Add regex extractor

This commit is contained in:
2026-03-21 09:27:12 -07:00
parent ed169b3ab4
commit b6cb5870b4
10 changed files with 209 additions and 6 deletions

View File

@@ -0,0 +1,104 @@
use std::sync::Arc;
use pile_config::Label;
use regex::Regex;
use smartstring::{LazyCompact, SmartString};
use crate::{
extract::traits::{ExtractState, ListExtractor, ObjectExtractor},
value::PileValue,
};
struct RegexData {
regex: Arc<Regex>,
/// Captured substrings indexed by group index (0 = whole match).
captures: Vec<Option<Arc<SmartString<LazyCompact>>>>,
}
impl RegexData {
fn new(regex: Arc<Regex>, input: &str) -> Option<Self> {
let caps = regex.captures(input)?;
let captures = caps
.iter()
.map(|m| m.map(|m| Arc::new(m.as_str().into())))
.collect();
Some(Self { regex, captures })
}
}
/// Exposes named capture groups as object fields.
pub struct RegexExtractor(Arc<RegexData>);
impl RegexExtractor {
/// Run `regex` against `input`. Returns `None` if there is no match.
pub fn new(regex: Arc<Regex>, input: &str) -> Option<Self> {
Some(Self(Arc::new(RegexData::new(regex, input)?)))
}
}
#[async_trait::async_trait]
impl ObjectExtractor for RegexExtractor {
async fn field(
&self,
_state: &ExtractState,
name: &Label,
args: Option<&str>,
) -> Result<Option<PileValue>, std::io::Error> {
if args.is_some() {
return Ok(None);
}
let Some(idx) = self
.0
.regex
.capture_names()
.position(|n| n == Some(name.as_str()))
else {
return Ok(None);
};
Ok(Some(
match self.0.captures.get(idx).and_then(|v| v.as_ref()) {
Some(s) => PileValue::String(s.clone()),
None => PileValue::Null,
},
))
}
async fn fields(&self) -> Result<Vec<Label>, std::io::Error> {
#[expect(clippy::unwrap_used)]
Ok(self
.0
.regex
.capture_names()
.flatten()
.map(|n| Label::new(n).unwrap())
.collect())
}
fn as_list(&self) -> Option<Arc<dyn ListExtractor>> {
Some(Arc::new(RegexExtractor(self.0.clone())))
}
}
#[async_trait::async_trait]
impl ListExtractor for RegexExtractor {
async fn get(
&self,
_state: &ExtractState,
idx: usize,
) -> Result<Option<PileValue>, std::io::Error> {
let raw_idx = idx + 1;
let Some(slot) = self.0.captures.get(raw_idx) else {
return Ok(None);
};
Ok(Some(match slot {
Some(s) => PileValue::String(s.clone()),
None => PileValue::Null,
}))
}
async fn len(&self, _state: &ExtractState) -> Result<usize, std::io::Error> {
Ok(self.0.captures.len().saturating_sub(1))
}
}