Avoid opening all files for reading on windows

It can be very expensive to do that, especially when it causes windows defender to read the files and scan them.

src/utils/mod.rs

@@ -85,15 +85,20 @@ pub fn get_dir_tree<P: AsRef<Path>>(
         None
     };
 
+    let mut fake_inode_counter = u64::max_value();
+    let mut examine_dir_args = ExamineDirMutArsg {
+        data: &mut data,
+        fake_inode_counter: &mut fake_inode_counter,
+        file_count_no_permission: &mut permissions,
+    };
     for b in top_level_names.iter() {
         examine_dir(
             b,
             apparent_size,
             &restricted_filesystems,
             ignore_directories,
-            &mut data,
-            &mut permissions,
             threads,
+            &mut examine_dir_args,
         );
     }
     (permissions == 0, data)
@@ -119,14 +124,19 @@ pub fn normalize_path<P: AsRef<Path>>(path: P) -> PathBuf {
     path.as_ref().components().collect::<PathBuf>()
 }
 
+struct ExamineDirMutArsg<'a> {
+    data: &'a mut HashMap<PathBuf, u64>,
+    file_count_no_permission: &'a mut u64,
+    fake_inode_counter: &'a mut u64,
+}
+
 fn examine_dir<P: AsRef<Path>>(
     top_dir: P,
     apparent_size: bool,
     filesystems: &Option<HashSet<u64>>,
     ignore_directories: &Option<Vec<PathBuf>>,
-    data: &mut HashMap<PathBuf, u64>,
-    file_count_no_permission: &mut u64,
     threads: Option<usize>,
+    mut_args: &mut ExamineDirMutArsg,
 ) {
     let top_dir = top_dir.as_ref();
     let mut inodes: HashSet<(u64, u64)> = HashSet::new();
@@ -136,9 +146,12 @@ fn examine_dir<P: AsRef<Path>>(
     if let Some(threads_to_start) = threads {
         iter = iter.num_threads(threads_to_start);
     }
+
     'entry: for entry in iter {
         if let Ok(e) = entry {
-            let maybe_size_and_inode = get_metadata(&e, apparent_size);
+            let maybe_size_and_inode =
+                get_metadata(&e, apparent_size, &mut mut_args.fake_inode_counter);
+
             if let Some(dirs) = ignore_directories {
                 let path = e.path();
                 let parts = path.components().collect::<Vec<std::path::Component>>();
@@ -156,13 +169,13 @@ fn examine_dir<P: AsRef<Path>>(
             match maybe_size_and_inode {
                 Some((size, inode, device)) => {
                     if !should_ignore_file(apparent_size, filesystems, &mut inodes, inode, device) {
-                        process_file_with_size_and_inode(top_dir, data, e, size)
+                        process_file_with_size_and_inode(top_dir, mut_args.data, e, size)
                     }
                 }
-                None => *file_count_no_permission += 1,
+                None => *mut_args.file_count_no_permission += 1,
             }
         } else {
-            *file_count_no_permission += 1
+            *mut_args.file_count_no_permission += 1
         }
     }
 }

src/utils/platform.rs

@@ -12,7 +12,11 @@ fn get_block_size() -> u64 {
 }
 
 #[cfg(target_family = "unix")]
-pub fn get_metadata(d: &DirEntry, use_apparent_size: bool) -> Option<(u64, u64, u64)> {
+pub fn get_metadata(
+    d: &DirEntry,
+    use_apparent_size: bool,
+    _fake_inode_counter: &mut u64,
+) -> Option<(u64, u64, u64)> {
     use std::os::unix::fs::MetadataExt;
     d.metadata.as_ref().unwrap().as_ref().ok().map(|md| {
         if use_apparent_size {
@@ -24,18 +28,85 @@ pub fn get_metadata(d: &DirEntry, use_apparent_size: bool) -> Option<(u64, u64,
 }
 
