PHP : Function Reference : Regular Expression Functions (POSIX Extended) : eregi PHP Examples Tutorials References

There seems to be a lot of confusion about escaping special characters within a bracket expression enclosed in [] The reference is the regex man page at http://www.tin.org/bin/man.cgi?section=7&topic=regex , Linked from the "Regular Expression Functions (POSIX Extended)" page in the PHP manual at http://php.planetmirror.com/manual/en/ref.regex.php The simple rule is there should be no backslash `\' in a bracket expression unless you are matching it literally. ALL special characters (the few exceptions are listed below), lose their special significance in a bracket expression. This includes .[$()|*+?{\ all of which have special meaning outside of a bracket expression, inside they will match LITERALLY! The backslash character can never escape anything in a bracket expression including itself! It is always interpreted literally. Here are the exceptions ^ at the start of the bracket expression will match on all but the set of characters within the bracket expression - (hyphen) Indicates range, so [a-z] will match a single character in the range a-z (case insensitive for eregi). To match a literal `-' it must be at the start or end of the bracket expression or the end of a range, or enclosed in [. .] to be the start of a range. Examples: [0-9-] [-0-9] Both the above will match a single character from the set comprising the digits and hyphen `-' [!--] is the same as [!"#$%&'()*+,-] i.e. it matches a single character from the range of characters from `!' to hyphen `-' [[.-.]-9] is the same as [-./0123456789] or [./0123456789-] and will match a single digit from the range of characters from hyphen `-' to digit 9 ] The closing bracket indicates the end of a bracket expression but not if it immediately follows the opening bracket `[' or `[^' in that case the closing bracket is matched literally. Examples: [][0-9] is the same as []0-9[] and matches a single character from the set of digits and a literal closing bracket `]' and opening bracket `]' [^][0-9] is the same as [^]0-9[] and matches a single character from the group of all characters except the set of digits and literal `]' or `[' Look at the regex man page for more on the above and the following special cases [.characters.] A collating element [=characters=] An equivalence class [:alnum:] A character class That's it for special cases! ALL other characters match literally within a bracket expression and can never be escaped with `\' Here is a list of some of the postings here that have mistakenly tried to use special characters within a bracket expression. 28-Jan-2002 12:56, 09-Feb-2002 06:28, 24-Sep-2002 01:21, 08-Mar-2003 02:21, 11-Apr-2003 03:22 (uses `|' as if it was `or' it is not, it is a literal `|'), 27-Jun-2003 04:03, 06-Sep-2003 07:17, 20-Apr-2004 04:55, 24-Nov-2004 09:12, 25-Nov-2004 02:11, 06-Jan-2005 08:54, 24-Feb-2005 06:05 It seems people are copying each others mistakes so this needs to be cleared up. Most of the mistakes attempt to escape special characters within a bracket expression with backslash. That never works! A simple fix is to simply remove any backslashes within the bracket expression that are not meant to be matched literally. The attempt to use | as 'or' within a bracketed expression should be put in parentheses instead. I used a simple script attached to a form to test these results. Here's the script. <?php if(isset($_POST['pattern']) && isset($_POST['teststring'])) { $pattern = stripslashes($_POST['pattern']); $teststring = stripslashes($_POST['teststring']); $match = ereg($pattern, $teststring, $regs); if(!$match) $match = 0; echo(' The regular expression is: '.$pattern.''); echo(' The string for testing is: '.$teststring.''); echo(' The number of matching characters is: '.$match.''); } ?> Just use a form with input fields named "pattern" and "teststring" Hope this helps and clears up some misconceptions Jason Smart Monash University Student

henrik jensen

Looking for at regex to check if a file is an image file, this seems to work? Note: it does not check for illegeal filesystem names - it only looks at the filename extension. if (eregi ("(.)+\\.(jp(e){0,1}g$|gif$|png$)",$filename)){ // This is an imagefile } [ remove this_ from emailaddress ]

