refactor: cleanup code

du has -L flag which allows it to dereference or follow
symlinks. Clone this feature into dust.
https://github.com/bootandy/dust/issues/276

completions/_dust

@@ -35,6 +35,10 @@ _dust() {
 '--version[Print version information]' \
 '-p[Subdirectories will not have their path shortened]' \
 '--full-paths[Subdirectories will not have their path shortened]' \
+'(-L --dereference-links)-l[Ignore links]' \
+'(-L --dereference-links)--ignore-links[Ignore links]' \
+'(-l --ignore-links)-L[dereference sym links - Treat sym links as directories and go into them]' \
+'(-l --ignore-links)--dereference-links[dereference sym links - Treat sym links as directories and go into them]' \
 '-x[Only count the files and directories on the same filesystem as the supplied directory]' \
 '--limit-filesystem[Only count the files and directories on the same filesystem as the supplied directory]' \
 '-s[Use file length instead of blocks]' \

completions/_dust.ps1

@@ -41,6 +41,10 @@ Register-ArgumentCompleter -Native -CommandName 'dust' -ScriptBlock {
             [CompletionResult]::new('--version', 'version', [CompletionResultType]::ParameterName, 'Print version information')
             [CompletionResult]::new('-p', 'p', [CompletionResultType]::ParameterName, 'Subdirectories will not have their path shortened')
             [CompletionResult]::new('--full-paths', 'full-paths', [CompletionResultType]::ParameterName, 'Subdirectories will not have their path shortened')
+            [CompletionResult]::new('-l', 'l', [CompletionResultType]::ParameterName, 'Ignore links')
+            [CompletionResult]::new('--ignore-links', 'ignore-links', [CompletionResultType]::ParameterName, 'Ignore links')
+            [CompletionResult]::new('-L', 'L', [CompletionResultType]::ParameterName, 'dereference sym links - Treat sym links as directories and go into them')
+            [CompletionResult]::new('--dereference-links', 'dereference-links', [CompletionResultType]::ParameterName, 'dereference sym links - Treat sym links as directories and go into them')
             [CompletionResult]::new('-x', 'x', [CompletionResultType]::ParameterName, 'Only count the files and directories on the same filesystem as the supplied directory')
             [CompletionResult]::new('--limit-filesystem', 'limit-filesystem', [CompletionResultType]::ParameterName, 'Only count the files and directories on the same filesystem as the supplied directory')
             [CompletionResult]::new('-s', 's', [CompletionResultType]::ParameterName, 'Use file length instead of blocks')

completions/dust.bash

@@ -19,7 +19,7 @@ _dust() {
 
     case "${cmd}" in
         dust)
-            opts="-h -V -d -n -p -X -x -s -r -c -b -z -f -i -v -e -t -w -H -D --help --version --depth --number-of-lines --full-paths --ignore-directory --limit-filesystem --apparent-size --reverse --no-colors --no-percent-bars --min-size --skip-total --filecount --ignore_hidden --invert-filter --filter --file_types --terminal_width --si --only-dir <inputs>..."
+            opts="-h -V -d -n -p -X -l -L -x -s -r -c -b -z -f -i -v -e -t -w -H -D --help --version --depth --number-of-lines --full-paths --ignore-directory --ignore-links --dereference-links --limit-filesystem --apparent-size --reverse --no-colors --no-percent-bars --min-size --skip-total --filecount --ignore_hidden --invert-filter --filter --file_types --terminal_width --si --only-dir <inputs>..."
             if [[ ${cur} == -* || ${COMP_CWORD} -eq 1 ]] ; then
                 COMPREPLY=( $(compgen -W "${opts}" -- "${cur}") )
                 return 0

