ntropy

Query language

search (and delete) accept a small query language for narrowing which notes to work with. The fastest way to learn it is by example:

# What was I supposed to do for work that isn't done yet?
ntropy search tag:work and not status:done

# That meeting note where somebody said "deadline"...
ntropy search text:deadline and tag:meeting

# Everything still in progress, or anything that's on fire.
ntropy search 'status:"in progress" or tag:urgent'

# I know I wrote the phrase "borrow checker" somewhere in here.
ntropy search "borrow checker"

# A note that mentions both "borrow" and "checker", apart or in any order.
ntropy search borrow checker

# Just show me the whole pile.
ntropy search

Terms

A bare word or quoted phrase that is not followed by : searches the note body. For more precise matching, reach for a typed term:

  • tag:x matches hierarchically. x and each note tag are split on /, and the term matches when x's segments appear as a contiguous run anywhere in the tag's segments. tag:programming therefore matches programming, programming/rust, and area/programming. Tag matching is case-insensitive.
  • field:value checks frontmatter. The field's scalar value must equal value exactly, or, for a list-valued field, the list must contain value. The value match is case-sensitive, the field name is not: Status:done finds a note with a status field, and a note with both Status: open and status: done matches status:open and status:done. A number is compared as ntropy reads it: x: 2.0 is found by x:2.0 and not by x:2, and x: 1.50 by x:1.5. tag and text always mean the terms above, in any letter case, so a frontmatter field named tag or text cannot be queried.
  • text:pattern is a regex matched against the note body, compiled with smart case: an all-lowercase pattern matches regardless of case, but a pattern containing an uppercase letter only matches that case exactly.

Words, phrases and quoting

Each argument you type is read on its own. Several words inside one argument, as in ntropy search "borrow checker", are a phrase: the words in that order with any whitespace between them, a line break included, and every character literal, so "c++ tips" is not a pattern. Separate arguments, as in ntropy search borrow checker, find notes that contain every word, anywhere in the body and in any order. An argument that is a whole query, such as 'tag:work and not status:done', stays a query, and a phrase that holds a :, a parenthesis, a quote or the word and, or or not is one too. Write such a phrase with double quotes inside single quotes: ntropy search '"page not found"'.

Quote a value that contains spaces, such as status:"in progress". A dot needs no quotes, so tag:milestone/0.13 works as it is; other punctuation such as + and # does (tag:"c++"). A quoted string may also contain regex metacharacters, and a literal quote or backslash inside one can be written as \" or \\. A bare search word matches as written, a dot included: node.js finds "node.js" and not "nodexjs". A quoted string is a pattern, so "node.js" finds both.

Combining terms

Combine terms with and, or, and not. not binds tighter than and, which binds tighter than or, so add parentheses when that default order isn't what you mean. Two terms side by side are an and: tag:work status:open is tag:work and status:open, and a b or c is (a and b) or c.

ntropy search '(tag:work or tag:side-project) and not status:done'

Beside other arguments, an argument that holds a whole query is grouped: ntropy search 'a or b' c is (a or b) and c. Blank arguments are ignored.

A word that is not an operator is always a search word. A mistyped operator, as in tag:work nd status:done, is therefore not an error but a search for nd, and a doubled operator as in a and and b searches for the word and.