Extension:ParserFunctions/String functions

From mediawiki.org
Warning Warning: In 2013, it was decided that these functions will never be enabled on any Wikimedia wiki, because they are inefficient when used on a large scale (see phab:T8455 for some history). These functions do NOT work on Wikimedia wikis!

If you are here to write something on a Wikimedia project, you are looking for something else: if your home wiki has string functions, it probably uses Lua . For example, the English Wikipedia uses Module:String, which does some of the same things with wildly different syntax. There are also individual String-handling templates.

The ParserFunctions extension optionally defines various string functions if $wgPFEnableStringFunctions = true; is enabled.

These functions consist of len, pos, rpos, sub, count, replace, explode, and urldecode.

All of these functions operate in O(n) time complexity, making them safe against DoS attacks.

  1. Some parameters of these functions are limited through global settings to prevent abuse. See section Limits hereafter.
  2. For functions that are case sensitive, you may use the magic word {{lc:your_string_here}} as a workaround in some cases.
  3. To determine whether a MediaWiki server enables these functions, check the list of supported Extended parser functions in Special:Version.
  4. String length is limited by $wgPFStringLengthLimit variable, default to 1000.

#len[edit]

The #len function returns the length of the given string. The syntax is:

{{#len:string}}

The return value is always a number of characters in the source string (after expansions of template invocations, but before conversion to HTML). If no string is specified, the return value is zero.

  • This function is safe with UTF-8 multibyte characters. Example:
    • {{#len:Žmržlina}}8
  • Leading and trailing spaces or newlines are not counted, but intermediate spaces and newlines are taken into account. Examples:
    • {{#len:Icecream }}8
    • {{#len: a   b }}5 - 3 spaces between 2 characters
  • Characters given by reference are not converted, but counted according to their source form.
    • {{#len: }}6 - named characters references
    • {{#len: }}5 - numeric characters references, not ignored despite it designates a space here.
  • Tags such as ‎<nowiki> and other tag extensions will always have a length of zero, since their content is hidden from the parser. Example:
    • {{#len:<nowiki>This is a </nowiki>test}}4

#pos[edit]

The #pos function returns the position of a given search term within the string. The syntax is:

{{#pos:string|search term|offset}}

The offset parameter, if specified, tells a starting position where this function should begin searching.

If the search term is found, the return value is a zero-based integer of the first position within the string.

If the search term is not found, the function returns an empty string.

  • This function is case sensitive.
  • The maximum allowed length of the search term is limited through the $wgStringFunctionsLimitSearch global setting.
  • This function is safe with UTF-8 multibyte characters. Example: {{#pos:Žmržlina|žlina}} returns 3.
  • As with #len, ‎<nowiki> and other tag extensions are treated as having a length of 1 for the purposes of character position. Example: {{#pos:<nowiki>This is a </nowiki>test|test}} returns 1.

#rpos[edit]

The #rpos function returns the last position of a given search term within the string. The syntax is:

 {{#rpos:string|search term}}

If the search term is found, the return value is a zero-based integer of its last position within the string.

If the search term is not found, the function returns -1.

When using this to search for the last delimiter, add +1 to the result to retrieve position after the last delimiter. This also works when the delimiter is not found, because "-1 + 1" is zero, which is the beginning of the given value.
  • This function is case sensitive.
  • The maximum allowed length of the search term is limited through the $wgStringFunctionsLimitSearch global setting.
  • This function is safe with UTF-8 multibyte characters. Example: {{#rpos:Žmržlina|lina}} returns 4.
  • As with #len, ‎<nowiki> and other tag extensions are treated as having a length of 1 for the purposes of character position. Example: {{#rpos:<nowiki>This is a </nowiki>test|test}} returns 1.

#sub[edit]

The #sub function returns a substring from the given string. The syntax is:

{{#sub:string|start|length}}

The start parameter, if positive (or zero), specifies a zero-based index of the first character to be returned.

Example: {{#sub:Icecream|3}} returns cream.

{{#sub:Icecream|0|3}} returns Ice.

If the start parameter is negative, it specifies how many characters from the end should be returned.

Example: {{#sub:Icecream|-3}} returns eam.

The length parameter, if present and positive, specifies the maximum length of the returned string.

Example: {{#sub:Icecream|3|3}} returns cre.

If the length parameter is negative, it specifies how many characters will be omitted from the end of the string.

Example: {{#sub:Icecream|3|-3}} returns cr.

If the start parameter is negative, it specifies how many characters from the end should be returned. The length parameter, if present and positive, specifies the maximum length of the returned string from the starting point.

Example: {{#sub:Icecream|-3|2}} returns ea.

  • If the length parameter is zero, it is not used for truncation at all.
    • Example: {{#sub:Icecream|3|0}} returns cream, {{#sub:Icecream|0|3}} returns Ice.
  • If start denotes a position beyond the truncation from the end by negative length parameter, an empty string will be returned.
    • Example: {{#sub:Icecream|3|-6}} returns an empty string.
  • This function is safe with UTF-8 multibyte characters. Example: {{#sub:Žmržlina|3}} returns žlina.
  • As with #len, ‎<nowiki> and other tag extensions are treated as having a length of 1 for the purposes of character position. Example: {{#sub:<nowiki>This is a </nowiki>test|1}} returns test.

