Help:Extension:ParserFunctions

The ParserFunctions extension provides eleven additional parser functions to supplement the "magic words ", which are already present in MediaWiki. (It may be configured to provide additional parser functions for string handling; these string functions are documented elsewhere .) All the parser functions provided by this extension take the form:

{{#functionname: argument 1 | argument 2 | argument 3 ... }}

#expr

Type	Operators
Grouping (parentheses)	( )
Numbers	1234.5   e (2.718)   pi (3.142)
	binary operator e   unary +,-
Unary	not ceil trunc floor abs exp ln sin cos tan acos asin atan
Binary	^
	* / div mod
	+ -
Round	round
Logic	= != <> > < >= <=
	and
	or

This function evaluates a mathematical expression and returns the calculated value. This function is also available in Scribunto via the mw.ext.ParserFunctions.expr function.

{{#expr: expression }}

The available operators are listed to the right, in order of precedence. See Help:Calculation for more details of the function of each operator. The accuracy and format of the result returned will vary depending on the operating system of the server running the wiki and the number format of the site language.

When evaluating using boolean algebra, zero evaluates to false, and any nonzero value, positive or negative, evaluates to true:

{{#expr: 1 and -1 }} → 1

{{#expr: 1 and 0 }} → 0

{{#expr: 1 or -1 }} → 1

{{#expr: -1 or 0 }} → 1

{{#expr: 0 or 0 }} → 0

An empty input expression returns an empty string. Invalid expressions return one of several error messages, which can be caught using the #iferror function:

{{#expr: }} →

{{#expr: 1+ }} →

{{#expr: 1 = }} →

{{#expr: 1 foo 2 }} →

The order of addition and subtraction operands before or after a number is meaningful and may be treated as a positive or negative value instead of as an operand with an erroneous input:

{{#expr: +1 }} → 1

{{#expr: -1 }} → -1

{{#expr: + 1 }} → 1

{{#expr: - 1 }} → -1

Note, if using the output of magic words, you must raw-format them in order to remove commas and translate the numerals. For example, {{NUMBEROFUSERS}} results in 17,464,807, where we want 17464807, which can be obtained using {{formatnum :{{NUMBEROFUSERS}}|R}}. This is especially important in some languages, where numerals are translated. For example, in Bengali, {{NUMBEROFUSERS}} produces ৩০,০৬১.

{{#expr:{{NUMBEROFUSERS}}+100}} →

{{#expr:{{formatnum:{{NUMBEROFUSERS}}|R}}+100}} → 17464907

Warning:

The operator mod gives wrong results for some values of the second argument: {{#expr: 123 mod (2^64-1)}} → (produces an empty string; should be 123)

If you want to do calculations based on dates (ex. test whether current date and time is after some other date and time), first convert the time to number of seconds after January 1, 1970 using {{#time: xNU }}, then you can simply add and subtract dates as numbers.

Rounding

Rounds off the number on the left to a multiple of 1/10 raised to a power, with the exponent equal to the truncated value of the number given on the right.

To round up or down use unary ceil or floor respectively.

Test case	Result	Method of rounding
{{#expr: 1/3 round 5 }}	0.33333	Final digit is < 5, so no apparent rounding occurs
{{#expr: 1/6 round 5 }}	0.16667	Final digit is ≥ 5, so it is rounded up
{{#expr: 8.99999/9 round 5 }}	1	Again, the result is rounded up on the last digit, which results in additional rounding
{{#expr: 1234.5678 round -2 }}	1200	Rounded to nearest 100 because negative values round to the left of the decimal point
{{#expr: 1234.5678 round 2 }}	1234.57	Rounded to nearest 100th because positive values round to the right of the decimal point
{{#expr: 1234.5678 round 2.3 }}	1234.57	Decimals in the rounding index make no difference in the rounded result
{{#expr: trunc 1234.5678 }}	1234	Decimal portion truncated (chopped off)
Rounding to the nearest integer
{{#expr: 1/3 round 0 }}	0	Down to the nearest integer, which is zero
{{#expr: 1/2 round 0 }}	1	Up to the nearest integer, which is one
{{#expr: 3/4 round 0 }}	1	Up to the nearest integer, which is one
{{#expr: -1/3 round 0 }}	-0	Up to the nearest integer, which is zero
{{#expr: -1/2 round 0 }}	-1	Down to the nearest integer, which is negative one
{{#expr: -3/4 round 0 }}	-1	Down to the nearest integer, which is negative one
Rounding up or down with ceil and floor
{{#expr: ceil(1/3) }}	1	Up to the next larger integer, which is one
{{#expr: floor(1/3) }}	0	Down to the next smaller integer, which is zero
{{#expr: ceil(-1/3) }}	-0	Up to the next larger integer, which is zero
{{#expr: floor(-1/3) }}	-1	Down to the next smaller integer, which is negative one
{{#expr: ceil 1/3 }}	0.33333333333333	Not rounded, since 1 already is an integer Warning: Interpreted as (ceil 1)/3, not ceil(1/3), as you might expect

Strings

Expressions only work with number-like values, they cannot compare strings or characters. #ifeq can be used instead.

{{#expr: "a" = "a" }} →

{{#expr: a = a }} →

{{#ifeq: a | a | 1 | 0 }} → 1

#if

This function evaluates a test string and determines whether or not it is empty. A test string containing only white space is considered to be empty.

{{#if: test string | value if test string is not empty | value if test string is empty (or only white space) }}

{{#if: first parameter | second parameter | third parameter }}

This function first tests whether the first parameter is not empty. If the first parameter is not empty the function displays the second argument. If the first parameter is empty or contains only whitespace characters (spaces, newlines, etc.) it displays the third argument.

{{#if: | yes | no}} → no

{{#if: string | yes | no}} → yes

{{#if:      | yes | no}} → no

{{#if:


| yes | no}} → no

The test string is always interpreted as pure text, so mathematical expressions are not evaluated:

{{#if: 1==2 | yes | no }} → yes

{{#if: 0 | yes | no }} → yes

The last parameter (false) may be omitted:

{{#if: foo | yes }} → yes

{{#if: | yes }} →

{{#if: foo | | no}} →

The function may be nested. To do so, nest the inner #if function in its full form in place of a parameter of the enclosing #if function. Up to seven levels of nesting is possible, although that may depend on the wiki or a memory limit.

