Add hardlink tracking, and an option to disable it

Freaky · Freaky · commit 5b522946adb5 · 2020-02-22T05:13:28.000Z
diff --git a/src/aggregate.rs b/src/aggregate.rs
@@ -1,4 +1,4 @@
-use crate::{WalkOptions, WalkResult};
+use crate::{WalkOptions, WalkResult, InodeFilter};
 use failure::Error;
 use std::borrow::Cow;
 use std::{fmt, io, path::Path};
@@ -20,6 +20,7 @@ pub fn aggregate(
     let mut total = 0;
     let mut num_roots = 0;
     let mut aggregates = Vec::new();
+    let mut inodes = InodeFilter::default();
     for path in paths.into_iter() {
         num_roots += 1;
         let mut num_bytes = 0u64;
@@ -29,7 +30,7 @@ pub fn aggregate(
             match entry {
                 Ok(entry) => {
                     let file_size = match entry.metadata {
-                        Some(Ok(ref m)) if !m.is_dir() => {
+                        Some(Ok(ref m)) if !m.is_dir() && (options.count_links || inodes.add(m)) => {
                             if options.apparent_size {
                                 m.len()
                             } else {
diff --git a/src/common.rs b/src/common.rs
@@ -152,6 +152,7 @@ pub struct WalkOptions {
     /// for more information.
     pub threads: usize,
     pub byte_format: ByteFormat,
+    pub count_links: bool,
     pub apparent_size: bool,
     pub color: Color,
     pub sorting: TraversalSorting,
diff --git a/src/inodefilter.rs b/src/inodefilter.rs
@@ -0,0 +1,76 @@
+
+#![cfg_attr(windows, feature(windows_by_handle))]
+
+use std::collections::HashMap;
+
+#[derive(Debug, Default, Clone)]
+pub struct InodeFilter {
+    inner: HashMap<u64, u64>
+}
+
+impl InodeFilter {
+    #[cfg(unix)]
+    pub fn add(&mut self, metadata: &std::fs::Metadata) -> bool {
+        use std::os::unix::fs::MetadataExt;
+
+        self.add_inode(metadata.ino(), metadata.nlink())
+    }
+
+    #[cfg(windows)]
+    pub fn add(&mut self, metadata: &std::fs::Metadata) -> bool {
+        use std::os::windows::fs::MetadataExt;
+
+        if let (Some(inode), Some(nlinks)) = (metadata.file_index(), metadata.number_of_links()) {
+            self.add_inode(inode, nlinks as u64)
+        } else {
+            true
+        }
+    }
+
+    #[cfg(not(any(unix, windows)))]
+    pub fn add(&mut self, metadata: &std::fs::Metadata) -> bool {
+        true
+    }
+
+    pub fn add_inode(&mut self, inode: u64, nlinks: u64) -> bool {
+        if nlinks <= 1 {
+            return true;
+        }
+
+        match self.inner.get_mut(&inode) {
+            Some(count) => {
+                *count -= 1;
+
+                if *count == 0 {
+                    self.inner.remove(&inode);
+                }
+
+                false
+            },
+            None => {
+                self.inner.insert(inode, nlinks - 1);
+                true
+            }
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn it_filters_inodes() {
+        let mut inodes = InodeFilter::default();
+
+        assert!(inodes.add_inode(1, 2));
+        assert!(!inodes.add_inode(1, 2));
+
+        assert!(inodes.add_inode(1, 3));
+        assert!(!inodes.add_inode(1, 3));
+        assert!(!inodes.add_inode(1, 3));
+
+        assert!(inodes.add_inode(1, 1));
+        assert!(inodes.add_inode(1, 1));
+    }
+}
diff --git a/src/interactive/app_test/utils.rs b/src/interactive/app_test/utils.rs
@@ -165,6 +165,7 @@ pub fn initialized_app_and_terminal_with_closure<P: AsRef<Path>>(
             threads: 1,
             byte_format: ByteFormat::Metric,
             apparent_size: true,
+            count_links: false,
             color: Color::None,
             sorting: TraversalSorting::AlphabeticalByFileName,
         },
diff --git a/src/lib.rs b/src/lib.rs
@@ -5,8 +5,10 @@ extern crate jwalk;
 
 mod aggregate;
 mod common;
+mod inodefilter;
 
 pub mod traverse;
 
 pub use aggregate::aggregate;
 pub use common::*;
+pub(crate) use inodefilter::InodeFilter;
diff --git a/src/main.rs b/src/main.rs
@@ -30,6 +30,7 @@ fn run() -> Result<(), Error> {
             Color::None
         },
         apparent_size: opt.apparent_size,
+        count_links: opt.count_links,
         sorting: TraversalSorting::None,
     };
     let res = match opt.command {
diff --git a/src/options.rs b/src/options.rs
@@ -56,6 +56,10 @@ pub struct Args {
     #[structopt(short = "A", long = "apparent-size")]
     pub apparent_size: bool,
 
+    /// Count hard-linked files each time they are seen
+    #[structopt(short = "l", long = "count-links")]
+    pub count_links: bool,
+
     /// One or more input files or directories. If unset, we will use all entries in the current working directory.
     #[structopt(parse(from_os_str))]
     pub input: Vec<PathBuf>,
diff --git a/src/traverse.rs b/src/traverse.rs
@@ -1,4 +1,4 @@
-use crate::{get_size_or_panic, WalkOptions};
+use crate::{get_size_or_panic, WalkOptions, InodeFilter};
 use failure::Error;
 use petgraph::{graph::NodeIndex, stable_graph::StableGraph, Directed, Direction};
 use std::{ffi::OsString, path::PathBuf, time::Duration, time::Instant};
@@ -66,6 +66,7 @@ impl Traversal {
         let mut sizes_per_depth_level = Vec::new();
         let mut current_size_at_depth = 0;
         let mut previous_depth = 0;
+        let mut inodes = InodeFilter::default();
 
         let mut last_checked = Instant::now();
 
@@ -93,7 +94,7 @@ impl Traversal {
                             entry.file_name
                         };
                         let file_size = match entry.metadata {
-                                Some(Ok(ref m)) if !m.is_dir() => {
+                                Some(Ok(ref m)) if !m.is_dir() && (walk_options.count_links || inodes.add(m)) => {
                                     if walk_options.apparent_size {
                                         m.len()
                                     } else {