tomi

Keran's solution is a good one but I found it does not work for a small number of domains -- because of a bug in gethostbyname() that causes some valid domains not to resolve properly. The workaround is to test such domains with "www." added to the beginning. This script does all that and works perfectly for me. Enjoy! // Function to validate emails. function validate_email($email_raw) { // replace any ' ' and \n in the email $email_nr = eregi_replace("\n", "", $email_raw); $email = eregi_replace(" +", "", $email_nr); $email = strtolower( $email ); // do the eregi to look for bad characters if( !eregi("^[a-z0-9]+([_\\.-][a-z0-9]+)*". "@([a-z0-9]+([\.-][a-z0-9]+))*$",$email) ){ // okay not a good email $feedback = 'Error: "' . $email . '" is not a valid e-mail address!'; return $feedback; } else { // okay now check the domain // split the email at the @ and check what's left $item = explode("@", $email); $domain = $item["1"]; if ( ( gethostbyname($domain) == $domain ) ) { if ( gethostbyname("www." . $domain) == "www." . $domain ) { $feedback = 'Error: "' . $domain . '" is most probably not a valid domain!'; return $feedback; } // ? $feedback = "valid"; return $feedback; } else { $feedback = "valid"; return $feedback; } } }

lueck

Just a simple ip address validation function. The function does only check that the ip address is in a correct format (IPv4): function checkipaddress($checkip) { if (eregi("^[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}$", $checkip)) { for ($i = 1; $i <= 3; $i++) { if (!(substr($checkip, 0, strpos($checkip, ".")) >= "0" && substr($checkip, 0, strpos($checkip, ".")) <= "255")) { echo "Der ".$i.". Block der IP-Adresse hat kein gültiges Format!"; return false; } $checkip = substr($checkip, strpos($checkip, ".") + 1); } if (!($checkip >= "0" && $checkip <= "255")) { echo "Der 4. Block der IP-Adresse hat kein gültiges Format!"; return false; } } else { echo "Die IP-Adresse hat kein gültiges Format!"; return false; } return true; }

benjohnson{-at-}maine{-dot-}rr{-dot-}com

It's probably worth noting that eregi() (and most likely, the related variations) appears to have a 255-character limit with respect to the length of the input it will attempt to parse. If you try to do something like if (!eregi("^[a-zA-Z0-9]{0,256}$", $text)) { ... eregi() will return FALSE, irrespective of whether or not the input matches the pattern.

banerian

It should be noted that in the function validateEmail, the final verification relies on the mail server responding to the test $email with a reply code of 250. Nowadays, many/most/all servers will reply 250/Ok for any user@thier.domain regardless of whether or not the userid actually exists. mail servers tend to accept the mail (if it passes other checks), and if the user really does not exist, it just "bounces" the mail back to sender. thus the validateEmail function can give false positives...an unfortunate victim of having to deal with spam.

bkdotcom

in regards to using gethostbyname() to validate email addresses: Large flaw: Doesn't work if the host has a NS & MX record, but no A record.

joseph dot ragsdale

I'm posting this as a - hopefully - helpful example to those who are learning regular expressions. Apologies to the documentation team and moderators for posting yet another email address validation example, but it's hard to argue against it as most everyone will see regular expressions as being a solution to this problem. <?php // Example results: $email[] = 'foo@example.com'; // matched $email[] = 'foo.bar@example.co.uk'; // matched $email[] = 'foo_bar@example.com'; // matched $email[] = '_foo_bar@example.com'; // matched $email[] = 'foo@example.example'; // matched $email[] = '_-._@-.--'; // matched $email[] = '@.'; // not matched $email[] = '@.com'; // not matched $email[] = ' @ .com'; // not matched $email[] = '.bar@example.com'; // not matched $email[] = 'foo.@example.com'; // not matched $email[] = 'foo@example.x.y'; // not matched $regex = '^'. '[_a-z0-9-]+'. /* One or more underscore, alphanumeric, or hyphen charactures. */ '(\.[_a-z0-9-]+)*'. /* Followed by zero or more sets consisting of a period and one or more underscore, alphanumeric, or hyphen charactures. */ '@'. /* Followed by an "at" characture. */ '[a-z0-9-]+'. /* Followed by one or more alphanumeric or hyphen charactures. */ '(\.[a-z0-9-]{2,})+'. /* Followed by one or more sets consisting of a period and two or more alphanumeric or hyphen charactures. */ '$'; foreach ($email as $example) { if (eregi($regex, $example)) { echo $example . ' matched '; } else { echo $example . ' not matched '; } } ?>

