Validate label names

crates/pile-dataset/src/index/index_fts.rs

@@ -1,5 +1,5 @@
 use jsonpath_rust::JsonPath;
-use pile_config::{ConfigToml, DatasetFts};
+use pile_config::{ConfigToml, DatasetFts, Label};
 use serde_json::Value;
 use std::{path::PathBuf, sync::LazyLock};
 use tantivy::{
@@ -106,7 +106,7 @@ impl DbFtsIndex {
 	pub fn get_field(
 		&self,
 		json: &Value,
-		field_name: &str,
+		field_name: &Label,
 	) -> Result<Option<String>, std::io::Error> {
 		let field = match self.cfg.schema.get(field_name) {
 			Some(x) => x,
@@ -124,7 +124,7 @@ impl DbFtsIndex {
 						warn!(
 							message = "Path returned more than one value, this is not supported. Skipping.",
 							?path,
-							field = field_name
+							field = field_name.to_string()
 						);
 						continue;
 					}
@@ -139,7 +139,7 @@ impl DbFtsIndex {
 					warn!(
 						message = "Invalid path, skipping",
 						?path,
-						field = field_name,
+						field = field_name.to_string(),
 						?error
 					);
 					continue;
@@ -150,7 +150,7 @@ impl DbFtsIndex {
 				Value::Null => {
 					trace!(
 						message = "Skipping field, is null",
-						field = field_name,
+						field = field_name.to_string(),
 						path,
 						value = ?val
 					);
@@ -174,7 +174,7 @@ impl DbFtsIndex {
 						} else if x.len() > 1 {
 							debug!(
 								message = "Skipping field, is array with more than one element",
-								field = field_name,
+								field = field_name.to_string(),
 								path,
 								value = ?val
 							);
@@ -182,7 +182,7 @@ impl DbFtsIndex {
 						} else {
 							debug!(
 								message = "Skipping field, is empty array",
-								field = field_name,
+								field = field_name.to_string(),
 								path,
 								value = ?val
 							);
@@ -192,7 +192,7 @@ impl DbFtsIndex {
 					Value::Null => {
 						trace!(
 							message = "Skipping field, is null",
-							field = field_name,
+							field = field_name.to_string(),
 							path,
 							value = ?val
 						);
@@ -201,7 +201,7 @@ impl DbFtsIndex {
 					Value::Object(_) => {
 						trace!(
 							message = "Skipping field, is object",
-							field = field_name,
+							field = field_name.to_string(),
 							path,
 							value = ?val
 						);

crates/pile-dataset/src/lib.rs

@@ -1,6 +1,9 @@
 mod traits;
 pub use traits::*;
 
+mod misc;
+pub use misc::*;
+
 pub mod index;
 pub mod item;
 pub mod source;

crates/pile-dataset/src/misc.rs

@@ -0,0 +1,61 @@
+use chrono::{DateTime, Utc};
+use std::fs;
+use std::path::Path;
+
+/// Returns the age of a path as a chrono DateTime.
+/// - If the path doesn't exist, returns None
+/// - If it's a file, returns the modified time
+/// - If it's a directory, returns the LATEST modified time of all files within
+pub fn path_age(path: impl AsRef<Path>) -> Option<DateTime<Utc>> {
+	let path = path.as_ref();
+
+	// Check if path exists
+	if !path.exists() {
+		return None;
+	}
+
+	let metadata = fs::metadata(path).ok()?;
+
+	if metadata.is_file() {
+		// For files, return the modified time
+		let modified = metadata.modified().ok()?;
+		Some(modified.into())
+	} else if metadata.is_dir() {
+		// For directories, find the latest modified time of all files
+		find_latest_modified(path)
+	} else {
+		None
+	}
+}
+
+fn find_latest_modified(dir: &Path) -> Option<DateTime<Utc>> {
+	let mut latest: Option<DateTime<Utc>> = None;
+
+	// Read directory entries
+	let entries = fs::read_dir(dir).ok()?;
+
+	for entry in entries.flatten() {
+		let path = entry.path();
+		let metadata = entry.metadata().ok()?;
+
+		if metadata.is_file() {
+			if let Ok(modified) = metadata.modified() {
+				let dt: DateTime<Utc> = modified.into();
+				latest = Some(match latest {
+					Some(prev) if prev > dt => prev,
+					_ => dt,
+				});
+			}
+		} else if metadata.is_dir() {
+			// Recursively check subdirectories
+			if let Some(dir_latest) = find_latest_modified(&path) {
+				latest = Some(match latest {
+					Some(prev) if prev > dir_latest => prev,
+					_ => dir_latest,
+				});
+			}
+		}
+	}
+
+	latest
+}