From Facebook Developer Wiki (
Jump to: navigation, search

The ParserFunctions extension provides eleven additional parser functions to supplement the "magic words", which are already present in MediaWiki. All the parser functions provided by this extension take the form:

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



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 = != <> > < >= <=

This function evaluates a mathematical expression and returns the calculated value.

{{#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

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+ }}Expression error: Missing operand for +.
{{#expr: 1 foo 2 }}Expression error: Unrecognised word "foo".
Warning Warning: The operator mod gives wrong results for some values of the second argument:
{{#expr: 123 mod (2^32-1)}}123 (should be 123)

Depending on the specification and configuration of the server running the wiki there may also be other errors:

{{#expr: 20060618093259 mod 10000}}3259 in most cases, but may occasionally give -6357. See bug 6356.


{{#if: test string | value if true | value if false }}

This function tests whether the first parameter is 'non-empty'. It evaluates to false if the test string is empty or contains only whitespace characters (spaces, newlines, etc).

{{#if: | yes | no}}no
{{#if: string | yes | no}}yes
{{#if:      | yes | no}}no

| yes | 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

Either or both the return values may be omitted:

{{#if: foo | yes }} yes
{{#if: | yes }}
{{#if: foo | | no}}

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


This parser function compares two strings and determines whether they are identical.

{{#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 | yes | no}}yes
{{#ifeq: 0 | -0 | yes | no}}yes
{{#ifeq: 1e3 | 1000 | yes | no}}yes
{{#ifeq: {{#expr:10^3}} | 1000 | yes | no}}yes

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

{{#ifeq: foo | bar | yes | no}}no
{{#ifeq: foo | Foo | yes | no}}no
{{#ifeq: "01" | "1" | yes | no}}no
{{#ifeq: 10^3 | 1000 | yes | no}}no
Warning Warning: Numerical comparisons with #ifeq and #switch are not equivalent with comparisons in expressions:

{{#ifeq:12345678901234567|12345678901234568|1|0}} gives 0

because PHP compares here two numbers of type integer, while

{{#ifexpr:12345678901234567=12345678901234568|1|0}} gives 1

because MediaWiki converts literal numbers in expressions to type float, which, for large integers like these, involves rounding.
Warning Warning: Content inside parser tags (such as <nowiki>) is temporarily replaced by a unique code. This affects comparisons:

{{#ifeq: <nowiki>foo</nowiki> | <nowiki>foo</nowiki> | yes | no}}no
{{#ifeq: <math>foo</math> | <math>foo</math> | yes | no}}yes
{{#ifeq: {{#tag:math|foo}} | {{#tag:math|foo}} | yes | no}}yes
{{#ifeq: [[foo]] | [[foo]] | yes | no}}yes
It 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.


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: <strong class="error">a</strong> | error | correct }}error


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 ordinary string; it is not equal to zero, so we get value if true).

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: 1 > 0 | | no}}
{{#ifexpr: 1 < 0 | | no}} no
{{#ifexpr: 1 > 0 }}


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: XXXHelp:Extension:ParserFunctionsXXX | exists | doesn't exist }}doesn't exist

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

{{#ifexist: Special:Watchlist | exists | doesn't exist }}exists
{{#ifexist: No such special page | exists | doesn't exist }}doesn't exist (because the CheckUser extension is not installed on this wiki)
{{#ifexist: MediaWiki:Copyright | exists | doesn't exist }}doesn't exist (because MediaWiki:Copyright has not been customised)

#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, the page is categorised into Category:Pages with too many expensive parser function calls, and any further #ifexist: functions automatically return false, whether the target page exists or not.

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 }}doesn't exist

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


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 }}Error: Invalid depth in path: "Help:Foo/bar/baz/../../../../quok" (tried to access a node above the root node).


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


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

#switch allows an editor to add information in one template and this information will be visible in several other templates which all have different formatting.


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.