{{#if: test string | value if test string is not empty | {{#if: test string | value if test string is not empty | value if test string is empty (or only white space) }} }}

You can also use a parameter as the test string in your #if statement. You need to ensure you add the | (pipe symbol) after the name of the variable. (So that if the parameter does not have a value, it evaluates to an empty string instead of the string "{{{1}}}".)

{{#if:{{{1|}}}|You entered text in variable 1|There is no text in variable 1 }}

See Help:Parser functions in templates for more examples of this parser function.

#ifeq

This parser function compares two input strings, determines whether they are identical, and returns one of two strings based on the result. If more comparisons and output strings are required, consider using #switch.

{{#ifeq: string 1 | string 2 | value if identical | value if different }}

If both strings are valid numerical values, the strings are compared numerically:

{{#ifeq: 01 | 1 | equal | not equal}} → equal

{{#ifeq: 0 | -0 | equal | not equal}} → equal

{{#ifeq: 1e3 | 1000 | equal | not equal}} → equal

{{#ifeq: {{#expr:10^3}} | 1000 | equal | not equal}} → equal

Otherwise the comparison is made as text; this comparison is case sensitive:

{{#ifeq: foo | bar | equal | not equal}} → not equal

{{#ifeq: foo | Foo | equal | not equal}} → not equal

{{#ifeq: "01" | "1" | equal | not equal}} → not equal  (compare to similar example above, without the quotes)

{{#ifeq: 10^3 | 1000 | equal | not equal}} → not equal  (compare to similar example above, with #expr returning a valid number first)

As a practical example, consider an existing template Template:Timer using the parser to choose between two standard times, short and long. It takes the parameter as the first input to compare against the string "short" – there is no convention for the order, but it is simpler to read if the parameter goes first. The template code is defined as:

{{#ifeq: {{{1|}}} | short | 20 | 40 }}

the following ensue:

{{timer|short}} → 20

{{timer|20}} → 40

{{timer}} → 40

Warning:

#ifexpr does not report equivalent numerical comparisons with the #ifeq and #switch parsers. These latter two are more accurate than #ifexpr, and do not return equivalent results. Consider these comparison with the final digit changed: {{#ifeq: 12345678901234567 | 12345678901234568 | equal | not equal}} → not equal {{#switch: 12345678901234567 | 12345678901234568 = equal | not equal}} → not equal Because PHP used in #ifeq and #switch compares two numbers of type integer, it returns the expected result correctly. Whereas with #ifexpr and the same numbers: {{#ifexpr: 12345678901234567 = 12345678901234568 | equal | not equal}} → equal With the different digit, the result of equal is actually incorrect. This behaviour in #ifexpr is caused because MediaWiki converts literal numbers in expressions to type float, which, for large integers like these, involves rounding.

Warning:

When used inside a parser function, any parser tags and other parser functions must be temporarily replaced with a unique code . This affects comparisons: {{#ifeq: <nowiki>foo</nowiki> | <nowiki>foo</nowiki> | equal | not equal}} → not equal {{#ifeq: <math>foo</math> | <math>foo</math> | equal | not equal}} → not equal {{#ifeq: {{#tag:math|foo}} | {{#tag:math|foo}} | equal | not equal}} → not equal {{#ifeq: [[foo]] | [[foo]] | equal | not equal}} → equal If the strings to be compared are given as equal calls to the same template containing such tags, then the condition is true, but in the case of two templates with identical content containing such tags it is false.

Warning:

Literal comparisons to page-name magic words may fail depending on site configuration. For example, {{FULLPAGENAME}}, depending on wiki, may capitalize the first letter, and will replace all underscores with spaces. To work around this, apply the magic word to both parameters: {{#ifeq: {{FULLPAGENAME: L'Aquila}} | {{FULLPAGENAME}} | equal | not equal}} → equal

#iferror

This function takes an input string and returns one of two results; the function evaluates to true if the input string contains an HTML object with class="error", as generated by other parser functions such as #expr, #time and #rel2abs, template errors such as loops and recursions, and other "failsoft" parser errors.

{{#iferror: test string | value if error | value if correct }}

One or both of the return strings can be omitted. If the correct string is omitted, the test string is returned if it is not erroneous. If the error string is also omitted, an empty string is returned on an error:

{{#iferror: {{#expr: 1 + 2 }} | error | correct }} → correct

{{#iferror: {{#expr: 1 + X }} | error | correct }} → error

{{#iferror: {{#expr: 1 + 2 }} | error }} → 3

{{#iferror: {{#expr: 1 + X }} | error }} → error

{{#iferror: {{#expr: 1 + 2 }} }} → 3

{{#iferror: {{#expr: 1 + X }} }} → ‌

{{#iferror: {{#expr: . }} | error | correct }} → correct

{{#iferror: <strong class="error">a</strong> | error | correct }} → error

#ifexpr

This function evaluates a mathematical expression and returns one of two strings depending on the boolean value of the result:

{{#ifexpr: expression | value if true | value if false }}

The expression input is evaluated exactly as for #expr above, with the same operators being available. The output is then evaluated as a boolean expression.

An empty input expression evaluates to false:

{{#ifexpr: | yes | no}} → no

As mentioned above, zero evaluates to false and any nonzero value evaluates to true, so this function is equivalent to one using #ifeq and #expr only:

{{#ifeq: {{#expr: expression }} | 0 | value if false | value if true }}

except for an empty or wrong input expression (an error message is treated as an empty string; it is not equal to zero, so we get value if true).

{{#ifexpr: = | yes | no }} →

comparing

{{#ifeq: {{#expr: = }} | 0 | no | yes }} → yes

Either or both the return values may be omitted; no output is given when the appropriate branch is left empty:

{{#ifexpr: 1 > 0 | yes }} → yes

{{#ifexpr: 1 < 0 | yes }} →

{{#ifexpr: 0 = 0 | yes }} → yes

{{#ifexpr: 1 > 0 | | no}} →

{{#ifexpr: 1 < 0 | | no}} → no

{{#ifexpr: 1 > 0 }} →

#ifexist

This function takes an input string, interprets it as a page title, and returns one of two values depending on whether or not the page exists on the local wiki.

{{#ifexist: page title | value if exists | value if doesn't exist }}

The function evaluates to true if the page exists, whether it contains content, is visibly blank (contains meta-data such as category links or magic words , but no visible content), is blank, or is a redirect . Only pages that are redlinked evaluate to false, including if the page used to exist but has been deleted.

