utf_normalizer.php

这些都是我以前学习是用到的源码

<?php
/** 
*
* @package phpBB3
* @version $Id: utf_normalizer.php,v 1.10 2006/11/19 20:26:40 davidmj Exp $ 
* @copyright (c) 2005 phpBB Group 
* @license http://opensource.org/licenses/gpl-license.php GNU Public License 
*
*/

/**
* Some Unicode characters encoded in UTF-8
*
* Preserved for compatibility
*/
define('UTF8_REPLACEMENT', "\xEF\xBF\xBD");
define('UTF8_MAX', "\xF4\x8F\xBF\xBF");
define('UTF8_FFFE', "\xEF\xBF\xBE");
define('UTF8_FFFF', "\xEF\xBF\xBF");
define('UTF8_SURROGATE_FIRST', "\xED\xA0\x80");
define('UTF8_SURROGATE_LAST', "\xED\xBF\xBF");
define('UTF8_HANGUL_FIRST', "\xEA\xB0\x80");
define('UTF8_HANGUL_LAST', "\xED\x9E\xA3");

define('UTF8_CJK_FIRST', "\xE4\xB8\x80");
define('UTF8_CJK_LAST', "\xE9\xBE\xBB");
define('UTF8_CJK_B_FIRST', "\xF0\xA0\x80\x80");
define('UTF8_CJK_B_LAST', "\xF0\xAA\x9B\x96");

// Unset global variables
unset($GLOBALS['utf_jamo_index'], $GLOBALS['utf_jamo_type'], $GLOBALS['utf_nfc_qc'], $GLOBALS['utf_combining_class'], $GLOBALS['utf_canonical_comp'], $GLOBALS['utf_canonical_decomp'], $GLOBALS['utf_nfkc_qc'], $GLOBALS['utf_compatibility_decomp']);

// NFC_QC and NFKC_QC values
define('UNICODE_QC_MAYBE', 0);
define('UNICODE_QC_NO', 1);

// Contains all the ASCII characters appearing in UTF-8, sorted by frequency
define('UTF8_ASCII_RANGE', "\x20\x65\x69\x61\x73\x6E\x74\x72\x6F\x6C\x75\x64\x5D\x5B\x63\x6D\x70\x27\x0A\x67\x7C\x68\x76\x2E\x66\x62\x2C\x3A\x3D\x2D\x71\x31\x30\x43\x32\x2A\x79\x78\x29\x28\x4C\x39\x41\x53\x2F\x50\x22\x45\x6A\x4D\x49\x6B\x33\x3E\x35\x54\x3C\x44\x34\x7D\x42\x7B\x38\x46\x77\x52\x36\x37\x55\x47\x4E\x3B\x4A\x7A\x56\x23\x48\x4F\x57\x5F\x26\x21\x4B\x3F\x58\x51\x25\x59\x5C\x09\x5A\x2B\x7E\x5E\x24\x40\x60\x7F\x00\x01\x02\x03\x04\x05\x06\x07\x08\x0B\x0C\x0D\x0E\x0F\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1A\x1B\x1C\x1D\x1E\x1F");

// Contains all the tail bytes that can appear in the composition of a UTF-8 char
define('UTF8_TRAILING_BYTES', "\xA9\xA0\xA8\x80\xAA\x99\xA7\xBB\xAB\x89\x94\x82\xB4\xA2\xAE\x83\xB0\xB9\xB8\x93\xAF\xBC\xB3\x81\xA4\xB2\x9C\xA1\xB5\xBE\xBD\xBA\x98\xAD\xB1\x84\x95\xA6\xB6\x88\x8D\x90\xB7\xBF\x92\x85\xA5\x97\x8C\x86\xA3\x8E\x9F\x8F\x87\x91\x9D\xAC\x9E\x8B\x96\x9B\x8A\x9A");

