All page names need to be in English.
en da  de  fr  it  ja  km  nl  ru  zh


From TYPO3Wiki
Jump to: navigation, search

<< Back to Extension manuals page


This is the wiki version of the manual for realurl

notice - Note

Please check what content of this page is new.

Everything that is published in the manual

must be removed from this page.


Real URL

Extension Key: realurl

Copyright 2003-2004

Martin Poelstra, <> & Kasper Skårhøj <>

This document is published under the Open Content License.

The content of this document is related to TYPO3

- a GNU/GPL CMS/Framework available from


What does it do?

The extension provides automatic transformation of URLs with GET parameter in the frontend (like "index.php?id=123&type=0&L=1") into a virtual path, a so called "Speaking URL" (like "dutch/contact/company-info/page.html") and back again. The objective is that URLs shall be as human readable as possible.

The extension is very flexible and can provide from simple translation of page IDs to encoding of almost any possible combination of GET parameters.

Typed URL TYPO3 id and type
http: // id=0, type=0
http: // id=123, type=0
http: // id=123, type=2


TYPO3 works with page-IDs. This works great, however the URLs are very ugly ("...index.php?id=123&type=0&L=1...." etc.). There are workarounds (simulateStaticDocuments), but that's just a fake: the ID must still be supplied in the URL, which is not desirable. Furthermore, only the page-title is shown, not the complete 'path' (or 'rootline') to the page.

Normally, you type in the path and filename of a document, but TYPO3 works exclusively with page-IDs. The RealURL-extension provides a way to translate between page-IDs and (virtual) URLs that are easy to read and remember.

The extension requires the Apache module "mod_rewrite" to rewrite the virtual URLs of the site to the TYPO3 frontend engine.

Generally it will work out-of-the-box but you will have to address the issue that all media referenced in the HTML page has to have either absolute URLs or the <base> tag set. Both methods has advantages and drawbacks but the bottomline is that you might have to fix your templates/coding various places to be compatible.

- Screenshot ( extensions encoding)


  • Supports various schemes for coding the page path, including userdefined schemes
  • Pagetitles can contain spaces and characters like /.,&@ etc, the URL will still be nice.
  • URLS are generated as nice-looking lowercase paths
  • If a page is renamed, the old URL can still be used (see below in the Users Manual), so if the page was indexed by e.g. Google, it can still be found.
  • Offers advanced translation of almost any set of GET parameters to/from virtual URL
  • Translation between a GET query string ("...&tx_myext[blabla]=123&type=2...") and a virtual URL (".../123/2/") is transparent to TYPO3 and all extensions; The only requirement is that the internal TYPO3 link generation functions are used ("tslib_cObj::typolink", "t3lib_tstemplate::linkData")
  • URLs are cached, so translating between URLs and IDs is very fast.
  • It can handle different frames, or other pagetypes
  • URLs are multilingual: if you're browsing in Dutch, you'll see Dutch URLs
  • Once configured the systems works fully automatic, creating new and updating existing URLs
  • You can easily see where shortcuts are pointing to, as the 'target' URL is generated, instead of the URL to the shortcut itself.
  • It automatically handles multiple domains in the database
  • It automatically handles installations of TYPO3 in directories other than the root of the website too

Rawurlencoding of remaining parameter GEt vars names

- Disable "simulateStaticDocuments"!!




To install this extension, four steps must be taken:

  1. Install it in the Extension Manager
  2. Configure Apache
  3. Modify your templates for Real URLs
  4. Configure the extension in typo3conf/localconf.php

Install the extension

This is documented very well in the usual TYPO3 docs: just click the little gray sphere with the plus-sign and when it asks for any changes to commit, let it make them. It's not doing anything yet though.

Configure Apache

RealURLs work by providing 'virtual paths' to 'virtual files'. These don't actually exist on the file-system, so you must tell Apache to let a PHP-script handle the request if it can't find the file. This way, all URLs to pages (like http: // will be 'redirected' to /index.php, which will handle the translation of the URL into GET parameters. Real files (like images, the TYPO3 backend, static html-files, etc.) will still be handled by Apache itself though.

You should put the supplied sample .htaccess file (called _.htaccess) in the root of your TYPO3-installation.

Alternatively, you could include the following lines in your httpd.conf, probably in the VirtualHost-section. Here is an example:

	DocumentRoot /var/www/typo3/dev/testsite-3/
	ServerName www.test1.intra
	RewriteEngine On
	RewriteRule ^/typo3$ - [L]
	RewriteRule ^/typo3/.*$ - [L]
	RewriteCond %{REQUEST_FILENAME} !-f
	RewriteCond %{REQUEST_FILENAME} !-d
	RewriteCond %{REQUEST_FILENAME} !-l
	RewriteRule .* /index.php

If you put it into a .htaccess file it has to look slightly different, basically stripping the leading slashes ("/"):

RewriteEngine On
RewriteRule ^typo3$ - [L]
RewriteRule ^typo3/.*$ - [L]
RewriteCond %{REQUEST_FILENAME} !-f
RewriteCond %{REQUEST_FILENAME} !-d
RewriteCond %{REQUEST_FILENAME} !-l
RewriteRule .* index.php

This will tell Apache that it should rewrite every URL that's not a filename, directory or symlink. It leaves everything starting with /typo3/ alone too.

Notice: For this to work you need the Apache module "mod_rewrite" and AllowOverride=All in the Directory directive

TypoScript configuration

Like with "simulateStaticDocuments" you need to activate the generation of the virtual file/path names in the TypoScript record ? otherwise your website will not utilize the new URL encoding method.

However that is trivial; just place these five lines in the main TypoScript template record of your website:

config.simulateStaticDocuments = 0
config.baseURL = http: //
config.tx_realurl_enable = 1
config.uniqueLinkVars = 1
# dont forget to set the allowed range - otherwise anything else could be inserted
config.linkVars = L(0-3)

Line 1 simply disables "simulateStaticDocuments" - "realurl" is incompatible with simulateStaticDocuments and will simply not work if it has been enabled. This line should remind you of this fact.

Line 2 makes the frontend output a "<base>" tag in the header of the pages. This is required because relative references to images, stylesheets etc. will break when the virtual paths are used unless this has been set. Please see below for a detail discussion of why this is needed. (Note: config.baseURL won't work anymore in 3.8.1, you need to enter a base URL, e.g. baseURL = Neither can you leave out the http://, nor the slash at the end. ).

Line 3 enables the encoding of URLs as the virtual paths, the "Speaking URLs".

Line 4 makes clear that, in case the &L parameter is used more than once in the URL, only the last occurence will be used

Line 5 adds the L parameter to every link in Typo3

Here is an example-TS for two diff. domains

config.baseURL = localhost.tld/
[hostname = sub1.localhost.tld]
config.baseURL = sub1.localhost.tld/
[hostname = sub2.localhost.tld]
config.baseURL = sub2.localhost.tld/

If this is not working, try:

config.baseURL = localhost.tld/
[globalString = ENV:HTTP_HOST=sub1.localhost.tld]
config.baseURL = sub1.localhost.tld/
[globalString = ENV:HTTP_HOST=sub2.localhost.tld]
config.baseURL = sub2.localhost.tld/

Configure the extension

Finally, you probably want to configure the way URLs are encoded. For simple needs this is quite easy and the more advanced URLs you want to encode the more configuration you need - simple isn't it.

Configuration is done in "localconf.php" with the variable $TYPO3_CONF_VARS['EXTCONF']['realurl']

Please see the section later dealing with configuration options. It also offers a lot of examples.

URL en/decoding background

This section provides a bit of background information about how URLs are encoded and decoded in the system.