mmc

I'm not sure about "\." being the same as "." above. Anyway, the mentioned regex would not recognize .museum-names, and generally isn't future safe. Also, it doesn't verify that usernames and hostnames cannot start with "-._". I would recommend a more general (=future safe) expression and then instead check the hostname. An example: if( !eregi("^[a-z0-9]+([_\\.-][a-z0-9]+)*" ."@([a-z0-9]+([\.-][a-z0-9]+))*$", $mail, $regs) ) { echo "Error: '$mail' isn't a valid mail address!\n"; } elseif( gethostbyname($regs[2]) == $regs[2] ) { echo "Error: Can't find the host '$regs[2]'! \n"; } Note: I had to split the regex for it to fit this note. Also note: The reason I'm using gethostbyname() and not getmxrr() or such is that getmxrr() doesn't work on Win2000/XP.

opedroso

I you are trying to match accented characters (e.g. you want to match "Café Olé"), you can use the following regex: "[a-zà-ü ]+" This takes advantage of the fact that in the iso-8859-1 "Latin 1", the "à" is the first accented char and the "ü", the last one. By the way, the regex above will match this string as a single match: "aeiou àèìòù áéíóú âêîôû ãõ ñ äëïöü ç" Beware that your local code page may include a few more char codes in the range "[à-ü]", so it may match more than you expect.

krum

I was searching for a fast way to check all fields of a form, so I can include the $_POST information directly to the sql query. So I wrote another script in my class in order to check the submitted data. So here it is: <? function check_post () { foreach ($_POST as $key => $val) { $$key = $val; } $data = Array(); $data["^[a-z0-9!@#$%^&*]"] = Array ("first_name", "last_name", "nick", "pass", "description", "user", "message", "event", "name", "title", "text"); $data['^[-!#$%&\'*+\\./0-9=?A-Z^_`a-z{|}~]+' .'@'.'[-!#$%&\'*+\\/0-9=?A-Z^_`a-z{|}~]+\.' .'[-!#$%&\'*+\\./0-9=?A-Z^_`a-z{|}~]+$'] = Array("mail"); $data["^[0-9]{1,4}"] = Array("day", "month", "owner", "type", "forum_id", "topic_id"); foreach ($data as $key => $val) { foreach ($val as $k => $c) { if ( isset($$c) ) { if ( !eregi($key, $$c) ) { echo "You have used an not allowed character in " .$$c. " field\n"; exit; } } } } } ?> I hope it will help someone ^_^

ian

I was looking for something to validate a URL, and as usual turned to 'php.net' as it provides an invaluable source of information. I couldn't find exactly what I wanted so came up with something of my own which I hope will be of use to others. <?php function validateURL($URL) { $domain = "([[:alpha:]][-[:alnum:]]*[[:alnum:]]) (\.[[:alpha:]][-[:alnum:]]*[[:alpha:]])+"; $dir = "(/[[:alpha:]][-[:alnum:]]*[[:alnum:]])*"; $page = "(/[[:alpha:]][-[:alnum:]]*\.[[:alpha:]]{3,5})?"; $getstring = "(\?([[:alnum:]][-_%[:alnum:]]*=[-_%[:alnum:]]+) (&([[:alnum:]][-_%[:alnum:]]*=[-_%[:alnum:]]+))*)?"; $pattern = "^".$domain.$dir.$page.$getstring."$"; return eregi($pattern, $URL); } ?> Be sure to enter each pattern on a single line. The function accepts a URL and validates the domain, directory path, page and any GET parameters. You can see I have split the patterns for validating each section for easier reading and maintaining. Domain Name: This can be made of of many parts each seperated by a '.' (dot). each part must start with an alpha char. followed by any alphanumeric chars., or '-'. With the exception of the first part each part must end in an alpha char.. e.g. www.google.co.uk and www9.g-88gle.co.uk are valid, but 9ww.google.3o.uk and www.google4.co.uk3 are not. Directory Path: Each directory must start with a '/' followed by an alpha char., then any alpha numeric char., or '-', and ending in an alphanumeric char. e.g. /path1/to-page is valid, but /path1-/2-page is not. Page: This is like the directory path but it must end with a '.' (dot) followed by 3-5 alpha chars. e.g. /phpinfo.php and /page.shtml are valid, but /phpinfo.php4 and /pageshtml are not. Get String: This follows the page and begins with a '?'. This is then followed by key/value pairs each seperated by an '&'. e.g. ?key1=value1&key2=value2 Each key, or value must start and end with an alpahnumeric char. I have allowed for a '%' to be specified for URL encoded chars., and also '_' (underscore) is allowed. Well thats it hope it is of some use. It is NOT perfect, for instance in the 'getstring' I would like to ensure that encoded characters (i.e. %20) consist of a % symbol followed by two characters with values '0-9A-F' but couldn't get anything I tried to work. I would welcome any suggestions or improvements, there are bound to be some. Note: The patterns are based on an earlier post by 'fusillo at NOSPAM dot netgang dot it' for matching domain names. Thanks to him.

