Search - General Questions

[shaneholloman]

Question 1 - Search "Fields"

Lunr has the notion of:

idx.search('title:foo* bar')

In mkdocs-material' case, this works well, awesome

However, I couldn't find any other "fields" that mkdocs-material is currently leveraging.
May we get a list of fields we can explicitly search within?

Question 2 - Searching Fenced Code

For instance:

sudo anything here

Is not visible in search. This is where Question 1 comes to play

For example I would like to search a command I forgot like

sudo -H pip install --ignore-installed six --upgrade six

By using a search like

code:sudo -H pip install *

Again this works great if I know the available fields.
So for titles:

title:sudo -H pip install

Is great, but only if that was in a title (h1, h2, h3 etc.), which it sometimes is, but normally isn't.

[squidfunk]

Question 1 - Search "Fields"

Searchable fields are:

• title
• text
• tags (currently Insiders only)

Note that the syntax demands for prefixing every word with the field, i.e. title:foo title:bar, because title:foo barmeans "documents with foo in title and bar in any field". This will, however, yield an OR search. If you want an AND search, you would have to search like this:

+title:foo +title:bar

... or let Material for MkDocs preprocess the query string and use quotes as known from Google:

"title:foo title:bar"

Why is this not documented? I think you very much see how complex and ugly the syntax is and most users just don't bother. My goal is to provide a search implementation that returns good results with simple keyword search and presents them in a way so that you can quickly skim through the results, thus omitting the need for explanation.

Question 2 - Searching Fenced Code

Code blocks are just converted to regular text when MkDocs is building the index, so there're no remnants of "hey, this is text inside a code block". One would have to monkey-patch the search plugin to somehow set markers and allow for dedicated indexing of code examples. If you look into search/search_index.json, which is what MkDocs generates, you'll see that we have no information your feature request could be based on.