// Constants used by the Hangul [de]composition algorithms
define('UNICODE_HANGUL_SBASE', 0xAC00);
define('UNICODE_HANGUL_LBASE', 0x1100);
define('UNICODE_HANGUL_VBASE', 0x1161);
define('UNICODE_HANGUL_TBASE', 0x11A7);
define('UNICODE_HANGUL_SCOUNT', 11172);
define('UNICODE_HANGUL_LCOUNT', 19);
define('UNICODE_HANGUL_VCOUNT', 21);
define('UNICODE_HANGUL_TCOUNT', 28);
define('UNICODE_HANGUL_NCOUNT', 588);
define('UNICODE_JAMO_L', 0);
define('UNICODE_JAMO_V', 1);
define('UNICODE_JAMO_T', 2);

/**
* Unicode normalization routines
*
* @package phpBB3
*/
class utf_normalizer
{
	/**
	* Validate, cleanup and normalize a string
	*
	* The ultimate convenience function! Clean up invalid UTF-8 sequences,
	* and convert to Normal Form C, canonical composition.
	*
	* @param	string	&$str	The dirty string
	* @return	string			The same string, all shiny and cleaned-up
	*/
	function cleanup(&$str)
	{
		// The string below is the list of all autorized characters, sorted by frequency in latin text
		$pos = strspn($str, "\x20\x65\x69\x61\x73\x6E\x74\x72\x6F\x6C\x75\x64\x5D\x5B\x63\x6D\x70\x27\x0A\x67\x7C\x68\x76\x2E\x66\x62\x2C\x3A\x3D\x2D\x71\x31\x30\x43\x32\x2A\x79\x78\x29\x28\x4C\x39\x41\x53\x2F\x50\x22\x45\x6A\x4D\x49\x6B\x33\x3E\x35\x54\x3C\x44\x34\x7D\x42\x7B\x38\x46\x77\x52\x36\x37\x55\x47\x4E\x3B\x4A\x7A\x56\x23\x48\x4F\x57\x5F\x26\x21\x4B\x3F\x58\x51\x25\x59\x5C\x09\x5A\x2B\x7E\x5E\x24\x40\x60\x7F\x0D");
		$len = strlen($str);

		if ($pos == $len)
		{
			// ASCII strings with no special chars return immediately
			return;
		}

		// Note: we do not check for $GLOBALS['utf_canonical_decomp']. It is assumed they are always loaded together
		if (!isset($GLOBALS['utf_nfc_qc']))
		{
			global $phpbb_root_path, $phpEx;
			include($phpbb_root_path . 'includes/utf/data/utf_nfc_qc.' . $phpEx);
		}

		// Replace any byte in the range 0x00..0x1F, except for \r, \n and \t
		// We replace those characters with a 0xFF byte, which is illegal in UTF-8 and will in turn be replaced with a UTF replacement char
		$str = strtr(
			$str,
			"\x00\x01\x02\x03\x04\x05\x06\x07\x08\x0B\x0C\x0E\x0F\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1A\x1B\x1C\x1D\x1E\x1F",
			"\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF"
		);

		$str = utf_normalizer::recompose($str, $pos, $len, $GLOBALS['utf_nfc_qc'], $GLOBALS['utf_canonical_decomp']);
	}

	/**
	* Validate and normalize a UTF string to NFC
	*
	* @param	string	&$str	Unchecked UTF string
	* @return	string			The string, validated and in normal form
	*/
	function nfc(&$str)
	{
		$pos = strspn($str, UTF8_ASCII_RANGE);
		$len = strlen($str);

		if ($pos == $len)
		{
			// ASCII strings return immediately
			return;
		}

		if (!isset($GLOBALS['utf_nfc_qc']))
		{
			global $phpbb_root_path, $phpEx;
			include($phpbb_root_path . 'includes/utf/data/utf_nfc_qc.' . $phpEx);
		}

		$str = utf_normalizer::recompose($str, $pos, $len, $GLOBALS['utf_nfc_qc'], $GLOBALS['utf_canonical_decomp']);
	}