The general principle is that the encoding and decoding should be totally transparent to the system. This means that any extension will work with "realurl" as long as they use the general link generation functions inside TYPO3 as they should do already. You can also use "simulateStaticDocuments" as a test - if it worked with that, it will (most likely) work with the "realurl" extension as well.

The implementation of this transparency is done by encoding the virtual URL strictly on the basis of the GET parameters given to the encoder method. And when a HTTP request is made to a virtual URL it is decoded into a set of GET parameters which is written back to the global variables HTTP_GET_VARS / _GET - and thus any application in the system will see the parameters as it they were passed as true GET parameters.


URL with GET parameters -> Speaking URL -> HTML page

The encoding of the URLs happens by using a hook in the method t3lib_tstemplate::linkData(). This is configured in "realurl/ext_localconf.php":
PHP script:
= 'EXT:realurl/class.tx_realurl.php:&tx_realurl->encodeSpURL';


HTTP request with Speaking URL -> URL is decoded, overriding values in HTTP_GET_VARS -> page rendered as always

The decoding of the URLs happens by using a hook in tslib_fe::checkAlternativeIdMethods(). This is configured like this:
PHP script:
= 'EXT:realurl/class.tx_realurl.php:&tx_realurl->decodeSpURL';

The syntax of a "Speaking URL"

Before we go on to the configuration section it's important to understand how the virtual path (Speaking URL) is decoded by the system. Lets settle an example and break it up into pieces:


This URL requests page id 123 with language "1" (danish) and type "1" (probably a content frame in a frameset) and on that page the display of a mininews item with id "456" is requested while the "mode" of the mininews menu is "1" (list). This parameter based URL could be translated into this Speaking URL:

The configuration of "realurl" in 'localconf.php' needed to do this magic is as follows:
PHP script:
   1: $TYPO3_CONF_VARS['EXTCONF']['realurl']['_DEFAULT'] = array(
   2:     'preVars' => array(
   3:         array(
   4:             'GETvar' => 'L',
   5:             'valueMap' => array(
   6:                 'dk' => '1',
   7:             ),
   8:             'noMatch' => 'bypass',
   9:         ),
  10:     ),
  11:     'fileName' => array (
  12:         'index' => array(
  13:             'page.html' => array(
  14:                 'keyValues' => array (
  15:                     'type' => 1,
  16:                 )
  17:             ),
  18:             '_DEFAULT' => array(
  19:                 'keyValues' => array(
  20:                 )
  21:             ),
  22:         ),
  23:     ),
  24:     'postVarSets' => array(
  25:         '_DEFAULT' => array (
  26:             'news' => array(
  27:                 array(
  28:                     'GETvar' => 'tx_mininews[mode]',
  29:                     'valueMap' => array(
  30:                         'list' => 1,
  31:                         'details' => 2,
  32:                     )
  33:                 ),
  34:                 array(
  35:                     'GETvar' => 'tx_mininews[showUid]',
  36:                 ),
  37:             ),
  38:         ),
  39:     ),
  40: );

In order to understand how this configuration can translate a speaking URL back to GET parameters we have to first look at how the Speaking URLs are divided into sections. This is the general syntax:

	[TYPO3_SITE_URL] [preVars] [pagePath] [fixedPostVars] [postVarSets] [fileName]

Each of these sections (except [fileName]) can consist of one or more segments divided by "/". Thus "news/list/456/" is a sequence of three segments, namely "news", "list" and "456"

Taking the speaking URL from above (http: // as an example we can now break it up into these sections:

SectionPart from Example URLGeneral descriptionComment accord. to config
[TYPO3_SITE_URL]http: // part of the URL - the base URL of the site and basically where the "index.php" script of the frontend is located - is stripped of and is of no interest to the resolve of the parameters.
[preVars]dk/This section can contain from zero to any number of segments divided by "/". Each segment is bound to a GET-var by configuration in the key "preVars" (see example above)

The number of segments in the pre-vars section is determined exactly by the arrays in "preVars" configuration.

Typical usage would be to bind a pre-var to the "L" GET parameter so the language of the site is represented by the first segment of the virtual path.
In the configuration above there is only one pre-var configured and that is bound to the GET var "L". Further a mapping table tells that the value "dk" should be translated to a "1" for the GET var value when decoded. Also, if the segment does not match "dk/" it is just ignored. meaning that the default language version of the danish page ".../dk/123/" would be just ".../123/"
[pagePath]123/The page path determining the page ID of the page. The default method is to just show the page ID, but through configuration you can translate say "contact/company_info/" into a page ID. The number of segments of the path used to resolve the page ID depends on the method used.In this example the default method is used (not configured at all) and that means the raw page id (or alias if there were any) is displayed; 123. This is not "Speaking URL" behaviour, sorry for this weak example...
[fixedPostVars] Fixed post vars is a sequence of fixed bindings between GET-vars and path segments, just as the pre-vars are. This is normally not useful to configure for a whole site since general parameters to pass around should probably be set as pre-vars, but you can configure fixed post vars for a single page where some application runs and in that case may come in handy. Typical usage is to apply this for a single page ID running a specific application in the system.Not used in this example.
[postVarSets]news/list/456/postVarSets are sequences of GET-var bindings (in pre-var style) initiated by the first segment of the path being an identification keyword for the sequence.

Decoding of postVarSets will continue until all remaining segments of the virtual path has been translated. This method can be used to effectively encode groups of GET vars (sets), typically for various plugins used on the website.

Typical usage is to configure postVarSets for each plugin on the website.
In this example there is a single post var set ("news/list/456/") where the keyword "news/" (first segment) identifies the following sequence ("list/456/") to be mapped to the GET vars tx_mininews[mode] and tx_mininews[showUid] respectively.
[fileName]page.htmlThe filename is always identified as the segment of the virtual path after the last slash ("/"). In the "fileName" configuration a filename can be mapped to a number of GET vars that will be set if the filename matches the index key in the array. Typical usage is to use the filename to encode the "type" or "print" GET vars of a site.In this example the "type" value "1" is mapped to the virtual filename "page.html". In any other case the type value will be set to blank.

Configuration directives

Configuration of "realurl" is done in the array $TYPO3_CONF_VARS['EXTCONF']['realurl'] which again contains arrays. The configuration directives are broken down into these tables describing options as they are grouped together in arrays within the configuration array.

To support your understanding of the options please read the background information presented previously in this document and look at the examples available.


[host-name]->siteCfg or pointer to other key with ->siteCfg in same arrayConfiguration of Speaking URL coding based on current host-name for the website. Offers you the possibility of individual configuration for multiple domains in the same database.

If the value of an entry is a string, then the system will expect it to point to another key in on the same level and look for ->siteCfg there.

Hostname is found by t3lib_div::getIndpEnv('TYPO3_HOST_ONLY') and always in lowercase.
_DEFAULT->siteCfg or pointer to other key with ->siteCfg in same arrayConfiguration of default Speaking URL coding if no matches was found for the specific HOST name.
PHP script:
   1: $TYPO3_CONF_VARS['EXTCONF']['realurl'] = array(
   2:     '_DEFAULT' => array(
   3:         ...
   4:     ),
   5:     '' => array (
   6:         ...
   7:     ),
   8:     '' => '',
   9:     '' => '',
  10:     '' => '_DEFAULT',
  11:     'localhost' => '_DEFAULT',
  12: );

In this example the keys "_DEFAULT" and "" is assumed to contain proper configuration of "realurl". If the hostname turns out to be "" or "" the configuration of "" is used. If the hostname is "" or "localhost" then the "_DEFAULT" configuration is used (which is redundant since it would be defaulted to anyways!)


init->initGeneral configuration of the extension
redirects [path]Redirect URLHere you can specify virtual paths that should not be processed into GET vars but rather trigger a HTTP redirect header directly to the URL entered as value. If such a match happens the script issues a location-header and exits.
preVars [0..x]->partDefConfiguration of pre-variables; a fixed set of variables bound to the initial segments of the virtual path. See description in previous section of this document.
pagePath->pagePathConfiguration of the id-to-path transformation method. See description in previous section of this document.
fixedPostVars [pageIndex] [0..x]->partDefConfiguration of a fixed post-variable set which does not need a keyword to trigger its interpretation. Basically like pre-vars, just set after the pagePath.

