Searching and Finding

Entering search queries

To start a search, enter a query into the short query input field and type enter or click the search icon. By default, the names, summary, tags, content, namengram, summaryngram, and contentngram fields are searched.

The search results view provides a form for refining the search through ajax updates. A transaction is started each time a character is added or removed in the search field. If keying is rapid, it is possible that results will processed out of order. The Whoosh query shows the last term processed.

Clicking the Search Options link displays alternatives for modifying the search. Ajax updates will be made whenever a radio button or checkbox is changed.

Below the search form is the query processed by Whoosh, and Whoosh generated suggestions for additional searches by input, item name, and item content.

Finally, the search hits are presented. These are ordered by the whoosh scoring number. Each hit will contain the item name and some meta data. If available, the item summary and partial item content with the search term highlighted will be shown.

Simple search queries

Just enter one or more words into the query input field and hit Enter.

If your query consists of multiple words, it will only find documents containing ALL those words. You can use AND, OR, NOT to refine your search. “AND” is the default.

Examples

Search for “wiki”:

wiki

Search for documents containing “wiki” AND “moin”:

wiki moin

Explicit alternative (does the same as above):

wiki AND moin

Search for documents containing “wiki” OR “moin”:

wiki OR moin

Search for documents containing “wiki” and NOT “bad”:

wiki NOT bad

Explicit alternative (does the same as above):

wiki AND NOT bad

Group terms using ():

wiki AND NOT (bad OR worse)

Redirect to best match

If you know the target item name, start the search term with a back-slash character. Only the names and namengram fields will be searched. If there is a hit, the browser will be redirected to the highest scoring hit.

Examples

Search for a specific item name and immediately redirect browser to best match:

\Home
\Home/subitem
\users/JoeDoe

Using wildcards

If you want to enter word fragments or if you are not sure about spelling or word form, you can use wildcards for the parts you do not know:

Wildcard	Matches
`?`	one arbitrary character
`*`	any count of arbitrary characters

Examples

Search for something like wiki, wika, wikb, …:

wik?

Search for something like wiki, willi, wi, …:

w*i

You can also use it for poor man’s language independent word stemming.

Matches on clean, cleaner, cleanest, cleaning, …:

clean*

Using regular expressions

Regular expressions enable even more flexibility for specifying search terms.

See https://en.wikipedia.org/wiki/Regular_expression for basics about regexes.

See https://docs.python.org/3/library/re.html about python’s regex implementation, which we use for MoinMoin.

You need to use this syntax when entering regexes: r”yourregex”

Examples

Search for hello or hallo:

r"h[ae]llo"

Search for words starting with foo:

r"^foo"
r"\Afoo"

Search for something like wiki, wika, wikb, …:

r"wik."

Search for something like wiki, willi, wi, …:

r"w.*i"

Searching in specific fields

If not specified otherwise, moin will search in names, tags, summary, comment and content fields. Three fields with n-gram support are also searched by default: namengram, summaryngram and contentngram.

N-gram indexing is a powerful method for getting fast, “search as you type” functionality. A tokinizer splits words within ngram content fields into strings of 3 to 6 characters. These small strings may be matched against search terms that are tokinized into strings of 3 to 6 characters.

To specify the field to search in, just use the fieldname:searchterm syntax. If embedded spaces are desired then do: fieldname:”search term”. Separate multiple terms with a space: content:foo tags:Foo is the same as content:foo AND tags:Foo.

The following table includes fields that may be useful for searching.

Field name	Field value
`acl` **	access control list (see below)
`address`	submitter IP address, e.g. 127.0.0.1
`comment`	editor comment on save, rename, etc.
`content`	document contents, e.g. This is some example content.
`contentngram` **	document contents, tokenized by 3 to 6 characters.
`contenttype`	document type: text, image, audio, moinwiki, jpg, …
`itemlinks` **	link targets of the document, e.g. OtherItem
`itemtransclusions` **	transclusion targets of the document, e.g. OtherItem
`language`	(main) language of the document contents, e.g. en
`mtime`	document modification (submission) date, 2011-08-07
`namengram` **	document names, tokenized by 3 to 6 characters.
`names`	document names, e.g. Home, MyWikiPage
`namespace`	namespace:”” for default or namespace:users
`name_exact`	same as `name`, but is not tokenized
`name_old`	name_old:* for all renamed items
`summary`	summary text, if provided by author
`summaryngram` **	summary text, tokenized by 3 to 6 characters.
`tags`	tags of the document, e.g. important, hard, todo
`username`	submitter user name, e.g. JoeDoe
`wikiname`	wiki name, e.g. ITWiki, EngineeringWiki, SalesWiki

** These fields exist only in the current revisions index, see Notes below.

Examples

Search in metadata fields:

contenttype:text
contenttype:image/jpeg
tags:todo
mtime:2022-01-08  # use ISO 8601 dates, not time; `mtime:2022-01 works
address:127.0.0.1
username:JoeDoe

Search items with an item ACL that explicitly gives Joe read rights:

acl:Joe:+read

Limiting search to a specific wiki, for example in a wiki farm’s shared index:

wikiname:SomeWiki  # requires correct caps

Notes

There are two indexes. The smaller index is used by default. It only indexes the current revision of each item. The larger index is used when the All radio button under the Search Options link is selected. The larger indexes all revisions of all items including revisions of deleted items. As noted in the table above the larger index omits several fields to save space.

By default, all namespaces and all wikinames are searched, including the userprofiles index. Because the userprofiles index is normally read restricted, hits will be blocked and included as n items are not shown because read permission was denied at the bottom of the page.

Items with transcluded content do not contain the transcluded content within the item’s index. An item containing “foo” within its content and trancluding an item with “bar” within its content cannot be matched by searching for “foo AND bar”. Both items will be matched by searching for “foo OR bar”.

Moin only uses an indexed search. Keep in mind that this has some special properties:

By using an index, the search is fast

Because it is only using an index, it can only find what was put there

If you use wildcards or regexes, it will still use the index, but in a different, slower way

For example:

create an item with “FooBar” in the name, content, summary, tag, and comment fields

search for “ooba” - the namengram, summaryngram, and contentngram will match

search for “FooBar”: names, namengram, tags, summary, summaryngram, content, contentngram, and comment will match

search for “foobar”: names, namengram, summary, summaryngram, content, contentngram, and comment will match

More information

See the Whoosh query language docs.