{{#ifexist: Help:Extension:ParserFunctions | exists | doesn't exist }} → exists

{{#ifexist: XXHelp:Extension:ParserFunctionsXX | exists | doesn't exist }} → doesn't exist

The function evaluates to true for system messages that have been customized, and for special pages that are defined by the software.

{{#ifexist: Special:Watchlist | exists | doesn't exist }} → exists

{{#ifexist: Special:CheckUser | exists | doesn't exist }} → exists (because the Checkuser extension is installed on this wiki)

{{#ifexist: MediaWiki:Copyright | exists | doesn't exist }} → exists (because MediaWiki:Copyright has been customized)

If a page checks a target using #ifexist:, then that page will appear in the Special:WhatLinksHere list for the target page. So if the code {{#ifexist:Foo }} were included live on this page (Help:Extension:ParserFunctions), Special:WhatLinksHere/Foo will list Help:Extension:ParserFunctions.

On wikis using a shared media repository, #ifexist: can be used to check if a file has been uploaded to the repository but not to the wiki itself:

{{#ifexist: File:Example.png | exists | doesn't exist }} → doesn't exist

{{#ifexist: Image:Example.png | exists | doesn't exist }} → doesn't exist

{{#ifexist: Media:Example.png | exists | doesn't exist }} → exists

If a local description page has been created for the file, the result is exists for all of the above.

#ifexist: does not work with interwiki links.

ifexist limits

#ifexist: is considered an "expensive parser function"; only a limited number of which can be included on any one page (including functions inside transcluded templates). When this limit is exceeded, any further #ifexist: functions automatically return false, whether the target page exists or not, and the page is categorized into Category:Pages with too many expensive parser function calls. The name of the tracking category may vary depending on the content language of your wiki.

For some use cases it is possible to emulate the ifexist effect with css, by using the selectors a.new (to select links to unexisting pages) or a:not(.new) (to select links to existing pages). Furthermore, since the number of expensive parser functions that can be used on a single page is controlled by $wgExpensiveParserFunctionLimit , one can also increase the limit in LocalSettings.php if needed.

ifexist and wanted pages

A page that does not exist and is tested for using #ifexist will end up on the Wanted Pages.

#rel2abs

This function converts a relative file path into an absolute filepath.

{{#rel2abs: path }}

{{#rel2abs: path | base path }}

Within the path input, the following syntax is valid:

• . → the current level
• .. → go up one level
• /foo → go down one level into the subdirectory /foo

If the base path is not specified, the full page name of the page will be used instead:

{{#rel2abs: /quok | Help:Foo/bar/baz }} → Help:Foo/bar/baz/quok

{{#rel2abs: ./quok | Help:Foo/bar/baz }} → Help:Foo/bar/baz/quok

{{#rel2abs: ../quok | Help:Foo/bar/baz }} → Help:Foo/bar/quok

{{#rel2abs: ../. | Help:Foo/bar/baz }} → Help:Foo/bar

Invalid syntax, such as /. or /./, is ignored. Since no more than two consecutive full stops are permitted, sequences such as these can be used to separate successive statements:

{{#rel2abs: ../quok/. | Help:Foo/bar/baz }} → Help:Foo/bar/quok

{{#rel2abs: ../../quok | Help:Foo/bar/baz }} → Help:Foo/quok

{{#rel2abs: ../../../quok | Help:Foo/bar/baz }} → quok

{{#rel2abs: ../../../../quok | Help:Foo/bar/baz }} →

#switch

See also : w:Help:Switch parser function

This function compares one input value against several test cases, returning an associated string if a match is found.

{{#switch: comparison string
 | case = result
 | case = result
 | ...
 | case = result
 | default result
}}

Examples:

{{#switch: baz | foo = Foo | baz = Baz | Bar }} → Baz

{{#switch: foo | foo = Foo | baz = Baz | Bar }} → Foo

{{#switch: zzz | foo = Foo | baz = Baz | Bar }} → Bar

#switch with partial transclusion tags can affect a configuration file that enables an editor unfamiliar with template coding to view and edit configurable elements.

Default

The default result is returned if no case string matches the comparison string:

{{#switch: test | foo = Foo | baz = Baz | Bar }} → Bar

In this syntax, the default result must be the last parameter and must not contain a raw equals sign (an equals sign without {{}}). If it does, it will be treated as a case comparison, and no text will display if no cases match. This is because the default value has not been defined (is empty). If a case matches however, its associated string will be returned.

{{#switch: test | Bar | foo = Foo | baz = Baz }} →

{{#switch: test | foo = Foo | baz = Baz | B=ar }} →

{{#switch: test | test = Foo | baz = Baz | B=ar }} → Foo

Alternatively, the default result may be explicitly declared with a case string of "#default".

{{#switch: comparison string
 | case = result
 | case = result
 | ...
 | case = result
 | #default = default result
}}

Default results declared in this way may be placed anywhere within the function:

{{#switch: test | foo = Foo | #default = Bar | baz = Baz }} → Bar

If the default parameter is omitted and no match is made, no result is returned:

{{#switch: test | foo = Foo | baz = Baz }} →

Grouping results

It is possible to have 'fall through' values, where several case strings return the same result string. This minimizes duplication.

{{#switch: comparison string
 | case1 = result1
 | case2 
 | case3 
 | case4 = result234
 | case5 = result5
 | case6 
 | case7 = result67
 | #default = default result
}}

Here cases 2, 3 and 4 all return result234; cases 6 and 7 both return result67. The "#default = " in the last parameter may be omitted in the above case.

Use with parameters

The function may be used with parameters as the test string. In this case, it is not necessary to place the pipe after the parameter name, because it is very unlikely that you will choose to set a case to be the string "{{{parameter name}}}". (This is the value the parameter will default to if the pipe is absent and the parameter doesn't exist or have a value. See Help:Parser functions in templates .)

{{#switch: {{{1}}} | foo = Foo | baz = Baz | Bar }}

In the above case, if {{{1}}} equals foo, the function will return Foo. If it equals baz, the function will return Baz. If the parameter is empty or does not exist, the function will return Bar.

As in the section above, cases can be combined to give a single result.

