WMDE Technical Wishes/AdvancedSearch/Functional scope

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search

The AdvancedSearch extension aims to make existing search options more visible and accessible for everyone. It focuses on the most common search options to prevent an overloaded interface and aims to use labels that are precise and easy-to-understand.

Here’s an overview of the parameters the extension already offers, is planning to add and cannot offer. Please note that all parameters that are not integrated in the extension can still be entered manually in the search field at the top. The list doesn't cover 100% of all existing keywords, but all keywords we already have discussed/looked into by now.

As always, feedback is very welcome!

Use case Search field Implementation Description and remarks included
full text search These words default search Searches without any restriction for words in the title or text. Green tickY yes
exact search Exactly this text "" Searches for a specific sequence of characters. Punctuation marks are ignored. Green tickY yes
excluding words Not these words - Excludes pages containing these words from the search. Green tickY yes
one of the given words One of these words OR Searches for pages containing at least one of the given terms. Green tickY yes
search in page title Page title contains intitle: Searches for pages with titles that contain this word. Green tickY yes
search for pages with specific templates Only pages with this template hastemplate: Searches for pages containing specific templates. Green tickY yes
searching for files of a particular type File type filetype:

filemime:

Searches only for files of the selected type (e.g. jpeg) or of a predefined bundle (e.g. image). Allows specification of width and height for images/videos. Green tickY yes
search in categories and subcategories deepcategory: Searches in categories and subcategories. Green tickY yes
search in categories / incategory: Searches in categories, but not in subcategories. This leads to very confusing search results. Instead of implementing this keyword, we were adding support for the new deep category search keyword that was implemented by the WMF's search team in 2018. Red XN not planned
search for content in a specific language inlanguage: Searches for content in a specific language. This option will only be visible for wikis that have the Translate extension installed. Green tickY yes
searching for subpages of a page subpageOf: Searches for all subpages of a page. The results are unordered, i.e. different hierarchy levels are mixed together in one list. This keyword works in combination with all namespaces (unlike prefix: search). Green tickY yes
searching for subpages or in page titles via prefix / prefix: prefix: will not be integrated in the extension because of its side effects: The value of this keyword always includes a namespace, e.g. “Wikipedia” in prefix:Wikipedia:Technische Wünsche. Having namespaces both here and in the namespace selection bar at the bottom (“Search In”) would create a very confusing user interface. Red XN not planned
prefer pages with certain templates / boost-templates: Ranks images/articles with certain templates higher in the search results. Not planned because this isn’t a commonly used keyword. Red XN not planned
prefer pages that were recently edited / prefer-recent: Ranks recently edited pages higher in the search results. Not planned because this isn’t a commonly used keyword. Red XN not planned
searching for links to a specific page / linksto: Searches for pages which link to a specific page. Not planned because this isn’t a commonly used keyword. Red XN not planned
searching in the source code  / insource: Searches in the source code of a page, e.g. to find markup. Searching in source code usually includes characters such as ~@#&*()-+{}[]|\<>?.\.

In order for the search to work, these characters must be entered as regular expressions. However, searches with regular expressions have serious performance problems. They create a heavy load on the search backend and often don’t deliver any results (i.e. they time out). Because of these performance problems insource: cannot be integrated. We don’t consider it reasonable to integrate insource: without regular expressions, because this would limit the applicability of the source code search severely.

Red XN not planned
searching for approximate matches (fuzzy search) / ~ Searches for approximate matches. This will not be integrated because fuzzy search is a complex concept and there is no short, easily understandable label that can describe the concept in the Advanced Search interface. Red XN not planned