More string extractors

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

@@ -4,6 +4,20 @@ use std::sync::Arc;
 
 use crate::{extract::traits::ObjectExtractor, value::PileValue};
 
+fn parse_name(s: &str) -> (&str, Option<&str>) {
+	match s.find('(') {
+		None => (s, None),
+		Some(i) => {
+			let name = &s[..i];
+			let rest = &s[i + 1..];
+			match rest.strip_suffix(')') {
+				Some(args) => (name, Some(args)),
+				None => (name, None),
+			}
+		}
+	}
+}
+
 pub struct StringExtractor {
 	item: Arc<SmartString<LazyCompact>>,
 }
@@ -17,24 +31,49 @@ impl StringExtractor {
 #[async_trait::async_trait]
 impl ObjectExtractor for StringExtractor {
 	async fn field(&self, name: &Label) -> Result<Option<PileValue>, std::io::Error> {
-		Ok(match name.as_str() {
-			"trim" => Some(PileValue::String(Arc::new(
+		let (name, args) = parse_name(name.as_str());
+		Ok(match (name, args) {
+			("trim", None) => Some(PileValue::String(Arc::new(
 				self.item.as_str().trim().into(),
 			))),
 
-			"upper" => Some(PileValue::String(Arc::new(
+			("upper", None) => Some(PileValue::String(Arc::new(
 				self.item.as_str().to_lowercase().into(),
 			))),
 
-			"lower" => Some(PileValue::String(Arc::new(
+			("lower", None) => Some(PileValue::String(Arc::new(
 				self.item.as_str().to_uppercase().into(),
 			))),
 
-			"nonempty" => Some(match self.item.is_empty() {
+			("nonempty", None) => Some(match self.item.is_empty() {
 				true => PileValue::Null,
 				false => PileValue::String(self.item.clone()),
 			}),
 
+			("trimprefix", Some(prefix)) => Some(PileValue::String(Arc::new(
+				self.item
+					.as_str()
+					.strip_prefix(prefix)
+					.unwrap_or(self.item.as_str())
+					.into(),
+			))),
+
+			("trimsuffix", Some(suffix)) => Some(PileValue::String(Arc::new(
+				self.item
+					.as_str()
+					.strip_suffix(suffix)
+					.unwrap_or(self.item.as_str())
+					.into(),
+			))),
+
+			("split", Some(by)) => Some(PileValue::Array(Arc::new(
+				self.item
+					.as_str()
+					.split(by)
+					.map(|s| PileValue::String(Arc::new(s.into())))
+					.collect(),
+			))),
+
 			_ => None,
 		})
 	}
@@ -49,3 +88,134 @@ impl ObjectExtractor for StringExtractor {
 		]);
 	}
 }
+
+#[cfg(test)]
+mod tests {
+	use super::*;
+
+	fn extractor(s: &str) -> StringExtractor {
+		StringExtractor::new(&Arc::new(s.into()))
+	}
+
+	#[expect(clippy::unwrap_used)]
+	async fn field(ext: &StringExtractor, name: &str) -> Option<PileValue> {
+		ext.field(&Label::new(name).unwrap()).await.unwrap()
+	}
+
+	fn string(v: Option<PileValue>) -> Option<String> {
+		match v? {
+			PileValue::String(s) => Some(s.as_str().to_owned()),
+			_ => panic!("expected string"),
+		}
+	}
+
+	fn array(v: Option<PileValue>) -> Vec<String> {
+		match v.expect("expected Some") {
+			PileValue::Array(arr) => arr
+				.iter()
+				.map(|v| match v {
+					PileValue::String(s) => s.as_str().to_owned(),
+					_ => panic!("expected string element"),
+				})
+				.collect(),
+			_ => panic!("expected array"),
+		}
+	}
+
+	#[tokio::test]
+	async fn trim() {
+		assert_eq!(
+			string(field(&extractor("  hi  "), "trim").await),
+			Some("hi".into())
+		);
+	}
+
+	#[tokio::test]
+	async fn trim_no_args() {
+		assert!(field(&extractor("x"), "trim(foo)").await.is_none());
+	}
+
+	#[tokio::test]
+	async fn nonempty_with_content() {
+		assert!(matches!(
+			field(&extractor("hello"), "nonempty").await,
+			Some(PileValue::String(_))
+		));
+	}
+
+	#[tokio::test]
+	async fn nonempty_empty_string() {
+		assert!(matches!(
+			field(&extractor(""), "nonempty").await,
+			Some(PileValue::Null)
+		));
+	}
+
+	#[tokio::test]
+	async fn trimprefix_present() {
+		assert_eq!(
+			string(field(&extractor("foobar"), "trimprefix(foo)").await),
+			Some("bar".into())
+		);
+	}
+
+	#[tokio::test]
+	async fn trimprefix_absent() {
+		assert_eq!(
+			string(field(&extractor("foobar"), "trimprefix(baz)").await),
+			Some("foobar".into())
+		);
+	}
+
+	#[tokio::test]
+	async fn trimprefix_no_args() {
+		assert!(field(&extractor("foobar"), "trimprefix").await.is_none());
+	}
+
+	#[tokio::test]
+	async fn trimsuffix_present() {
+		assert_eq!(
+			string(field(&extractor("foobar"), "trimsuffix(bar)").await),
+			Some("foo".into())
+		);
+	}
+
+	#[tokio::test]
+	async fn trimsuffix_absent() {
+		assert_eq!(
+			string(field(&extractor("foobar"), "trimsuffix(baz)").await),
+			Some("foobar".into())
+		);
+	}
+
+	#[tokio::test]
+	async fn split_basic() {
+		assert_eq!(
+			array(field(&extractor("a,b,c"), "split(,)").await),
+			vec!["a", "b", "c"]
+		);
+	}
+
+	#[tokio::test]
+	async fn split_no_match() {
+		assert_eq!(
+			array(field(&extractor("abc"), "split(,)").await),
+			vec!["abc"]
+		);
+	}
+
+	#[tokio::test]
+	async fn split_no_args() {
+		assert!(field(&extractor("abc"), "split").await.is_none());
+	}
+
+	#[tokio::test]
+	async fn split_unclosed_paren() {
+		assert!(field(&extractor("abc"), "split(,").await.is_none());
+	}
+
+	#[tokio::test]
+	async fn unknown_field() {
+		assert!(field(&extractor("abc"), "bogus").await.is_none());
+	}
+}