{{#switch: {{{1}}} | foo | zoo | roo = Foo | baz = Baz | Bar }}

Here, if {{{1}}} equals foo, zoo or roo, the function will return Foo. If it equals baz, the function will return Baz. If the parameter is empty or does not exist, the function will return Bar.

Additionally, the default result can be omitted if you do not wish to return anything if the test parameter value does not match any of the cases.

{{#switch: {{{1}}} | foo = Foo | bar = Bar }}

In this case, the function returns an empty string unless {{{1}}} exists and equals foo or bar, in which case it returns Foo or Bar, respectively.

This has the same effect as declaring the default result as empty.

{{#switch: {{{1}}} | foo | zoo | roo = Foo | baz = Baz | }}

If for some reason you decide to set a case as "{{{parameter name}}}", the function will return that case's result when the parameter doesn't exist or doesn't have a value. The parameter would have to exist and have a value other than the string "{{{parameter name}}}" to return the function's default result.

(when {{{1}}} doesn't exist or is empty):

{{#switch: {{{1}}} | {{{1}}} = Foo | baz = Baz | Bar }} → Foo

(when {{{1}}} has the value "test"):

{{#switch: {{{1}}} | {{{1}}} = Foo | baz = Baz | Bar }} → Bar

(when {{{1}}} has the value "{{{1}}}"):

{{#switch: {{{1}}} | {{{1}}} = Foo | baz = Baz | Bar }} → Foo

In this hypothetical case, you would need to add the pipe to the parameter ({{{1|}}}).

Comparison behavior

As with #ifeq, the comparison is made numerically if both the comparison string and the case string being tested are numeric; or as a case-sensitive string otherwise:

{{#switch: 0 + 1 | 1 = one | 2 = two | three}} → three

{{#switch: {{#expr: 0 + 1}} | 1 = one | 2 = two | three}} → one

{{#switch: a | a = A | b = B | C}} → A

{{#switch: A | a = A | b = B | C}} → C

A case string may be empty:

{{#switch: | = Nothing | foo = Foo | Something }} → Nothing

Once a match is found, subsequent cases are ignored:

{{#switch: b | f = Foo | b = Bar | b = Baz | }} → Bar

Warning:

Numerical comparisons with #switch and #ifeq are not equivalent to comparisons in expressions (see also above): {{#switch: 12345678901234567 | 12345678901234568 = A | B}} → B {{#ifexpr: 12345678901234567 = 12345678901234568 | A | B}} → A

Raw equal signs

"Case" strings cannot contain raw equals signs. To work around this, create a template {{=}} containing a single equals sign: =, or replace equals sign with html code =.

Example:

{{#switch: 1=2

 | 1=2 = raw

 | 1<nowiki>=</nowiki>2 = nowiki

 | 1{{=}}2 = template

 | default

}} → template

{{#switch: 1=2

 | 1=2 = html

 | default

}} → html

For a simple real life example of the use of this function, check Template:NBA color. Two complex examples can be found at Template:Extension and w:Template:BOTREQ.

Replacing #ifeq

#switch can be used to reduce expansion depth.

For example:

• {{#switch:{{{1}}} |condition1=branch1 |condition2=branch2 |condition3=branch3 |branch4}}

is equivalent to

• {{#ifeq:{{{1}}}|condition1 |branch1 |{{#ifeq:{{{1}}}|condition2 |branch2 |{{#ifeq:{{{1}}}|condition3 |branch3 |branch4}}}}}}

i.e. deep nesting, linear:

{{#ifeq:{{{1}}}|condition1
  |<!--then-->branch1
  |<!--else-->{{#ifeq:{{{1}}}|condition2
                |<!--then-->branch2
                |<!--else-->{{#ifeq:{{{1}}}|condition3
                              |<!--then-->branch3
                              |<!--else-->branch4}}}}}}

On the other hand, the switch replacement could be complicated/impractical for IFs nested in both branches (shown with alternatives of indentation, indented on both sides), making full symmetrical tree:

{{#ifeq:{{{1}}}|condition1
 |<!--then-->branch1t{{
  #ifeq:{{{1}}}|condition2
   |<!--then-->branch1t2t{{#ifeq:{{{1}}}|condition4|<!--then-->branch1t2t4t|<!--else-->branch1t2t4e}}
   |<!--else-->branch1t2e{{#ifeq:{{{1}}}|condition5|<!--then-->branch1t2e5t|<!--else-->branch1t2e5e}}
  }}
 |<!--else-->branch1e{{#ifeq:{{{1}}}|condition3
   |<!--then-->branch1e3t{{#ifeq:{{{1}}}|condition6|branch1e3t6t|branch1e3t6e}}
   |<!--else-->branch1e3e{{
    #ifeq:{{{1}}}|condition7
     |branch1e3e7t
     |branch1e3e7t
    }}
  }}
}}