webmaster

I notice a lot of queries involving regular expressions and is_float() regarding currency validation. If you have a version of php that uses the std C ctype.h functions then it will ALWAYS work with the following code. function check_price($price) { $data=split('[.]',$price); if ( count($data) != 2 ) return "false"; if ( ctype_digit($data[0]) && ctype_digit($data[1]) && $data[0][0] != '0') return "true"; else return "false"; } Peter Lorimer http:www.textedit.co.uk webmaster@textedit.co.uk

info

i needed a function to find hyperlinks containing a url as text of the hyperlink, exceeding a given maximum length. here my function to finds too long hyperlinks and insert <br />s where needed into the linktext: function breakTooLongLinks($text,$maxLen) { //find hyperlinks that contain too many chars & insert s where neccessary $pattern= '[>]www[.].*/*(.doc|.pdf|.htm|.html|.shtml|.php|.asp)(</a>|</A>)'; $match= eregi($pattern, $text, $regs); if ($match) { foreach ($regs as $link) { if (strlen($link)>$maxLen) { $linkParts= explode('/',$link); $linkRepl= array(); $replI=0; $curLinkPart=''; foreach($linkParts as $linkPart) { $curLinkPart.= $linkPart.'/'; if (strlen($curLinkPart)>$maxLen) { $linkRepl[]= $curLinkPart; $curLinkPart=''; } } $linkRepl= implode('<br />',$linkRepl); $text= str_replace($link, $linkRepl, $text); } } } return $text; }

keran

I couldn't get any of the email validation items above to actually work (maybe I'm thick) :) So I adapted a couple and came up with this, and it seems to work // function to validate email function validate_email($email_raw) { // replace any ' ' and \n in the email $email_nr = eregi_replace("\n", "", $email_raw); $email = eregi_replace(" +", "", $email_nr); // do the ergei to look for bad characters if( !eregi("^[a-z0-9]+([_\\.-][a-z0-9]+)*". "@([a-z0-9]+([\.-][a-z0-9]+))*$",$email) ){ // okay not a good email $feedback = "Error: $email isn't a valid mail address!"; return $feedback; } else { // okay now check the domain // split the email at the @ and check what's left $item = explode("@", $email); $domain = $item["1"]; if( gethostbyname($domain) == $domain) { $feedback = "Error: $domain isn't a valid domain!"; return $feedback; } else { $feedback = "valid"; return $feedback; } } }

steve

hope this helps with some validation problems, simple text validation: //$firstname = "somename"; // valid or //$firstname = "somenam3"; // not valid //lets validate a field entry & trim off any white spaces $firstname = trim($firstname); //set lenght of the field to a max 12 characters $len = "0,12"; $field = $firstname; //call function if (is_valid($field, $len)){ //if field entry valid then set $valid_firstname = $field; //set record to yes $valid_record = "TRUE"; echo "$field is a valid name"; } else{ //set record to null $valid_record = "FALSE"; echo " $field is not a valid name"; } //process to see if valid record if ($valid_record == "FALSE"){ echo " & is not a valid record"; } else{ //assume record is valid echo " & is a valid record"; } //validate field entry function function is_valid($field, $len) { if(eregi("^[[:alpha:]]{{$len}}$", $field)) return TRUE; else return FALSE; }