See description in previous section of this document.

Notice: "pageIndex" allows you to specify this for individual pages or all using "_DEFAULT" keyword. See notice below this table.
postVarSets [pageIndex] [keyword]->postVarSetConfiguration of sets of post-variables; sets of post-variables are triggered by a keyword in the virtual path.

See description in previous section of this document. Notice: "pageIndex" allows you to specify this for individual pages or all using "_DEFAULT" keyword. See notice below this table.

Important: The order in which the postVarSets occur is of great importance since the first keyword definition that contains a definition for a single available GET-var will be chosen. You should arrange the postVarSets strategically.
fileName->fileNameConfiguration of filename significance; filenames can be bound to specific values of GET parameters. See description in previous section of this document.

Notice: In the table above there is defined an array key, "pageIndex", for "fixedPostVars and postVarSets. This works mostly in the same way as the host-name pointer did for the outer level. The key can be

  1. either a page id, eg. "123"
  2. or the keyword "_DEFAULT".

The value of a key should be an array (according to the definition above) but if it is a string it is interpreted as a pointer to another key on the same level.

A pointer cannot be set for "_DEFAULT". Further, only page ids can be used (internally page aliases given as parameters will be resolved first!).

Example structure
PHP script:
   1: array(
   2:     'init' => array(
   3:         ...
   4:     ),
   5:     'redirects' => array(
   6:          => 'cms/',    // If default URL, redirect to subdir "cms/"
   7:         'test/' => 'http: //www.test.test/',    // If subdir is "test/" then redirect to URL
   8:         'myFolder/mySubfolder/myFile.html' => 'test/index.php',
   9:     ),
  10:     'preVars' => array(
  11:         array(
  12:             ...
  13:         ),
  14:         array(
  15:             ...
  16:         ),
  17:     ),
  18:     'pagePath' => array(
  19:         ...
  20:     ),
  21:     'fixedPostVars' => array(
  22:         '1383' => array (
  23:             array(
  24:                 ...
  25:             ),
  26:             array(
  27:                 ...
  28:             ),
  29:         ),
  30:         '123' => '1383'
  31:     ),
  32:     'postVarSets' => array(
  33:         '_DEFAULT' => array (
  34:             'consultancy' => array(
  35:                 ...
  36:             ),
  37:             'admin' => array(
  38:                 ...
  39:             )
  40:         ),
  41:     ),
  42:     'fileName' => array(
  43:         ...
  44:     )
  45: );

This skeleton will help you to understand the structure defined in the table above for the "->siteCfg" level in the configuration. Notice the examples for redirects.


General configuration of the "realurl" extension

doNotRawUrlEncodeParameterNamesbooleanDisable rawurlencoding of non-translated GET parameter names during encoding.


During the encoding of Speaking URLs from GET parameters any GET parameters that cannot be translated into a Speaking URL will be set as GET parameters again. During this process the parameter name will be rawurlencoded as it actually should according to the RFCs covering this topic.

This means that a parameter like "tx_myext[hello]=world" will become "tx_myext%5Bhello%5D=world" instead - which is not as nice visually but more correct technically.
enableCHashCachebooleanIf set, "cHash" GET parameters are stored in a cache table if they are the only parameters left as GET vars. This allows you to get rid of those remaining parameters that some plugins might use to enable caching of their parameter based content.
respectSimulateStaticURLsbooleanIf set, all requests where the Speaking URL path is only a single document with no path prefix (eg. "123.1.html") are ignored as Speaking URLs. This flag can enable backwards compatibility with old URLs using simulateStaticDocuments on the site.
appendMissingSlashboolean / stringIf set, a trailing slash will be added internally to the path if it was not set by the user. For instance someone writes "http: //" with no slash in the end. "contact" will be interpreted as the filename by realurl - and the user wanted it to be the directory. So this option fixes that problem.

Keyword: "ifNotFile"

You can specify the option as the keyword "ifNotFile". If you use that string as value the slash gets prepended only if the last part of the path doesn't look like a filename (based on the existence of a dot "." character).
adminJumpToBackendbooleanIf set, then the "admin" mode will not show edit icons in the frontend but rather direct the user to the backend, going directly to the page module for editing of the current page id.
postVarSet_failureModestring (keyword)Keyword: "redirect_goodUpperDir". Will compose a URL from the parts successfully mapped and redirect to that.

Keyword: "ignore": A silent accept of the remaining parts.

Default (blank value) is a 404 page not found from TYPO3s frontend API.
enableUrlDecodeCachebooleanIf true, caching of URL decoding is enabled. The cache table is flushed when "all cache" is flushed in TYPO3.
enableUrlEncodeCachebooleanIf true, caching of URL encoding is enabled. The cache table is flushed when "all cache" is flushed in TYPO3.
emptyUrlReturnValuestringIf the URL is empty it usually is meant to be a link to the frontpage.

If you set this value to a string, that will be the URL returned if the URL is otherwise empty.

If you set this value true (PHP boolean, "TRUE"), then it will return the baseURL set in TSFE.

Setting it to "./" should work as a reference to the root as well. But it is not so beautiful.


Definition of mapping between a segment of the virtual path and a GET variable or otherwise.

typestring keywordBy default the array defines a GETvar mapping but there are alternatives which are configured by setting the type key:
  • "action" : If set, the segment can define various actions, like setting frontend editing or redirecting to a certain URL, setting some parameters.
type = "action"
index[segment] index["_DEFAULT"]->actionConfig (or blank value which will just accept the segment but take no action on it)The index array defines various actions and the first segment of the path is used as key to look up which action in the array to take. See ->actionConfig for more details and examples.
Default type
GETvar The GET var name for which this processing is done.


The value of the GETvar will pass through a transformation defined by the other configuration options here. Basically this is the flow:

  • First, check if "cond['prevValueInList']" allows processing and if not, return accordingly.
  • The value is translated through the "valueMap" if entries matches
  • If no entries in "valueMap" matched, "noMatch" is consulted for an action on this situation.
  • If noMatch did not trigger, we look for a lookup table and if defined we make a look up translation ("lookUpTable").
  • If no lookup table was defined to translate the value, we look for the "valueDefault" and if set, apply that value.
  • If none of these actions captured the value we just pass it through in its raw form.
cond['prevValueInList']list of valuesIf this key is set the segment will be processed only if the previous value is found in the comma list of this value! Otherwise it will be bypassed.
valueMaparray of "segment" => "Get var value" (translation table)The valueMap is a static translation table where each key represents a segment from the speaking URL and the value for the key is the true value of that parameter. When URLs are encoded this table is reversed and if there are duplicate values the last entry is used as speaking URL value.
noMatchstring keywordKeyword that defines the action if the value did not match any entry in the valueMap array.
  • "bypass" : means that if the path segment did NOT match an entry in "valueMap" the segment will be re-applied to the stack and we return/break (and thus preserves the parameter for the next segment)
  • "null" : means that if the value is NOT found in the valuemap the getParameter is not set at all.
lookUpTable->lookUpTableConfiguration of a database table by which to perform translation from id to alias string.
valueDefaultstringDefault value to set if the path segment did not match any entries in "valueMap" or was otherwise captured for translation. Notice: Default values are applied AFTER any "noMatch" settings are processed (and others, see flow description for key "GETvar")
PHP script:
   1:     'preVars' => array(
   2:         array(
   3:             'GETvar' => 'no_cache',
   4:             'valueMap' => array(
   5:                 'no_cache' => 1,
   6:             ),
   7:             'noMatch' => 'bypass',
   8:         ),
   9:         array(
  10:             'GETvar' => 'L',
  11:             'valueMap' => array(
  12:                 'dk' => '1',
  13:                 'danish' => '1',
  14:                 'uk' => '2',
  15:                 'english' => '2',
  16:             ),
  17:             'noMatch' => 'bypass',
  18:         ),
  19:     ),

The above example shows a configuration that allows two prevars in a path BUT they are both optional (due to the "noMatch" => "bypass" setting).

