vrl/stdlib/
find.rs

1use crate::compiler::prelude::*;
2use std::sync::LazyLock;
3
4static DEFAULT_FROM: LazyLock<Value> = LazyLock::new(|| Value::Integer(0));
5
6static PARAMETERS: LazyLock<Vec<Parameter>> = LazyLock::new(|| {
7    vec![
8        Parameter::required("value", kind::BYTES, "The string to find the pattern in."),
9        Parameter::required(
10            "pattern",
11            kind::BYTES | kind::REGEX,
12            "The regular expression or string pattern to match against.",
13        ),
14        Parameter::optional("from", kind::INTEGER, "Offset to start searching.")
15            .default(&DEFAULT_FROM),
16    ]
17});
18
19#[allow(clippy::cast_possible_wrap)]
20fn find(value: Value, pattern: Value, from: Value) -> Resolved {
21    // TODO consider removal options
22    #[allow(clippy::cast_sign_loss, clippy::cast_possible_truncation)]
23    let from = from.try_integer()? as usize;
24
25    Ok(FindFn::find(value, pattern, from)?
26        .map_or(Value::Null, |value| Value::Integer(value as i64)))
27}
28
29#[derive(Clone, Copy, Debug)]
30pub struct Find;
31
32impl Function for Find {
33    fn identifier(&self) -> &'static str {
34        "find"
35    }
36
37    fn usage(&self) -> &'static str {
38        "Determines from left to right the start position of the first found element in `value` that matches `pattern`. Returns `-1` if not found."
39    }
40
41    fn category(&self) -> &'static str {
42        Category::String.as_ref()
43    }
44
45    fn return_kind(&self) -> u16 {
46        kind::INTEGER
47    }
48
49    fn parameters(&self) -> &'static [Parameter] {
50        PARAMETERS.as_slice()
51    }
52
53    fn examples(&self) -> &'static [Example] {
54        &[
55            example! {
56                title: "Match text",
57                source: r#"find("foobar", "bar")"#,
58                result: Ok("3"),
59            },
60            example! {
61                title: "Match text at start",
62                source: r#"find("foobar", "foo")"#,
63                result: Ok("0"),
64            },
65            example! {
66                title: "Match regex",
67                source: r#"find("foobar", r'b.r')"#,
68                result: Ok("3"),
69            },
70            example! {
71                title: "No matches",
72                source: r#"find("foobar", "baz")"#,
73                result: Ok("null"),
74            },
75            example! {
76                title: "With an offset",
77                source: r#"find("foobarfoobarfoo", "bar", 4)"#,
78                result: Ok("9"),
79            },
80        ]
81    }
82
83    fn compile(
84        &self,
85        _state: &state::TypeState,
86        _ctx: &mut FunctionCompileContext,
87        arguments: ArgumentList,
88    ) -> Compiled {
89        let value = arguments.required("value");
90        let pattern = arguments.required("pattern");
91        let from = arguments.optional("from");
92
93        Ok(FindFn {
94            value,
95            pattern,
96            from,
97        }
98        .as_expr())
99    }
100}
101
102#[derive(Debug, Clone)]
103struct FindFn {
104    value: Box<dyn Expression>,
105    pattern: Box<dyn Expression>,
106    from: Option<Box<dyn Expression>>,
107}
108
109impl FindFn {
110    fn find_regex_in_str(value: &str, regex: &ValueRegex, offset: usize) -> Option<usize> {
111        regex.find_at(value, offset).map(|found| found.start())
112    }
113
114    fn find_bytes_in_bytes(value: &Bytes, pattern: &Bytes, offset: usize) -> Option<usize> {
115        if pattern.len() > value.len() {
116            return None;
117        }
118        for from in offset..=(value.len() - pattern.len()) {
119            let to = from + pattern.len();
120            if value[from..to] == *pattern {
121                return Some(from);
122            }
123        }
124        None
125    }
126
127    fn find(value: Value, pattern: Value, offset: usize) -> ExpressionResult<Option<usize>> {
128        match pattern {
129            Value::Bytes(bytes) => Ok(Self::find_bytes_in_bytes(
130                &value.try_bytes()?,
131                &bytes,
132                offset,
133            )),
134            Value::Regex(regex) => Ok(Self::find_regex_in_str(
135                &value.try_bytes_utf8_lossy()?,
136                &regex,
137                offset,
138            )),
139            other => Err(ValueError::Expected {
140                got: other.kind(),
141                expected: Kind::bytes() | Kind::regex(),
142            }
143            .into()),
144        }
145    }
146}
147
148impl FunctionExpression for FindFn {
149    fn resolve(&self, ctx: &mut Context) -> Resolved {
150        let value = self.value.resolve(ctx)?;
151        let pattern = self.pattern.resolve(ctx)?;
152        let from = self
153            .from
154            .map_resolve_with_default(ctx, || DEFAULT_FROM.clone())?;
155
156        find(value, pattern, from)
157    }
158
159    fn type_def(&self, _: &state::TypeState) -> TypeDef {
160        TypeDef::integer().infallible()
161    }
162}
163
164#[cfg(test)]
165mod tests {
166    use regex::Regex;
167
168    use crate::value;
169
170    use super::*;
171
172    test_function![
173        find => Find;
174
175        str_matching_end {
176            args: func_args![value: "foobar", pattern: "bar"],
177            want: Ok(value!(3)),
178            tdef: TypeDef::integer().infallible(),
179        }
180
181        str_matching_beginning {
182            args: func_args![value: "foobar", pattern: "foo"],
183            want: Ok(value!(0)),
184            tdef: TypeDef::integer().infallible(),
185        }
186
187        str_matching_middle {
188            args: func_args![value: "foobar", pattern: "ob"],
189            want: Ok(value!(2)),
190            tdef: TypeDef::integer().infallible(),
191        }
192
193        str_too_long {
194            args: func_args![value: "foo", pattern: "foobar"],
195            want: Ok(value!(null)),
196            tdef: TypeDef::integer().infallible(),
197        }
198
199        regex_matching_end {
200            args: func_args![value: "foobar", pattern: Value::Regex(Regex::new("bar").unwrap().into())],
201            want: Ok(value!(3)),
202            tdef: TypeDef::integer().infallible(),
203        }
204
205        regex_matching_start {
206            args: func_args![value: "foobar", pattern: Value::Regex(Regex::new("fo+z?").unwrap().into())],
207            want: Ok(value!(0)),
208            tdef: TypeDef::integer().infallible(),
209        }
210
211        wrong_pattern {
212            args: func_args![value: "foobar", pattern: Value::Integer(42)],
213            want: Err("expected string or regex, got integer"),
214            tdef: TypeDef::integer().infallible(),
215        }
216    ];
217}
vrl/stdlib/find.rs

vrl/stdlib/
find.rs
