Windows: Do not filter archive directories

I do not have a windows box so this is hard to test. But my guess is
that issues like this have been caused by archive directories being
ignored. This may fix this.
https://github.com/bootandy/dust/issues/153

Cargo.lock

@@ -148,7 +148,7 @@ checksum = "fea41bba32d969b513997752735605054bc0dfa92b4c56bf1189f2e174be7a10"
 
 [[package]]
 name = "du-dust"
-version = "0.8.1"
+version = "0.8.1-alpha.2"
 dependencies = [
  "ansi_term",
  "assert_cmd",
@@ -340,9 +340,9 @@ dependencies = [
 
 [[package]]
 name = "regex"
-version = "1.5.5"
+version = "1.5.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1a11647b6b25ff05a515cb92c365cec08801e83423a235b51e231e1808747286"
+checksum = "d07a8629359eb56f1e2fb1652bb04212c072a87ba68546a04065d525673ac461"
 dependencies = [
  "aho-corasick",
  "memchr",

Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "du-dust"
 description = "A more intuitive version of du"
-version = "0.8.1"
+version = "0.8.1-alpha.2"
 authors = ["bootandy <bootandy@gmail.com>", "nebkor <code@ardent.nebcorp.com>"]
 edition = "2018"
 readme = "README.md"

README.md

@@ -32,7 +32,7 @@ Because I want an easy way to see where my disk is being used.
 
 #### Windows:
 * Windows GNU version - works
-* Windows MSVC - requires: [VCRUNTIME140.dll](https://docs.microsoft.com/en-gb/cpp/windows/latest-supported-vc-redist?view=msvc-170)
+* Windows MSVC - requires: VCRUNTIME140.dll
 
 #### Download
 

ci/how2publish.txt

@@ -3,6 +3,7 @@
 # tag a commit and push (increment version in Cargo.toml first):
 #   git tag v0.4.5
 #   git push origin v0.4.5
+# pre release notation: "0.8.1-alpha.1"
 
 # cargo publish to put it in crates.io
 

src/dir_walker.rs

@@ -34,7 +34,7 @@ pub fn walk_it(dirs: HashSet<PathBuf>, walk_data: WalkData) -> (Vec<Node>, bool)
     let top_level_nodes: Vec<_> = dirs
         .into_iter()
         .filter_map(|d| {
-            let n = walk(d, &permissions_flag, &walk_data, 0);
+            let n = walk(d, &permissions_flag, &walk_data);
             match n {
                 Some(n) => {
                     let mut inodes: HashSet<(u64, u64)> = HashSet::new();
@@ -73,7 +73,6 @@ fn clean_inodes(
         size: x.size + new_children.iter().map(|c| c.size).sum::<u64>(),
         children: new_children,
         inode_device: x.inode_device,
-        depth: x.depth,
     });
 }
 
@@ -109,12 +108,7 @@ fn ignore_file(entry: &DirEntry, walk_data: &WalkData) -> bool {
     (is_dot_file && walk_data.ignore_hidden) || is_ignored_path
 }
 
-fn walk(
+fn walk(dir: PathBuf, permissions_flag: &AtomicBool, walk_data: &WalkData) -> Option<Node> {
-    dir: PathBuf,
-    permissions_flag: &AtomicBool,
-    walk_data: &WalkData,
-    depth: usize,
-) -> Option<Node> {
     let mut children = vec![];
 
     if let Ok(entries) = fs::read_dir(dir.clone()) {
@@ -132,7 +126,7 @@ fn walk(
                     if !ignore_file(entry, walk_data) {
                         if let Ok(data) = entry.file_type() {
                             if data.is_dir() && !data.is_symlink() {
-                                return walk(entry.path(), permissions_flag, walk_data, depth + 1);
+                                return walk(entry.path(), permissions_flag, walk_data);
                             }
                             return build_node(
                                 entry.path(),
@@ -143,7 +137,6 @@ fn walk(
                                 data.is_symlink(),
                                 data.is_file(),
                                 walk_data.by_filecount,
-                                depth,
                             );
                         }
                     }
@@ -165,7 +158,6 @@ fn walk(
         false,
         false,
         walk_data.by_filecount,
-        depth,
     )
 }
 