Normally a URL in the default language would look like this:


Then, if the L=1 GETvar is set, the URL will be like this:


Finally, if the first segment matches "no_cache" the "no_cache=1" GET var is set and the interpretation of the language GETvar is moved to segment 2:


The concept of bypassing non-matched values opens the possibility of error if two values from neighbouring configurations matches. For instance errors would result from having a language labeled "no_cache" since that is a keyword in the configuration of the first segment!

Removing the "noMatch" setting will yield these URLs instead:


A better solution would be to set a default value for the language:

PHP script:
   1:     'preVars' => array(
   2:         array(
   3:             'GETvar' => 'no_cache',
   4:             'valueMap' => array(
   5:                 'no_cache' => 1,
   6:             ),
   7:             'noMatch' => 'bypass',
   8:         ),
   9:         array(
  10:             'GETvar' => 'L',
  11:             'valueMap' => array(
  12:                 'dk' => '1',
  13:                 'danish' => '1',
  14:                 'uk' => '2',
  15:                 'english' => '2',
  16:             ),
  17:             'valueDefault' => 'uk',
  18:         ),
  19:     ),

This would yield this result:


It still maintains the bypass-setting for the "no_cache" parameter but that might just fit in nicely.

Example: "fixedPostVars"
PHP script:
   1:     'fixedPostVars' => array(
   2:         'testPlaceHolder' => array (
   3:             array(
   4:                 'GETvar' => 'tx_extrepmgm_pi1[mode]',
   5:                 'valueMap' => array (
   6:                     'new' => 1,
   7:                     'categories' => 2,
   8:                     'popular' => 3,
   9:                     'reviewed' => 4,
  10:                     'state' => 7,
  11:                     'list' => 5,
  12:                 )
  13:             ),
  14:             array(
  15:                 'cond' => array(
  16:                     'prevValueInList' => '2'
  17:                  ),
  16:                 'GETvar' => 'tx_extrepmgm_pi1[display_cat]',
  18:                 'valueMap' => array (
  19:                     'docs' => 10,
  20:                 ),
  21:             ),
  22:             array(
  23:                 'GETvar' => 'tx_extrepmgm_pi1[showUid]',
  24:                 'lookUpTable' => array(
  25:                             'table' => 'tx_extrep_keytable',
  26:                             'id_field' => 'uid',
  27:                             'alias_field' => 'extension_key',
  28:                             'addWhereClause' => ' AND NOT deleted'
  29:                         )
  30:             ),
  31:             array(
  32:                 'GETvar' => 'tx_extrepmgm_pi1[cmd]',
  33:             )
  34:         ),
  35:         '1383' => 'testPlaceHolder',
  36:     ),

This configuration shows how "fixedPostVars" can be used like "preVars" but after the page path. Typically it would be used on a single page where a known plugin runs. In the above example this is the case; page id "1383" is pointed to the configuration "alias" named "testPlaceHolder". The example is designed for the Extension Repository.

The configuration sets up a sequence of 3-4 segments in the virtual path. The first is the main menu where integer values defining the mode is mapped to nice alias strings. The second segment is the category id to display but notice how the cond["prevValueInList"] is set to "2" - this means that only if the previous variable was "2" then will this segment be interpreted, otherwise bypassed! Finally there is the extension uid, here configured for translation to/from the extension keys. That is a safe process since the extension keys are unique. Finally the "command" which defines the menu level when displaying single extensions.

This configuration allows for a URL like this (4 segments in "fixedPostVars" sequence):

http: //

or (only 3 segments in "fixedPostVars" sequence, since first segment was not "categories" / 2)

http: //


Defines a table to use for look up in translation of id to alias strings for GETvars.

tablestringTable name
id_fieldstringFieldname of the field holding the id, typically an integer, eg. "uid"
alias_fieldstringFieldname of the field holding the alias string matched to the id. Should be unique.
addWhereClausestringAdditional where clause to add to the look up query. Has to be set automatically, even for "deleted" fields since the lookup might take place before any table configuration is accessible!
Example value is: "AND NOT deleted"
maxLengthintegerDefines the maximum accepted length of the alias value. If the alias value is longer than this integer the original value will be returned instead.
Default value is "100"
useUniqueCachebooleanIf set, the translation from id to alias is automatically stored in a look-up table where the uniqueness of the alias is ensured; When storing it is simply checked if the alias is already associated with another ID (in the same table/fields combination) and if so a unique alias is created, typically the requested alias with numbers appended.
useUniqueCache_conf['strtolower']booleanIf set, the aliases are converted strtolower()
useUniqueCache_conf['spaceCharacter']stringNormally, this defaults to an underscore (_), which is used to replace spaces and such in an URL. You can set this to e.g. a hyphen (-) if you want to.
                        'lookUpTable' => array(
                            'table' => 'user_3dsplmxml_bfsbrand',
                            'id_field' => 'xml_id',
                            'alias_field' => 'xml_title',
                            'maxLength' => 10,
                            'addWhereClause' => ' AND NOT deleted'


typestring keywordThe action is identified by one of these keywords:
  • "admin" : Enables the Frontend Editing features (similar feature to ->postVarSet, type = "admin".
  • "redirect" : Redirects you to another URL with optional markers filled with the value of the path segment and the remaining path. Useful for site shortcuts
  • "notfound" : Throws a 404 error
  • "bypass" : Bypass AND adds the key to the stack again!
  • default: Passthrough(without adding key to stack).
Notice about the "_DEFAULT" key: If the "_DEFAULT" type is not "bypass" (meaning that the _DEFAULT action is to do something) the URL encoding method will look for the first occurence of an action of no type (passthrough without adding key to stack) and use that as segment value.
Only type "redirect"
urlstringThe URL to redirect to. There are two markers you can use in the URL:
  • ###INDEX### - Inserts the current segment.
  • ###REMAIN_PATH### - Inserts the remaining path from this point (including any filename)
The values are rawurlencoded()
Example: Redirects and required prefixes
   1:     'redirects' => array(
   2:          => 'cms/',
   3:         'mailinglist/' => 'http: //',
   4:     ),
   5:     'preVars' => array (
   6:         array(
   7:             'type' => 'action',        // "type" action
   8:             'index' => array(
   9:                 'cms' => ,    // Just bypass
  10:                 'admin' => array(
  11:                     'type' => 'admin'    // Jump BE login OR setting frontend edit flags...
  12:                 ),
  13:                 'search' => array(
  14:                     'type' => 'redirect',    // Redirect...
  15:                     'url' => 'index.php?id=1344&tx_indexedsearch[sword]=###REMAIN_PATH###',
  16:                 ),
  17:                 'ext' => array(
  18:                     'type' => 'redirect',    // Redirect...
  19:                     'url' => 'cms/1383/list/###REMAIN_PATH###/index.html',
  20:                 ),
  21:                 '_DEFAULT' => array(
  22:                     'type' => 'notfound'    // If key was not found in index, throw "404" not found.
  23:                 ),
  24:             ),
  25:         ),
  26:     ),

In this example the the first segment of the URL is configured to be an action. The segment is required since the "type" of the "_DEFAULT" index is set to "notfound" meaning that if none of the other keys are matched you will see a "Page not found" error.

