User talk:Mathglot/Regex

All regular expressions are standard PCRE unless otherwise stated. A few might be Cirrus regexes used by Wikipedia's regex editor; see also Elastic search Cirrus regex syntax. Template:Regex can be used to safely test regexes.

Further information: Help:Searching/Regex, Template:Template usage, and mw:Help:CirrusSearch § Regular expression searches

Regex match

Further information: Template:Regex and Template:Template usage

Doc links

Regex flavors:

• Lua: mw:Lua manual#Patterns (also § Ustring patterns), Help:Lua for beginners#Understanding patterns, Help:Lua for beginners#Note on Lua patterns versus regular expressions,
  • No alternation, or *very* limited: see mw:Help:CirrusSearch/Logical operators)
  • No repetition of multi-char patterns; i.e., no way to match choo, choo-choo or choo-choo-choo
  • No min-max repetitions, like {3,5}; e.g., do %d%d%d%d?%d? instead of PCRE: [0-9]{3,5}
• AWB uses A .NET flavor of regex; see Wikipedia:AutoWikiBrowser/Regular expression
• Advanced search uses: mw:Help:CirrusSearch: including Cirrus regex
  • Help:Searching/Regex#Metacharacters
  • Help:Searching/Features – stemming, fuzzy, linksto, hastemplate, incategory|a|b, morelike, sorting, boosting,

Piped links

1. Match piped link in any namespace (e.g, could be 'File:' in first part)
   • \[\[([^]]+)\|([^]]+)\]\]
2. Match piped link in current namespace only (not containing colon in first part):
   • \[\[([^]:]+)\|([^]]+)\]\]
   • \[\[([^|\]]*?)\|([^]]+)\]\]
3. Piped or unpiped link in current namespace:
   • \[\[([^|\]]+)\|?([^|\]]+)?\]\]

Piped links with possible WP:NOPIPE or WP:NOTBROKEN issues

• \[\[([^\|]{6,})([^\|]+)\|\1[^\]]?\]\]

Urls

Bareurl:

• Unnamed: <ref>\s*\[\s*[^]]+\s*\]\s*</ref>
  • Cirrus: : insource:/\<ref\>\s*\[\s*[^] ]+\s*\]\s*\<\/ref\>/

• External url in reference (could be a complete, plaintext ref with a linked title):
  • <ref>\s*\[\s*https?:\/\/*[^] ]+\s+[^]]+\]\s*</ref>
  • Cirrus: : insource:/\<ref\>\s*\[\s*https?:[^] ]+\s+[^]]+\]\s*\<\/ref\>/

Reference

• <ref(name\s*=|s*"?([-\s\w\d]+)?"?)\s*>([^<]+)</ref>

Reference with lastN but no matching firstN

Assumes lastN occurs first, then firstN

• <ref[^>]*?>[^>]*?\|last(\d)?(?!.*?\|first\1?).*?<\/ref>

Charts

.tab file creation

3-col table tsv to json data

• SRCH: ^(\S+)\s(\S+)\s(\S+)\s
• RPLC: [\n\t"\1",\n\t\2,\n\t\3\n\t],

Citation template creation

JSTOR

See also: User:Mathglot/sandbox/Templates/Cite MLA

JSTOR cite, and intermediate steps to get there; e.g.:

JSTOR MLA

On a JSTOR page, click Cite, and select MLA:

Match a name, in First M. [N.] Last format:

• ^[A-Z][a-zA-Z]+(?: [A-Z]\.){0,2} [A-Z][a-zA-Z]+\.?

Match two names in that format, joined by , and :

• ^([A-Z][a-zA-Z]+(?: [A-Z]\.){0,2} [A-Z][a-zA-Z]+\.?)(?:, and ([A-Z][a-zA-Z]+(?: [A-Z]\.){0,2} [A-Z][a-zA-Z]+\.?))?