completions/dust.elv

@@ -38,6 +38,10 @@ set edit:completion:arg-completer[dust] = {|@words|
             cand --version 'Print version information'
             cand -p 'Subdirectories will not have their path shortened'
             cand --full-paths 'Subdirectories will not have their path shortened'
+            cand -l 'Ignore links'
+            cand --ignore-links 'Ignore links'
+            cand -L 'dereference sym links - Treat sym links as directories and go into them'
+            cand --dereference-links 'dereference sym links - Treat sym links as directories and go into them'
             cand -x 'Only count the files and directories on the same filesystem as the supplied directory'
             cand --limit-filesystem 'Only count the files and directories on the same filesystem as the supplied directory'
             cand -s 'Use file length instead of blocks'

completions/dust.fish

@@ -8,6 +8,8 @@ complete -c dust -s w -l terminal_width -d 'Specify width of output overriding t
 complete -c dust -s h -l help -d 'Print help information'
 complete -c dust -s V -l version -d 'Print version information'
 complete -c dust -s p -l full-paths -d 'Subdirectories will not have their path shortened'
+complete -c dust -s l -l ignore-links -d 'Ignore links'
+complete -c dust -s L -l dereference-links -d 'dereference sym links - Treat sym links as directories and go into them'
 complete -c dust -s x -l limit-filesystem -d 'Only count the files and directories on the same filesystem as the supplied directory'
 complete -c dust -s s -l apparent-size -d 'Use file length instead of blocks'
 complete -c dust -s r -l reverse -d 'Print tree upside down (biggest highest)'

src/cli.rs

@@ -34,6 +34,20 @@ pub fn build_cli() -> Command<'static> {
                 .multiple_occurrences(true)
                 .help("Exclude any file or directory with this name"),
         )
+        .arg(
+            Arg::new("ignore_links")
+                .short('l')
+                .long("ignore-links")
+                .conflicts_with("dereference_links")
+                .help("Ignore links"),
+        )
+        .arg(
+            Arg::new("dereference_links")
+                .short('L')
+                .long("dereference-links")
+                .conflicts_with("ignore_links")
+                .help("dereference sym links - Treat sym links as directories and go into them"),
+        )
         .arg(
             Arg::new("limit_filesystem")
                 .short('x')
@@ -136,6 +150,6 @@ pub fn build_cli() -> Command<'static> {
             Arg::new("only_dir")
                 .short('D')
                 .long("only-dir")
-                .help("Only directories will be displayed."), 
+                .help("Only directories will be displayed."),
         )
 }

src/dir_walker.rs

@@ -26,18 +26,19 @@ pub struct WalkData<'a> {
     pub use_apparent_size: bool,
     pub by_filecount: bool,
     pub ignore_hidden: bool,
+    pub ignore_links: bool,
+    pub follow_links: bool,
 }
 
 pub fn walk_it(dirs: HashSet<PathBuf>, walk_data: WalkData) -> (Vec<Node>, bool) {
     let permissions_flag = AtomicBool::new(false);
 
-    let mut inodes = HashSet::new();
     let top_level_nodes: Vec<_> = dirs
         .into_iter()
         .filter_map(|d| {
             clean_inodes(
                 walk(d, &permissions_flag, &walk_data, 0)?,
-                &mut inodes,
+                &mut HashSet::new(),
                 walk_data.use_apparent_size,
             )
         })
@@ -142,24 +143,27 @@ fn walk(
 
                     // return walk(entry.path(), permissions_flag, ignore_directories, allowed_filesystems, use_apparent_size, by_filecount, ignore_hidden);
 