	/**
	* Validate and normalize a UTF string to NFKC
	*
	* @param	string	&$str	Unchecked UTF string
	* @return	string			The string, validated and in normal form
	*/
	function nfkc(&$str)
	{
		$pos = strspn($str, UTF8_ASCII_RANGE);
		$len = strlen($str);

		if ($pos == $len)
		{
			// ASCII strings return immediately
			return;
		}

		if (!isset($GLOBALS['utf_nfkc_qc']))
		{
			global $phpbb_root_path, $phpEx;
			include($phpbb_root_path . 'includes/utf/data/utf_nfkc_qc.' . $phpEx);
		}

		if (!isset($GLOBALS['utf_canonical_comp']))
		{
			global $phpbb_root_path, $phpEx;
			include($phpbb_root_path . 'includes/utf/data/utf_canonical_comp.' . $phpEx);
		}

		$str = utf_normalizer::recompose($str, $pos, $len, $GLOBALS['utf_nfkc_qc'], $GLOBALS['utf_compatibility_decomp']);
	}

	/**
	* Validate and normalize a UTF string to NFD
	*
	* @param	string	&$str	Unchecked UTF string
	* @return	string			The string, validated and in normal form
	*/
	function nfd(&$str)
	{
		$pos = strspn($str, UTF8_ASCII_RANGE);
		$len = strlen($str);

		if ($pos == $len)
		{
			// ASCII strings return immediately
			return;
		}

		if (!isset($GLOBALS['utf_canonical_decomp']))
		{
			global $phpbb_root_path, $phpEx;
			include($phpbb_root_path . 'includes/utf/data/utf_canonical_decomp.' . $phpEx);
		}

		$str = utf_normalizer::decompose($str, $pos, $len, $GLOBALS['utf_canonical_decomp']);
	}

	/**
	* Validate and normalize a UTF string to NFKD
	*
	* @param	string	&$str	Unchecked UTF string
	* @return	string			The string, validated and in normal form
	*/
	function nfkd(&$str)
	{
		$pos = strspn($str, UTF8_ASCII_RANGE);
		$len = strlen($str);

		if ($pos == $len)
		{
			// ASCII strings return immediately
			return;
		}

		if (!isset($GLOBALS['utf_compatibility_decomp']))
		{
			global $phpbb_root_path, $phpEx;
			include($phpbb_root_path . 'includes/utf/data/utf_compatibility_decomp.' . $phpEx);
		}

		$str = utf_normalizer::decompose($str, $pos, $len, $GLOBALS['utf_compatibility_decomp']);
	}


	/**
	* Recompose a UTF string
	*
	* @param	string	$str			Unchecked UTF string
	* @param	integer	$pos			Position of the first UTF char (in bytes)
	* @param	integer	$len			Length of the string (in bytes)
	* @param	array	&$qc			Quick-check array, passed by reference but never modified
	* @param	array	&$decomp_map	Decomposition mapping, passed by reference but never modified
	* @return	string					The string, validated and recomposed
	*
	* @access	private
	*/
	function recompose($str, $pos, $len, &$qc, &$decomp_map)
	{
		global $utf_combining_class, $utf_canonical_comp, $utf_jamo_type, $utf_jamo_index;

		// Load some commonly-used tables
		if (!isset($utf_jamo_index, $utf_jamo_type, $utf_combining_class))
		{
			global $phpbb_root_path;
			include($phpbb_root_path . 'includes/utf/data/utf_normalizer_common.php');
		}

		// Buffer the last ASCII char before the UTF-8 stuff if applicable
		$tmp = '';
		$i = $tmp_pos = $last_cc = 0;

		$buffer = ($pos) ? array(++$i => $str[$pos - 1]) : array();

		// UTF char length array
		// This array is used to determine the length of a UTF character.
		// Be $c the result of ($str[$pos] & "\xF0") --where $str is the string we're operating on and $pos
		// the position of the cursor--, if $utf_len_mask[$c] does not exist, the byte is an ASCII char.
		// Otherwise, if $utf_len_mask[$c] is greater than 0, we have a the leading byte of a multibyte character
		// whose length is $utf_len_mask[$c] and if it is equal to 0, the byte is a trailing byte.
		$utf_len_mask = array(
			// Leading bytes masks
			"\xC0" => 2, "\xD0" => 2, "\xE0" => 3, "\xF0" => 4,
			// Trailing bytes masks
			"\x80" => 0, "\x90" => 0, "\xA0" => 0, "\xB0" => 0
		);

