Update readme

Several tests for recursive dirs, hard & soft links.
Tests use the tempfile project.

Personally I find the tests hard to read. Am considering adding a
'--no-color' output option as this will make the tests much more
readable. (Currently we have to call format_string to get the matching
colors and if a test fails the diff is very hard to read).

.travis.yml

@@ -1,42 +1,76 @@
+# Based on the "trust" template v0.1.2
+# https://github.com/japaric/trust/tree/v0.1.2
+
+dist: trusty
+language: rust
+services: docker
+sudo: required
+
+# TODO Rust builds on stable by default, this can be
+# overridden on a case by case basis down below.
 
 env:
- global:
-   - PROJECT_NAME=dust
-   - CHANNEL=stable
+  global:
+    # TODO Update this to match the name of your project.
+    - CRATE_NAME=dust
 
-language: rust
-rust:
-  - stable
 matrix:
-  fast_finish: true
-    - os: osx
-      env: TARGET=i686-apple-darwin
-    - os: linux
-      env: TARGET=i686-unknown-linux-gnu
-    - os: osx
-      env: TARGET=x86_64-apple-darwin
-    - os: linux
-      env: TARGET=x86_64-unknown-linux-gnu
-cache: cargo
+  # TODO These are all the build jobs. Adjust as necessary. Comment out what you
+  # don't need
+  include:
+    # Linux
+    - env: TARGET=x86_64-unknown-linux-gnu
+
+    # OSX
+    - env: TARGET=x86_64-apple-darwin
+      os: osx
 
 before_install:
-   - export PATH="$PATH:$HOME/.cargo/bin"
+  - set -e
+  - rustup self update
+
+install:
+  - sh ci/install.sh
+  - source ~/.cargo/env || true
 
 script:
-    - cargo build --release --target $TARGET
-    - cargo test --target $TARGET
+  - bash ci/script.sh
+
+after_script: set +e
+
+before_deploy:
+  - sh ci/before_deploy.sh
 
 deploy:
-  provider: releases
+  # TODO update `api_key.secure`
+  # - Create a `public_repo` GitHub token. Go to: https://github.com/settings/tokens/new
+  # - Encrypt it: `travis encrypt 0123456789012345678901234567890123456789
+  # - Paste the output down here
   api_key:
-     secure: "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"
-  file: ${PROJECT_NAME}-${TRAVIS_TAG}-${TARGET}.tar.gz
-  skip_cleanup: true
+    secure: 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
+  file_glob: true
+  file: $CRATE_NAME-$TRAVIS_TAG-$TARGET.*
   on:
-    repo: bootandy/dust
+    # TODO Here you can pick which targets will generate binary releases
+    # In this example, there are some targets that are tested using the stable
+    # and nightly channels. This condition makes sure there is only one release
+    # for such targets and that's generated using the stable channel
+    condition: $TRAVIS_RUST_VERSION = stable
     tags: true
+  provider: releases
+  skip_cleanup: true
+
+cache: cargo
+before_cache:
+  # Travis can't cache files that are not readable by "others"
+  - chmod -R a+r $HOME/.cargo
+
+branches:
+  only:
+    # release tags
+    - /^v\d+\.\d+\.\d+.*$/
+    - master
 
 notifications:
   email:
     on_success: never
-

Cargo.toml

@@ -1,8 +1,11 @@
 [package]
 name = "dust"
-version = "0.1.0"
-authors = ["bootandy <bootandy@gmail.com>"]
+version = "0.2.3"
+authors = ["bootandy <bootandy@gmail.com>", "nebkor <code@ardent.nebcorp.com>"]
 
 [dependencies]
 ansi_term = "0.11"
 clap = "2.31"
+assert_cli = "0.5"
+tempfile = "3"
+

README.md

@@ -4,6 +4,12 @@
 # Dust
 du + rust = dust. A rust alternative to du
 
+[Releases](https://github.com/bootandy/dust/releases)
+To install:
+ * Download linux / mac binary from [Releases](https://github.com/bootandy/dust/releases)
+ * unzip file: tar -xvf <file>
+ * copy file to search path: sudo mv dust /usr/local/bin/
+
 Unlike du, dust is meant to give you an instant overview of which directories are using disk space without requiring sort or head. Dust does not count file system blocks; it uses file sizes instead. Dust will print a maximum of 1 'Did not have permissions message'.
 
