10 releases
0.4.2 | Oct 25, 2023 |
---|---|
0.4.1 | Sep 25, 2023 |
0.4.0-alpha.4 | Aug 7, 2023 |
0.4.0-alpha.3 | Jul 12, 2023 |
0.2.0 | Jun 5, 2023 |
#129 in #query-language
2,946 downloads per month
Used in sikula
16KB
255 lines
SiKuLa
Simple Query Language - [ˈziːˈkuːˈlaː]
Rationale
Another query language, are you serious?
Actually it isn't that new. But naming it "Query language similar to GitHub's search syntax" (QLSTGHSS) wasn't a real option.
Think of it more as an implementation of a familiar syntax.
What's the difference then?
They are subtle. But I don't want to spoil the surprise. Or maybe I am just too lazy documenting it. 🤷
Example
Assuming you have an enum defined for searching e-mails:
use sikula::prelude::*;
#[derive(Search, Clone, Debug, PartialEq, Eq)]
enum DeriveResource<'a> {
/// Standard qualifier: `author:someone`
#[search(sort, scope)]
Author(&'a str),
/// Default primary: `warranty`
#[search(default)]
Subject(Primary<'a>),
/// Non-default primary: `warranty in:message`, to search in both: `warranty in:message in:subject`
#[search(scope)]
Message(Primary<'a>),
/// Predicate: `is:read`
Read,
/// Numeric qualifier example:
/// * `size:100` (equals)
/// * `size:>=100` (size greater than or equals 100)
/// * `size:100..200` (size between 100 inclusive and 200 exclusive)
/// * `size:*..200` (size up to 200 exclusive)
#[search(sort)]
Size(Ordered<usize>),
#[search(sort)]
Sent(Ordered<time::OffsetDateTime>),
Label(Qualified<'a, &'a str>),
}
The Query
derive provides the trait implementation. The #[query(scope)]
attribute flags the variant Subject
as Body
scopes for the primary search terms, marking Subject
as the default if none was selected.
In general, there are three types of terms: Primary, Qualifiers, Predicates. Predicates are simple "is this condition true" style of filters. If an enum variant doesn't have any value, it is a predicate.
Qualifiers are additional matching criteria, which depend on the type of the value.
With the #[query(sort)]
flag, a field can be used for sorting the result.
Now, you can do the following queries:
Query | Retrieves all entries… |
---|---|
foo |
… containing "foo" in the "subject" |
foo in:subject in:message |
… containing "foo" in either "subject" or "body" |
foo in:subject in:message is:read |
… containing "foo" in either "subject" or "body" being "read" |
foo bar |
… containing "foo" and "bar" in the subject |
size:>10000 |
… having a size greater than 10000 |
size:100..200 |
… having a size between 100 (inclusive) and 200 (exclusive) |
-is:read |
… being "not read" |
foo sort:sent |
… containing "foo" in the subject, sorted by "sent" ascending |
foo -sort:sent |
… containing "foo" in the subject, sorted by "sent" descending |
sender:"Max Mustermann" |
… having a sender of Max Mustermann |
sender:"Max Mustermann" sender:"Eva Mustermann" |
… having a sender of Max Mustermann and Eva Mustermann (most likely no results will be found) |
sender:"Max Mustermann","Eva Mustermann" |
… having a sender of Max Mustermann or Eva Mustermann |
foo OR bar |
… containing "foo" or "bar" in the "subject" |
foo AND bar |
… containing "foo" and "bar" in the "subject" |
foo OR bar AND baz |
… containing either "foo" or ( "bar" and "baz" ) in the "subject" |
(foo OR bar) AND baz |
… containing ( "foo" or "bar" ) and "baz" in the "subject" |
foo OR bar baz |
… containing ( "foo" or "bar" ) and "baz" in the "subject" |
For testing more examples with the resource above, you can run the cli
example:
cargo run --example cli --features time -- -is:read AND foo
Which will give you a structured output of the parsed query:
Input: '-is:read AND foo'
Query {
terms: And(
[
Not(
Match(
Read,
),
),
Match(
Subject(
Partial(
"foo",
),
),
),
],
),
sorting: [],
}
Dependencies
~1–1.5MB
~29K SLoC