#time

Code	Description	Current output
Year
Y	4-digit year.	2021
y	2-digit year.	21
L	1 if it's a leap year, 0 if not.	0
o	ISO-8601 year of the specified week.	2021
Month
n	Month index, not zero-padded.	4
m	Month index, zero-padded.	04
M	An abbreviation of the month name, in the site language.	Apr
F	The full month name in the site language.	April
xg	Output the full month name in the genitive form for site languages that distinguish between genitive and nominative forms. This option is useful for many Slavic languages like Polish, Russian, Belarusian, Czech, Slovak, Slovene, Ukrainian, etc.	For Polish: {{#time:F Y|June 2010|pl}} → czerwiec 2010 (nominative) {{#time:d xg Y|20 June 2010|pl}} → 20 czerwca 2010 (genitive)
Day of the month or the year
j	Day of the month, not zero-padded.	11
d	Day of the month, zero-padded.	11
z	Day of the year (January 1 = 0).  Note: To get the ISO day of the year add 1.	100
Week and day of the week
W	ISO 8601 week number, zero-padded.	14
N	ISO 8601 day of the week (Monday = 1, Sunday = 7).	7
w	Number of the day of the week (Sunday = 0, Saturday = 6).	0
D	An abbreviation for the day of the week. Rarely internationalized.	Sun
l	The full weekday name. Rarely internationalized.	Sunday
Hour
a	"am" during the morning (00:00:00 → 11:59:59), "pm" otherwise (12:00:00 → 23:59:59).	pm
A	Uppercase version of a above.	PM
g	Hour in 12-hour format, not zero-padded.	7
h	Hour in 12-hour format, zero-padded.	07
G	Hour in 24-hour format, not zero-padded.	19
H	Hour in 24-hour format, zero-padded.	19
Minutes and seconds
i	Minutes past the hour, zero-padded.	38
s	Seconds past the minute, zero-padded.	23
U	Unix time. Seconds since January 1 1970 00:00:00 GMT.	1618169903
Timezone (as of 1.22wmf2)
e	Timezone identifier.	UTC
I	Whether or not the date is in daylight savings time.	0
O	Difference to Greenwich time (GMT)	+0000
P	Difference to Greenwich time (GMT), with colon	+00:00
T	Timezone abbreviation.	UTC
Z	Timezone offset in seconds.	0
Miscellaneous
t	Number of days in the current month.	30
c	ISO 8601 formatted date, equivalent to Y-m-d"T"H:i:s+00:00.	2021-04-11T19:38:23+00:00
r	RFC 5322 formatted date, equivalent to D, j M Y H:i:s +0000, with weekday name and month name not internationalized.	Sun, 11 Apr 2021 19:38:23 +0000
Non-Gregorian calendars
Islamic
xmj	Day of the month.	28
xmF	Full month name.	Sha'aban
xmn	Month index.	8
xmY	Full year.	1442
Iranian (Jalaly)
xit	Number of days in the month.	31
xiz	Day of the year.	21
xij	Day of the month.	22
xiF	Full month name.	Farvardin
xin	Month index.	1
xiY	Full year.	1400
xiy	2-digit year.	00
Hebrew
xjj	Day of the month.	29
xjF	Full month name.	Nisan
xjt	Number of days in month.	30
xjx	Genitive form of the month name.	Nisan
xjn	Month number.	7
xjY	Full year.	5781
Thai solar
xkY	Full year in Thai solar calendar.  Note: For years before 1941 the dates in Jan-Mar range are not calculated properly.	2564
Minguo/Juche year
xoY	Full year.	110
Japanese nengo
xtY	Full year.	令和3
Flags
xn	Format the next numeric code as a raw ASCII number.	In the Hindi language, {{#time:H, xnH}} produces ०६, 06.
xN	Like xn, but as a toggled flag, which endures until the end of the string or until the next appearance of xN in the string.
xr	Format the next number as a roman numeral. Only works for numbers up to 10,000 (up to 3,000 in pre MediaWiki 1.20).	{{#time:xrY}} → MMXXI
xh	Format the next number as a Hebrew numeral.	{{#time:xhY}} → ב'כ"א

This parser function takes a date and/or time (in the Gregorian calendar) and formats it according to the syntax given. A date/time object can be specified; the default is the value of the magic word {{CURRENTTIMESTAMP}} – that is, the time the page was last rendered into HTML.

{{#time: format string }}

{{#time: format string | date/time object }}

{{#time: format string | date/time object | language code }}

{{#time: format string | date/time object | language code | local }}

The list of accepted formatting codes is given in the table to the right. Any character in the formatting string that is not recognized is passed through unaltered; this applies also to blank spaces (the system does not need them for interpreting the codes). There are also two ways to escape characters within the formatting string:

1. A backslash followed by a formatting character is interpreted as a single literal character
2. Characters enclosed in double quotes are considered literal characters, and the quotes are removed.

In addition, the digraph xx is interpreted as a single literal "x".

{{#time: Y-m-d }} → 2021-04-11

{{#time: [[Y]] m d }} → 2021 04 11

{{#time: [[Y (year)]] }} → 2021 (21UTCpmSun, 11 Apr 2021 19:38:23 +0000)

{{#time: [[Y "(year)"]] }} → 2021 (year)

{{#time: i's" }} → 38'23"

The date/time object can be in any format accepted by PHP's strtotime() function. Both absolute (eg 20 December 2000) and relative (eg +20 hours) times are accepted.

{{#time: r|now}} → Sun, 11 Apr 2021 19:38:23 +0000

{{#time: r|+2 hours}} → Sun, 11 Apr 2021 21:38:23 +0000

{{#time: r|now + 2 hours}} → Sun, 11 Apr 2021 21:38:23 +0000

{{#time: r|20 December 2000}} → Wed, 20 Dec 2000 00:00:00 +0000

{{#time: r|December 20, 2000}} → Wed, 20 Dec 2000 00:00:00 +0000

{{#time: r|2000-12-20}} → Wed, 20 Dec 2000 00:00:00 +0000

{{#time: r|2000 December 20}} →

The language code in ISO 639-3 (?) allows the string to be displayed in the chosen language

{{#time:d F Y|1988-02-28|nl}} → 28 februari 1988

{{#time:l|now|uk}} → неділя

{{#time:d xg Y|20 June 2010|pl}} → 20 czerwca 2010

The local parameter specifies if the date/time object refers to the local timezone or to UTC.

This is a boolean parameters: its value is determined by casting the value of the argument (see the official PHP documentation for details on how string are cast to boolean values).

Please note that, if the variable $wgLocaltimezone is set to UTC, there is no difference in the output when local is set to true or false.

See the following examples for details:

{{#time: Y F d H:i:s|now|it|0}} → 2021 aprile 11 19:38:23

{{#time: Y F d H:i:s|now|it|1}} → 2021 aprile 11 19:38:23

{{#time: Y F d H:i:s|+2 hours||0}} → 2021 April 11 21:38:23

{{#time: Y F d H:i:s|+2 hours||1}} → 2021 April 11 21:38:23

{{#time:c|2019-05-16T17:05:43+02:00|it}} → 2019-05-16T15:05:43+00:00

{{#time:c|2019-05-16T17:05:43+02:00|it|0}} → 2019-05-16T15:05:43+00:00

{{#time:c|2019-05-16T17:05:43+02:00|it|true}} → 2019-05-16T15:05:43+00:00

If you've calculated a Unix timestamp, you may use it in date calculations by pre-pending an @ symbol.