 
@@ -11,32 +17,35 @@ Dust will list the 15 biggest sub directories and will smartly recurse down the
 
 ```
 Usage: dust <dir>
+Usage: dust <dir> <another_dir> <and_more>
+Usage: dust -s <dir> (apparent-size - shows the length of the file as opposed to the amount of disk space it uses)
 Usage: dust -n 30  <dir>  (Shows 30 directories not 15)
 ```
 
 
 ```
-dust .
- 161M  .
- 160M └── ./target
- 123M    ├── ./target/debug
-  83M    │  ├── ./target/debug/deps
-  16M    │  │  ├── ./target/debug/deps/libclap-82e6176feef5d4b7.rlib
- 8.6M    │  │  └── ./target/debug/deps/dust-993f7d919d92f0f8.dSYM
- 8.6M    │  │     └── ./target/debug/deps/dust-993f7d919d92f0f8.dSYM/Contents
- 8.6M    │  │        └── ./target/debug/deps/dust-993f7d919d92f0f8.dSYM/Contents/Resources
-  27M    │  ├── ./target/debug/incremental
-  12M    │  └── ./target/debug/build
-  20M    ├── ./target/x86_64-apple-darwin
-  20M    │  └── ./target/x86_64-apple-darwin/debug
-  20M    │     └── ./target/x86_64-apple-darwin/debug/deps
-  16M    │        └── ./target/x86_64-apple-darwin/debug/deps/libclap-7e3f8513c52cd558.rlib
-  16M    └── ./target/release
-  13M       └── ./target/release/deps
+djin:git/dust> dust
+  65M  .
+  65M └─┬ ./target
+  49M   ├─┬ ./target/debug
+  26M   │ ├─┬ ./target/debug/deps
+  21M   │ │ └── ./target/debug/deps/libclap-9e6625ac8ff074ad.rlib
+  13M   │ ├── ./target/debug/dust
+ 8.9M   │ └─┬ ./target/debug/incremental
+ 6.7M   │   ├─┬ ./target/debug/incremental/dust-2748eiei2tcnp
+ 6.7M   │   │ └─┬ ./target/debug/incremental/dust-2748eiei2tcnp/s-ezd6jnik5u-163pyem-1aab9ncf5glum
+ 3.0M   │   │   └── ./target/debug/incremental/dust-2748eiei2tcnp/s-ezd6jnik5u-163pyem-1aab9ncf5glum/dep-graph.bin
+ 2.2M   │   └─┬ ./target/debug/incremental/dust-1dlon65p8m3vl
+ 2.2M   │     └── ./target/debug/incremental/dust-1dlon65p8m3vl/s-ezd6jncecv-1xsnfd0-4dw9l1r2th2t
+  15M   └─┬ ./target/release
+ 9.2M     ├─┬ ./target/release/deps
+ 6.7M     │ └── ./target/release/deps/libclap-87bc2534ea57f044.rlib
+ 5.9M     └── ./target/release/dust
 ```
-
 Performance: dust is currently about 4 times slower than du.
 
+Alternatives:
+ * [NCDU](https://dev.yorhel.nl/ncdu)
+ * du -d 1 -h | sort -h
 
-
-
+Note: Apparent-size is calculated slightly differently in dust to gdu. In dust each hard link is counted as using file_length space. In gdu only the first entry is counted.

ci/before_deploy.ps1

@@ -0,0 +1,23 @@
+# This script takes care of packaging the build artifacts that will go in the
+# release zipfile
+
+$SRC_DIR = $PWD.Path
+$STAGE = [System.Guid]::NewGuid().ToString()
+
+Set-Location $ENV:Temp
+New-Item -Type Directory -Name $STAGE
+Set-Location $STAGE
+
+$ZIP = "$SRC_DIR\$($Env:CRATE_NAME)-$($Env:APPVEYOR_REPO_TAG_NAME)-$($Env:TARGET).zip"
+
+# TODO Update this to package the right artifacts
+Copy-Item "$SRC_DIR\target\$($Env:TARGET)\release\dust" '.\'
+
+7z a "$ZIP" *
+
+Push-AppveyorArtifact "$ZIP"
+
+Remove-Item *.* -Force
+Set-Location ..
+Remove-Item $STAGE
+Set-Location $SRC_DIR