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

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 = result2
 | case5 = result3
 | case6 
 | case7 = result4
 | #default = default result

Here cases 2, 3 and 4 all return result2; cases 6 and 7 both return result4

Comparison behaviour

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 Warning: Numerical comparisons with #switch and #ifeq are not equivalent with 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: =.


{{#switch: 1=2
 | 1=2 = raw
 | 1<nowiki>=</nowiki>2 = nowiki
 | 1&#61;2 = html
 | 1{{=}}2 = template
 | default }}template

Note Note: For a simple real life example of the use of this function, check Template:NBA color. A complex example can be found at Template:Extension.


Code Description Current output
Y 4-digit year. 2017
y 2-digit year. 17
L 1 or 0 whether it's a leap year or not 0
o ¹ ISO-8601 year number. ² 2017 ³

¹ Requires PHP 5.1.0 and newer and rev:45208
² This has the same value as Y, except that if the ISO week number (W) belongs to the previous or next year, that year is used instead.
³ Will output literal o if ¹ not fulfilled

n Month index, not zero-padded. 8
m Month index, zero-padded. 08
M An abbreviation of the month name, in the site language. Aug
F The full month name in the site language. August
xg Output the full month name in the genitive form for site languages that distinguish between genitive and nominative forms. For Polish:


{{#time:d F Y|20 June 2010}} → 20 czerwiec 2010


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

W ISO 8601 week number, zero-padded. 33
j Day of the month, not zero-padded. 19
d Day of the month, zero-padded. 19
z Day of the year (January 1 = 0)
Note Note: To get the ISO day of the year add 1.
D An abbreviation for the day of the week. Rarely internationalised. Sat
l The full weekday name. Rarely internationalised. Saturday
N ISO 8601 day of the week (Monday = 1, Sunday = 7). 6
w number of the day of the week (Sunday = 0, Saturday = 6). 6
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. 4
h Hour in 12-hour format, zero-padded. 04
G Hour in 24-hour format, not zero-padded. 16
H Hour in 24-hour format, zero-padded. 16
Minutes and seconds
i Minutes past the hour, zero-padded. 49
s Seconds past the minute, zero-padded. 22
U Seconds since January 1 1970 00:00:00 GMT. 1503161362
t Number of days in the current month. 31
c ISO 8601 formatted date, equivalent to Y-m-dTH:i:s+00:00. 2017-08-19T16:49:22+00:00
r RFC 2822 formatted date, equivalent to D, j M Y H:i:s +0000, with weekday name and month name not internationalised. Sat, 19 Aug 2017 16:49:22 +0000
Non-Gregorian calendars
xij Day of the month 28
xiF Full month name Mordad
xin Month index 5
xiY Full year 1396
xjj Day of the month 27
xjF Full month name Av
xjx Genitive form of the month name Av
xjn Month number 11
xjY Full year 5777
Thai solar
xkY Full year 2560
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 3000. {{#time:xrY}} → MMXVII

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 }}

The list of accepted formatting codes is given in the table to the right. Any character in the formatting string that is not recognised 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 }}2017-08-19
{{#time: [[Y]] m d }}2017 08 19
{{#time: [[Y (year)]] }}2017 (17epmSat, 19 Aug 2017 16:49:22 +0000)
{{#time: [[Y "(year)"]] }}2017 (year)
{{#time: i's" }}49'22"

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}}Sat, 19 Aug 2017 16:49:22 +0000
{{#time: r|+2 hours}}Sat, 19 Aug 2017 18:49:22 +0000
{{#time: r|now + 2 hours}}Sat, 19 Aug 2017 18:49:22 +0000

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

{{#time: U | now }}1503161362
{{#time: r|@1503161362}}Sat, 19 Aug 2017 16:49:22 +0000

Warning Warning: The range of acceptable input is 1 January 0111 → 31 December 9999. For the years 100 through 110 the output in 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 }}Error: Invalid time.
{{#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, even when written with leading zeros:
{{#time: d F Y | 1 Jan 0069 }}01 January 0069

{{#time: d F Y | 1 Jan 0070 }}01 January 0070

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 }}2017

A four-digit number is interpreted as hours and minutes if possible, and otherwise as year:
{{#time: Y m d H:i:s | 1959 }}1959 08 19 00:00:00 Input is treated as a time rather than a year.
{{#time: Y m d H:i:s | 1960 }}1960 08 19 00:00:00 Since 19:60 is not a valid time, 1960 is treated as a year.

