Add regex extractor

crates/pile-value/src/extract/mod.rs

@@ -1,4 +1,5 @@
 pub mod item;
 pub mod misc;
+pub mod regex;
 pub mod string;
 pub mod traits;

crates/pile-value/src/extract/regex.rs

@@ -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))
+	}
+}

crates/pile-value/src/extract/string.rs

@@ -1,9 +1,13 @@
 use pile_config::Label;
+use regex::Regex;
 use smartstring::{LazyCompact, SmartString};
 use std::sync::Arc;
 
 use crate::{
-	extract::traits::{ExtractState, ObjectExtractor},
+	extract::{
+		regex::RegexExtractor,
+		traits::{ExtractState, ObjectExtractor},
+	},
 	value::PileValue,
 };
 
@@ -67,6 +71,18 @@ impl ObjectExtractor for StringExtractor {
 					.collect(),
 			))),
 
+			("regex", Some(pattern)) => {
+				let Ok(re) = Regex::new(pattern) else {
+					return Ok(None);
+				};
+				Some(
+					match RegexExtractor::new(Arc::new(re), self.item.as_str()) {
+						Some(ext) => PileValue::ObjectExtractor(Arc::new(ext)),
+						None => PileValue::Null,
+					},
+				)
+			}
+
 			_ => None,
 		})
 	}
@@ -78,6 +94,9 @@ impl ObjectExtractor for StringExtractor {
 			Label::new("upper").unwrap(),
 			Label::new("lower").unwrap(),
 			Label::new("nonempty").unwrap(),
+			Label::new("trimprefix").unwrap(),
+			Label::new("trimsuffix").unwrap(),
+			Label::new("split").unwrap(),
 		]);
 	}
 }

crates/pile-value/src/extract/traits.rs

@@ -35,6 +35,11 @@ pub trait ObjectExtractor: Send + Sync {
 	/// and [None] for all others.
 	async fn fields(&self) -> Result<Vec<pile_config::Label>, std::io::Error>;
 
+	/// Return a list view of this extractor, if supported.
+	fn as_list(&self) -> Option<std::sync::Arc<dyn ListExtractor>> {
+		None
+	}
+
 	/// Convert this to a JSON value.
 	async fn to_json(&self, state: &ExtractState) -> Result<serde_json::Value, std::io::Error> {
 		let keys = self.fields().await?;

crates/pile-value/src/value/value.rs

@@ -86,7 +86,9 @@ impl PileValue {
 			Self::String(_) => Arc::new(VecExtractor::default()),
 			Self::Blob { .. } => Arc::new(VecExtractor::default()),
 			Self::ListExtractor(e) => e.clone(),
-			Self::ObjectExtractor(_) => Arc::new(VecExtractor::default()),
+			Self::ObjectExtractor(e) => e
+				.as_list()
+				.unwrap_or_else(|| Arc::new(VecExtractor::default())),
 			Self::Item(_) => Arc::new(VecExtractor::default()),
 		}
 	}