ci/before_deploy.sh

@@ -0,0 +1,33 @@
+# This script takes care of building your crate and packaging it for release
+
+set -ex
+
+main() {
+    local src=$(pwd) \
+          stage=
+
+    case $TRAVIS_OS_NAME in
+        linux)
+            stage=$(mktemp -d)
+            ;;
+        osx)
+            stage=$(mktemp -d -t tmp)
+            ;;
+    esac
+
+    test -f Cargo.lock || cargo generate-lockfile
+
+    # TODO Update this to build the artifacts that matter to you
+    cross rustc --bin dust --target $TARGET --release -- -C lto
+
+    # TODO Update this to package the right artifacts
+    cp target/$TARGET/release/dust $stage/
+
+    cd $stage
+    tar czf $src/$CRATE_NAME-$TRAVIS_TAG-$TARGET.tar.gz *
+    cd $src
+
+    rm -rf $stage
+}
+
+main

ci/install.sh

@@ -0,0 +1,27 @@
+set -ex
+
+main() {
+    local target=
+    if [ $TRAVIS_OS_NAME = linux ]; then
+        target=x86_64-unknown-linux-musl
+        sort=sort
+    else
+        target=x86_64-apple-darwin
+        sort=gsort  # for `sort --sort-version`, from brew's coreutils.
+    fi
+
+    # This fetches latest stable release
+    local tag=$(git ls-remote --tags --refs --exit-code https://github.com/japaric/cross \
+                       | cut -d/ -f3 \
+                       | grep -E '^v[0.1.0-9.]+$' \
+                       | $sort --version-sort \
+                       | tail -n1)
+    curl -LSfs https://japaric.github.io/trust/install.sh | \
+        sh -s -- \
+           --force \
+           --git japaric/cross \
+           --tag $tag \
+           --target $target
+}
+
+main

ci/script.sh

@@ -0,0 +1,24 @@
+# This script takes care of testing your crate
+
+set -ex
+
+# TODO This is the "test phase", tweak it as you see fit
+main() {
+    cross build --target $TARGET
+    cross build --target $TARGET --release
+
+    if [ ! -z $DISABLE_TESTS ]; then
+        return
+    fi
+
+    cross test --target $TARGET
+    cross test --target $TARGET --release
+
+    cross run --target $TARGET
+    cross run --target $TARGET --release
+}
+
+# we don't run the "test phase" when doing deploys
+if [ -z $TRAVIS_TAG ]; then
+    main
+fi