Match two names in that format, joined by , and , where the first is Last, First [M. [N.]] (this is JSTOR format):

• ^([A-Z][a-zA-Z]+(?:'[A-Z][a-zA-Z]+)?(?:-[A-Z][a-zA-Z]+)?, [A-Z][a-zA-Z]+(?: [A-Z]\.){0,2},)(?: and ([A-Z][a-zA-Z]+(?: [A-Z]\.){0,2} [A-Z][a-zA-Z]+\.))?

Match JSTOR MLA entirely (needs slight work to eliminate commas in output):

• SRCH: ^([A-Z][a-zA-Z]+(?:'[A-Z][a-zA-Z]+)?(?:-[A-Z][a-zA-Z]+)?, [A-Z][a-zA-Z]+(?: [A-Z]\.){0,2},)(?: and ([A-Z][a-zA-Z]+(?: [A-Z]\.){0,2} [A-Z][a-zA-Z]+\.))?\s“([^”]+)”\s([^,]+),\svol\.\s(\d+),\sno\.\s(\d+),\s(\d{4}), (pp?\.)\s([-–—\d]+)\.\sJSTOR,\s(http.*?)\.\sAccessed\s(.*?)\.
• RPLC: 1=\1; 2=\2; 3=\3; 4=\4; 5=\5; 6=\6; 7=\7; 8=\8; 9=\9; 10=\10; 11=\11.

Single-author Jstor MLA. Matches: Berk, Laura E. “Vygotsky’s Theory: The Importance of Make-Believe Play.” Young Children, vol. 50, no. 1, 1994, pp. 30–38. JSTOR, http://www.jstor.org/stable/42728322. Accessed 4 Apr. 2025.

• SRCH: ^([A-Z][a-zA-Z]+(?:'[A-Z][a-zA-Z]+)?(?:-[A-Z][a-zA-Z]+)?),\s([A-Z][a-zA-Z]+(?: [A-Z]\.)\s*){0,3}“([^”]+)”\s([^,]+),\svol\.\s(\d+),\sno\.\s(\d+),\s(\d{4}), (pp?)\.\s([-–—\d]+)\.\sJSTOR,\s(http.*?)\.\sAccessed\s(.*?)\.$
• RPLC: {{cite journal |last=\1 |first=\2 |date=\7 |title=\3 |journal=\4 |vol=\5 |issue=\6 |publisher= |location= |doi= |issn= |oclc= |\8=\9 |url=\10 |access-date=\11}}

Ref tags

Match all ref tags, reused named, named, or unnamed:

• SRCH: (?:<ref(?:\s+name="?[^">]+"?)?>[^<]+</ref>|<ref(?:\s+name="?[^">]+"?)?\s*\/>)

Remove blanks between text and ref tag (don't use \s here which matches newline):

• SRCH:  +<ref
• RPLC: <ref

Supply missing blank between trailing end-ref and next text word:

• SRCH: </ref>([^\s.<])
• RPLC: </ref> \1

Move punctuation trailing a ref tag to before the tag (with 'dot matches newline' or PCRE 's' modifier):

• SRCH: <ref([^<]+)</ref>([.,:;?!])
• RPLC: \2<ref\1</ref>

cite book, from Google bibliographic info

This is the bibliographic info that appears at the end of every Google books page.

• Search:

Title\t(.*)
(.*)
Author\t(.*)
Publisher\t(.*), (\d\d\d\d)
ISBN\t([^,]+),? ?(\d+)?
Length\t(\d+) pages$

• Replace: {{cite book |author=\3 |last= |first= |date=\5 |title=\1 |publisher=\4 |series=\2 |isbn=\7 <!--isbn2=\6--> |oclc= |page= <!--total-pages=\8--> |url=}}

---

• Search:

Title\t(.*)
Author\t(.*)
Publisher\t(.*), (\d\d\d\d)
ISBN\t([^,]+),? ?(\d+)?
Length\t(\d+) pages$

• Replace: {{cite book |author=\2 |last= |first= |date=\4 |title=\1 |publisher=\3 |isbn=\6 <!--isbn2=\5--> |oclc= |page= <!--total-pages=\7--> |url=}}

---

• Search:

