Open main menu
This page is a translated version of the page Help:Extension:ParserFunctions and the translation is 52% complete.

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

扩展:解析器函数 扩展提供11个附加解析器函数以补充MediaWiki固有的“魔术字”。 (它有可能会配置到增加的解析器函数以支持运行:这些字符串函数已在其他文档页面记载。) 所有的解析器函数由下列的表格提供。

{{#函数名: 参数1 | 参数2 | 参数3... }}


类型 运算符号
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。 这个函数也通过mw.ext.ParserFunctions.expr函数中在扩展:Scribunto 有效。

{{#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 = }}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

注意:使用魔术字输出的数值时,首先要将其格式化以去除逗号,获得纯数字。例如:{{NUMBEROFUSERS}}输出17,204,344,而期望的输出值是17204344,使用{{formatnum :{{NUMBEROFUSERS}}|R}}便可以实现。对于某些语言,格式化数字尤为重要。例如孟加拉语中,{{NUMBEROFUSERS}}的输出结果是৩০,০৬১。

{{#expr:{{NUMBEROFUSERS}}+100}} Expression error: Unrecognized punctuation character ",".
  警告: 运算符mod會對第二參數的某些值給出錯誤的结果:
{{#expr: 123 mod (2^64-1)}}Division by zero. (輸出一個空值;應該是123)
如果你想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年1月1日的秒数,再对秒数进行加减运算。


a round b是指将a四舍五入至(1/10)b的整数倍。


案例 结果 凑整方法
{{#expr: 1/3 round 5 }} 0.33333 最终的数字小于5,所以没有明显的取整发生
{{#expr: 1/6 round 5 }} 0.16667 最终的数字大于5,所以被向上取整
{{#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 舍入至最近的整百数,负值在小数点左边。
{{#expr: 1234.5678 round 2 }} 1234.57 舍入至最近的整1/100数,因为正值会舍入至小数点右边。
{{#expr: 1234.5678 round 2.3 }} 1234.57 舍入指数的小数点在舍入结果中不起作用
{{#expr: trunc 1234.5678 }} 1234 Decimal portion truncated (chopped off)
{{#expr: 1/3 round 0 }} 0 向下取整至0
{{#expr: 1/2 round 0 }} 1 向上取整至1
{{#expr: 3/4 round 0 }} 1 向上取整至1
{{#expr: -1/3 round 0 }} -0 向上取整至0
{{#expr: -1/2 round 0 }} -1 向下取整至-1
{{#expr: -3/4 round 0 }} -1 向下取整至-1
{{#expr: ceil(1/3) }} 1 向上取整至1
{{#expr: floor(1/3) }} 0 向下取整至0
{{#expr: ceil(-1/3) }} -0 向上取整至0
{{#expr: floor(-1/3) }} -1 向下取整至-1
{{#expr: ceil 1/3 }} 0.33333333333333 没有取整,因为1本来就是整数
  警告: 注意会被解析为(ceil 1)/3,而不是你想象中的ceil(1/3)


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

{{#expr: "a" = "a" }}Expression error: Unrecognized punctuation character """.
{{#expr: a = a }}Expression error: Unrecognized word "a".
{{#ifeq: a | a | 1 | 0 }}1



{{#if: 测试字符串 | 当测试字符串不是空值时的值 | 当测试字符串是空值(或只包括空格)时的值 }}
{{#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


{{#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.

{{#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,會先回傳一個有效的整數)

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 }}


  警告: #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


{{#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.

  警告: 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.
  警告: 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


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Checkuser 扩展已经安装于此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: 不会对维基内链起作用。

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 $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.


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 }}


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
  警告: 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


"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 &#61;.


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

{{#switch: 1=2
 | 1&#61;2 = html
 | default

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


#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}}}}}}

i.e. deep nesting, linear:


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:



代码 描述 当前输出
Y 4位数年。 2019
y 2位数年。 19
L 如果是闰年则为1,否则为0。 0
o[note 1] 指定周的ISO-8601年。[note 2] 2019[note 3]
  1. 需要 PHP 5.1.0 及更新版本和 rev:45208.
  2. 这拥有与Y相同的值,除了当ISO周数(W)属于去年或明年时,会改用那一年。
  3. 如果未满足注释1,将输出文字“o”。
n 月指数,非零填充。 9
m 月指数,零填充。 09
M 月名的缩写,使用网站语言。 9月
F 使用网站语言的完整月名。 9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:F Y|June 2010|pl}} → czerwiec 2010
{{#time:d xg Y|20 June 2010|pl}} → 20 czerwca 2010
Day of the month or the year
j 月内一日,非零填充。 19
d 月内一日,零填充。 19
z Day of the year (January 1 = 0).
  注意: To get the ISO day of the year add 1.
Week and day of the week
W ISO 8601周数,零填充。 38
N ISO 8601 一周中的某天 (Monday = 1, Sunday = 7). 4
w Number of the day of the week (Sunday = 0, Saturday = 6). 4
D An abbreviation for the day of the week. Rarely internationalized.
l The full weekday name. Rarely internationalized. 星期四
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. 9
h Hour in 12-hour format, zero-padded. 09
G Hour in 24-hour format, not zero-padded. 21
H Hour in 24-hour format, zero-padded. 21
i Minutes past the hour, zero-padded. 17
s Seconds past the minute, zero-padded. 34
U Unix时间。 Seconds since January 1 1970 00:00:00 GMT. 1568927854
e 时区标识符。 UTC
I Whether or not the date is in daylight savings time. 0
O 与格林尼治时间(GMT)的差值 +0000
P Difference to Greenwich time (GMT), with colon +00:00
T Timezone abbreviation. UTC
Z 时区偏移(秒)。 0
t Number of days in the current month. 30
c ISO 8601格式化日期,与Y-m-d"T"H:i:s+00:00等效。 2019-09-19T21:17:34+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. Thu, 19 Sep 2019 21:17:34 +0000
xmj 月内一日。 19
xmF 月的全名。 穆哈兰姆月
xmn 月索引。 1
xmY 全年。 1441
xit 月内日数。 31
xiz 年内一日。 182
xij 月内一日。 28
xiF 月的全名。 Shahrivar
xin 月索引。 6
xiY 全年。 1398
xiy 2位数年。 98
xjj 月内一日。 19
xjF 月的全名。 以禄月
xjt 月内日数。 29
xjx Genitive form of the month name. Elul
xjn 月数。 12
xjY 全年。 5779
xkY Full year in Thai solar calendar.
  注意: For years before 1941 the dates in Jan-Mar range are not calculated properly.
xoY 全年。 108
xtY 全年。 令和元
xn Format the next numeric code as a raw ASCII number. 在印地语中,{{#time:H, xnH}}产生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}} → MMXIX
xh Format the next number as a Hebrew numeral. {{#time:xhY}} → ב'י"ט

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 }}2019-09-19
{{#time: [[Y]] m d }}2019 09 19
{{#time: [[Y (year)]] }}2019 (19UTCpmThu, 19 Sep 2019 21:17:34 +0000)
{{#time: [[Y "(year)"]] }}2019 (year)
{{#time: i's" }}17'34"

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}}Thu, 19 Sep 2019 21:17:35 +0000
{{#time: r|+2 hours}}Thu, 19 Sep 2019 23:17:35 +0000
{{#time: r|now + 2 hours}}Thu, 19 Sep 2019 23:17:35 +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}}Error: Invalid time.

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

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}}2019 settembre 19 21:17:35
{{#time: Y F d H:i:s|now|it|1}}2019 settembre 19 21:17:35
{{#time: Y F d H:i:s|+2 hours||0}}2019 9月 19 23:17:35
{{#time: Y F d H:i:s|+2 hours||1}}2019 9月 19 23:17:35

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

{{#time: U | now }}1568927855
{{#time: r | @1568927854 }}Thu, 19 Sep 2019 21:17:34 +0000
  警告: 若不在數字格式的時間戳值的前面包含@前綴,結果通常會出錯或是未預期的數值:
{{#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 (正確)
{{#time: r | 999 }}Error: Invalid time. (不支援的年分格式)
{{#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 (正確)
{{#time: r | 1000 }}Fri, 19 Sep 1000 00:00:00 +0000 (被解讀為某年的目前月分和日期)
{{#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 (正確)
{{#time: r | 9999 }}Sun, 19 Sep 9999 00:00:00 +0000 (被解讀為某年的目前月分和日期)
{{#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 (正確)
{{#time: r | 10000 }}Error: Invalid time. (不支援的年分格式)
  警告: 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 3月 0100
{{#time: r | 29 Feb 0100 }}Mon, 01 Mar 0100 00:00:00 +0000 (錯誤,100被解讀為2000,但那是閏年)
{{#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, except when written in 4-digit format with leading zeros:
{{#time: d F Y | 1 Jan 6 }}01 1月 2006
{{#time: d F Y | 1 Jan 06 }}01 1月 2006
{{#time: d F Y | 1 Jan 006 }}01 1月 2006
{{#time: d F Y | 1 Jan 0006 }}01 1月 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 }}2019
  警告: 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 }}2019 06 19 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 09 19 00:00:00 Gives the start of the day, but the current day of the year.


{{#time: Y m d H:i:s | 1959 }}1959 09 19 00:00:00


{{#time: Y m d H:i:s | 195909 }}2019 09 19 19:59:09 輸入被視為時間而不是年+月代碼。
{{#time: Y m d H:i:s | 196009 }}Error: Invalid time. 雖然19:60:09不是有效時間,但196009不被解讀為1960年9月。


{{#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}}2019-8月



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 }} ==> 5:17 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 }} ==> 9:17 PM

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

  • {{#time:g:i A | +330 minutes }} ==> 2:47 AM
  • {{#time:g:i A | +19800 seconds }} ==> 2:47 AM

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


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:

  警告: 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时间。自1970年1月1日 00:00:00(GMT)以来的秒数。
Z 时区偏移(秒)。
{{#time: U}}1568927854
{{#timel: U}}1568927854
{{#time: Z}}0
{{#timel: Z}}0


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 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 {{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&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. 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
  警告: 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
  警告: 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.
  警告: 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
  警告: This function does not degrade gracefully if the input exceeds 255 characters. If the input string is 256 characters long or more, the whole string is returned.



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.
  警告: 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.


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


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: &#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


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



  1. 在2011年r86805之前,情況並非如此。
  2. phabricator.wikimedia.org的ExtParserFunctions.php