src/display.rs

@@ -0,0 +1,131 @@
+extern crate ansi_term;
+
+use self::ansi_term::Colour::Fixed;
+use dust::Node;
+
+static UNITS: [char; 4] = ['T', 'G', 'M', 'K'];
+
+pub fn draw_it(permissions: bool, heads: &Vec<Node>, to_display: &Vec<&Node>) -> () {
+    if !permissions {
+        eprintln!("Did not have permissions for all directories");
+    }
+
+    for d in to_display {
+        if heads.contains(d) {
+            display_node(d, &to_display, true, "")
+        }
+    }
+}
+
+fn display_node<S: Into<String>>(
+    node_to_print: &Node,
+    to_display: &Vec<&Node>,
+    is_first: bool,
+    indentation_str: S,
+) {
+    let mut is = indentation_str.into();
+    print_this_node(node_to_print, is_first, is.as_ref());
+
+    is = is.replace("└─┬", "  ");
+    is = is.replace("└──", "  ");
+    is = is.replace("├──", "│ ");
+    is = is.replace("├─┬", "│ ");
+
+    let printable_node_slashes = node_to_print.name().matches('/').count();
+
+    let mut num_siblings = to_display.iter().fold(0, |a, b| {
+        if node_to_print.children().contains(b)
+            && b.name().matches('/').count() == printable_node_slashes + 1
+        {
+            a + 1
+        } else {
+            a
+        }
+    });
+
+    let mut is_biggest = true;
+    for node in to_display {
+        if node_to_print.children().contains(node) {
+            let has_display_children = node.children()
+                .iter()
+                .fold(false, |has_kids, n| has_kids || to_display.contains(&n));
+
+            let has_children = node.children().len() > 0 && has_display_children;
+            if node.name().matches('/').count() == printable_node_slashes + 1 {
+                num_siblings -= 1;
+
+                let tree_chars = {
+                    if num_siblings == 0 {
+                        if has_children {
+                            "└─┬"
+                        } else {
+                            "└──"
+                        }
+                    } else {
+                        if has_children {
+                            "├─┬"
+                        } else {
+                            "├──"
+                        }
+                    }
+                };
+                display_node(&node, to_display, is_biggest, is.to_string() + tree_chars);
+                is_biggest = false;
+            }
+        }
+    }
+}
+
+fn print_this_node(node: &Node, is_biggest: bool, indentation: &str) {
+    let pretty_size = format!("{:>5}", human_readable_number(node.size()),);
+    println!(
+        "{}",
+        format_string(node.name(), is_biggest, pretty_size.as_ref(), indentation)
+    )
+}
+
+pub fn format_string(dir_name: &str, is_biggest: bool, size: &str, indentation: &str) -> String {
+    format!(
+        "{} {} {}",
+        if is_biggest {
+            Fixed(196).paint(size)
+        } else {
+            Fixed(7).paint(size)
+        },
+        indentation,
+        dir_name,
+    )
+}
+
+fn human_readable_number(size: u64) -> String {
+    for (i, u) in UNITS.iter().enumerate() {
+        let marker = 1024u64.pow((UNITS.len() - i) as u32);
+        if size >= marker {
+            if size / marker < 10 {
+                return format!("{:.1}{}", (size as f32 / marker as f32), u);
+            } else {
+                return format!("{}{}", (size / marker), u);
+            }
+        }
+    }
+    return format!("{}B", size);
+}
+
+mod tests {
+    #[allow(unused_imports)]
+    use super::*;
+
+    #[test]
+    fn test_human_readable_number() {
+        assert_eq!(human_readable_number(1), "1B");
+        assert_eq!(human_readable_number(956), "956B");
+        assert_eq!(human_readable_number(1004), "1004B");
+        assert_eq!(human_readable_number(1024), "1.0K");
+        assert_eq!(human_readable_number(1536), "1.5K");
+        assert_eq!(human_readable_number(1024 * 512), "512K");
+        assert_eq!(human_readable_number(1024 * 1024), "1.0M");
+        assert_eq!(human_readable_number(1024 * 1024 * 1024 - 1), "1023M");
+        assert_eq!(human_readable_number(1024 * 1024 * 1024 * 20), "20G");
+        assert_eq!(human_readable_number(1024 * 1024 * 1024 * 1024), "1.0T");
+    }
+}

src/lib.rs

@@ -0,0 +1,68 @@
+use std::cmp::{Eq, Ord, Ordering, PartialEq, PartialOrd};
+
+#[derive(Clone, Debug)]
+pub struct Node {
+    name: String,
+    size: u64,
+    children: Vec<Node>,
+}
+
+impl Node {
+    pub fn new<S: Into<String>>(name: S, size: u64, children: Vec<Node>) -> Self {
+        Node {
+            children: children,
+            name: name.into(),
+            size: size,
+        }
+    }
+
+    pub fn children(&self) -> &Vec<Node> {
+        &self.children
+    }
+
+    pub fn name(&self) -> &String {
+        &self.name
+    }
+
+    pub fn size(&self) -> u64 {
+        self.size
+    }
+}
+
+impl Ord for Node {
+    fn cmp(&self, other: &Self) -> Ordering {
+        if self.size > other.size {
+            Ordering::Less
+        } else if self.size < other.size {
+            Ordering::Greater
+        } else {
+            let my_slashes = self.name.matches('/').count();
+            let other_slashes = other.name.matches('/').count();
+
+            if my_slashes > other_slashes {
+                Ordering::Greater
+            } else if my_slashes < other_slashes {
+                Ordering::Less
+            } else {
+                if self.name < other.name {
+                    Ordering::Less
+                } else if self.name > other.name {
+                    Ordering::Greater
+                } else {
+                    Ordering::Equal
+                }
+            }
+        }
+    }
+}
+impl PartialOrd for Node {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+impl PartialEq for Node {
+    fn eq(&self, other: &Self) -> bool {
+        (&self.name, self.size) == (&other.name, other.size)
+    }
+}
+impl Eq for Node {}

src/main.rs

@@ -1,77 +1,19 @@
-// test:
-// recursive dirs that link to each other.
-// Pass in bad dir name
-// num to search for is less than num available
-// admin files.
-//
-extern crate ansi_term;
 #[macro_use]
 extern crate clap;
+extern crate assert_cli;
+extern crate dust;
 