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 }}2017 08 19 19:59:09 Input is treated as a time rather than a year+month code.
{{#time: Y m d H:i:s | 196009 }}Error: Invalid time. Although 19:60:09 is not a valid time, 196009 is not interpreted as September 1960.

Warning 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 | January 1 }}2017 01 01 00:00:00
{{#time: Y m d H:i:s | February 2007 }}2007 02 01 00:00:00 Goes to the start of the month, not the current day.

The function performs a certain amount of date mathematics:

{{#time: d F Y | January 0 2008 }}31 December 2007
{{#time: d F | January 32 }}Error: Invalid time.
{{#time: d F | February 29 2008 }}29 February
{{#time: d F | February 29 2007 }}01 March

The total length of the format strings of the calls of #time is limited to 6000 characters [1].


This function is identical to {{#time: ... }}, except that it uses the local time of the wiki (as set in $wgLocaltimezone) when no date is given.

{{#time: Y-m-d }}2017-08-19
{{#timel: Y-m-d }}2017-08-19
{{#time: Y F d h:i:s}}2017 August 19 04:49:22
{{#timel: Y F d h:i:s}}2017 August 19 04:49:22


This function separates a pagetitle 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 parameter is not specified, it defaults to "0", which returns all the segments. If the first segment 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
{{#titleparts: Talk:Foo/bar/baz/quok | 2 }}Talk:Foo/bar
{{#titleparts: Talk:Foo/bar/baz/quok | 2 | 2 }}bar/baz

Negative values are accepted for both values. Negative values for number of segments effectively 'strips' segments from the end of the string. Negative values for first segment translates to "add this value to the total number of segments", loosely equivalent to "count from the right":

{{#titleparts: Talk:Foo/bar/baz/quok | -1 }}Talk:Foo/bar/baz
{{#titleparts: Talk:Foo/bar/baz/quok | | -1 }} quok
{{#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

The string is split a maximum of 25 times; further slashes are ignored. 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

Warning Warning: You can use #titleparts as a small "string parser & converter", but consider that it returns the first substring capitalized. If lower case is needed, use lc: function to control output.

{{#titleparts: one/two/three/four|1|1 }}One
{{#titleparts: one/two/three/four|1|2 }}two
{{lc: {{#titleparts: one/two/three/four|1|1 }} }}one

General points


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 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.


Parser functions will mangle wikitable syntax, treating all the raw pipe characters as parameter divisors. To avoid this, most wikis create the template Template:! with its contents only a raw pipe character (|). 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. Alternatively, raw HTML table syntax can be used, although this is less intuitive and more error-prone.

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, adding any non-whitespace characters (including the HTML encoding for a whitespace character, invisible Unicode characters such as the zero-width space or direction marks, or sequences recognised and stripped by the MediaWiki parser such as <nowiki />) will prevent further stripping:

{{#ifeq: foo            |            foo | equal | not equal }}
{{#ifeq: &#32;foo           &#32; | &#32;           foo&#32; | equal | not equal }}
not equal
{{#ifeq: <nowiki />foo           <nowiki /> | <nowiki />           foo<nowiki /> | equal | not equal }}
not equal
{{#ifeq: <nowiki />foo<nowiki /> | <nowiki />foo<nowiki /> | equal | not equal }}
not equal
{{#ifeq: foo | foo | equal | not equal }}

See also

Language: English
Personal tools
Graph API