--- fenglin/lib/str_process.inc.php	2025/03/24 05:46:49	1.11
+++ fenglin/lib/str_process.inc.php	2025/10/31 10:31:55	1.24
@@ -1,12 +1,74 @@
-<?
-function split_line($str, $prefix = "", $width = 76, $lines_limit = PHP_INT_MAX)
+<?php
+function str_length(string $str, bool $skip_ctrl_seq = false) : int
+{
+	$len = strlen($str);
+	$ret = 0;
+
+	for ($i = 0; $i < $len; $i++)
+	{
+		$c = $str[$i];
+
+		if ($c == "\r" || $c == "\7") // skip
+		{
+			continue;
+		}
+
+		if ($skip_ctrl_seq && $c == "\033" && $i + 1 < $len && $str[$i + 1] == "[") // Skip control sequence
+		{
+			for ($i = $i + 2; 
+				$i < $len && (ctype_digit($str[$i]) || $str[$i] == ';' || $str[$i] == '?');
+				$i++)
+				;
+
+			if ($i < $len && $str[$i] == 'm') // valid
+			{
+				// skip
+			}
+			else if ($i < $len && ctype_alpha($str[$i]))
+			{
+				// unsupported ANSI CSI command
+			}
+			else
+			{
+				$i--;
+			}
+
+			continue;
+		}
+
+		// Process UTF-8 Chinese characters
+		$v1 = ord($c);
+		if ($v1 & 0x80) //head of multi-byte character
+		{
+			$v2 = ($v1 & 0x70) << 1;
+			while ($v2 & 0x80)
+			{
+				$i++;
+				$c .= $str[$i];
+				$v2 = ($v2 & 0x7f) << 1;
+			}
+
+			$ret += 2;
+		}
+		else
+		{
+			$ret++;
+		}
+	}
+
+	return $ret;
+}
+
+function split_line(string $str, string $prefix = "", int $width = PHP_INT_MAX, int $lines_limit = PHP_INT_MAX, string $end_of_line = "\n") : string
 {
 	if ($width <= 0)
-		return $str;
-	
+	{
+		$width = PHP_INT_MAX;
+	}
+
 	$result = "";
 	$len = strlen($str);
-	$prefix_len = strlen($prefix);
+	$prefix_len = str_length($prefix);
 
 	$lines_count = 0;
 
@@ -21,10 +83,15 @@ function split_line($str, $prefix = "",
 		{
 			continue;
 		}
-				
+
 		if ($c == "\n")
 		{
-			$result .= ($line . "\n");
+			if ($lines_count + 1 >= $lines_limit)
+			{
+				break;
+			}
+
+			$result .= ($line . $end_of_line);
 			$lines_count++;
 			$line = $prefix;
 			$line_len = $prefix_len;
@@ -33,21 +100,25 @@ function split_line($str, $prefix = "",
 
 		// Process UTF-8 Chinese characters
 		$v1 = ord($c);
-		if (($v1 & 0b10000000) == 0b10000000) //head of multi-byte character
+		if ($v1 & 0x80) //head of multi-byte character
 		{
-			$v2 = ($v1 & 0b01111000) << 1;
-			while ($v2 & 0b10000000)
+			$v2 = ($v1 & 0x70) << 1;
+			while ($v2 & 0x80)
 			{
 				$i++;
-				$v3 = $str[$i];
-				$c .= $v3;
-				$v2 = ($v2 & 0b01111111 ) << 1;
+				$c .= $str[$i];
+				$v2 = ($v2 & 0x7f) << 1;
 			}
 
 			// Each UTF-8 CJK character should use two character length for display
 			if ($line_len + 2 > $width)
 			{
-				$result .= ($line . "\n");
+				if ($lines_count + 1 >= $lines_limit)
+				{
+					break;
+				}
+
+				$result .= ($line . $end_of_line);
 				$lines_count++;
 				$line = $prefix;
 				$line_len = $prefix_len;
@@ -59,15 +130,20 @@ function split_line($str, $prefix = "",
 			$line_len++;
 		}
 
-		$line .= $c;
-
-		if ($line_len >= $width)
+		if ($line_len > $width)
 		{
-			$result .= ($line . "\n");
+			if ($lines_count + 1 >= $lines_limit)
+			{
+				break;
+			}
+
+			$result .= ($line . $end_of_line);
 			$lines_count++;
 			$line = $prefix;
-			$line_len = $prefix_len;
+			$line_len = $prefix_len + 1;
 		}
+
+		$line .= $c;
 	}
 
 	if ($lines_count < $lines_limit)
@@ -77,4 +153,3 @@ function split_line($str, $prefix = "",
 
 	return $result;
 }
-?>