کد:
<?php
// windows-1256 to utf8 convertor 0.1
error_reporting(E_ALL);
function convert_w1256_to_utf_8($text) {
$converted_text=preg_replace_callback("/(:\&\#([0-9]{1,4})\;|([\x80-\xFF]))/","replace_content",$text);
return $converted_text;
}
function replace_content($matches_content) {
//windows-1256 character set
$windows1256 = array(
'€', # 0x20AC EURO SIGN
'پ', # 0x067E ARABIC LETTER PEH
'‚', # 0x201A SINGLE LOW-9 QUOTATION MARK
'ƒ', # 0x0192 LATIN SMALL LETTER F WITH HOOK
'„', # 0x201E DOUBLE LOW-9 QUOTATION MARK
'…', # 0x2026 HORIZONTAL ELLIPSIS
'', # 0x2020 DAGGER
'‡', # 0x2021 DOUBLE DAGGER
'ˆ', # 0x02C6 MODIFIER LETTER CIRCUMFLEX ACCENT
'‰', # 0x2030 PER MILLE SIGN
'ٹ', # 0x0679 ARABIC LETTER TTEH
'‹', # 0x2039 SINGLE LEFT-POINTING ANGLE QUOTATION MARK
'Œ', # 0x0152 LATIN CAPITAL LIGATURE OE
'چ', # 0x0686 ARABIC LETTER TCHEH
'ژ', # 0x0698 ARABIC LETTER JEH
'ڈ', # 0x0688 ARABIC LETTER DDAL
'گ', # 0x06AF ARABIC LETTER GAF
'‘', # 0x2018 LEFT SINGLE QUOTATION MARK
'’', # 0x2019 RIGHT SINGLE QUOTATION MARK
'“', # 0x201C LEFT DOUBLE QUOTATION MARK
'”', # 0x201D RIGHT DOUBLE QUOTATION MARK
'•', # 0x2022 BULLET
'–', # 0x2013 EN DASH
'—', # 0x2014 EM DASH
'ک', # 0x06A9 ARABIC LETTER KEHEH
'™', # 0x2122 TRADE MARK SIGN
'ڑ', # 0x0691 ARABIC LETTER RREH
'›', # 0x203A SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
'œ', # 0x0153 LATIN SMALL LIGATURE OE
'', # 0x200C ZERO WIDTH NON-JOINER
'', # 0x200D ZERO WIDTH JOINER
'ں', # 0x06BA ARABIC LETTER NOON GHUNNA
' ', # 0x00A0 NO-BREAK SPACE
'،', # 0x060C ARABIC COMMA
'¢', # 0x00A2 CENT SIGN
'£', # 0x00A3 POUND SIGN
'¤', # 0x00A4 CURRENCY SIGN
'¥', # 0x00A5 YEN SIGN
'¦', # 0x00A6 BROKEN BAR
'§', # 0x00A7 SECTION SIGN
'¨', # 0x00A8 DIAERESIS
'©', # 0x00A9 COPYRIGHT SIGN
'ھ', # 0x06BE ARABIC LETTER HEH DOACHASHMEE
'«', # 0x00AB LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
'¬', # 0x00AC NOT SIGN
'', # 0x00AD SOFT HYPHEN
'®', # 0x00AE REGISTERED SIGN
'¯', # 0x00AF MACRON
'°', # 0x00B0 DEGREE SIGN
'±', # 0x00B1 PLUS-MINUS SIGN
'²', # 0x00B2 SUPERSCRIPT TWO
'³', # 0x00B3 SUPERSCRIPT THREE
'´', # 0x00B4 ACUTE ACCENT
'µ', # 0x00B5 MICRO SIGN
'¶', # 0x00B6 PILCROW SIGN
'·', # 0x00B7 MIDDLE DOT
'¸', # 0x00B8 CEDILLA
'¹', # 0x00B9 SUPERSCRIPT ONE
'؛', # 0x061B ARABIC SEMICOLON
'»', # 0x00BB RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
'¼', # 0x00BC VULGAR FRACTION ONE QUARTER
'½', # 0x00BD VULGAR FRACTION ONE HALF
'¾', # 0x00BE VULGAR FRACTION THREE QUARTERS
'؟', # 0x061F ARABIC QUESTION MARK
'ہ', # 0x06C1 ARABIC LETTER HEH GOAL
'ء', # 0x0621 ARABIC LETTER HAMZA
'آ', # 0x0622 ARABIC LETTER ALEF WITH MADDA ABOVE
'أ', # 0x0623 ARABIC LETTER ALEF WITH HAMZA ABOVE
'ؤ', # 0x0624 ARABIC LETTER WAW WITH HAMZA ABOVE
'إ', # 0x0625 ARABIC LETTER ALEF WITH HAMZA BELOW