@@ -180,7 +172,6 @@ mod tests {
             size: 10,
             children: vec![],
             inode_device: Some((5, 6)),
-            depth: 0,
         }
     }
 

src/filter.rs

@@ -5,10 +5,18 @@ use std::collections::HashMap;
 use std::collections::HashSet;
 use std::path::PathBuf;
 
+pub fn get_by_depth(top_level_nodes: Vec<Node>, n: usize) -> Option<DisplayNode> {
+    if top_level_nodes.is_empty() {
+        // perhaps change this, bring back Error object?
+        return None;
+    }
+    let root = get_new_root(top_level_nodes);
+    Some(build_by_depth(&root, n - 1))
+}
+
 pub fn get_biggest(
     top_level_nodes: Vec<Node>,
     n: usize,
-    depth: usize,
     using_a_filter: bool,
 ) -> Option<DisplayNode> {
     if top_level_nodes.is_empty() {
@@ -22,14 +30,14 @@ pub fn get_biggest(
     let mut allowed_nodes = HashSet::new();
 
     allowed_nodes.insert(&root.name);
-    heap = add_children(using_a_filter, &root, depth, heap);
+    heap = add_children(using_a_filter, &root, heap);
 
     for _ in number_top_level_nodes..n {
         let line = heap.pop();
         match line {
             Some(line) => {
                 allowed_nodes.insert(&line.name);
-                heap = add_children(using_a_filter, line, depth, heap);
+                heap = add_children(using_a_filter, line, heap);
             }
             None => break,
         }
@@ -69,20 +77,17 @@ pub fn get_all_file_types(top_level_nodes: Vec<Node>, n: usize) -> Option<Displa
 
 fn add_children<'a>(
     using_a_filter: bool,
-    file_or_folder: &'a Node,
+    line: &'a Node,
-    depth: usize,
     mut heap: BinaryHeap<&'a Node>,
 ) -> BinaryHeap<&'a Node> {
-    if depth > file_or_folder.depth {
     if using_a_filter {
-            file_or_folder.children.iter().for_each(|c| {
+        line.children.iter().for_each(|c| {
             if c.name.is_file() || c.size > 0 {
                 heap.push(c)
             }
         });
     } else {
-            file_or_folder.children.iter().for_each(|c| heap.push(c));
+        line.children.iter().for_each(|c| heap.push(c));
-        }
     }
     heap
 }
@@ -106,6 +111,29 @@ fn build_by_all_file_types(top_level_nodes: Vec<Node>, counter: &mut HashMap<Str
     }
 }
 
+fn build_by_depth(node: &Node, depth: usize) -> DisplayNode {
+    let new_children = {
+        if depth == 0 {
+            vec![]
+        } else {
+            let mut new_children: Vec<_> = node
+                .children
+                .iter()
+                .map(|c| build_by_depth(c, depth - 1))
+                .collect();
+            new_children.sort();
+            new_children.reverse();
+            new_children
+        }
+    };
+
+    DisplayNode {
+        name: node.name.clone(),
+        size: node.size,
+        children: new_children,
+    }
+}
+
 fn get_new_root(top_level_nodes: Vec<Node>) -> Node {
     if top_level_nodes.len() > 1 {
         let total_size = top_level_nodes.iter().map(|node| node.size).sum();
@@ -114,7 +142,6 @@ fn get_new_root(top_level_nodes: Vec<Node>) -> Node {
             size: total_size,
             children: top_level_nodes,
             inode_device: None,
-            depth: 0,
         }
     } else {
         top_level_nodes.into_iter().next().unwrap()