Anyways, the example takes offset in the website and this is the consequences of the above configuration:

URLWhat happens
http: // URL would result in a 404 error if it wasn't for the redirect configuration in line 1-4 of the configuration. Here the system is ordered to redirect to "cms/" in case there is no virtual path found.
http: // will show page ID 1420. The action key "cms" doesn't do anything - it just bypasses the processing to the next level which is the page ID resolving. By configuring such a prefix you basically define all of your site to "run" from the virtual directory "cms/". Configuration is in line 9
http: // will also show page ID 1420 but activate the frontend editing icons in the interface - and if the user is not currently logged in as a backend user a redirect to the backend login form will happen. Configuration of this feature is in line 10-12 When using the Frontend Editing trigger ("admin") in "realurl" the "admin/" segment of the path will be carried around in the links on the site so you stay in admin mode until you remove it again. But this is also means that caching is disabled for all pages so page links are never stored with the "admin/" action in!
http: // redirect to http: // which is the page where the indexed_search plugin is running and automatically performing a search for the words after "search/". This is configured in line 13-16; notice how the search word (the remaining path) is automatically inserted in the URL by the marker string "###REMAIN_PATH###"
http: // redirect to http: // where "lang" is inserted by the marker "###REMAIN_PATH###" just like with the "search" action before. The principles are the same. Configuration in line 17-20
http: // redirect to http: // according to line 3 in the configuration
Example: Language prefix and "admin" action
   1:     'preVars' => array (
   2:         array(
   3:             'type' => 'action',        // "type" action
   4:             'index' => array(
   5:                 'admin' => array(
   6:                     'type' => 'admin'    // Jump BE login OR setting frontend edit flags...
   7:                 ),
   8:                 'search' => array(
   9:                     'type' => 'redirect',    // Redirect...
  10:                     'url' => 'index.php?id=1344&tx_indexedsearch[sword]=###REMAIN_PATH###',
  11:                 ),
  12:                 'ext' => array(
  13:                     'type' => 'redirect',    // Redirect...
  14:                     'url' => 'cms/1383/list/###REMAIN_PATH###/index.html',
  15:                 ),
  16:                 '_DEFAULT' => array(
  17:                     'type' => 'bypass'    // If key was not found in index, throw "404" not found.
  18:                 ),
  19:             ),
  20:         ),
  21:         array(
  22:             'GETvar' => 'L',
  23:             'valueMap' => array(
  24:                 'dk' => '1',
  25:             ),
  26:             'noMatch' => 'bypass',
  27:         ),
  28:     ),

In this example two preVars are configured, the first is an action containing almost the same actions as the previous example except that the "_DEFAULT" configuration is of the "bypass" type which means that if none of the key matches the first segment the interpreter will simply move on to the next preVar configuration for that segment (bypassing and adding segment value to stack again).

In addition there is a language prefix configured as well.

The result of this configuration should be clear from looking at the examples in the following table:

URLWhat happens
http: // page 1420
http: // page 1420 with Frontend Editing icons
http: // page 1420 in danish (&L=1)
http: // page 1420 in danish (&L=1) with Frontend Editing icons


Configuration of the method that en/decodes the id to/from a "page path"

typestringSetting the method used for encoding/deconding of the ID

The default simply is to set the page id/alias as a single entry in the virtual path.

"user" : Calls external class for generation.
Only type "user":
userFuncfunction referenceFunction reference to handle the id encoding.

Examples can be found in class.tx_realurl_dummy.php An full fledged implementation is found in "class.tx_realurl_advanced.php". See more details later in documentation about this. Example value:

Example of page id translation to path:
    'pagePath' => array(
        'type' => 'user',
        'userFunc' => 'EXT:realurl/class.tx_realurl_advanced.php:&tx_realurl_advanced->main',
        'spaceCharacter' => '-',
        'languageGetVar' => 'L',
        'expireDays' => 3

Calls a user function which will render a true Speaking URL of the page titles and not just output the page id numbers. See a thorough description of this class later in this document, chapter "class.tx_realurl_advanced.php"

By the configuration above URLs will look like this (before / after):

Example of dummy setup:
    'pagePath' => array(
        'type' => 'user',
        'userFunc' => 'EXT:realurl/class.tx_realurl_advanced.php:&tx_realurl_dummy->main',

Calls a dummy class which does exactly what the main class does: Outputs the page id/alias and nothing more. But if you want to implement your own schemes this class is a useful offset for you!


typestring keywordBy default a postVarSet consists of
  • a keyword that identifies the following sequence in the virtual path
  • a sequence of one or more segments in the path which is mapped to GETvars

An example (from previously) is "news/list/456/" where the keyword "news/" (first segment) identifies the following sequence ("list/456/") to be mapped to the GET-vars tx_mininews[mode] and tx_mininews[showUid] respectively (according to configuration). The configuration of the sequence is done by a numeric array of ->partDef. Other modes: There are other modes than the default mode and they can be triggered by setting the "type" key to one of the following keywords. In these alternative cases there might not be any sequence of segments after the keyword, but still the keyword is triggering the alternative mode so that will always be around. "single" : Using this keyword you bind the keyword to represent an exact amount of GETvars with exact values. This works precisely like filenames are bound to GETvars (see ->fileName)

"admin" : Using this keyword the backend will enable frontend editing mode for the user, showing the context sensitive edit icons in the frontend. If the user is not logged in as a backend user there will be a redirect to the backend login form where the user can login and after successful login the user will be redirected to the previous page.
Only default type:
[0..x]->partDefThe configuration of the sequence associated with the "keyword" that defines this postVarSet.
Only "single" type:
keyValuesarray of [GETvar] => [string value]The "keyValues" array defines one or more GET variables with specific values. The keyword of the postVarSet is matched if if all the GET-vars configured in ["keyValues"] is found in the remaining pool of GET vars that needs translations and if the values match exactly!
Example: Frontend edit
   1:     'postVarSets' => array (
   2:         '_DEFAULT' => array(
   3:             ....
   4:             'edit_now' => array(
   5:                 'type' => 'admin'
   6:             )
   7:         ),
   8:     )

Adding lines 4-6 in the above codesnippet to the postVarSets of a configuration will enable frontend edit mode if users append ".../edit_now/" to the virtual path. Of course you can choose any "admin-directory" you like.

One warning here; If the keyword is appended to a URL where a previous postVarSet sequence is not yet finished then the keyword will of course be seen as a parameter of that postVarSet and not as the keyword triggering the frontend edit mode as you wanted. Therefore you might want to use the same feature but for pre-vars instead.

Example: PostVarSets
   1:     'postVarSets' => array(
   2:         '_DEFAULT' => array (
   3:             'plaintext' => array(
   4:                 'type' => 'single',    // Special feature of postVars
   5:                 'keyValues' => array (
   6:                     'type' => 99
   7:                 )
   8:             ),
   9:             'ext' => array(
  10:                 array(
  11:                     'GETvar' => 'tx_myExt[p1]',
  12:                 ),
  13:                 array(
  14:                     'GETvar' => 'tx_myExt[p2]',
  15:                 ),
  16:                 array(
  17:                     'GETvar' => 'tx_myExt[p3]',
  18:                 ),
  19:             ),
  20:             'news' => array(
  21:                 array(
  22:                     'GETvar' => 'tx_mininews[mode]',
  23:                     'valueMap' => array(
  24:                         'list' => 1,
  25:                         'details' => 2,
  26:                     )
  27:                 ),
  28:                 array(
  29:                     'GETvar' => 'tx_mininews[showUid]',
  30:                 ),
  31:             ),
  32:         ),
  33:     ),

This example shows how three sets of postVarSets has been configured of which two of them are the default type (keyword + sequence of GETvars) while the third is of the "single" type, mapping to a fixed GETvar value.