default.nix

@@ -0,0 +1,35 @@
+{
+  pkgs ? import <nixpkgs> { },
+}:
+
+pkgs.rustPlatform.buildRustPackage {
+  pname = "pile";
+  version = "0.0.1";
+  src = ./.;
+
+  cargoLock.lockFile = ./Cargo.lock;
+
+  buildInputs = [
+    pkgs.pdfium-binaries
+    pkgs.openssl
+  ]
+  ++ pkgs.lib.optionals pkgs.stdenv.isDarwin [
+    pkgs.darwin.apple_sdk.frameworks.Security
+    pkgs.darwin.apple_sdk.frameworks.SystemConfiguration
+  ];
+
+  nativeBuildInputs = [
+    pkgs.pkg-config
+    pkgs.makeWrapper
+  ];
+
+  postInstall = ''
+    wrapProgram $out/bin/pile \
+      --prefix LD_LIBRARY_PATH : ${pkgs.lib.makeLibraryPath [ pkgs.pdfium-binaries ]}
+  '';
+
+  meta = {
+    description = "pile - flexible file indexing";
+    mainProgram = "pile";
+  };
+}

flake.nix

@@ -0,0 +1,72 @@
+{
+  description = "pile - personal data indexer";
+
+  inputs = {
+    nixpkgs.url = "github:NixOS/nixpkgs/nixos-unstable";
+    rust-overlay = {
+      url = "github:oxalica/rust-overlay";
+      inputs.nixpkgs.follows = "nixpkgs";
+    };
+    flake-utils.url = "github:numtide/flake-utils";
+  };
+
+  outputs = { self, nixpkgs, rust-overlay, flake-utils }:
+    flake-utils.lib.eachDefaultSystem (system:
+      let
+        overlays = [ (import rust-overlay) ];
+        pkgs = import nixpkgs { inherit system overlays; };
+
+        rustToolchain = pkgs.rust-bin.stable."1.85.0".default;
+
+        rustPlatform = pkgs.makeRustPlatform {
+          cargo = rustToolchain;
+          rustc = rustToolchain;
+        };
+
+      in {
+        packages.default = rustPlatform.buildRustPackage {
+          pname = "pile";
+          version = "0.0.1";
+          src = ./.;
+
+          cargoLock.lockFile = ./Cargo.lock;
+
+          buildInputs = [
+            pkgs.pdfium-binaries
+            pkgs.openssl
+          ] ++ pkgs.lib.optionals pkgs.stdenv.isDarwin [
+            pkgs.darwin.apple_sdk.frameworks.Security
+            pkgs.darwin.apple_sdk.frameworks.SystemConfiguration
+          ];
+
+          nativeBuildInputs = [
+            pkgs.pkg-config
+            pkgs.makeWrapper
+          ];
+
+          # pdfium-render loads libpdfium dynamically at runtime
+          postInstall = ''
+            wrapProgram $out/bin/pile \
+              --prefix LD_LIBRARY_PATH : ${pkgs.lib.makeLibraryPath [ pkgs.pdfium-binaries ]}
+          '';
+
+          meta = {
+            description = "pile - personal data indexer and search tool";
+            homepage = "https://github.com/markus/pile";
+            license = pkgs.lib.licenses.mit;
+            mainProgram = "pile";
+          };
+        };
+
+        devShells.default = pkgs.mkShell {
+          buildInputs = [
+            rustToolchain
+            pkgs.pdfium-binaries
+            pkgs.openssl
+            pkgs.pkg-config
+          ];
+
+          LD_LIBRARY_PATH = pkgs.lib.makeLibraryPath [ pkgs.pdfium-binaries ];
+        };
+      });
+}