src/main.rs

@@ -10,7 +10,7 @@ use self::display::draw_it;
 use clap::{crate_version, Arg};
 use clap::{Command, Values};
 use dir_walker::{walk_it, WalkData};
-use filter::{get_all_file_types, get_biggest};
+use filter::{get_all_file_types, get_biggest, get_by_depth};
 use regex::Regex;
 use std::cmp::max;
 use std::path::PathBuf;
@@ -99,6 +99,9 @@ fn get_regex_value(maybe_value: Option<Values>) -> Vec<Regex> {
 }
 
 fn main() {
+    let default_height = get_height_of_terminal();
+    let def_num_str = default_height.to_string();
+
     let options = Command::new("Dust")
         .about("Like du but more intuitive")
         .version(crate_version!())
@@ -109,7 +112,7 @@ fn main() {
                 .long("depth")
                 .help("Depth to show")
                 .takes_value(true)
-                .default_value(usize::MAX.to_string().as_ref())
+                .conflicts_with("number_of_lines"),
         )
         .arg(
             Arg::new("number_of_lines")
@@ -117,6 +120,7 @@ fn main() {
                 .long("number-of-lines")
                 .help("Number of lines of output to show. (Default is terminal_height - 10)")
                 .takes_value(true)
+                .default_value(def_num_str.as_ref()),
         )
         .arg(
             Arg::new("display_full_paths")
@@ -184,6 +188,7 @@ fn main() {
                 .multiple_occurrences(true)
                 .conflicts_with("filter")
                 .conflicts_with("types")
+                .conflicts_with("depth")
                 .help("Exclude filepaths matching this regex. To ignore png files type: -v \"\\.png$\" "),
         )
         .arg(
@@ -194,6 +199,7 @@ fn main() {
                 .number_of_values(1)
                 .multiple_occurrences(true)
                 .conflicts_with("types")
+                .conflicts_with("depth")
                 .help("Only include filepaths matching this regex. For png files type: -e \"\\.png$\" "),
         )
         .arg(
@@ -230,36 +236,26 @@ fn main() {
     let filter_regexs = get_regex_value(options.values_of("filter"));
     let invert_filter_regexs = get_regex_value(options.values_of("invert_filter"));
 
+    let number_of_lines = match options.value_of_t("number_of_lines") {
+        Ok(v) => v,
+        Err(_) => {
+            eprintln!("Ignoring bad value for number_of_lines");
+            default_height
+        }
+    };
+
     let terminal_width = match options.value_of_t("width") {
         Ok(v) => v,
         Err(_) => get_width_of_terminal(),
     };
 
-    let depth = match options.value_of_t("depth") {
+    let depth = options.value_of("depth").and_then(|depth| {
-        Ok(v) => v,
+        depth
-        Err(_) => {
+            .parse::<usize>()
-            eprintln!("Ignoring bad value for depth");
+            .map(|v| v + 1)
-            usize::MAX
+            .map_err(|_| eprintln!("Ignoring bad value for depth"))
-        }
+            .ok()
-    };
+    });
-    // If depth is set we set the default number_of_lines to be max
-    // instead of screen height
-    let default_height = if depth != usize::MAX {
-        usize::MAX
-    } else {
-        get_height_of_terminal()
-    };
-
-    let number_of_lines = match options.value_of("number_of_lines") {
-        Some(v) => match v.parse::<usize>() {
-            Ok(num_lines) => num_lines,
-            Err(_) => {
-                eprintln!("Ignoring bad value for number_of_lines");
-                default_height
-            }
-        },
-        None => default_height,
-    };
 
     let no_colors = init_color(options.is_present("no_colors"));
     let use_apparent_size = options.is_present("display_apparent_size");
@@ -301,10 +297,10 @@ fn main() {
     let tree = {
         match (depth, summarize_file_types) {
             (_, true) => get_all_file_types(top_level_nodes, number_of_lines),
-            (depth, _) => get_biggest(
+            (Some(depth), _) => get_by_depth(top_level_nodes, depth),
+            (_, _) => get_biggest(
                 top_level_nodes,
                 number_of_lines,
-                depth,
                 options.values_of("filter").is_some()
                     || options.value_of("invert_filter").is_some(),
             ),

src/node.rs

@@ -12,7 +12,6 @@ pub struct Node {
     pub size: u64,
     pub children: Vec<Node>,
     pub inode_device: Option<(u64, u64)>,
-    pub depth: usize,
 }
 