-use std::collections::HashSet;
-
-use ansi_term::Colour::Fixed;
+use self::display::draw_it;
 use clap::{App, AppSettings, Arg};
+use utils::{find_big_ones, get_dir_tree};
 
-use std::cmp;
-use std::cmp::Ordering;
-use std::fs;
-use std::fs::ReadDir;
-use std::io;
+mod display;
+mod utils;
 
-#[derive(Clone, Debug)]
-struct Node {
-    dir: Dir,
-    children: Vec<Node>,
-}
-impl Ord for Node {
-    fn cmp(&self, other: &Self) -> Ordering {
-        if self.dir.size > other.dir.size {
-            Ordering::Less
-        } else if self.dir.size < other.dir.size {
-            Ordering::Greater
-        } else {
-            let my_slashes = self.dir.name.matches("/").count();
-            let other_slashes = other.dir.name.matches("/").count();
-
-            if my_slashes > other_slashes {
-                Ordering::Greater
-            } else if my_slashes < other_slashes {
-                Ordering::Less
-            } else {
-                if self.dir.name < other.dir.name {
-                    Ordering::Less
-                } else if self.dir.name > other.dir.name {
-                    Ordering::Greater
-                } else {
-                    Ordering::Equal
-                }
-            }
-        }
-    }
-}
-impl PartialOrd for Node {
-    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
-        Some(self.cmp(other))
-    }
-}
-impl PartialEq for Node {
-    fn eq(&self, other: &Self) -> bool {
-        (&self.dir.name, self.dir.size) == (&other.dir.name, other.dir.size)
-    }
-}
-impl Eq for Node {}
-
-#[derive(Clone, Debug)]
-struct Dir {
-    name: String,
-    size: u64,
-}
-
-static DEFAULT_NUMBER_OF_LINES: &'static str = &"15";
+static DEFAULT_NUMBER_OF_LINES: &'static str = "15";
 
 fn main() {
-    let options = App::new("Trailing args example")
+    let options = App::new("Dust")
         .setting(AppSettings::TrailingVarArg)
         .arg(
             Arg::with_name("number_of_lines")
@@ -80,6 +22,11 @@ fn main() {
                 .takes_value(true)
                 .default_value(DEFAULT_NUMBER_OF_LINES),
         )
+        .arg(
+            Arg::with_name("use_apparent_size")
+                .short("s")
+                .help("If set will use file length. Otherwise we use blocks"),
+        )
         .arg(Arg::with_name("inputs").multiple(true))
         .get_matches();
 
@@ -90,275 +37,12 @@ fn main() {
         }
     };
     let number_of_lines = value_t!(options.value_of("number_of_lines"), usize).unwrap();
+    let use_apparent_size = options.is_present("use_apparent_size");
 
-    let (permissions, results) = get_dir_tree(filenames);
-    let slice_it = find_big_ones(&results, number_of_lines);
-    display(permissions, slice_it);
+    let (permissions, node_per_top_level_dir) = get_dir_tree(&filenames, use_apparent_size);
+    let slice_it = find_big_ones(&node_per_top_level_dir, number_of_lines);
+    draw_it(permissions, &node_per_top_level_dir, &slice_it);
 }
 