In order to understand this configuration and the effect it has you should study these commented examples based on the above configuration. Each example consists of two lines where the first is the URL with GETparameters and the second is the Speaking URL version of the first.


Above, the postVarSet "ext" is used to encode the GET parameters. The sequence is initiated by the keyword "ext" and the following three segments of the virtual path is mapped to the three GET-vars configured for the keyword and in the order they appear in the configuration above (line 10-18)


The above two examples shows what happens if only one or two of the parameters are used - basically the empty values are stripped off from the end of the path. The first example would actually render "123/ext/aaa///" and the second would be "123/ext/aaa/bbb//" but since the empty values are in the end of the path we can safely strip them off as the example shows.


In this example only "tx_myExt[p1]" and "tx_myExt[p3]" is used and since the sequence requires "p2" to be in between we have to accept the empty segment of the virtual path.


In the above example the mininews parameters are encoded, using the keyword "news". Notice that the "tx_mininews[mode]" GETvar has a mapping table which allows automated translation between the value "1" and "list" used in the virtual URL. This feature (and other similar options) allows to create truely speaking URLs even for parameters that are ID numbers.


In this example we have two postVarSets, namely "ext" and "news". As you can see this is no problem at all as long as the sequences contains the correct amount of segments so the next keyword gets registered.

Notice that the "ext" keyword gets listed first. This is because the "ext" postVarSet is the first one found in the configuration and therefore is triggered before the "news" postVarSet.


In this example we have left out two parameters from the previous URL and that means there are two empty segments in the virtual path. There is no way around this since the sequence length has to be respected and in case of the "news" postVarSet the "mode" was defined before the "showUid" parameter which means it is not possible to strip of the the empty value.


This example is only different to the above examples by showing what happens to an unknown GET var when a URL is encoded; quite simply that GET var is appended to the final URL - what else!


Configuration of the significance of the filename in the virtual path.

index[filename]["keyValues"]array of [GETvar] => [string value]The "index" is an array of virtual filenames (eg. "page.html") which is associated with one or more GET variables with specific values.

A filename is chosen during encoding if all the GET-vars configured in index[filename]["keyValues"] is found in the remaining pool of GET vars that needs translations and if the values match exactly! The filename value "_DEFAULT" is used if no match was found.

Important: The order in which the filenames occur is of great importance since the first filename that matches will be chosen. You should arrange the filenames strategically. See example below.
defaultToHTMLsuffixOnPrevbooleanIf set, then the last directory part of the virtual path being made will be turned into the filename suffixed ".html" IF the filename part is non-existing.

For example, "workplace-learning-solutions/companion-solutions/" would be turned into "workplace-learning-solutions/companion-solutions.html" and the basepart of the filename (stripping the ".html" extension) will still be perceived as the last part of the virtual path.

This approach is useful if you want to simulate HTML documents even if you don't configure any fileName mappings.
Example: Multiple filenames for a frameset
   1:     'fileName' => array (
   2:         'index' => array(
   3:             'print.html' => array(
   4:                 'keyValues' => array (
   5:                     'print' => 1,
   6:                     'type' => 1,
   7:                 )
   8:             ),
   9:             'page.html' => array(
  10:                 'keyValues' => array (
  11:                     'type' => 1,
  12:                 )
  13:             ),
  14:             'top.html' => array(
  15:                 'keyValues' => array (
  16:                     'type' => 2,
  17:                 )
  18:             ),
  19:             '_DEFAULT' => array(
  20:                 'keyValues' => array(
  21:                 )
  22:             ),
  23:         ),
  24:     ),

This example could be configuration for a frames-based website. When there should be no "type" value the default key is used (which renders no filename of course) but if the &type value is 1 or 2 either "page.html" or "top.html" is used; those filenames will then represent the GET var "&type=1" and "&type=2" respectively during decoding.

Also notice how the "&print=1" parameter has been encoded into a filename! The idea is that if the filename is "print.html" then two GET vars are set; both "&type=1" and "&print=1". But you must be very careful how you arrange the filenames; if "print.html" was entered below "page.html" then it would never be used since the first match wins and "page.html" would be found to match exactly with "&type=1" and thus the "&print=1" GET var would be appended the URL (like "page.html?print=1") instead of being encoded into the filename ("print.html").

Example: Default filename
    'fileName' => array (
        'index' => array(
            'index.html' => array(
                'keyValues' => array(

This example shows a configuration that will prefix the filename "index.html" by default no matter what.

Handling relative links with Speaking URLs

By default, TYPO3 generates all links to other pages as http: //, so all pages seem to be in one (filesystem-) directory: the root of the website. The problem is, that many extensions (and TYPO3 core code) rely on images, javascripts, etc. to be in a directory relative to the TYPO3-root, like "typo3/ext/indexed_search/pi/res/pages.gif". This approach doesn't work when the path is constantly changing.

For example, a file "fileadmin/my_image.jpg" referenced from "index.php?id=123" will be found because "index.php" is in the root of the website where also the "fileadmin/" folder is. But as soon as the URL "index.php?id=123" is encoded into a Speaking URL, say "contact/company_address/", then your browser will try to find the image in "contact/company_address/fileadmin/my_image.jpg" where is obviously isn't located.

So to solve this problem you

  1. either have to prefix all relative references with an absolute path to the site root
  2. or set the <base> tag in the HTML files header to the site root.


There is a TypoScript-setup directive to set an absolute prefix to all links and images (config.absRefPrefix), but sadly enough that isn't implemented in all places (the indexed-search and front-end-editing for example), so that doesn't work too well.

Please don't use config.absRefPrefix. It has some nasty properties that render RealURLs complete unusable sometimes. The only problem is that the 404-page of TYPO3 doesn't have the <base>-tag, so it doesn't show the TYPO3-logo :)

Support for this might be allowed when the bugs are fixed but generally it will require all code generating reference to use this method and that cannot be guaranteed for all extensions of course.

<base> tag

There is a very simple solution in HTML though: just supply the <base>-tag in the <head> of your pages, like:

<base href="http: //">

To make your TypoScript templates RealURL-enabled, you should therefore include this statement in your HTML-templates, or use the following TypoScript snippet:

config.baseURL = 1

This will automatically read what the current base URL is on your website (using t3lib_div::getIndpEnv('TYPO3_SITE_URL')) and create a <base> tag in the header of the HTML output in the frontend. Be careful: Due to security-concerns this behaviour is no longer supported in version newer than TYPO3 3.8.

If for some reason you want to set another URL instead of the automatically created one can just set it like this:

config.baseURL = http: //

The <base> tag method seems to work flawlessly in TYPO3 except in two cases where you have a link like <a href="#anchor_point"....> - this will not work because it refers to the site root but obviously is an 'internal' reference in the current document.

However you can solve this situation in a simple way:

config.prefixLocalAnchors = all

This will set the needed prefix for all occurencies of '< a href="#...."....' in the page; basically anywhere a local anchor is generated. This substitution happens by a ereg_replace on the general page content after rendering. See TSref for details.

The other situation is specific for MSIE; When you set the "document.location" via JavaScript, MSIE will understand relative URLs as relative to the current URL, not the <base> URL. Therefore you will have to prefix the base URL. You can find that value in $GLOBALS['TSFE']->baseUrl (or use t3lib_div::getIndpEnv("TYPO3_SITE_URL")).

Warning: It might be dangerous to enable "config.baseURL = 1" (the automatic creation of base URL) if you are able to access the website from an IP address or internal network. In such a case the base URL could be "http: //" which a) will not work for anyone visiting your site from outside and b) reveal information about the server directory structure (security). This problem will not appear if pages are not cached when visited from the internal network but if pages are cached during internal visits to the site the internal base URLs will be on the pages that people from outside see, hence the problem! Actually the same is true if pages are generated and cached by "index.php?id=123" and then later requested by a speaking URL - then a page generated from a "wrong" location will show the cached content of the wrong location.