{{#time: U | now }} → 1618169903

{{#time: r | @1618169903 }} → Sun, 11 Apr 2021 19:38:23 +0000

Warning:

Without the @ prefix before numeric timestamp values, the result is an error most of the time, or is an unexpected value: {{#time: r | 1970-01-01 00:16:39 }} → Thu, 01 Jan 1970 00:16:39 +0000 {{#time: U | 1970-01-01 00:16:39 }} → 999 {{#time: r | @999 }} → Thu, 01 Jan 1970 00:16:39 +0000 (correct) {{#time: r | 999 }} → (unsupported year format) {{#time: r | 1970-01-01 00:16:40 }} → Thu, 01 Jan 1970 00:16:40 +0000 {{#time: U | 1970-01-01 00:16:40 }} → 1000 {{#time: r | @1000 }} → Thu, 01 Jan 1970 00:16:40 +0000 (correct) {{#time: r | 1000 }} → Fri, 11 Apr 1000 00:00:00 +0000 (interpreted as a year with current month and day of the month) {{#time: r | 1970-01-01 02:46:39 }} → Thu, 01 Jan 1970 02:46:39 +0000 {{#time: U | 1970-01-01 02:46:39 }} → 9999 {{#time: r | @9999 }} → Thu, 01 Jan 1970 02:46:39 +0000 (correct) {{#time: r | 9999 }} → Sun, 11 Apr 9999 00:00:00 +0000 (interpreted as a year with current month and day of the month) {{#time: r | 1970-01-01 02:46:40 }} → Thu, 01 Jan 1970 02:46:40 +0000 {{#time: U | 1970-01-01 02:46:40 }} → 10000 {{#time: r | @10000 }} → Thu, 01 Jan 1970 02:46:40 +0000 (correct) {{#time: r | 10000 }} → (unsupported year format)

Warning:

The range of acceptable input is 1 January 0111 → 31 December 9999. For the years 100 through 110 the output is inconsistent, Y and leap years are like the years 100-110, r, D, l and U are like interpreting these years as 2000-2010. {{#time: d F Y | 29 Feb 0100 }} → 01 March 0100 (correct, no leap year), but {{#time: r | 29 Feb 0100 }} → Mon, 01 Mar 0100 00:00:00 +0000 (wrong, even if 100 is interpreted as 2000, because that is a leap year) {{#time: d F Y | 15 April 10000 }} → {{#time: r | 10000-4-15 }} → Sat, 15 Apr 2000 10:00:00 +0000 Year numbers 0-99 are interpreted as 2000-2069 and 1970-1999, except when written in 4-digit format with leading zeros: {{#time: d F Y | 1 Jan 6 }} → 01 January 2006 {{#time: d F Y | 1 Jan 06 }} → 01 January 2006 {{#time: d F Y | 1 Jan 006 }} → 01 January 2006 {{#time: d F Y | 1 Jan 0006 }} → 01 January 0006 (4-digit format) The weekday is supplied for the years 100-110 and from 1753, for the years 111-1752 the r-output shows "Unknown" and the l-output "<>". As a consequence, the r-output is not accepted as input for these years.

Full or partial absolute dates can be specified; the function will "fill in" parts of the date that are not specified using the current values:

{{#time: Y | January 1 }} → 2021

Warning:

The fill-in feature is not consistent; some parts are filled in using the current values, others are not: {{#time: Y m d H:i:s | June }} → 2021 06 11 00:00:00 Gives the start of the day, but the current day of the month and the current year. {{#time: Y m d H:i:s | 2003 }} → 2003 04 11 00:00:00 Gives the start of the day, but the current day of the year. There's exception case of the filled day: {{#time: Y m d H:i:s | June 2003 }} → 2003 06 01 00:00:00 Gives the start of the day and the start of the month.

A four-digit number is always interpreted as a year, never as hours and minutes:

{{#time: Y m d H:i:s | 1959 }} → 1959 04 11 00:00:00

A six-digit number is interpreted as hours, minutes and seconds if possible, but otherwise as an error (not, for instance, a year and month):

{{#time: Y m d H:i:s | 195909 }} → 2021 04 11 19:59:09 Input is treated as a time rather than a year+month code.

{{#time: Y m d H:i:s | 196009 }} → Although 19:60:09 is not a valid time, 196009 is not interpreted as September 1960.

The function performs a certain amount of date mathematics:

{{#time: d F Y | January 0 2008 }} → 31 December 2007

{{#time: d F | January 32 }} →

{{#time: d F | February 29 2008 }} → 29 February

{{#time: d F | February 29 2007 }} → 01 March

{{#time:Y-F|now -1 months}} → 2021-March

The total length of the format strings of the calls of #time is limited to 6000 characters.

Time Zone issue

There is a bug in this #time parser function (more specifically in PHP DateTime) that does not allow the passing-in of non-integers as relative time zone offsets. This issue does not apply when using an on-the-hour time zone, such as EDT. For example:

• {{#time:g:i A | -4 hours }} → 3:38 PM

However, India is on a +5.5 hours time offset from UTC, and thus using its time zone will not normally allow the correct calculation of a relative time zone offset. Here's what happens:

• {{#time:g:i A | +5.5 hours }} → 7:38 PM

To workaround this issue, simply convert the time into minutes or seconds, like this:

• {{#time:g:i A | +330 minutes }} → 1:08 AM
• {{#time:g:i A | +19800 seconds }} → 1:08 AM

(Tim Starling, the developer of this function, provided the exact syntax for this solution.)

#timel

This function is identical to {{#time: ... }}, when the local parameter is set to true, so it always uses the local time of the wiki (as set in $wgLocaltimezone ).

Syntax of the function is:

{{#timel: format string }}

{{#timel: format string | date/time object }}

{{#timel: format string | date/time object | language code }}

Please note that, if the variable $wgLocaltimezone is set to UTC, there is no difference in the output when local is set to true or false

Example of the use of #time and #timel parser functions from a server where the timezone is not UTC

For instance, see the following examples:

{{#time:c|now|it}} → 2021-04-11T19:38:24+00:00

{{#time:c|now|it|0}} → 2021-04-11T19:38:24+00:00

{{#time:c|now|it|1}} → 2021-04-11T19:38:24+00:00

{{#timel:c|now|it}} → 2021-04-11T19:38:24+00:00

Warning Example from https://no.wikipedia.org/wiki/Maldiskusjon:Sommertid

Warning:

Be aware that U for both time and timel will return the same number of seconds since 1970-01-01 00:00:00 UTC on Wikipedias with different timezones than UTC (formerly known as GMT) U Unix time. Seconds since January 1 1970 00:00:00 GMT. Z Timezone offset in seconds. {{#time: U}} → 1618169903 {{#timel: U}} → 1618169903 {{#time: Z}} → 0 {{#timel: Z}} → 0

#titleparts

This function separates a page title into segments based on slashes, then returns some of those segments as output.