c00lways

hodsfords: i love your expression, and i've came out with a solutions which does not need to set the number of times {1,3} for the domain. it can accept unlimited number of times, but @ least 1 time .com / .xxx $exp = "^[a-z0-9]+[a-z0-9\?\.\+-_]*" . @[a-z0-9_-]+(\.[a-z0-9_-]+)*\.[a-z]+$";

gullevek

for the comment from 'joseph dot ragsdale at xconduit dot com' The regex has one bug thought. It elminates <user>@jp.t.ne.jp addresses which are completly valid as it needs at least 2 chars in the domain part. This regex fixes it: ^[_a-z0-9-]+(\.[_a-z0-9-]+)* @[a-z0-9-]+(\.[a-z0-9-]{1,})*\.([a-z]{2,}){1}$ [but the regex in one line]

emil

At m.hoppe: Perhaps this will help you verify an email address? I found it on http://www.regular-expressions.info <?php $email = "user@example.com"; if (eregi("^[A-Z0-9._%-]+@[A-Z0-9._%-]+\.[A-Z]{2,4}$", $email)) { echo "'$email' is a valid email!"; } ?>

An additional note to one of these expressions above: I use "^[a-z0-9]+([_.-][a-z0-9]+)*@([a-z0-9] +([.-][a-z0-9]+)*)+\\.[a-z]{2,4}$"

gorbiz

About the email verification posted by "emil at janesten dot com" I figured the maximum length of top domains is more than 4, at least 6, (ex .museum). Just replace {2,4} with {2,6}, like this: <?php $email = "user@example.com"; if (eregi("^[A-Z0-9._%-]+@[A-Z0-9._%-]+\.[A-Z]{2,6}$", $email)) { echo "'$email' is a valid email!"; } ?>

marcelo

A very sample code: checking numbers: <?php // author: marcelo at orionlab dot net // checking numbers $number = 465324; // match 0,1,2,3,4,5,6,7,8,9 if( eregi("[0-9]",$number) ) { printf("The variable contains numbers!"); } ?>

tuxx

A small note to one of these expressions above: Inside a character class ([...]) one does not need to escape the fullstop, therefore [-\\._] would become [-._]

alexm _a_t_ factory7.com

A little interesting and hopefully on topic point I found while browsing the internet at: http://www.clemburg.com/x1713.html#AEN1717 According to RFC 822 (see http://www.rfc-editor.org/), email addresses are case-sensitive in the part before the "@" sign, except for the special address "POSTMASTER", which is not case-sensitive. The part after the "@" sign is a host name, and is not case-sensitive. Note that many organizations will implement an aliasing facility that recognizes alternative forms for an email address (e.g., providing "John.Doe@my.organization.com", "john.doe@my.organization.com", "doe@my.organization.com", "Doe@my.organization.com" etc.). However, this is not required by the standard as specified in RFC 822. so you may want to keep that in mind.... i.e. breakup the username/alias (ereg) and domain (eregi) parts of the email

dompie

@ mmc at nospam dot dk Nice thinking!! But just a slight remark. This function isn't so general as you say.. f.e.: example@yahoo.co.uk. Here is $regs[2] == yahoo.co , so 'Error: Can't find the host ' Therefor just a little change in the code: <? if( !eregi("^[a-z0-9]+([_\\.-][a-z0-9]+)*" ."@([a-z0-9]+([\.-][a-z0-9]{1,})+)*$", $mail, $regs) ) { echo "Error: '$mail' isn't a valid mail address!\n"; } elseif( gethostbyname($regs[2]) == $regs[2] ) { echo "Error: Can't find the host '$regs[2]'! \n"; } ?> Now you have a fine working mail validation.. (as far as i know =) )