-                    if !ignore_file(entry, walk_data) {
+                    if ignore_file(entry, walk_data) {
-                        if let Ok(data) = entry.file_type() {
+                        return None;
-                            return if data.is_dir() && !data.is_symlink() {
+                    }
-                                walk(entry.path(), permissions_flag, walk_data, depth + 1)
+                    if let Ok(data) = entry.file_type() {
-                            } else {
+                        if data.is_symlink() && walk_data.ignore_links {
-                                build_node(
+                            return None;
-                                    entry.path(),
-                                    vec![],
-                                    walk_data.filter_regex,
-                                    walk_data.invert_filter_regex,
-                                    walk_data.use_apparent_size,
-                                    data.is_symlink(),
-                                    data.is_file(),
-                                    walk_data.by_filecount,
-                                    depth,
-                                )
-                            };
                         }
+                        if data.is_dir() || (walk_data.follow_links && data.is_symlink()) {
+                            return walk(entry.path(), permissions_flag, walk_data, depth + 1);
+                        }
+                        return build_node(
+                            entry.path(),
+                            vec![],
+                            walk_data.filter_regex,
+                            walk_data.invert_filter_regex,
+                            walk_data.use_apparent_size,
+                            data.is_symlink(),
+                            data.is_file(),
+                            walk_data.by_filecount,
+                            depth,
+                        );
                     }
                 } else {
                     permissions_flag.store(true, atomic::Ordering::Relaxed);

src/filter.rs

@@ -20,20 +20,13 @@ pub fn get_biggest(
 
     let mut heap = BinaryHeap::new();
     let number_top_level_nodes = top_level_nodes.len();
-
     let root = get_new_root(top_level_nodes);
     let mut allowed_nodes = HashSet::new();
 
     allowed_nodes.insert(root.name.as_path());
+    heap = add_children(using_a_filter, min_size, only_dir, &root, depth, heap);
 
-    if number_top_level_nodes > 1 {
+    for _ in number_top_level_nodes..n {
-        heap = add_children(using_a_filter, min_size, only_dir, &root, usize::MAX, heap);
-    } else {
-        heap = add_children(using_a_filter, min_size, only_dir, &root, depth, heap);
-    }
-
-    let number_of_lines_in_output = n - number_top_level_nodes;
-    for _ in 0..number_of_lines_in_output {
         let line = heap.pop();
         match line {
             Some(line) => {

src/main.rs

@@ -152,6 +152,8 @@ fn main() {
 
     let by_filecount = options.is_present("by_filecount");
     let limit_filesystem = options.is_present("limit_filesystem");
+    let ignore_links = options.is_present("ignore_links");
+    let follow_links = options.is_present("dereference_links");
 
     let simplified_dirs = simplify_dir_names(target_dirs);
     let allowed_filesystems = limit_filesystem
@@ -167,6 +169,8 @@ fn main() {
         filter_regex: &filter_regexs,
         invert_filter_regex: &invert_filter_regexs,
         allowed_filesystems,
+        ignore_links,
+        follow_links,
         use_apparent_size: config.get_apparent_size(&options),
         by_filecount,
         ignore_hidden: config.get_ignore_hidden(&options),

tests/test_flags.rs

@@ -73,7 +73,6 @@ pub fn test_ignore_dir() {
     let output = build_command(vec!["-c", "-X", "dir_substring", "tests/test_dir2/"]);
     assert!(!output.contains("dir_substring"));
 }
-// Add test for multiple dirs - with -d 0 and maybe -d 1 check the
 
 #[test]
 pub fn test_with_bad_param() {

tests/tests_symlinks.rs

@@ -17,18 +17,6 @@ fn build_temp_file(dir: &TempDir) -> PathBuf {
     file_path
 }
 
-fn link_it(link_path: PathBuf, file_path_s: &str, is_soft: bool) -> String {
-    let link_name_s = link_path.to_str().unwrap();
-    let mut c = Command::new("ln");
-    if is_soft {
-        c.arg("-s");
-    }
-    c.arg(file_path_s);
-    c.arg(link_name_s);
-    assert!(c.output().is_ok());
-    return link_name_s.into();
-}
-
 #[cfg_attr(target_os = "windows", ignore)]
 #[test]
 pub fn test_soft_sym_link() {
@@ -38,7 +26,13 @@ pub fn test_soft_sym_link() {
     let file_path_s = file.to_str().unwrap();
 
     let link_name = dir.path().join("the_link");
-    let link_name_s = link_it(link_name, file_path_s, true);
+    let link_name_s = link_name.to_str().unwrap();
+    let c = Command::new("ln")
+        .arg("-s")
+        .arg(file_path_s)
+        .arg(link_name_s)
+        .output();
+    assert!(c.is_ok());
 