 #[cfg(target_family = "windows")]
-pub fn get_metadata(d: &DirEntry, _use_apparent_size: bool) -> Option<(u64, u64, u64)> {
-    use winapi_util::file::information;
-    use winapi_util::Handle;
+pub fn get_metadata(
+    d: &DirEntry,
+    _use_apparent_size: bool,
+    fake_inode_counter: &mut u64,
+) -> Option<(u64, u64, u64)> {
+    // On windows opening the file to get size, file ID and volume can be very
+    // expensive because 1) it causes a few system calls, and more importantly 2) it can cause
+    // windows defender to scan the file.
+    // Therefore we try to avoid doing that for common cases, mainly those of
+    // plain files:
 
-    let h = Handle::from_path_any(d.path()).ok()?;
-    let info = information(&h).ok()?;
+    // The idea is to make do with the file size that we get from the OS for
+    // free as part of iterating a folder. Therefore we want to make sure that
+    // it makes sense to use that free size information:
 
-    Some((
-        info.file_size(),
-        info.file_index(),
-        info.volume_serial_number(),
-    ))
+    // Volume boundaries:
+    // The user can ask us not to cross volume boundaries. If the DirEntry is a
+    // plain file and not a reparse point or other non-trivial stuff, we assume
+    // that the file is located on the same volume as the directory that
+    // contains it.
+
+    // File ID:
+    // This optimization does deprive us of access to a file ID. As a
+    // workaround, we just make one up that hopefully does not collide with real
+    // file IDs.
+    // Hard links: Unresolved. We don't get inode/file index, so hard links
+    // count once for each link. Hopefully they are not too commonly in use on
+    // windows.
+
+    // Size:
+    // We assume (naively?) that for the common cases the free size info is the
+    // same as one would get by doing the expensive thing. Sparse, encrypted and
+    // compressed files are not included in the common cases, as one can image
+    // there being more than view on their size.
+
+    // Savings in orders of magnitude in terms of time, io and cpu have been
+    // observed on hdd, windows 10, some 100Ks files taking up some hundreds of
+    // GBs:
+    // Consistently opening the file: 30 minutes.
+    // With this optimization:         8 sec.
+
+    fn get_metadata_expensive(d: &DirEntry) -> Option<(u64, u64, u64)> {
+        use winapi_util::file::information;
+        use winapi_util::Handle;
+
+        let h = Handle::from_path_any(d.path()).ok()?;
+        let info = information(&h).ok()?;
+
+        Some((
+            info.file_size(),
+            info.file_index(),
+            info.volume_serial_number(),
+        ))
+    }
+
+    match d.metadata {
+        Some(Ok(ref md)) => {
+            use std::os::windows::fs::MetadataExt;
+            const FILE_ATTRIBUTE_ARCHIVE: u32 = 0x20u32;
+            const FILE_ATTRIBUTE_READONLY: u32 = 0x1u32;
+            const FILE_ATTRIBUTE_HIDDEN: u32 = 0x2u32;
+            const FILE_ATTRIBUTE_SYSTEM: u32 = 0x4u32;
+            const FILE_ATTRIBUTE_NORMAL: u32 = 0x80u32;
+            const FILE_ATTRIBUTE_DIRECTORY: u32 = 0x10u32;
+
+            let attr_filtered = md.file_attributes()
+                & !(FILE_ATTRIBUTE_HIDDEN | FILE_ATTRIBUTE_READONLY | FILE_ATTRIBUTE_SYSTEM);
+            if attr_filtered == FILE_ATTRIBUTE_ARCHIVE
+                || attr_filtered == FILE_ATTRIBUTE_DIRECTORY
+                || md.file_attributes() == FILE_ATTRIBUTE_NORMAL
+            {
+                *fake_inode_counter -= 1;
+                Some((md.len(), *fake_inode_counter, 0xcafe_cafe_u64))
+            } else {
+                get_metadata_expensive(&d)
+            }
+        }
+        _ => get_metadata_expensive(&d),
+    }
 }
 
 #[cfg(target_family = "unix")]