add method to File to get extents on-disk

Author: agrif

src/file.rs

@@ -25,6 +25,17 @@ pub struct File<'a, IO: ReadWriteSeek, TP, OCC> {
     fs: &'a FileSystem<IO, TP, OCC>,
 }
 
+/// An extent containing a file's data on disk.
+///
+/// This is created by the `extents` method on `File`, and represents
+/// a byte range on the disk that contains a file's data. All values
+/// are in bytes.
+#[derive(Clone, Debug)]
+pub struct Extent {
+    pub offset: u64,
+    pub size: u32,
+}
+
 impl<'a, IO: ReadWriteSeek, TP, OCC> File<'a, IO, TP, OCC> {
     pub(crate) fn new(
         first_cluster: Option<u32>,
@@ -74,6 +85,39 @@ impl<'a, IO: ReadWriteSeek, TP, OCC> File<'a, IO, TP, OCC> {
         }
     }
 
+    /// Get the extents of a file on disk.
+    ///
+    /// This returns an iterator over the byte ranges on-disk occupied by
+    /// this file.
+    pub fn extents(&mut self) -> impl Iterator<Item=Result<Extent, Error<IO::Error>>> + 'a {
+
+        let fs = self.fs;
+        let cluster_size = fs.cluster_size();
+        let mut bytes_left = match self.size() {
+            Some(s) => s,
+            None => return None.into_iter().flatten(),
+        };
+        let first = match self.first_cluster {
+            Some(f) => f,
+            None => return None.into_iter().flatten(),
+        };
+
+        Some(core::iter::once(Ok(first)).chain(fs.cluster_iter(first))
+             .map(move |cluster_err| {
+                 match cluster_err {
+                     Ok(cluster) => {
+                         let size = cluster_size.min(bytes_left);
+                         bytes_left -= size;
+                         Ok(Extent {
+                             offset: fs.offset_from_cluster(cluster),
+                             size: size,
+                         })
+                     },
+                     Err(e) => Err(e),
+                 }
+             })).into_iter().flatten()
+    }
+
     pub(crate) fn abs_pos(&self) -> Option<u64> {
         // Returns current position relative to filesystem start
         // Note: when between clusters it returns position after previous cluster