{{#titleparts: pagename | number of segments to return | first segment to return }}

If the number of segments to return parameter is not specified, it defaults to "0", which returns all the segments from the first segment to return (included). If the first segment to return parameter is not specified or is "0", it defaults to "1":

{{#titleparts: Talk:Foo/bar/baz/quok }} → Talk:Foo/bar/baz/quok

{{#titleparts: Talk:Foo/bar/baz/quok | 1 }} → Talk:Foo See also {{ROOTPAGENAME }}.

{{#titleparts: Talk:Foo/bar/baz/quok | 2 }} → Talk:Foo/bar

{{#titleparts: Talk:Foo/bar/baz/quok | 2 | 2 }} → bar/baz

{{#titleparts: Talk:Foo/bar/baz/quok | | 2 }} → bar/baz/quok

{{#titleparts: Talk:Foo/bar/baz/quok | | 5 }} →

Negative values are accepted for both values. Negative values for the number of segments to return parameter effectively 'strips' segments from the end of the string. Negative values for the first segment to return translates to "start with this segment counting from the right":

{{#titleparts: Talk:Foo/bar/baz/quok | -1 }} → Talk:Foo/bar/baz Strips one segment from the end of the string. See also {{BASEPAGENAME}}.

{{#titleparts: Talk:Foo/bar/baz/quok | -4 }} → Strips all 4 segments from the end of the string

{{#titleparts: Talk:Foo/bar/baz/quok | -5 }} → Strips 5 segments from the end of the string (more than exist)

{{#titleparts: Talk:Foo/bar/baz/quok | | -1 }} → quok Returns last segment. See also {{SUBPAGENAME}}.

{{#titleparts: Talk:Foo/bar/baz/quok | -1 | 2 }} → bar/baz Strips one segment from the end of the string, then returns the second segment and beyond

{{#titleparts: Talk:Foo/bar/baz/quok | -1 | -2 }} → baz Start copying at the second last element; strip one segment from the end of the string

Before processing, the pagename parameter is HTML-decoded: if it contains some standard HTML character entities, they will be converted to plain characters (internally encoded with UTF-8, i.e. the same encoding as in the MediaWiki source page using this parser function).

For example, any occurrence of ", ", or " in pagename will be replaced by ".

No other conversion from HTML to plain text is performed, so HTML tags are left intact at this initial step even if they are invalid in page titles.

Some magic keywords or parser functions of MediaWiki (such as {{PAGENAME }} and similar) are known to return strings that are needlessly HTML-encoded, even if their own input parameter was not HTML-encoded:

The titleparts parser function can then be used as a workaround, to convert these returned strings so that they can be processed correctly by some other parser functions also taking a page name in parameter (such as {{PAGESINCAT: }} but which are still not working properly with HTML-encoded input strings.

For example, if the current page is Category:Côte-d'Or, then:

• {{#ifeq: {{FULLPAGENAME}} | Category:Côte-d'Or | 1 | 0 }}, and {{#ifeq: {{FULLPAGENAME}} | Category:Côte-d'Or | 1 | 0 }} are both returning 1; (the #ifeq parser function does perform the HTML-decoding of its input parameters).
• {{#switch: {{FULLPAGENAME}} | Category:Côte-d'Or = 1 | #default = 0 }}, and {{#switch: {{FULLPAGENAME}} | Category:Côte-d'Or = 1 | #default = 0 }} are both returning 1; (the #switch parser function does perform the HTML-decoding of its input parameters).
• {{#ifexist: {{FULLPAGENAME}} | 1 | 0 }}, {{#ifexist: Category:Côte-d'Or | 1 | 0 }}, or even {{#ifexist: Category:Côte-d'Or | 1 | 0 }} will all return 1 if that category page exists (the #ifexist parser function does perform the HTML-decoding of its input parameters);
• {{PAGESINCAT: Côte-d'Or }} will return a non-zero number, if that category contains pages or subcategories, but:
• {{PAGESINCAT: {{CURRENTPAGENAME}} }}, may still unconditionally return 0, just like:
• {{PAGESINCAT: {{PAGENAME|Category:Côte-d'Or}} }}
• {{PAGESINCAT: {{PAGENAME|Category:Côte-d'Or}} }}

The reason of this unexpected behavior is that, with the current versions of MediaWiki, there are two caveats:

• {{FULLPAGENAME}}, or even {{FULLPAGENAME|Côte-d'Or}} may return the actually HTML-encoded string Category:Côte-d'Or and not the expected Category:Côte-d'Or, and that:
• {{PAGESINCAT: Côte-d'Or }} unconditionally returns 0 (the PAGESINCAT magic keyword does not perform any HTML-decoding of its input parameter).

The simple workaround using titleparts (which will continue to work if the two caveats are fixed in a later version of MediaWiki) is:

• {{PAGESINCAT: {{#titleparts: {{CURRENTPAGENAME}} }} }}
• {{PAGESINCAT: {{#titleparts: {{PAGENAME|Category:Côte-d'Or}} }} }}
• {{PAGESINCAT: {{#titleparts: {{PAGENAME|Category:Côte-d'Or}} }} }}, that all return the actual number of pages in the same category.

Then the decoded pagename is canonicalized into a standard page title supported by MediaWiki, as much as possible:

1. All underscores are automatically replaced with spaces:

   {{#titleparts: Talk:Foo/bah_boo|1|2}} → bah boo Not bah_boo, despite the underscore in the original.

2. The string is split a maximum of 25 times; further slashes are ignored and the 25th element will contain the rest of the string. The string is also limited to 255 characters, as it is treated as a page title:

   {{#titleparts: a/b/c/d/e/f/g/h/i/j/k/l/m/n/o/p/q/r/s/t/u/v/w/x/y/z/aa/bb/cc/dd/ee | 1 | 25 }} → y/z/aa/bb/cc/dd/ee

   If for whatever reason you needed to push this function to its limit, although very unlikely, it is possible to bypass the 25 split limit by nesting function calls:

   {{#titleparts: {{#titleparts: a/b/c/d/e/f/g/h/i/j/k/l/m/n/o/p/q/r/s/t/u/v/w/x/y/z/aa/bb/cc/dd/ee| 1 | 25 }} | 1 | 2}} → z

3. Finally the first substring is capitalized according to the capitalization settings of the local wiki (if that substring also starts by a local namespace name, that namespace name is also normalized).