Making extensions compatible with "config.baseURL"

If you set "config.baseURL" and subsequently "config.prefixLocalAnchors = all" then extensions might still produce wrong local anchors. That is if extensions are including un-cached page content by USER_INT or USER_EXT cObjects that content is not processed! (unless "config.prefixLocalAnchors" is set to "output"). For such extensions there should be inherent support for "realurl" and that can be done (with full backwards compatibility) by prefixing all local anchors made by the result of this:


or in recent TYPO3 versions:



Make sure you include either configuration it in ALL page-types that are generated!

User annotations:

More Working Samples


Special thanks to the mailing list folks for providing all them tips. The provided localconf.php modification handles tt_news.

Code Listing:


RewriteEngine On
RewriteRule ^typo3$ - [L]
RewriteRule ^typo3/.*$ - [L]

RewriteCond %{REQUEST_FILENAME} !-f
RewriteCond %{REQUEST_FILENAME} !-d
RewriteCond %{REQUEST_FILENAME} !-l
RewriteRule (\.html|/)$ /index.php


$TYPO3_CONF_VARS['SC_OPTIONS']['t3lib/class.t3lib_tstemplate.php']['linkData-PostProc'][] = 'EXT:realurl/class.tx_realurl.php:&tx_realurl->encodeSpURL';
$TYPO3_CONF_VARS['SC_OPTIONS']['tslib/class.tslib_fe.php']['checkAlternativeIdMethods-PostProc'][] = 'EXT:realurl/class.tx_realurl.php:&tx_realurl->decodeSpURL';

$TYPO3_CONF_VARS['SC_OPTIONS']['t3lib/class.t3lib_tcemain.php']['clearAllCache_additionalTables']['tx_realurl_urldecodecache'] = 'tx_realurl_urldecodecache';
$TYPO3_CONF_VARS['SC_OPTIONS']['t3lib/class.t3lib_tcemain.php']['clearAllCache_additionalTables']['tx_realurl_urlencodecache'] = 'tx_realurl_urlencodecache';

$TYPO3_CONF_VARS['FE']['addRootLineFields'].= ',tx_realurl_pathsegment';

$TYPO3_CONF_VARS['EXTCONF']['realurl']['_DEFAULT'] = array(
'init' => array(
'enableCHashCache' => 1
'preVars' => array(
'GETvar' => 'no_cache',
'valueMap' => array(
'no_cache' => 1,
'noMatch' => 'bypass',
'GETvar' => 'L',
'valueMap' => array(
'no' => '1',
'noMatch' => 'bypass',
'fileName' => array (
'index' => array(
'backend.php' => array(
'keyValues' => array (
'type' => 100,
'print' => array(
'keyValues' => array(
'type' => 98,
'postVarSets' => array(
'_DEFAULT' => array (
'article' => array(
'GETvar' => 'tx_ttnews[tt_news]',
'GETvar' => 'tx_ttnews[backPid]',
'category' => array(
'GETvar' => 'tx_ttnews[cat]',
'neste' => array(
'GETvar' => 'tx_ttnews[pointer]',
'pagePath' => array(
'type' => 'user',
'userFunc' => 'EXT:realurl/class.tx_realurl_advanced.php:&tx_realurl_advanced->main',
'spaceCharacter' => '-',
'languageGetVar' => 'L',
'expireDays' => 3

Root template setup

config.simulateStaticDocuments = 0
config.baseURL = 1
config.tx_realurl_enable = 1
# config.absRefPrefix = /

No Language -> a little bit different

Date: 29-08-2004 15:03 by maholtz

instead of

// Leave empty to disable a language-prefix in the URL
$TYPO3_CONF_VARS["FE"]["realurl"]["langCodes"] = array(0 => );

you should write

// disable a language-prexix in the URL
$TYPO3_CONF_VARS["FE"]["realurl"]["langCodes"] = array();

apache redirection may slow your site

Date: 16-08-2004 15:25 by luite

If you follow the above example to configure your apache redirecting there is an unwelcome side effect: _everything_ is redirected to index.php (except for existing files). Now let's suppose you add a reference (by mistake) to a non-existing file, say "/fileadmin/mistake.css", this file is not found and therefor redirected to index.php. Now you've got a css style filled with some webpage content, effectively bringing your webbrowser to it's knees. So let's do the redirecting just a bit smarter. In the following example only .html files and directories that do not exist are redirected to index.php, other non-existent files will give a 404 not found page!

Code Listing:

RewriteCond %{REQUEST_FILENAME} !-f
RewriteCond %{REQUEST_FILENAME} !-d
RewriteCond %{REQUEST_FILENAME} !-l
RewriteRule (\.html|/)$ /index.php

Use nav_title instead of pagetitle

Date: 22-05-2004 02:25 by xonx

To use pages.nav_title instead of pages.title, you will have to change the code in class.ux_tslib_fe.php This example works for me (on 3.6.1), but it's just a quick and dirty hack. Based on version 0.1.4 Alter the following three lines (change 'title' to 'nav_title'):

Code Listing:

Line 399:
$paths[$i] = $this->encodeTitle($page["nav_title"]);
Line 692:
$query = "SELECT uid,nav_title FROM pages WHERE pid = $pid AND deleted = 0 AND doktype <> 255";
Line 707:
$query = "SELECT nav_title FROM pages_language_overlay WHERE pid = $l_id";


Date: 27-02-2004 16:42 by stfl_wolfgang

To make it work with TYPO3 in a directory add the first line and change the second one:

Code Listing:

RewriteBase /my/directory/
RewriteRule .* /index.php


RewriteRule .* index.php


The class tx_realurl_advanced offers advanced encoding of page IDs to paths including encoding in localized titles and cache management...

[martin, finish this section which is about your script... I have already imported some of your documentation but you SHOULD review it and make it more meaning ful in the new context. I didn't cut anything out because it was bad, so just add whatever old content that still fits...]

Page resolve methods, how does it work?

Let's first take a look at the simple case, where everything goes smooth and nothing goes wrong:

When you type an URL (or click on it), it is looked up in the so called URL-cache. Assuming it is found, we then know the page-ID (from the path) to generate the correct page. And we're done.

Now, some things can go wrong here: first of all, the paths generated by RealURL only contain a..z, 0..9 and underscores ('_'), so it's a good idea to strip the URL off all unwanted characters before we look it up. So we do that :)

Furthermore, if the URL isn't found in the URL-cache (e.g. when the cache is cleared), we have to search for it in the database. This is done by first examining the domain-part, and then searching for every 'directory' in the URL until we reach the destination-page. The language-part (if present) is also translated to a language-ID. This result is cached, so we can use it later on.

If the language wasn't given in the URL, a function is called to figure out what language will be most appropriate. I created some code which looks the IP-address up in the IP2Country-database (a table), which I imported into the TYPO3-database. I might create a separate extension for this, but for now you can uncomment the code if you want to use it. Look for getDefaultLangID().

When the page-ID is found, another check is done (after TYPO3 has loaded all kinds of information about the page): it checks to see if the requested URL corresponds to the 'real' URL of the page. This might be different due to changes in the page-title of the page, or one of it's parents. Or, one might have typed the URL to a page that is a shortcut to the real page. In those cases, the user is redirected to the real/official/new URL of the page and in case of a changed page-title, the old URL is marked as 'expiring'.

This makes it possible to change the page-title of a page (and thus it's URL), but still be able to reach that page through the old URL, which will still be used by e.g. Google.

A problem arises when you create another page, with the same title as a page that existed before, because that URL still points to the other page. Therefore, if RealURL notices that you typed an expiring URL, it searches the database like the URL wasn't found in the cache. If a page is found this way, that new URL + page-ID is cached instead. If no page is found, the cached result will be used.