Title\t(.*)
Author\t(.*)
Edition\t(.*)
Publisher\t(.*), (\d\d\d\d)
Original from\t.*
Digitized\t.*
ISBN\t([^,]+),? ?(\d+)?
Length\t(\d+) pages$

• Replace: {{cite book |author=\2 |last= |first= |date=\5 |title=\1 |edition=\3 |publisher=\4 |isbn=\7 <!--isbn2=\6--> |oclc= |page= <!--total-pages=\8--> |url=}}

---

• Search:

Title\t(.*)
(.*)
Editors?\t(.*)
Edition\t(.*)
Publisher\t(.*), (\d\d\d\d)
ISBN\t([^,]+),? ?(\d+)?
Length\t(\d+) pages$

• Replace: {{cite book |author= |last= |first= |editors=\3 |editor1-last= |editor1-first= |editor2-last= |editor2-first= |editor3-last= |editor3-first= |date=\6 |title=\1 |edition=\4 |publisher=\4 |series=\2 |isbn=\8 <!--isbn2=\7--> |oclc= |page=<!--total-pages=\9--> |url=}}

---

• Search:

Title\t(.*)
(.*)
Editors?\t(.*)
Edition\t(.*)
Publisher\t(.*), (\d\d\d\d)
ISBN\t([^,]+),? ?(\d+)?
Length\t(\d+) pages$

• Replace: {{cite book |last= |first= |editors=\3 |editor1-last= |editor1-first= |editor2-last= |editor2-first= |editor3-last= |editor3-first= |date=\6 |title=\1 |edition=\4 |series=\2 |publisher=\5 |isbn=\8 <!--isbn2=\7--> |oclc= |page= <!--total-pages=\9--> |url=}}

Citation template tweaking and reordering

Drop query params from Google books urls

But keep the page param &pg=PAnn

• Search: (https://books\.google\.\w\w/books\?id=[^&|}]+)(&pg=[\w\d]+)?[^]|}]+
• Replace: \1\2 

Fix MOS:REFPUNCT problems

• Search: <ref([^<]+)</ref>([.,;?!])
• Replace: \2<ref\1</ref>

Convert <ref>{{Harvtxt...}} (with page number; sans ref name) to {{sfn}}

• Search: <ref>{{harvtxt(?:\s*(\|\s*[^|]+)\s*)(.*?)\|\s*(\d\d\d\d)\|(p+=\d+\s*)}}\s*</ref>
• Replace: {{sfn\1\2|\3|\4}}

Citations with 'author=' to 'last=... first=...'

Assumes a regular CS1 or CS2 citation, with space before vertical bar, and '|author=' present:

• Search: \|author=([ \w]+)\s+(\w[^\|]+)\s+\|
• Replace: |last1=\2 |first1=\1 |

Alt (author or author1; name possibly wikilinked):

• Search: \|author1?=\[?\[?([ -\w]+)\]?\]?\s+(\w[^\|]+)\s+\|
• Replace: |last1=\2 |first1=\1 |

AuthorN-first (or last) to firstN or lastN

With or without the N present:

• Search: \|author(\d)?-(last|first)=
• Replace: |\2\1=

Move url to the back

• Search: \s*\|url=([^|\s]+)([^}]+)}}
• Replace: \2 |url=\1}}

Possible failure case: *<!--Chenntouf-->{{cite web |last1=Chenntouf |first=Tayeb |date=1999 |title="La dynamique de la frontière au Maghreb", Des frontières en Afrique du xiie au xxe siècle |url=https://unesdoc.unesco.org/in/documentViewer.xhtml?v=2.1.196&id=p::usmarcdef_0000139816&file=/in/rest/annotationSVC/DownloadWatermarkedAttachment/attach_import_c35456f4-f4da-4b4a-b938-9d61f48fa689?_=139816fre.pdf&locale=fr&multi=true&ark=/ark:/48223/pf0000139816/PDF/139816fre.pdf#%5B%7B%22num%22:605,%22gen%22:0%7D,%7B%22name%22:%22XYZ%22%7D,-250,769,0%5D |access-date=2020-07-17 |website=unesdoc.unesco.org}}