		$extra_check = array(
			"\xED" => 1, "\xEF" => 1, "\xC0" => 1, "\xC1" => 1, "\xE0" => 1, "\xF0" => 1,
			"\xF4" => 1, "\xF5" => 1, "\xF6" => 1, "\xF7" => 1, "\xF8" => 1, "\xF9" => 1,
			"\xFA" => 1, "\xFB" => 1, "\xFC" => 1, "\xFD" => 1, "\xFE" => 1, "\xFF" => 1
		);

		$utf_validation_mask = array(
			2	=> "\xE0\xC0",
			3	=> "\xF0\xC0\xC0",
			4	=> "\xF8\xC0\xC0\xC0"
		);

		$utf_validation_check = array(
			2	=> "\xC0\x80",
			3	=> "\xE0\x80\x80",
			4	=> "\xF0\x80\x80\x80"
		);

		// Main loop
		do
		{
			// STEP 0: Capture the current char and buffer it
			$c = $str[$pos];
			$c_mask = $c & "\xF0";

			if (isset($utf_len_mask[$c_mask]))
			{
				// Byte at $pos is either a leading byte or a missplaced trailing byte
				if ($utf_len = $utf_len_mask[$c_mask])
				{
					// Capture the char
					$buffer[++$i & 7] = $utf_char = substr($str, $pos, $utf_len);

					// Let's find out if a thorough check is needed
					if (isset($qc[$utf_char]))
					{
						// If the UTF char is in the qc array then it may not be in normal form. We do nothing here, the actual processing is below this "if" block
					}
					else if (isset($utf_combining_class[$utf_char]))
					{
						if ($utf_combining_class[$utf_char] < $last_cc)
						{
							// A combining character that is NOT canonically ordered
						}
						else
						{
							// A combining character that IS canonically ordered, skip to the next char
							$last_cc = $utf_combining_class[$utf_char];

							$pos += $utf_len;
							continue;
						}
					}
					else
					{
						// At this point, $utf_char holds a UTF char that we know is not a NF[K]C_QC and is not a combining character.
						// It can be a singleton, a canonical composite, a replacement char or an even an ill-formed bunch of bytes. Let's find out
						$last_cc = 0;

						// Check that we have the correct number of trailing bytes
						if (($utf_char & $utf_validation_mask[$utf_len]) != $utf_validation_check[$utf_len])
						{
							// Current char isn't well-formed or legal: either one or several trailing bytes are missing, or the Unicode char
							// has been encoded in a five- or six- byte sequence
							if ($utf_char[0] >= "\xF8")
							{
								if ($utf_char[0] < "\xFC")
								{
									$trailing_bytes = 4;
								}
								else if ($utf_char[0] > "\xFD")
								{
									$trailing_bytes = 0;
								}
								else
								{
									$trailing_bytes = 5;
								}
							}
							else
							{
								$trailing_bytes = $utf_len - 1;
							}

							$tmp .= substr($str, $tmp_pos, $pos - $tmp_pos) . UTF8_REPLACEMENT;
							$pos += strspn($str, UTF8_TRAILING_BYTES, ++$pos, $trailing_bytes);
							$tmp_pos = $pos;

							continue;
						}

						if (isset($extra_check[$c]))
						{
							switch ($c)
							{
								// Note: 0xED is quite common in Korean
								case "\xED":
									if ($utf_char >= "\xED\xA0\x80")
									{
										// Surrogates (U+D800..U+DFFF) are not allowed in UTF-8 (UTF sequence 0xEDA080..0xEDBFBF)
										$tmp .= substr($str, $tmp_pos, $pos - $tmp_pos) . UTF8_REPLACEMENT;
										$pos += $utf_len;
										$tmp_pos = $pos;
										continue 2;
									}
								break;

