Infra: Add search to the PEP index #4083

Activity

1. hugovk

   added

   infraCore infrastructure for building and rendering PEPsCore infrastructure for building and rendering PEPs

   on Oct 27, 2024
2. hugovk

   changed the title [-]Add search to the PEP index[/-] [+]Meta: Add search to the PEP index[/+] on Oct 27, 2024
3. hugovk commented on Oct 27, 2024

   

   hugovk

   on Oct 27, 2024

   Member

   More actions

   Agreed that it would be useful, it's something I often miss.

   👍React with 👍2Reacted by Nice Zombies and Carol Willing
4. hugovk

   changed the title [-]Meta: Add search to the PEP index[/-] [+]Infra: Add search to the PEP index[/+] on Oct 27, 2024
5. willingc commented on Oct 31, 2024

   

   willingc

   on Oct 31, 2024

   Contributor

   More actions

   Nice idea @nineteendo.
6. miketheman commented on Jan 30, 2025

   

   miketheman

   on Jan 30, 2025

   Member

   More actions

   I spent some time looking into this today, and here's what I've found so far:

   sphinx search was intentionally disabled in #1934 (comment) , recommending folks use public search engines with site-scoping.
   While that's not a bad solution, it's not advertised anywhere user-facing, and implementing a search box on the site that would use a public (maybe even popular) search engine may be counter to some folks' desires to keep their searches private, and I'd hate to drag people into a weird search.

   I tried reenabling sphinx search, and it's not too bad of a change, but since the general approach sphinx "wants" is to have a theme that extends an existing one (like basic) to inject the right JS snippets, so in order to make this work we'd need to either vendor the JS or figure our a smart JS loader from the pre-packed code - I couldn't figure that out yet.

   Largely, it comes from here:
   

   peps/pep_sphinx_extensions/pep_processor/html/pep_html_builder.py

   Lines 9 to 23 in 1fb889b

   	class FileBuilder(StandaloneHTMLBuilder):
   	copysource = False # Prevent unneeded source copying - we link direct to GitHub
   	search = False # Disable search
   	# Things we don't use but that need to exist:
   	indexer = None
   	relations = {}
   	_script_files = _css_files = []
   	globalcontext = {"script_files": [], "css_files": []}
   	def prepare_writing(self, _doc_names: set[str]) -> None:
   	self.docwriter = HTMLWriter(self)
   	_opt_parser = OptionParser([self.docwriter], defaults=self.env.settings, read_config_files=True)
   	self.docsettings = _opt_parser.get_default_values()
   	self._orig_css_files = self._orig_js_files = []

   I'd be happy to continue to poke at this if there's continued interest in having a PEP web search, and don't want to invest the effort if it's not going to pass a review ;)

   👍React with 👍2Reacted by Nice Zombies and Hugo van Kemenade
7. AA-Turner commented on Jan 30, 2025

   

   AA-Turner

   on Jan 30, 2025

   Last edited by AA-Turner

   Member

   More actions

   Indeed, disabling search was intentional, both to improve performance and as Sphinx's search isn't that great, so readers are better served by using their favourite search engine (there are many discussions about this on Sphinx's side, notably https://github.com/orgs/sphinx-doc/discussions/12419, but the précis is that improving search is costly and will take a not insignificant amount of effort to do right).

   A
8. miketheman commented on Jan 31, 2025

   

   miketheman

   on Jan 31, 2025

   Member

   More actions

   Thanks @AA-Turner - kinda makes sense.

   I took another approach for this - using pagefind - and it seems to be a workable idea - check it out! #4247

   🎉React with 🎉2Reacted by Hugo van Kemenade and Nice Zombies
9. hugovk

   closed this as completedin #4247on Jan 24, 2026