     let c = format!(" ├── {}", link_name_s);
     let b = format!(" ┌── {}", file_path_s);
@@ -47,7 +41,7 @@ pub fn test_soft_sym_link() {
     let mut cmd = Command::cargo_bin("dust").unwrap();
     // Mac test runners create long filenames in tmp directories
     let output = cmd
-        .args(["-p", "-c", "-s", "-w", "999", dir_s])
+        .args(["-p", "-c", "-s", "-w 999", dir_s])
         .unwrap()
         .stdout;
 
@@ -67,14 +61,19 @@ pub fn test_hard_sym_link() {
     let file_path_s = file.to_str().unwrap();
 
     let link_name = dir.path().join("the_link");
-    link_it(link_name, file_path_s, false);
+    let link_name_s = link_name.to_str().unwrap();
+    let c = Command::new("ln")
+        .arg(file_path_s)
+        .arg(link_name_s)
+        .output();
+    assert!(c.is_ok());
 
     let file_output = format!(" ┌── {}", file_path_s);
     let dirs_output = format!("─┴ {}", dir_s);
 
     let mut cmd = Command::cargo_bin("dust").unwrap();
     // Mac test runners create long filenames in tmp directories
-    let output = cmd.args(["-p", "-c", "-w", "999", dir_s]).unwrap().stdout;
+    let output = cmd.args(["-p", "-c", "-w 999", dir_s]).unwrap().stdout;
 
     // The link should not appear in the output because multiple inodes are now ordered
     // then filtered.
@@ -83,34 +82,6 @@ pub fn test_hard_sym_link() {
     assert!(output.contains(file_output.as_str()));
 }
 
-#[cfg_attr(target_os = "windows", ignore)]
-#[test]
-pub fn test_hard_sym_link_no_dup_multi_arg() {
-    let dir = Builder::new().tempdir().unwrap();
-    let dir_link = Builder::new().tempdir().unwrap();
-    let file = build_temp_file(&dir);
-    let dir_s = dir.path().to_str().unwrap();
-    let dir_link_s = dir_link.path().to_str().unwrap();
-    let file_path_s = file.to_str().unwrap();
-
-    let link_name = dir_link.path().join("the_link");
-    let link_name_s = link_it(link_name, file_path_s, false);
-
-    let mut cmd = Command::cargo_bin("dust").unwrap();
-
-    // Mac test runners create long filenames in tmp directories
-    let output = cmd
-        .args(["-p", "-c", "-w", "999", "-b", dir_link_s, dir_s])
-        .unwrap()
-        .stdout;
-
-    // The link or the file should appear but not both
-    let output = str::from_utf8(&output).unwrap();
-    let has_file_only = output.contains(file_path_s) && !output.contains(&link_name_s);
-    let has_link_only = !output.contains(file_path_s) && output.contains(&link_name_s);
-    assert!(has_file_only || has_link_only);
-}
-
 #[cfg_attr(target_os = "windows", ignore)]
 #[test]
 pub fn test_recursive_sym_link() {
@@ -118,7 +89,14 @@ pub fn test_recursive_sym_link() {
     let dir_s = dir.path().to_str().unwrap();
 
     let link_name = dir.path().join("the_link");
-    let link_name_s = link_it(link_name, dir_s, true);
+    let link_name_s = link_name.to_str().unwrap();
+
+    let c = Command::new("ln")
+        .arg("-s")
+        .arg(dir_s)
+        .arg(link_name_s)
+        .output();
+    assert!(c.is_ok());
 
     let a = format!("─┬ {}", dir_s);
     let b = format!(" └── {}", link_name_s);
@@ -129,8 +107,7 @@ pub fn test_recursive_sym_link() {
         .arg("-c")
         .arg("-r")
         .arg("-s")
-        .arg("-w")
+        .arg("-w 999")
-        .arg("999")
         .arg(dir_s)
         .unwrap()
         .stdout;