bobocop

<?php // Completely update for match RFC 2822 and RFC 1035 // http://www.faqs.org/rfcs/rfc2822.html // http://www.faqs.org/rfcs/rfc1035.html // Example results: $email[] = 'foo@example.com'; // matched $email[] = 'foo.bar@example.co.uk'; // matched $email[] = 'foo_bar@example.com'; // matched $email[] = '_foo_bar@example.com'; // matched $email[] = 'foo@example.example'; // matched $email[] = '%#a+f.*&654_-._@ee.xx'; // matched $email[] = 'foo@abc-123.xx'; // matched $email[] = 'a@a.a.a.a.aa'; // matched $email[] = 'a@a9.aa'; // matched $email[] = 'a!b#c$d%e^f&g*h\'i+j-k{l|m}n_/@op.qr'; //matched $email[] = ''; //separator $email[] = 'foo@-example.com'; // not matched $email[] = 'foo@example-.com'; // not matched $email[] = '%#af.*&@a%#b.xx'; // not matched $email[] = 'a@a.99.00.a.aa'; // not matched $email[] = '_-._@-.--'; // not matched $email[] = 'any..thing@bla.bla'; // not matched $email[] = '@.'; // not matched $email[] = '@.com'; // not matched $email[] = '@exam@exam.com'; // not matched $email[] = ' @ .com'; // not matched $email[] = '.bar@example.com'; // not matched $email[] = 'foo.@example.com'; // not matched $email[] = 'foo@example.x'; // not matched $atom = '[-a-z0-9!#$%&\'*+/=?^_`{|}~]'; // allowed characters for part before "at" character $domain = '([a-z]([-a-z0-9]*[a-z0-9]+)?)'; // allowed characters for part after "at" character $regex = '^' . $atom . '+' . // One or more atom characters. '(\.' . $atom . '+)*'. // Followed by zero or more dot separated sets of one or more atom characters. '@'. // Followed by an "at" character. '(' . $domain . '{1,63}\.)+'. // Followed by one or max 63 domain characters (dot separated). $domain . '{2,63}'. // Must be followed by one set consisting a period of two '$'; // or max 63 domain characters. foreach ($email as $example) { if (strlen($example) == 0): echo '  '; else: if (eregi($regex, $example)): echo $example . ' matched '; else: echo '<strong>'. $example . ' not matched</strong> '; endif; endif; } ?>

selman17@yahoo com

<?php $iletisim="selman17@yahoo.com http://www.yahoo.com selman@yahoo.com"; $ilet=explode(" ",$iletisim); $iletisim=""; for($i=0;$i<sizeof($ilet);$i++){ if(eregi("^.[a-zA-Z0-9_.\.-]+@[a-z0-9_]+\\.[a-z]+",$ilet[$i])){ $ilet[$i]= str_replace("$ilet[$i]","<a href='mailto:$ilet[$i]'>$ilet[$i]</a> ",$ilet[$i]); } if(eregi("http://.[a-zA-Z0-9_.\.-]",$ilet[$i])){ $ilet[$i]= str_replace("$ilet[$i]","<a href='$ilet[$i]' target='_blank'>$ilet[$i]</a> ",$ilet[$i]); } $iletisim .=$ilet[$i]; } /* this will out put <a href="mailto:selman17@yahoo.com">selman17@yahoo.com</a> <a href="http://www.yahoo.com" target="_blank">http://www.yahoo.com</a> <a href="mailto:selman@yahoo.com">selman17@yahoo.com</a> */ ?>

kidproto

"^[a-zA-Z0-9_\-]+(:?\.svg|\.wmf|\.zip|\.tar\.gz|\.tgz)$" That is the proper file check for a well formatted beginning and ending filename for those select files. The one in this list is incorrect for at least php 4.3.0 which is what I'm using. The (:?) makes sure that the parentheses aren't used for storing the matched regex.

eregi

Case insensitive regular expression match (PHP 4, PHP 5)
int eregi ( string pattern, string string [, array &regs] )

Example 1896. eregi() example

Related Examples ( Source code ) » eregi

Code Examples / Notes » eregi

Change Language