								// Note: 0xEF is quite common in Japanese
								case "\xEF":
									if ($utf_char == "\xEF\xBF\xBE" || $utf_char == "\xEF\xBF\xBF")
									{
										// U+FFFE and U+FFFF are explicitly disallowed (UTF sequence 0xEFBFBE..0xEFBFBF)
										$tmp .= substr($str, $tmp_pos, $pos - $tmp_pos) . UTF8_REPLACEMENT;
										$pos += $utf_len;
										$tmp_pos = $pos;
										continue 2;
									}
								break;

								case "\xC0":
								case "\xC1":
									if ($utf_char <= "\xC1\xBF")
									{
										// Overlong sequence: Unicode char U+0000..U+007F encoded as a double-byte UTF char
										$tmp .= substr($str, $tmp_pos, $pos - $tmp_pos) . UTF8_REPLACEMENT;
										$pos += $utf_len;
										$tmp_pos = $pos;
										continue 2;
									}
								break;

								case "\xE0":
									if ($utf_char <= "\xE0\x9F\xBF")
									{
										// Unicode char U+0000..U+07FF encoded in 3 bytes
										$tmp .= substr($str, $tmp_pos, $pos - $tmp_pos) . UTF8_REPLACEMENT;
										$pos += $utf_len;
										$tmp_pos = $pos;
										continue 2;
									}
								break;

								case "\xF0":
									if ($utf_char <= "\xF0\x8F\xBF\xBF")
									{
										// Unicode char U+0000..U+FFFF encoded in 4 bytes
										$tmp .= substr($str, $tmp_pos, $pos - $tmp_pos) . UTF8_REPLACEMENT;
										$pos += $utf_len;
										$tmp_pos = $pos;
										continue 2;
									}
								break;

								default:
									// Five- and six- byte sequences do not need being checked for here anymore
									if ($utf_char > UTF8_MAX)
									{
										// Out of the Unicode range
										if ($utf_char[0] < "\xF8")
										{
											$trailing_bytes = 3;
										}
										else if ($utf_char[0] < "\xFC")
										{
											$trailing_bytes = 4;
										}
										else if ($utf_char[0] > "\xFD")
										{
											$trailing_bytes = 0;
										}
										else
										{
											$trailing_bytes = 5;
										}

										$tmp .= substr($str, $tmp_pos, $pos - $tmp_pos) . UTF8_REPLACEMENT;
										$pos += strspn($str, UTF8_TRAILING_BYTES, ++$pos, $trailing_bytes);
										$tmp_pos = $pos;
										continue 2;
									}
								break;
							}
						}

						// The char is a valid starter, move the cursor and go on
						$pos += $utf_len;
						continue;
					}
				}
				else
				{
					// A trailing byte came out of nowhere, we will advance the cursor and treat the this byte and all following trailing bytes as if
					// each of them was a Unicode replacement char
					$spn = strspn($str, UTF8_TRAILING_BYTES, $pos);
					$tmp .= substr($str, $tmp_pos, $pos - $tmp_pos) . str_repeat(UTF8_REPLACEMENT, $spn);

					$pos += $spn;
					$tmp_pos = $pos;
					continue;
				}


				// STEP 1: Decompose current char

				// We have found a character that is either:
				//  - in the NFC_QC/NFKC_QC list
				//  - a non-starter char that is not canonically ordered
				//
				// We are going to capture the shortest UTF sequence that satisfies these two conditions:
				//
				//  1 - If the sequence does not start at the begginning of the string, it must begin with a starter,
				// and that starter must not have the NF[K]C_QC property equal to "MAYBE"
				//
				//  2 - If the sequence does not end at the end of the string, it must end with a non-starter and be
				// immediately followed by a starter that is not on the QC list
				//
				$utf_seq = array();
				$last_cc = 0;
				$lpos = $pos;
				$pos += $utf_len;

				if (isset($decomp_map[$utf_char]))
				{
					$_pos = 0;
					$_len = strlen($decomp_map[$utf_char]);

					do
					{
						$_utf_len =& $utf_len_mask[$decomp_map[$utf_char][$_pos] & "\xF0"];

						if (isset($_utf_len))
						{
							$utf_seq[] = substr($decomp_map[$utf_char], $_pos, $_utf_len);
							$_pos += $_utf_len;
						}
						else
						{
							$utf_seq[] = $decomp_map[$utf_char][$_pos];
							++$_pos;
						}
					}
					while ($_pos < $_len);