   {{#titleparts: talk:a/b/c }} → Talk:A/b/c

Warning:

You can use #titleparts as a small "string parser and converter", but consider that it returns the first substring capitalized: {{#titleparts: one/two/three/four|1|1 }} → One {{#titleparts: one/two/three/four|1|2 }} → two If lower case is needed, use lc: function to control output: {{lc: {{#titleparts: one/two/three/four|1|1 }} }} → one You can prepend a 'dummy' slash at the beginning of the string to get the correct first substring capitalization (uppercase or lowercase). Use 2 instead of 1 for first segment to return: {{#titleparts: /one/two/three/four|1|2 }} → one {{#titleparts: /One/two/three/four|1|2 }} → One

Warning:

Certain characters that are illegal in a page title will cause #titleparts to not parse the string: {{#titleparts: {one/two} | 1 | 1 }} → {one/two}. Does not produce the expected: {one {{#titleparts: [[page]]/123 | 1 | 2 }} → page/123. Does not work because brackets are illegal in page titles and this parser function does not process links embedded in its input pagename parameter, even when they use the MediaWiki syntax, or any other HTML or MediaWiki tags. {{#titleparts: red/#00FF00/blue | 1 | 3 }} → "". Does not work because "#" is also illegal in page titles.

Warning:

If any part of the title is just "." or "..", #titleparts will not parse the string: {{#titleparts: one/./three | 1 | 1 }} → one/./three. The whole string is returned. It does not produce the expected: one

Warning:

This function does not degrade gracefully if the input exceeds 255 bytes in UTF-8. If the input string is 256 bytes or more, the whole string is returned.

StringFunctions

All of these functions (len, pos, rpos, sub, replace, explode) are integrated from the StringFunctions extension, but are only available if an administrator sets $wgPFEnableStringFunctions = true; in LocalSettings.php.

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:string}} 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

The #len parser function was merged from the StringFunctions extension as of version 1.2.0.

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

The #pos parser function was merged from the StringFunctions extension as of version 1.2.0.

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

The #rpos parser function was merged from the StringFunctions extension as of version 1.2.0.

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

The #sub parser function was merged from the StringFunctions extension as of version 1.2.0.

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.

#replace

The #replace parser function was merged from the StringFunctions extension as of version 1.2.0.

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

The #explode parser function was merged from the StringFunctions extension as of version 1.2.0.

The #explode function splits the given string into pieces and then returns one of the pieces. 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 returned, 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| |2|3}} returns you tolerate this

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

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

{{#urldecode:value}}

Notes:

• 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

This module defines three global settings:

• $wgStringFunctionsLimitSearch
• $wgStringFunctionsLimitReplace
• $wgStringFunctionsLimitPad

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

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

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.

General points

Substitution

Parser functions can be substituted by prefixing the hash character with subst::

{{subst:#ifexist: Help:Extension:ParserFunctions | [[Help:Extension:ParserFunctions]] | Help:Extension:ParserFunctions }} → the code [[Help:Extension:ParserFunctions]] will be inserted in the wikitext since the page Help:Extension:ParserFunctions exists.

Warning:

The results of substituted parser functions are undefined if the expressions contain unsubstituted volatile code such as variables or other parser functions. For consistent results, all the volatile code in the expression to be evaluated must be substituted. See Help:Substitution.

Substitution does not work within <ref>…</ref> ; you can use {{subst:#tag:ref|…}} for this purpose.

Redirects

Especially {{#time:…|now-…}} could be handy in redirects to pages including dates, but this does not work.

Escaping pipe characters in tables

Parser functions will mangle wikitable syntax and pipe characters (|), treating all the raw pipe characters as parameter dividers. To avoid this, most wikis used a template Template:! with its contents only a raw pipe character (|), since MW 1.24 a {{!}} magic word replaced this kludge. This 'hides' the pipe from the MediaWiki parser, ensuring that it is not considered until after all the templates and variables on a page have been expanded. It will then be interpreted as a table row or column separator. Alternatively, raw HTML table syntax can be used, although this is less intuitive and more error-prone.

You can also escape the pipe character for display as a plain, uninterpreted character using an HTML entity: | .

Description	You type	You get
Escaping pipe character as table row/column separator	{{!}}	|
Escaping pipe character as a plain character	|	|

Stripping whitespace

Whitespace, including newlines, tabs, and spaces, is stripped from the beginning and end of all the parameters of these parser functions. If this is not desirable, comparison of strings can be done after putting them in quotation marks.

{{#ifeq: foo           |           foo | equal | not equal }} → equal

{{#ifeq: "foo          " | "          foo" | equal | not equal }} → not equal

To prevent the trimming of then and else parts, see m:Template:If. Some people achieve this by using <nowiki > </nowiki> instead of spaces.

foo{{#if:|| bar }}foo → foobarfoo

foo{{#if:||<nowiki /> bar <nowiki />}}foo → foo bar foo

However, this method can be used to render a single whitespace character only, since the parser squeezes multiple whitespace characters in a row into one.

<span style="white-space: pre;">foo{{#if:||<nowiki/> bar <nowiki/>}}foo</span>

→

foo bar foo

In this example, the white-space: pre style is used to force the whitespace to be preserved by the browser, but even with it the spaces are not shown. This happens because the spaces are stripped by the software, before being sent to the browser.

It is possible to workaround this behavior replacing whitespaces with   (breakable space) or   (non-breakable space), since they are not modified by the software:

<span style="white-space: pre;">foo{{#if:||&#32;&#32;&#32;bar&#32;&#32;&#32;}}foo</span> → foo bar foo

foo{{#if:||   bar   }}foo → foo   bar   foo

See also

• Help:Parser functions in templates
• m:Help:Calculation
• m:Help:Newlines and spaces
• m:Help:Comparison between ParserFunctions syntax and TeX syntax
• Help:Magic words
• Parser function hooks , an (incomplete) list of parser functions added by core and extensions.
• Module:String obsoleting Extension:StringFunctions
• Extension:PhpTags
• Parser functions for Wikibase (the extensions that enables Wikidata): d:Special:MyLanguage/Wikidata:How to use data on Wikimedia projects