Swap last with first

• Search: \|first=([^|]+)\s\|last=([^|]+)\s
• Replace: |last=\2 |first=\1

Swap editor-last with editor-first

• Search: \s*\|editor-first=([^|]+)\s*\|editor-last=([^|]+)\s*\|
• Replace: |editor-last=\2 |editor-first=\1|

Swap editorN-last with editorN-first

• Search: \s*\|editor(\d)-first\s*=\s*([^|]+)\s*\|editor\1-last\s*=\s*([^|]+)\s*\|
• Replace: |editor\1-last=\3|editor\1-first=\2|

Swap lastn with firstn

• Search: \|title=([^|]+)\s?\|(last\d?)=([^|]+)\s?\|(first\d?)=([^|]+)\s?
• Replace: |\2=\3 |\4=\5 |title=\1

Move last-first before title

• Search: \|title=([^|]+)\s*\|last=([^|]+)\s\|first=([^|]+)\s
• Replace: |last=\2 |first=\3 |title=\1

Move year after first

• Search: ^(.*?)\|first([^|]+)(.*?)\s*\|year=(\d+)(.*?)$
• Replace: \1|first\2|year=\4 \3\5

Punctuation after citation, to before

Sfn:

• Search: ({{sfn[^}]+}})([-–—,;!\?\.])
• Replace: <nowiki\2\1</nowiki>

Swap |first=X |last=y around so last is first in citation

• Search: \|first(\d)=([^|]+)\s\|last\1=([^|]+)\s
• Replace: *|last\1=\3 |first\1=\2

plain refs to cite web

Text sources which don't use {{cite web}} may be transformed by a series of regex replaces, if the format is reasonably standard. For example, this change by this series:

* => * {{cite web |last=
(\, ?)(.*)$ => |first=\2
\.\ +''(.*?)'' => |title=\1
first=([\s\w]+),\s+and\s+([\s\w]+),\s+([\s\w]+) => first=\1 |last2=\2 |first2=\3
\((\d{4})\) => |year=\1
\((\d{4})\)\s+([ \w]+)\. => |year=\1 |publisher=\2
\s+isbn\s+([-\d]{10,17}) => |isbn=\1
$ => |ref=harv }}

See also User:Mathglot/sandbox/Templates/Cite MLA (in progress...)

Updating named refs to template:R

Example: Holocaust denial, revision 843383121. Three steps:

1. change quoted named refs:
<ref name="([^"]+)"\s*\/> -> {{R|"\1"}}

2. change unquoted named refs (with or without trailing blanks before the slash)
<ref name=([^ #"'/=>?\\]+)\s*/> -> {{R|\1}}

3. combine consecutive R's
{{R\|([^}]+)}}\s*{{R\|([^}]+)}} -> {{R|\1|\2}} g(repeat till done)

Edit summary:

Minimize visual impact on the wikicode of [[WP:NAMEDREFS|named refs]] using [[Template:R]]. No change to rendered footnote section. Using global regex replace: 1: (change quoted named refs): s!<ref name="([^"]+)"\s*\/>!{{R|\1}}!g 2: (change unquoted named refs): s!<ref name=([^ #"'/=>?\\]+)\s*/>!{{R|\1}}!g 3: (combine consecutive Rs into one): s!{{R\|([^}]+)}}\s*{{R\|([^}]+)}}!{{R|\1|\2}}!g

Page history

from scrape of rendered page

Created page to parsed data

Only those with {{Expand}} in the early words that made it into the edit summary

SRCH:^(\d+:\d\d,\s\d+\s[JFMASOND][a-z]+\s\d{4})\s\|\s(\d+)\sdiff hist .*? N ([^+]+)\sCreated page with '{{Expand ([^|]+)\|\s*([^}]+)}}.*?$
RPLC:* ts=\1; rev=\2; ti=\3; lang=\4; src=\5

Add links to parsed data page

First use the § Created page to parsed data above, then apply this one to add links:

SRCH: ^\* ts=([^;]+); rev=([^;]+); ti=([^;]+); lang=([^;]+); src=(.*?)$
RPLC: * ts=\1; rev=[[Special:Diff/\2|\2]]; ti=[[\3]]; lang=\4; src=[[:{{#if: {{iso 639 name|fn=iso_639_name_exists|\4}} | {{iso 639 name|fn=iso_639_name_to_code|\4}}|\4}}:\5]]

from HTML page source

Article page history to parsed data

Turn article page history into a series of parsed lines:

• 1=ARTICLE_TITLE 2=REVISION 3=HH:MM 4=Month DD, YYYY 5=TOTAL_BYTES 6=BYTE_CHANGE

1. Go to article page history page
2. Rt-click, Page source
3. Select-all, copy, paste
4. Apply Search/Replace Regex below, with "dot matches newline"
5. Optional step to convert underscore to blank in article titles

SEARCH:
<li.*?index.php\?title=([^&]+)&oldid=(\d+)[^>]+>(\d\d:\d\d),\s(.*?)</a>.*?title="([,\d]+)\sbytes after change of this size">(.?\d+)</span>.*?</li>

To generate the following output, use this replacement:
1=ARTICLE_TITLE 2=REVISION 3=HH:MM 4=Month DD, YYYY 5=TOTAL_BYTES 6=BYTE_CHANGE

REPLACE:
1=\1 2=\2 3=\3 4=\4 5=\5 6=\6

To generate the following sample output, use this replace instead:

• 916661155 Risk_aversion diff 00:29 September 20, 2019; (change:-1b to 31,671 bytes)

REPLACE:
* [[Special:Permalink/\2|\2]] [[\1]] [[Special:Diff/\2|diff]] \3 \4; (change:\6b to \5 bytes)

To generate a six-column table row with this data, including one extra column for remarks, use this:
REPLACE:
|-
| [[\1]] || [[Special:Permalink/\2|\2]] || [[Special:Diff/\2|\6]] || \5 || \3 \4 || any remark here

Followed by optional underscore replacement. (s/_/ /gi).

To generate the following table row examples (table header/footer code added for context):

Article history for Example user
Article	Perm	Diff	Len	Timestamp	Remark
Risk aversion	916661155	-1	31,671	00:29 September 20, 2019	any remark
History of the provincial electoral map of Quebec	916660706	-1	26,988	00:26 September 20, 2019	other remark

User contribution history to parsed data

Turn article page history into a series of parsed lines:

• 1=REVISION 2=TITLE 3=TIMESTAMPE 4=BYTE_CHANGE 5=EDIT_SUMMARY

1. Go to user contrib history page
2. Rt-click, Page source
3. Select-all, copy, paste
4. Find '<h4 class="mw-index-pager-list-header-first' and cut everything above it.
5. Find '
   ' and cut everything below it
6. Apply Search/Replace Regex below, with "dot matches newline"
7. Optional step to convert underscore to blank in article titles

SEARCH: (options: dot matches newline)
^<li data-mw-revid="(\d+)".*?class="mw-changeslist-date" title="(.*?)">(.*?)</a>.*?size">(.*?)</strong>.*?parentheses">(.*?)</span>.*?</li>$

To generate the following output
1=REVISION 2=TITLE 3=TIMESTAMPE 4=BYTE_CHANGE 5=EDIT_SUMMARY
use this replacement:

REPLACE:
1=\1 2=\2 3=\3 4=\4 5=\5

To generate: rev=REVISION title=TITLE timestamp=TIMESTAMPE bytes=BYTE_CHANGE summary=EDIT_SUMMARY
REPLACE:
rev=\1 title=\2 time=\3 bytes=\4 summary=\5

To generate: rev=REVISION title=TITLE
SEARCH: (options: dot matches newline)
^<li data-mw-revid="(\d+)".*?title="([^"]+).*?</li>$
REPLACE:
rev=\1 title=\2

New contribs Translated pages to bullet list

From Special:contribs with 'new' pages box ticked; extracting pages with ContentTranslation tool summary:

1. Search: \)‎ \. \. N\s([^(]*?) ‎ \(Created by translating the page "([^"]+)"
2. Copy matches
3. Replace: * [[\1]] from [[es:\2]]

Other regex replace

Add leading hidden token to ref-named citations as prep for sorting the Bibliography

• Search: <!--{{sfn\|LAST\|YYYY\|p=}}--> *<ref name="([\(\)\w]+)\s+(\d+)">
• Replace: *<!--{{sfn|\1|\2|p=}}-->

Alphabetize citations in Bibliography

The technique is 1) add a leading token consisting of the (first) last name, 2) sort, 3) strip out the token. Only step 1 is shown:

• Search: ^\*\s*{{cite(.*?)\|\s*last1?\s*=\s*([^|]+)\s*(.*)$
• Replace: **<!--\2-->{{cite\1 |last1=\2\3

Convert glossary anchor to vanchor

SEARCH:
^;\s*{{Anchor\|([^\}]+)}}(?:[-<>\s,:\w\d]+)$
REPLACE:
;{{Vanchor|\1}}

Convert glossary &tl;term> to be in-linkable

SEARCH:
^{{term\s*\|(term\s*=\s*)?([^|{}]+)
REPLACE:
{{term|\1|2={{Vanchor|\2}}

ES: Convert glossary <term>s to be in-linkable via global regex replace s!^{{term\s*\|(term\s*=\s*)?([^|{}]+)!{{term|\1|2={{Vanchor|\2}}!g

Convert mu (Chinese acres)

SRCH: (\d[.,\d]+) mu(\b)
RPLC: \1 mu {{convert mu|\1|mu|ha}}\2   (to 'hectares')
RPLC: \1 mu {{convert mu|\1|mu|ha|abbr=on}}\2   (to 'ha')
RPLC: \1 mu {{cvt mu|\1|mu|ha}}\2   (to 'ha', via wrapper)

Parse wikilinks

Parse wikilinks, exclude colons to exclude namespaces (this will exclude wikilinks that have colons in the anchor):

$1 = Target article $2 = Anchor (#-fragments untested):

• \[\[([^:\|\]]+)\|?([^:\]]+)?\]\]

This saves the pipe (if there is one) in \2, so can use replace to generate lang-prefixed links, for example, if translating a nav template from en to fr, one could start like this:

• Search: \[\[([^:\|\]]+)(\|?[^:\]]+)?\]\]
• Replace: [[:en:\1\2]]

This adds superscript wikidata links to all wikilinks on a page so they can be easily translated:

• Search: \[\[([^:\|\]]+)(\|?[^:\]]+)?\]\]('')?
• Replace: [[\1\2]]\3<sup>[[[d:{{subst:wikidata|label|raw|page=\1}}#sitelinks-wikipedia|wd]]]</sup>

Interlanguage template transformation

1. Turn {{ca:GEC}} into {{sfn}}:
   • Search: {{GEC\|id=([\d]+)\|nom=([ \w]+).*?}}
   • Rplce: {{sfn|GEC|loc=[http://www.enciclopedia.cat/EC-GEC-\1.xml \2]}}

Convert italic markup to lang templates

See also: Template:Lang § Links, and MOS:FOREIGNITALIC

First, fix the links (two types, depending where italic markup is):

• piped (type 1): e.g., ''[[École navale|FOO]]'' ⟶ {{lang|fr|[[École navale|FOO]]}}
  • SRCH: (?<!')''\[\[([^ |]+)\|([^]]+)\]\]''(?<!') # handles the 2-pop case; excludes 3-pop, but also 5-pop; add (?:''')? for that
  • RPLC: {{lang|fr|[[\1|\2]]}}

• piped (type 2): e.g., [[École navale|''FOO'']] ⟶ {{lang|fr|[[École navale|FOO]]}}
  • SRCH: \[\[([^ |]+)\|((?<!')''([^']])+''(?<!')\]\]
  • RPLC: {{lang|fr|[[\1|\2]]}}

• unpiped (order matters; must be done after piped links)
  • SRCH: ''\[\[([^|]+)\]\]'' e.g., ''[[École navale]]'' ⟶ {{lang|fr|[[École navale]]}}
  • RPLC: {{lang|fr|[[\1]]}}