'ئ', # 0x0626 ARABIC LETTER YEH WITH HAMZA ABOVE
'ا', # 0x0627 ARABIC LETTER ALEF
'ب', # 0x0628 ARABIC LETTER BEH
'ة', # 0x0629 ARABIC LETTER TEH MARBUTA
'ت', # 0x062A ARABIC LETTER TEH
'ث', # 0x062B ARABIC LETTER THEH
'ج', # 0x062C ARABIC LETTER JEEM
'ح', # 0x062D ARABIC LETTER HAH
'خ', # 0x062E ARABIC LETTER KHAH
'د', # 0x062F ARABIC LETTER DAL
'ذ', # 0x0630 ARABIC LETTER THAL
'ر', # 0x0631 ARABIC LETTER REH
'ز', # 0x0632 ARABIC LETTER ZAIN
'س', # 0x0633 ARABIC LETTER SEEN
'ش', # 0x0634 ARABIC LETTER SHEEN
'ص', # 0x0635 ARABIC LETTER SAD
'ض', # 0x0636 ARABIC LETTER DAD
'×', # 0x00D7 MULTIPLICATION SIGN
'ط', # 0x0637 ARABIC LETTER TAH
'ظ', # 0x0638 ARABIC LETTER ZAH
'ع', # 0x0639 ARABIC LETTER AIN
'غ', # 0x063A ARABIC LETTER GHAIN
'ـ', # 0x0640 ARABIC TATWEEL
'ف', # 0x0641 ARABIC LETTER FEH
'ق', # 0x0642 ARABIC LETTER QAF
'ك', # 0x0643 ARABIC LETTER KAF
'', # 0x00E0 LATIN SMALL LETTER A WITH GRAVE
'ل', # 0x0644 ARABIC LETTER LAM
'â', # 0x00E2 LATIN SMALL LETTER A WITH CIRCUMFLEX
'م', # 0x0645 ARABIC LETTER MEEM
'ن', # 0x0646 ARABIC LETTER NOON
'ه', # 0x0647 ARABIC LETTER HEH
'و', # 0x0648 ARABIC LETTER WAW
'ç', # 0x00E7 LATIN SMALL LETTER C WITH CEDILLA
'è', # 0x00E8 LATIN SMALL LETTER E WITH GRAVE
'é', # 0x00E9 LATIN SMALL LETTER E WITH ACUTE
'ê', # 0x00EA LATIN SMALL LETTER E WITH CIRCUMFLEX
'ë', # 0x00EB LATIN SMALL LETTER E WITH DIAERESIS
'ى', # 0x0649 ARABIC LETTER ALEF MAKSURA
'ي', # 0x064A ARABIC LETTER YEH
'î', # 0x00EE LATIN SMALL LETTER I WITH CIRCUMFLEX
'ï', # 0x00EF LATIN SMALL LETTER I WITH DIAERESIS
'ً', # 0x064B ARABIC FATHATAN
'ٌ', # 0x064C ARABIC DAMMATAN
'ٍ', # 0x064D ARABIC KASRATAN
'َ', # 0x064E ARABIC FATHA
'ô', # 0x00F4 LATIN SMALL LETTER O WITH CIRCUMFLEX
'ُ', # 0x064F ARABIC DAMMA
'ِ', # 0x0650 ARABIC KASRA
'÷', # 0x00F7 DIVISION SIGN
'ّ', # 0x0651 ARABIC SHADDA
'ù', # 0x00F9 LATIN SMALL LETTER U WITH GRAVE
'ْ', # 0x0652 ARABIC SUKUN
'û', # 0x00FB LATIN SMALL LETTER U WITH CIRCUMFLEX
'ü', # 0x00FC LATIN SMALL LETTER U WITH DIAERESIS
'', # 0x200E LEFT-TO-RIGHT MARK
'', # 0x200F RIGHT-TO-LEFT MARK
'ے', # 0x06D2 ARABIC LETTER YEH BARREE
); #windows1256
return $windows1256[ord($matches_content[1])-0x80];
}
function get_web_content($web_site_address) {
$site_content=''; // to make notice happy
$file_handle=fopen($web_site_address,"r");
while ($line=fgets($file_handle,1000))
{
$site_content.=$line;
}
fclose ($file_handle);
return $site_content;
}
$web_site_address=$_REQUEST['q'];
if (!$web_site_address)
{
// example site
$web_site_address='[مهمان/کاربر گرامی برای دیدن لینک ها ابتدا باید عضو سایت شوید و لاگین کنید برای ثبت نام اینجا کلیک کنید]
}
$site_content=get_web_content($web_site_address);
$converted_content=convert_w1256_to_utf_8($site_content);
echo $converted_content;
?>
پاسخ با نقل قول
علاقه مندی ها (Bookmarks)