 #[allow(clippy::too_many_arguments)]
@@ -25,7 +24,6 @@ pub fn build_node(
     is_symlink: bool,
     is_file: bool,
     by_filecount: bool,
-    depth: usize,
 ) -> Option<Node> {
     match get_metadata(&dir, use_apparent_size) {
         Some(data) => {
@@ -52,7 +50,6 @@ pub fn build_node(
                 size,
                 children,
                 inode_device,
-                depth,
             })
         }
         None => None,

src/platform.rs

@@ -105,7 +105,6 @@ pub fn get_metadata(d: &Path, _use_apparent_size: bool) -> Option<(u64, Option<(
     use std::os::windows::fs::MetadataExt;
     match d.metadata() {
         Ok(ref md) => {
-            const FILE_ATTRIBUTE_ARCHIVE: u32 = 0x20u32;
             const FILE_ATTRIBUTE_READONLY: u32 = 0x1u32;
             const FILE_ATTRIBUTE_HIDDEN: u32 = 0x2u32;
             const FILE_ATTRIBUTE_SYSTEM: u32 = 0x4u32;
@@ -114,8 +113,7 @@ pub fn get_metadata(d: &Path, _use_apparent_size: bool) -> Option<(u64, Option<(
 
             let attr_filtered = md.file_attributes()
                 & !(FILE_ATTRIBUTE_HIDDEN | FILE_ATTRIBUTE_READONLY | FILE_ATTRIBUTE_SYSTEM);
-            if (attr_filtered & FILE_ATTRIBUTE_ARCHIVE) != 0
+            if attr_filtered == FILE_ATTRIBUTE_DIRECTORY
-                || (attr_filtered & FILE_ATTRIBUTE_DIRECTORY) != 0
                 || md.file_attributes() == FILE_ATTRIBUTE_NORMAL
             {
                 Some((md.len(), None))

src/utils.rs

@@ -7,10 +7,11 @@ use regex::Regex;
 
 pub fn simplify_dir_names<P: AsRef<Path>>(filenames: Vec<P>) -> HashSet<PathBuf> {
     let mut top_level_names: HashSet<PathBuf> = HashSet::with_capacity(filenames.len());
+    let mut to_remove: Vec<PathBuf> = Vec::with_capacity(filenames.len());
+
     for t in filenames {
         let top_level_name = normalize_path(t);
         let mut can_add = true;
-        let mut to_remove: Vec<PathBuf> = Vec::new();
 
         for tt in top_level_names.iter() {
             if is_a_parent_of(&top_level_name, tt) {
@@ -19,13 +20,14 @@ pub fn simplify_dir_names<P: AsRef<Path>>(filenames: Vec<P>) -> HashSet<PathBuf>
                 can_add = false;
             }
         }
-        for r in to_remove {
+        to_remove.sort_unstable();
-            top_level_names.remove(&r);
+        top_level_names.retain(|tr| to_remove.binary_search(tr).is_err());
-        }
+        to_remove.clear();
         if can_add {
             top_level_names.insert(top_level_name);
         }
     }
+
     top_level_names
 }
 
@@ -92,7 +94,6 @@ mod tests {
     fn test_simplify_dir_rm_subdir() {
         let mut correct = HashSet::new();
         correct.insert(["a", "b"].iter().collect::<PathBuf>());
-        assert_eq!(simplify_dir_names(vec!["a/b/c", "a/b", "a/b/d/f"]), correct);
         assert_eq!(simplify_dir_names(vec!["a/b", "a/b/c", "a/b/d/f"]), correct);
     }