• What's left, is unlinked:
  • SRCH: (?<!')''([^']+)''(?<!') e.g., ''École navale'' ⟶ {{lang|fr|École navale}}
  • RPLC: {{lang|fr|\1}}

FR - EN article translation preprocessing

1 <ref>{{(\w\w)}}\s*{{citation\|(.*?)</ref> -> {{efn|"{{lang|\1|\2}}"}}
2 ''{{lang\|de\|(.*?)}}'' -> {{lang|de|\1}}
3 {{citation\|(.*?)}} -> "\1"
4 <ref>\s*{{de}}\s*(.*?)\s*</ref> -> {{efn|{{lang|de|\1}}}}
5 <ref>{{harvsp\|(.*?)}}.</ref> -> {{sfn|\1}}

Substify and unsubstify

Per this VPT discussion neither of these 'ugly subst tricks' are advisable. Instead, the entire template should be within <includeonly>..</includeonly>.

Find expressions in need of safesubst (two left curlies, but not three or more)

• Search: (?<!{){{(?!{)

Substify via |safesubst

• Search: (?<!{){\{(?!\{)
• Replace: {{ {{{|safesubst:}}}
• ES:         s!(?<!{){\{(?!\{)<!{{ {{{|safesubst:}}}!g

Unsubstify via |safesubst

• Search: \s*{{\s*{{{\s*\|safesubst:\s*}}}
• Replace: {{
• ES:         s!\s*{{\s*{{{\s*\|safesubst:\s*}}}!{{!g

Substify via safesubst:<noinclude />

• Search: (?<!{){\{(?!\{)
• Replace: {{safesubst:<noinclude />

Unsubstify via safesubst:<noinclude />

• Search: {{safesubst:<noinclude />
• Replace: {{

Substify via safesubst:    Recommended

• Search: (?<!{){\{(?!\{)
• Replace: {{safesubst:

Unsubstify via safesubst:

• Search: {{safesubst:
• Replace: {{

Wikilink to {{ill}}

Aimed at Nav template translation, so handles bulleted links, optional pops or bolding, and specific lang prefix:

Unpiped links (e.g., * ''[[:fr:Documents maçonniques]]''):

• Search: ^\*\s*('*)?\[\[:fr:([^]]+)\]\]('*)?
• Replace: \1{{ill|ENGLISHNAME|fr|\2|v=sup}}\3

Piped links (.e.g., * ''[[:fr:Idées (revue, 1941-1944)|Idées]]''):

• Search: ^\*\s*('*)?\[\[:fr:([^]|]+)\|?([^]]+)?\]\]('*)?$
• Replace: \1{{ill|ENGLISHNAME|fr|\2|lt=\3|v=sup}}\4

For bios or proper names, duplicate the Foreign name in the English article field:

• Search: ^\*\s*('*)?\[\[:fr:([^]|]+)\|?([^]]+)?\]\]('*)?$
• Replace: * \1{{ill|\2|fr|\2|lt=\3|v=sup}}\4

Examples:

• * ''[[:fr:Le Juif et la France]]'' ⟶
  • * ''{{ill|Le Juif et la France|fr|Le Juif et la France|lt=|v=sup}}''
• * ''[[:fr:Combats]]'' ⟶
  • * ''{{ill|Combats|fr|Combats|lt=|v=sup}}''
• * ''[[:fr:Idées (revue, 1941-1944)|Idées]]'' ⟶
  • * ''{{ill|Idées (revue, 1941-1944)|fr|Idées (revue, 1941-1944)|lt=Idées|v=sup}}''
• * [[:fr:Publications antisémites en France]] ⟶
  • {{ill|Publications antisémites en France|fr|Publications antisémites en France|lt=|v=sup}}

Section demote

• Search: ^(={2,5})([^=].*?)\1
• Replace: =\1\2\1=

Subsection promote

• Search: ^=(={2,5})([^=].*?)\1=
• Replace: \1\2\1

Reflib section from last-first-year

• Search: ^\*\s*(.*?)\|last=([^|]+)\s\|first=([^|]+)\s*\|year=(\d+)(.*?)$
• Replace:

  == \2-\4 ==
  \1|last=\2 |first=\3|year=\4\5

Regionalize English: AE to BE

In Regex editor style, for direct editing

Zed to ess (recognize ⟶ recognise)

• SRCH: /((?:[a-z-[aeiuo]]{0,3}[aeiouy]{1,2}){1,}[a-z-[aeiuo]]{0,3}[iy])z((?:e|ed|es|er|ers|ing)\b)/g
• RPLC: $1s$2

Unlink bulleted external links in the body

Formatted with bullet (or subbullets), Lbracket, url, space, display txt, Rbracket, extra text

• SRCH: ^ *(\*+) *\[\s*(https?:\/\/\S+) +([^]]+)\] *(.*)$
• RPLC: \1 <nowiki>\2</nowiki> – \3 \4

JSON

Json to Lua

Some test data at

Strip first two double quotes before the first colon in every line

• SRCH: ^([^":]+)"([^":]+)":(.*)$
• RPLC: \1\2 =\3

(A follow-up simple regex is required to change (single) brackets to curlies.)

Cirrus searches

These are designed to be run as WP:Advanced search on Wikimedia:

See also: Template:Regex

Articles with lots of Talk page archives

Note: Only the Donald Trump page has more than 150 archives (202, at this writing). Find other articles that have over 100 archives:

• search: Talk: intitle:/\/Archive <100-1000>/ -intitle:Main_page -intitle:/<2000-2025>/ -intitle/[0-9]{5}/ -intitle:Donald_Trump
• 202 Talk:Donald Trump minthreadstoarchive = 1
• 139 Talk:Jesus (no param)
• 121 Talk:United States (no param)
• 111 Talk:Israel (no param)
• 104 Talk:Race and intelligence minthreadstoarchive = 1

PCRE tips

Dot matches newline (dotall)

To turn dotall on and off within a pattern, or just for a particular group:

• (?s) – turn dotall on in pattern
• (?-s) – turn dotall off in pattern
• (?s:...) – turn dotall on for a specific group:
• (?-s:...) – turn dotall off for a specific group:

Lookaround

Lookahead

• (?=) – Positive lookahead
  • q(?=u) – q followed by u; nothing saved
  • q(?=(u)) – q followed by u; u saved in \1
  • – does not match quit, and never matches any string (one char cannot be both u and i)
• (?!) Negative lookahead –
  • q(?!u) – q not followed by u

Lookbehind

• (?<=text) – Positive lookbehind –
• (?<!text) – Negative lookbehind –

Examples

• (?<=a)b against "thingamabob" – matches first b
• \b\w+(?<!s)\b – words not ending in s
  • \b\w*[^s\W]\b – same thing w/o lookbehind, but which is easier to understand?
  • note that \b\w+[^s]\b won't do it, e.g., cuz of "John's"
• Double requirements; e.g., length + contents:
  • a word that is six letters long and contains the three consecutive letters cat:
    • no lookahead: cat\w{3}|\wcat\w{2}|\w{2}cat\w|\w{3}cat
    • lookahead, #1: (?=\b\w{6}\b)\b\w*cat\w*\b
    • optimize, #2: (?=\b\w{6}\b)\w*cat\w* – by removing zero-length word boundaries
    • optimize, #3: (?=\b\w{6}\b)\w{0,3}cat\w* – in a successful match, there can never be more than 3 letters before "cat"
    • optimize, #4: \b(?=\w{6}\b)\w{0,3}cat\w* – since 1st \b is zero-length, there's no need to put it inside the lookahead
  • any word 6–12 letters long containing "cat", "dog" or "mouse":
    • \b(?=\w{6,12}\b)\w{0,9}(cat|dog|mouse)\w* – using optimizations from above

Conditionals

• (?ifthen|else) – then and else can be any regex
• (?(?=regex)then|else) – conditional with positive lookahead in if part
• Example: extract email headers