Jump to: navigation, search
This page is a translated version of the page Help:Extension:ParserFunctions and the translation is 30% complete.

Other languages:
العربية • ‎български • ‎বাংলা • ‎català • ‎Deutsch • ‎Zazaki • ‎English • ‎español • ‎فارسی • ‎suomi • ‎français • ‎galego • ‎italiano • ‎日本語 • ‎한국어 • ‎Lëtzebuergesch • ‎मराठी • ‎polski • ‎پښتو • ‎português • ‎português do Brasil • ‎русский • ‎svenska • ‎தமிழ் • ‎ไทย • ‎Türkçe • ‎Tiếng Việt • ‎中文

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:

{{#函数名: 参数1 | 参数2 | 参数3... }}


类型 操作者
分组(括弧) ( )
数字 1234.5   e (2.718)   pi (3.142)
二进制操作者 e   一元 +,-
一元 not ceil trunc floor abs exp ln sin cos tan acos asin atan
二进制 ^
* / div mod
+ -
取整 round
逻辑 = != <> > < >= <=

这个函数计算并返回数学表达式的结果。 This function is also available in Scribunto via the mw.ext.ParserFunctions.expr function.

{{#expr: 表达式 }}



{{#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


{{#expr: }}
{{#expr: 1+ }}Expression error: Missing operand for +.
{{#expr: 1 foo 2 }}Expression error: Unrecognized word "foo".


{{#expr: +1 }}1
{{#expr: -1 }}-1
{{#expr: + 1 }}1
{{#expr: - 1 }}-1


{{#expr:{{NUMBEROFUSERS}}+100}}Expression error: Unrecognized punctuation character ",".

Warning Warning:

如若要进行基于日期时间的计算(如检测当前日期时间是否超过指定日期时间),首先用{{#time: xNU }}将日期时间转换为距1970.1.1的秒数,再对秒数进行加减运算。


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.

案例 结果 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 since 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 Warning:



{{#if: 字符串 | value if test string is not empty | value if test string is empty (or only white space) }}
{{#if: 参数1 | 参数2 | 参数3 }}


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

| yes | no}}


{{#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. A depth of at least seven levels of nesting is possible, although that may depend on the wiki or a memory limit.

  • Example: {{#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.

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


{{#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


{{#ifeq: foo | bar | equal | not equal}}not equal
{{#ifeq: foo | Foo | equal | not equal}}not equal
{{#ifeq: "01" | "1" | equal | not equal}}not equal  (对比上面没有引号的例子)
{{#ifeq: 10^3 | 1000 | equal | not equal}}not equal  (对比上面带有#expr的例子)

As a practical example, given an existing template Template:Size used to set standard short and long sizes, defined as:

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

the following ensue:


Warning Warning:

Warning Warning:

Warning Warning:


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: 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 }}Expression error: Unexpected = operator.


{{#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: 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/zh | exists | doesn't exist }}exists
{{#ifexist: XXHelp:Extension:ParserFunctions/zhXX | 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 (因为用户查核拓展已经安装于此wiki)
{{#ifexist: MediaWiki:Copyright | exists | doesn't exist }}exists (因为MediaWiki:Copyright已经自定义)

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/zh), Special:WhatLinksHere/Foo will list Help:Extension:ParserFunctions/zh.

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 限制

#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 (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 $wgExpensiveParserFunctionLimitManual:$wgExpensiveParserFunctionLimit, one can also increase the limit in LocalSettings.php if needed.


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

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


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 = 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


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:

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

Note Note: 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 Template:BOTREQ.

Replacing #ifeq

#switch can be used to reduce expansion depth.


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


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




代码 描述 Current output
(Purge this page's cache to update)
Y 4位数年。 2015
y 2位数年。 15
L 如果是闰年则为1,否则为0。 0
o ¹ ISO-8601 year of the specified week. ² 2015 ³

¹ 需要 PHP 5.1.0 及更新版本和 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. 10
m Month index, zero-padded. 10
M An abbreviation of the month name, in the site language. 10月
F The full month name in the site language. 10月
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. 波兰语:


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


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

W ISO 8601周数,零填充。 40
j 月内一日,非零填充。 4
d 月内一日,零填充。 04
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 internationalized.
l The full weekday name. Rarely internationalized. 星期日
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
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. 04
s Seconds past the minute, zero-padded. 53
U Seconds since January 1 1970 00:00:00 GMT. 1443974693
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
t Number of days in the current month. 31
c ISO 8601 formatted date, equivalent to Y-m-d"T"H:i:s+00:00. 2015-10-04T16:04:53+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, 04 Oct 2015 16:04:53 +0000
xmj 月内一日。 20
xmF 月的全名。 Dhu al-Hijjah
xmn 月索引。 12
xmY 全年。 1436
xij 月内一日。 12
xiF 月的全名。 Mehr
xin 月索引。 7
xiY 全年。 1394
xiy 2位数年。 94
xjj 月内一日。 21
xjF 月的全名。 Tishrei
xjt 月内日数。 30
xjx Genitive form of the month name. Tishrei
xjn 月数。 1
xjY 全年。 5776
xkY Full year in Thai solar calendar.
Note 注意: For years before 1941 the dates in Jan-Mar range are not calculated properly.
xoY 全年。 104
xtY 全年。 平成27
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
(upto 3,000 in pre MediaWiki 1.20).
{{#time:xrY}} → MMXV
xh Format the next number as a Hebrew numeral. {{#time:xhY}} → ב'ט"ו

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 }}

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 }}2015-10-04
{{#time: [[Y]] m d }}2015 10 04
{{#time: [[Y (year)]] }}2015 (15UTCpmSun, 04 Oct 2015 16:04:53 +0000)
{{#time: [[Y "(year)"]] }}2015 (year)
{{#time: i's" }}04'53"

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, 04 Oct 2015 16:04:53 +0000
{{#time: r|+2 hours}}Sun, 04 Oct 2015 18:04:53 +0000
{{#time: r|now + 2 hours}}Sun, 04 Oct 2015 18:04:53 +0000

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:d xg Y|20 June 2010|pl}}20 czerwca 2010

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

{{#time: U | now }}1443974693
{{#time: r|@1443974693}}Sun, 04 Oct 2015 16:04:53 +0000

Warning Warning:

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 }}2015

Warning Warning:

With MediaWiki r86805 - Code Review, a four-digit number is always interpreted as a year, never as hours and minutes:
{{#time: Y m d H:i:s | 1959 }}1959 10 04 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 }}2015 10 04 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.

The function performs a certain amount of date mathematics:

{{#time: d F Y | January 0 2008 }}31 12月 2007
{{#time: d F | January 32 }}Error: Invalid time.
{{#time: d F | February 29 2008 }}29 2月
{{#time: d F | February 29 2007 }}01 3月
{{#time:Y-F|now -1 months}}2015-9月

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

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 }} ==> 12:04 PM

However, Venezuela is on a -4.5 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 | -4.5 hours }} ==> 1:04 AM

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

  • {{#time:g:i A | -270 minutes }} ==> 11:34 AM
  • {{#time:g:i A | -16200 seconds }} ==> 11:34 AM

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


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 }}2015-10-04
{{#timel: Y-m-d }}2015-10-04
{{#time: Y F d H:i:s}}2015 10月 04 16:04:53
{{#timel: Y F d H:i:s}}2015 10月 04 16:04:53


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 {{ROOTPAGENAMEHelp:Magic_words#Page_names}}.
{{#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

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 occurence of &quot;, &#34;, or &#x22; 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 {{PAGENAMEHelp:Magic words#Page names}} 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:Help:Magic words#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&apos;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&apos;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&apos;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&apos;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&apos;Or and not the expected Category:Côte-d'Or, and that:
  • {{PAGESINCAT: Côte-d&apos;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&apos;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. The list of substrings is not cleaned up hierarchically (unlike with the {{PAGENAMEHelp:Magic words#Page names}} magic keyword, the ./ and ../ relative path segments are left intact by #titleparts):
    {{#titleparts: ../c/d }}../c/d
    {{PAGENAME: ../c/d }} → '
  4. 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 Warning:

Warning Warning:

Warning Warning:

General points


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

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

Warning Warning:

Substitution does not work within <ref></ref>Extension:Cite#Substitution and embedded parser functions, you can use {{subst:#tag:ref|}} for this purpose.


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 create the template Template:! with its contents only a raw pipe character (|) or use the {{!}} magic word which is available since MW 1.24. 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: &#124; .

说明 您输入的 您输出的
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

For preventing trimming then- and else-parts, see m:Template:If. Some people use also <nowiki> </nowiki> instead of spaces.

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

But this method allows the rendering of only one whitespace, since the parser reduce multiple spaces to a single one.

<span style="white-space: pre;">foo{{#if:||<nowiki></nowiki>      bar      <nowiki></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 &#32; (breakable space) or &nbsp; (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:||&nbsp;&nbsp;&nbsp;bar&nbsp;&nbsp;&nbsp;}}foofoo   bar   foo