--- fenglin/lib/str_process.inc.php	2004/12/10 15:54:32	1.5
+++ fenglin/lib/str_process.inc.php	2025/10/31 06:13:35	1.23
@@ -1,75 +1,155 @@
-<?
-
-function split_line($str,$prefix="",$width=75)
+<?php
+function str_length(string $str, bool $skip_ctrl_seq = false) : int
 {
-	$lines = array();
-	$lines_count = 0;
 	$len = strlen($str);
+	$ret = 0;
 
-	$hz_ok = true;
-	
-	if ($width<=0)
-		return $str;
-	
-	$line = $prefix;
-	$line_count = strlen($line);
-	for($i=0;$i<$len;$i++)
+	for ($i = 0; $i < $len; $i++)
 	{
-		$line.=$str[$i];
-		$line_count++;
-		if (ord($str[$i])>127)
+		$c = $str[$i];
+
+		if ($c == "\r" || $c == "\7") // skip
+		{
+			continue;
+		}
+
+		if ($skip_ctrl_seq && $c == "\033" && isset($str[$i + 1]) && $str[$i + 1] == "[") // Skip control sequence
 		{
-			$hz_ok = (!$hz_ok);
+			for ($i = $i + 2; 
+				isset($str[$i]) && (ctype_digit($str[$i]) || $str[$i] == ';' || $str[$i] == '?');
+				$i++)
+				;
+
+			if (isset($str[$i]) && $str[$i] == 'm') // valid
+			{
+				// skip
+			}
+			else if (isset($str[$i]) && ctype_alpha($str[$i]))
+			{
+				// unsupported ANSI CSI command
+			}
+			else
+			{
+				$i--;
+			}
+
+			continue;
 		}
-		if (ord($str[$i])<=127)
+
+		// Process UTF-8 Chinese characters
+		$v1 = ord($c);
+		if ($v1 & 0x80) //head of multi-byte character
 		{
-			$hz_ok = true;
+			$v2 = ($v1 & 0x70) << 1;
+			while ($v2 & 0x80)
+			{
+				$i++;
+				$c .= $str[$i];
+				$v2 = ($v2 & 0x7f) << 1;
+			}
+
+			$ret += 2;
 		}
-		if ($line_count >= $width && $hz_ok)
+		else
 		{
-			$lines[$lines_count++] = $line;
-			$line = $prefix;
-			$line_count = strlen($line);
+			$ret++;
 		}
 	}
-	$lines[$lines_count++] = $line;
 
-	$result = "";
-	for($i=0;$i<$lines_count;$i++)
-	{
-		$result.=($lines[$i]."\r\n");
-	}
-	return $result;
+	return $ret;
 }
 
-function str_process($str,$prefix="",$width=75)
+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)
+	{
+		$width = PHP_INT_MAX;
+	}
+
 	$result = "";
-	
-	$lines = explode("\n",str_replace("\r\n","\n",$str));
+	$len = strlen($str);
+	$prefix_len = str_length($prefix);
+
+	$lines_count = 0;
 
-	foreach($lines as $line)
+	$line = $prefix;
+	$line_len = $prefix_len;
+	for ($i = 0; $i < $len && $lines_count < $lines_limit; $i++)
 	{
-		$result .= split_line($line,$prefix,$width);
+		$c = $str[$i];
+
+		// Skip special characters
+		if ($c == "\r" || $c == "\7")
+		{
+			continue;
+		}
+
+		if ($c == "\n")
+		{
+			if ($lines_count + 1 >= $lines_limit)
+			{
+				break;
+			}
+
+			$result .= ($line . $end_of_line);
+			$lines_count++;
+			$line = $prefix;
+			$line_len = $prefix_len;
+			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;
+			}
+
+			// Each UTF-8 CJK character should use two character length for display
+			if ($line_len + 2 > $width)
+			{
+				if ($lines_count + 1 >= $lines_limit)
+				{
+					break;
+				}
+
+				$result .= ($line . $end_of_line);
+				$lines_count++;
+				$line = $prefix;
+				$line_len = $prefix_len;
+			}
+			$line_len += 2;
+		}
+		else
+		{
+			$line_len++;
+		}
+
+		if ($line_len > $width)
+		{
+			if ($lines_count + 1 >= $lines_limit)
+			{
+				break;
+			}
+
+			$result .= ($line . $end_of_line);
+			$lines_count++;
+			$line = $prefix;
+			$line_len = $prefix_len + 1;
+		}
+
+		$line .= $c;
 	}
-	
-	return $result;
-}
 
-function str_b_lines($str,$max_line=1)
-{
-	$result = "";
-	$l = 0;
-	$lines = explode("\n",str_replace("\r\n","\n",$str));
-	foreach($lines as $line)
+	if ($lines_count < $lines_limit)
 	{
-		$result .= ($line."\r\n");
-		$l += (strlen($line)/256 + 1);
-		if ($l >= $max_line)
-			break;
+		$result .= $line;
 	}
 
 	return $result;
 }
-
-?>