vrl 0.32.0

use crate::compiler::prelude::*;
use std::sync::LazyLock;

static DEFAULT_FROM: LazyLock<Value> = LazyLock::new(|| Value::Integer(0));

static PARAMETERS: LazyLock<Vec<Parameter>> = LazyLock::new(|| {
    vec![
        Parameter::required("value", kind::BYTES, "The string to find the pattern in."),
        Parameter::required(
            "pattern",
            kind::BYTES | kind::REGEX,
            "The regular expression or string pattern to match against.",
        ),
        Parameter::optional("from", kind::INTEGER, "Offset to start searching.")
            .default(&DEFAULT_FROM),
    ]
});

#[allow(clippy::cast_possible_wrap)]
fn find(value: Value, pattern: Value, from: Value) -> Resolved {
    // TODO consider removal options
    #[allow(clippy::cast_sign_loss, clippy::cast_possible_truncation)]
    let from = from.try_integer()? as usize;

    Ok(FindFn::find(value, pattern, from)?
        .map_or(Value::Null, |value| Value::Integer(value as i64)))
}

#[derive(Clone, Copy, Debug)]
pub struct Find;

impl Function for Find {
    fn identifier(&self) -> &'static str {
        "find"
    }

    fn usage(&self) -> &'static str {
        "Determines from left to right the start position of the first found element in `value` that matches `pattern`. Returns `-1` if not found."
    }

    fn category(&self) -> &'static str {
        Category::String.as_ref()
    }

    fn return_kind(&self) -> u16 {
        kind::INTEGER
    }

    fn parameters(&self) -> &'static [Parameter] {
        PARAMETERS.as_slice()
    }

    fn examples(&self) -> &'static [Example] {
        &[
            example! {
                title: "Match text",
                source: r#"find("foobar", "bar")"#,
                result: Ok("3"),
            },
            example! {
                title: "Match text at start",
                source: r#"find("foobar", "foo")"#,
                result: Ok("0"),
            },
            example! {
                title: "Match regex",
                source: r#"find("foobar", r'b.r')"#,
                result: Ok("3"),
            },
            example! {
                title: "No matches",
                source: r#"find("foobar", "baz")"#,
                result: Ok("null"),
            },
            example! {
                title: "With an offset",
                source: r#"find("foobarfoobarfoo", "bar", 4)"#,
                result: Ok("9"),
            },
        ]
    }

    fn compile(
        &self,
        _state: &state::TypeState,
        _ctx: &mut FunctionCompileContext,
        arguments: ArgumentList,
    ) -> Compiled {
        let value = arguments.required("value");
        let pattern = arguments.required("pattern");
        let from = arguments.optional("from");

        Ok(FindFn {
            value,
            pattern,
            from,
        }
        .as_expr())
    }
}

#[derive(Debug, Clone)]
struct FindFn {
    value: Box<dyn Expression>,
    pattern: Box<dyn Expression>,
    from: Option<Box<dyn Expression>>,
}

impl FindFn {
    fn find_regex_in_str(value: &str, regex: &ValueRegex, offset: usize) -> Option<usize> {
        regex.find_at(value, offset).map(|found| found.start())
    }

    fn find_bytes_in_bytes(value: &Bytes, pattern: &Bytes, offset: usize) -> Option<usize> {
        if pattern.len() > value.len() {
            return None;
        }
        for from in offset..=(value.len() - pattern.len()) {
            let to = from + pattern.len();
            if value[from..to] == *pattern {
                return Some(from);
            }
        }
        None
    }

    fn find(value: Value, pattern: Value, offset: usize) -> ExpressionResult<Option<usize>> {
        match pattern {
            Value::Bytes(bytes) => Ok(Self::find_bytes_in_bytes(
                &value.try_bytes()?,
                &bytes,
                offset,
            )),
            Value::Regex(regex) => Ok(Self::find_regex_in_str(
                &value.try_bytes_utf8_lossy()?,
                &regex,
                offset,
            )),
            other => Err(ValueError::Expected {
                got: other.kind(),
                expected: Kind::bytes() | Kind::regex(),
            }
            .into()),
        }
    }
}

impl FunctionExpression for FindFn {
    fn resolve(&self, ctx: &mut Context) -> Resolved {
        let value = self.value.resolve(ctx)?;
        let pattern = self.pattern.resolve(ctx)?;
        let from = self
            .from
            .map_resolve_with_default(ctx, || DEFAULT_FROM.clone())?;

        find(value, pattern, from)
    }

    fn type_def(&self, _: &state::TypeState) -> TypeDef {
        TypeDef::integer().infallible()
    }
}

#[cfg(test)]
mod tests {
    use regex::Regex;

    use crate::value;

    use super::*;

    test_function![
        find => Find;

        str_matching_end {
            args: func_args![value: "foobar", pattern: "bar"],
            want: Ok(value!(3)),
            tdef: TypeDef::integer().infallible(),
        }

        str_matching_beginning {
            args: func_args![value: "foobar", pattern: "foo"],
            want: Ok(value!(0)),
            tdef: TypeDef::integer().infallible(),
        }

        str_matching_middle {
            args: func_args![value: "foobar", pattern: "ob"],
            want: Ok(value!(2)),
            tdef: TypeDef::integer().infallible(),
        }

        str_too_long {
            args: func_args![value: "foo", pattern: "foobar"],
            want: Ok(value!(null)),
            tdef: TypeDef::integer().infallible(),
        }

        regex_matching_end {
            args: func_args![value: "foobar", pattern: Value::Regex(Regex::new("bar").unwrap().into())],
            want: Ok(value!(3)),
            tdef: TypeDef::integer().infallible(),
        }

        regex_matching_start {
            args: func_args![value: "foobar", pattern: Value::Regex(Regex::new("fo+z?").unwrap().into())],
            want: Ok(value!(0)),
            tdef: TypeDef::integer().infallible(),
        }

        wrong_pattern {
            args: func_args![value: "foobar", pattern: Value::Integer(42)],
            want: Err("expected string or regex, got integer"),
            tdef: TypeDef::integer().infallible(),
        }
    ];
}