0         int        = -1      // default len for whole the number or string
 	len1         int        = 6       // number of decimal digits, if needed
@@ -214,7 +221,10 @@ pub struct BF_param {
 }
 
 pub fn format_str(s string, p BF_param) string {
-	dif := p.len0 - s.len
+	if p.len0 <= 0 {
+		return s
+	}
+	dif := p.len0 - utf8_str_visible_length(s)
 	if dif <= 0 {
 		return s
 	}