#count[edit]

The #count function returns the number of times a given substring appears within the provided text.

{{#count:string|substring}}

#replace[edit]

The #replace function returns the given string with all occurrences of a search term replaced with a replacement term.

{{#replace:string|search term|replacement term}}

If the search term is unspecified or empty, a single space will be searched for.

If the replacement term is unspecified or empty, all occurrences of the search term will be removed from the string.

  • This function is case-sensitive.
  • The maximum allowed length of the search term is limited through the $wgStringFunctionsLimitSearch global setting.
  • The maximum allowed length of the replacement term is limited through the $wgStringFunctionsLimitReplace global setting.
  • Even if the replacement term is a space, an empty string is used. This is a side-effect of the MediaWiki parser. To use a space as the replacement term, put it in nowiki tags.
    • Example: {{#replace:My_little_home_page|_|<nowiki> </nowiki>}} returns My little home page.
    • If this doesn't work, try {{#replace:My_little_home_page|_|<nowiki/> <nowiki/>}} with two self-closing tags.
    • Note that this is the only acceptable use of nowiki in the replacement term, as otherwise nowiki could be used to bypass $wgStringFunctionsLimitReplace, injecting an arbitrarily large number of characters into the output. For this reason, all occurrences of ‎<nowiki> or any other tag extension within the replacement term are replaced with spaces.
  • This function is safe with UTF-8 multibyte characters. Example: {{#replace:Žmržlina|ž|z}} returns Žmrzlina.
  • If multiple items in a single text string need to be replaced, one could also consider Extension:ReplaceSet . It adds a parser function for a sequence of replacements.
Case-insensitive replace

Currently the syntax doesn't provide a switch to toggle case-sensitivity setting. But you may make use of magic words of formatting as a workaround. (e.g. {{lc:your_string_here}}) For example, if you want to remove the word "Category:" from the string regardless of its case, you may type:

{{#replace:{{lc:{{{1}}}}}|category:|}}

But the disadvantage is that the output will become all lower-case. If you want to keep the casing after replacement, you have to use multiple nesting levels (i.e. multiple replace calls) to achieve the same thing.

#explode[edit]

The #explode function splits the given string into pieces and then returns one of the pieces. The pieces are 0-indexed. The syntax is:

{{#explode:string|delimiter|position|limit}}

The delimiter parameter specifies a string to be used to divide the string into pieces. This delimiter string is then not part of any piece, and when two delimiter strings are next to each other, they create an empty piece between them. If this parameter is not specified, a single space is used. The limit parameter is available in ParserFunctions only, not the standalone StringFunctions version, and allows you to limit the number of parts that the value is split into, with all remaining text included in the final part.

The position parameter specifies which piece is to be returned. Pieces are counted from 0. If this parameter is not specified, the first piece is used (piece with number 0). When a negative value is used as position, the pieces are counted from the end. In this case, piece number -1 means the last piece. Examples:

  • {{#explode:And if you tolerate this| |2}} returns you
  • {{#explode:String/Functions/Code|/|-1}} returns Code
  • {{#explode:Split%By%Percentage%Signs|%|2}} returns Percentage
  • {{#explode:And if you tolerate this thing and expect no more| |2|3}} returns you tolerate this thing and expect no more

The return value is the position-th piece. If there are fewer pieces than the position specifies, an empty string is returned.

  • This function is case sensitive.
  • The maximum allowed length of the delimiter is limited through $wgStringFunctionsLimitSearch global setting.
  • This function is safe with UTF-8 multibyte characters. Example: {{#explode:Žmržlina|ž|1}} returns lina.

#urldecode[edit]

#urldecode converts the escape characters from an 'URL encoded' string back to readable text. The syntax is:

{{#urldecode:value}}
  • This function works by directly exposing PHP's urldecode() function.
  • A character-code-reference can be found at www.w3schools.com.
  • The opposite, urlencode, has been integrated into MediaWiki as of version 1.18; for examples, see Help:Magic Words .
  • urldecode was merged from Stringfunctions in 2010, by commit 1b75afd18d3695bdb6ffbfccd0e4aec064785363

Limits[edit]

This module defines three global settings:

These are used to limit some parameters of some functions to ensure the functions operate in O(n) time complexity, and are therefore safe against DoS attacks.

$wgStringFunctionsLimitSearch[edit]

This setting is used by #pos, #rpos, #replace, and #explode. All these functions search for a substring in a larger string while they operate, which can run in O(n*m) and therefore make the software more vulnerable to DoS attacks. By setting this value to a specific small number, the time complexity is decreased to O(n).

This setting limits the maximum allowed length of the string being searched for.

The default value is 30 multibyte characters.

$wgStringFunctionsLimitReplace[edit]

This setting is used by #replace. This function replaces all occurrences of one string for another, which can be used to quickly generate very large amounts of data, and therefore makes the software more vulnerable to DoS attacks. This setting limits the maximum allowed length of the replacing string.

The default value is 30 multibyte characters.

See also[edit]