The other way around is much simpler: when an URL has to be generated for a certain page, it is looked up in the same URL-cache. If it isn't already there, the URL will be created by building the so-called RootLine for the page, filtering every page-title so that it contains only a..z, 0..9 and underscores and finally caching it. This process does take languages into account, so if you're browsing to the Dutch version of a page, you'll get a Dutch URL to it (using the Dutch page-titles).


You should create Domain-records on the pages where domains start. Even if you only have one domain, it's a good idea to create a Domain-record for it. There's one thing you should note:

If you have installed TYPO3 in the document-root of a host, you should create a domain-record named like 'http: //'. If, on the other hand, your TYPO3-installation is in a different directory, you should create a domain-record named something like 'http: //'. Slashes at the end don't matter that much.

Configure "realurl" to work with "tx_realurl_advanced" ID encoding

Simply set this configuration for the key "pagePath" in the configuration array:

    'pagePath' => array(
        'type' => 'user',
        'userFunc' => 'EXT:realurl/class.tx_realurl_advanced.php:&tx_realurl_advanced->main',
        'spaceCharacter' => '-',
        'languageGetVar' => 'L',
        'expireDays' => 3

The directives specific for "tx_realurl_advanced" are these:

Directive:Data Type:Description:
languageGetVarstringDefines which GET variable in the URL that defines language id; if set the path will take this language value into account and try to generate the path in localized version.
expireDaysintegerThe time the old URL of a page whose pagetitle changed will still be remembered (in days) (Temporarily disabled)
spaceCharacter Normally, this defaults to an underscore (_), which is used to replace spaces and such in an URL. You can set this to e.g. a hyphen (-) if you want to.
segTitleFieldListlist of fieldnamesThe prioritized order of field names from pages table (root line !) that is used when building the speaking URL.

Default is "tx_realurl_pathsegment,alias,nav_title,title" (for Alternative Page Language records this is only "nav_title, title").

Notice: If you specify user defined fields which are not currently in root line you will have to add them to the root line via "$GLOBALS['TYPO3_CONF_VARS']['FE']['addRootLineFields']"
disablePathCachebooleanWill disable the usage of path cache. The system will rely solely on forward-lookups on the fields in "segTitleFieldList".
firstHitPathCachebooleanIf set, then the path-cache look up is accepted only if it returns a result in the first hit. It will do so only when no postVarSets are used for the URL.
rootpage_idintegerDefines the root page uid of the site for which the configuration is made.

This setting is needed if you run multiple sites in the same database using real urls for more than one of the sites. The setting makes it possible for the "path-to-id" algorithms to distiguish the sites. For instance two sites might have the same page title on the first level which will generate the same speaking url path. In such a case two different IDs are associated with the same page path and something is needed to distinguish the look ups.

It is also necessary to set this if the "default" site is not the first page under the page tree root!

In order to configure different sites you simply use the possibility to make different configuration for different domains. See the main section about configuration.

It is important that you use the uid of the root page in the sites. Otherwise the fall-back look up of page paths will not work correctly.

See example below
encodeTitle_userProcuser functionAdditional user processing in "encodeTitle()".
dontResolveShortcutsbooleanIf set, page shortcuts are not resolved to their destination page.
excludePageIdsstringComma list of page ids to exclude from being realurl rendered. The list must NOT contain any spaces between page id numbers!
Example: Configuration of realurl for use on more than one domain in the same database
   1: $TYPO3_CONF_VARS['EXTCONF']['realurl'] = array(
   2:     '_DEFAULT' => array(
   3:         'pagePath' => array(
   4:             'type' => 'user',
   5:             'userFunc' => 'EXT:realurl/class.tx_realurl_advanced.php:&tx_realurl_advanced->main',
   6:             'rootpage_id' => 437
   7:         ),
   8:     ),
   9:     'www.test1.intra' => array(
  10:         'pagePath' => array(
  11:             'type' => 'user',
  12:             'userFunc' => 'EXT:realurl/class.tx_realurl_advanced.php:&tx_realurl_advanced->main',
  13:             'rootpage_id' => 111
  14:         ),
  15:     )
  16: );

Notice how the "rootpage_id" is set differently for these two cases! The _DEFAULT does not work for the later URL settings.

Known problems (class: tx_realurl_advanced)

  • Links to pages with type 'URL' will be generated as though they were normal pages. The link where they point to should be taken actually.
  • Pages that have subpages with the same page-title (actually: that resolve to the same URL) don't work yet. See TODO.

To-Do list (class: tx_realurl_advanced)

  • Extensive testing still has to be done, especially with changing page-titles, deleting a page, then creating another page with the same name, the multiple-domain-stuff, etc.
    • I have RealURL in production and regularily expire pages. The expire code is great, I changed a few page's names a few weeks ago and it still finds the correct PID under the old URL. The same holds for postvars, I changed the postvarset for tt_news once and the old URLs still work. This is great for external links. Only one bug found: When you have a page, e.g. /a/b/, and you delete "b" and afterwards create a new page "b" at the same URL, realurl will fail because the /a/b/ entry in the pathcache still points to the (nonexistant) old page "b". This should be fixed. --Alex
  • There's a function to find the desired language when it's not explicitly given in the URL. It needs some more work to be generally useable though. (Take a look at the source, there's code to figure out the language with the IP2Country database...)
  • The table should be automatically cleaned up once in a while, to remove expired URLs. Haven't looked into this yet. One possibility is to create a cron-job for it, another way is to let a random number decide that it is time to delete all old URLs in the function updateURLCache()
  • I could create an option to enforce usage of the https-protocol quite easily.
  • Making the pathPrefix-detection (the path/to/typo/) more elegant.
  • Fix bugs and rest of TODO-items in source (denoted by // !! TODO !! Bladiebla).

ToDo list

- For technical todo list see "doc/TODO.txt" in the extension

  • Include information on how to set this up with IIS, if possible at all. Can somebody please try this?
  • Some clever 404 page can be created, using the builtin indexed-search for example. The requested language (/nl/...) could help to provide the page in the requested language. Nothing done with this yet though.

User annotations:

Make sure you have

$TYPO3_CONF_VARS['FE']['pageNotFound_handling_statheader'] = 'HTTP/1.0 404 Not Found';

in localconf.php(or ext_localconf.php). If not, search engines will save your "Page Not Found" links.

Ref: http: //

Using alias instead of pagetitle?

Date: 28-03-2004 19:12 by mshigorin

Probably being able to conveniently specify which page parameters to use instead of page title (especially in need for those whose pagetitles don't do 7-bit but aliases can do) could help with more concise URLs without compromising the page titles (which are index-valuable).

Date: 03-03-2004 07:40 by iibscdev

sorry removed ... i've moved comment to RealURL forum/support

Date 12-01-2006 08:27 by allanj

The above message says there is a RealURL forum/support somewhere, could we have a link to it please......


  • 0.0.1: First upload
  • 0.0.4: First working version
  • 0.0.5: An icon was added, thanks to Netcreaters for creating it!
  • 0.0.6: Documentation in StarOffice format added
  • 0.0.7: Almost complete rewrite / revision of the code, implemented '/path_to_typo'-feature, implemented support for multiple domains, changed the code so that most of the configuration is now automatic, updated documentation
  • 0.1.0: First publicly available version
  • 0.1.1: pathPrefix didn't work correctly, so a hack was added to allow it too work now
  • 0.1.2: URLs in pages weren't rendered correctly on single-language-sites
  • 0.1.3: Added possibility to choose another character to replace a space (instead of an underscore), fixed another stupid bug regarding the rendering of some URLs
  • TOTAL re-organization by Kasper Skårhøj march 2004