-fn get_dir_tree(filenames: Vec<&str>) -> (bool, Vec<Node>) {
-    let mut permissions = true;
-    let mut results = vec![];
-    for b in filenames {
-        let mut new_name = String::from(b);
-        while new_name.chars().last() == Some('/') && new_name.len() != 1 {
-            new_name.pop();
-        }
-        let (hp, data) = examine_dir_str(new_name);
-        permissions = permissions && hp;
-        results.push(data);
-    }
-    (permissions, results)
-}
-
-fn examine_dir_str(loc: String) -> (bool, Node) {
-    let mut inodes: HashSet<u64> = HashSet::new();
-    let (hp, result) = examine_dir(fs::read_dir(&loc), &mut inodes);
-
-    // This needs to be folded into the below recursive call somehow
-    let new_size = result.iter().fold(0, |a, b| a + b.dir.size);
-    (
-        hp,
-        Node {
-            dir: Dir {
-                name: loc,
-                size: new_size,
-            },
-            children: result,
-        },
-    )
-}
-
-#[cfg(target_os = "linux")]
-fn get_metadata_blocks_and_inode(d: &std::fs::DirEntry) -> Option<(u64, u64)> {
-    use std::os::linux::fs::MetadataExt;
-    match d.metadata().ok() {
-        Some(md) => Some((md.len(), md.st_ino())),
-        None => None,
-    }
-}
-
-#[cfg(target_os = "unix")]
-fn get_metadata_blocks_and_inode(d: &std::fs::DirEntry) -> Option<(u64, u64)> {
-    use std::os::unix::fs::MetadataExt;
-    match d.metadata().ok() {
-        Some(md) => Some((md.len(), md.ino())),
-        None => None,
-    }
-}
-
-#[cfg(target_os = "macos")]
-fn get_metadata_blocks_and_inode(d: &std::fs::DirEntry) -> Option<(u64, u64)> {
-    use std::os::macos::fs::MetadataExt;
-    match d.metadata().ok() {
-        Some(md) => Some((md.len(), md.st_ino())),
-        None => None,
-    }
-}
-
-#[cfg(not(any(target_os = "linux", target_os = "unix", target_os = "macos")))]
-fn get_metadata_blocks_and_inode(_d: &std::fs::DirEntry) -> Option<(u64, u64)> {
-    match d.metadata().ok() {
-        Some(md) => Some((md.len(), 0)), //move to option not 0
-        None => None,
-    }
-}
-
-fn examine_dir(a_dir: io::Result<ReadDir>, inodes: &mut HashSet<u64>) -> (bool, Vec<Node>) {
-    let mut result = vec![];
-    let mut have_permission = true;
-
-    if a_dir.is_ok() {
-        let paths = a_dir.unwrap();
-        for dd in paths {
-            match dd {
-                Ok(d) => {
-                    let file_type = d.file_type().ok();
-                    let maybe_size_and_inode = get_metadata_blocks_and_inode(&d);
-
-                    match (file_type, maybe_size_and_inode) {
-                        (Some(file_type), Some((size, inode))) => {
-                            let s = d.path().to_string_lossy().to_string();
-                            if inodes.contains(&inode) {
-                                continue;
-                            }
-                            inodes.insert(inode);
-
-                            if d.path().is_dir() && !file_type.is_symlink() {
-                                let (hp, recursive) = examine_dir(fs::read_dir(d.path()), inodes);
-                                have_permission = have_permission && hp;
-                                let new_size = recursive.iter().fold(size, |a, b| a + b.dir.size);
-                                result.push(Node {
-                                    dir: Dir {
-                                        name: s,
-                                        size: new_size,
-                                    },
-                                    children: recursive,
-                                })
-                            } else {
-                                result.push(Node {
-                                    dir: Dir {
-                                        name: s,
-                                        size: size,
-                                    },
-                                    children: vec![],
-                                })
-                            }
-                        }
-                        (_, None) => have_permission = false,
-                        (_, _) => (),
-                    }
-                }
-                Err(_) => (),
-            }
-        }
-    } else {
-        have_permission = false;
-    }
-    (have_permission, result)
-}
-
-// We start with a list of root directories - these must be the biggest folders
-// We then repeadedly merge in the children of the biggest directory - Each iteration
-// the next biggest directory's children are merged in.
-fn find_big_ones<'a>(l: &'a Vec<Node>, max_to_show: usize) -> Vec<&Node> {
-    let mut new_l: Vec<&Node> = l.iter().map(|a| a).collect();
-    new_l.sort();
-
-    for processed_pointer in 0..max_to_show {
-        if new_l.len() == processed_pointer {
-            break;
-        }
-        // Must be a list of pointers into new_l otherwise b_list will go out of scope
-        // when it is deallocated
-        let mut b_list: Vec<&Node> = new_l[processed_pointer]
-            .children
-            .iter()
-            .map(|a| a)
-            .collect();
-        new_l.extend(b_list);
-        new_l.sort();
-        /*println!(
-            "{:?} -------------------",
-            new_l
-                .iter()
-                .map(|a| a.dir.size.to_string() + ": " + &a.dir.name)
-                .collect::<Vec<String>>()
-        );*/
-    }
-    if new_l.len() > max_to_show {
-        new_l[0..max_to_show + 1].to_vec()
-    } else {
-        new_l
-    }
-}
-
-fn display(permissions: bool, to_display: Vec<&Node>) -> () {
-    if !permissions {
-        eprintln!("Did not have permissions for all directories");
-    }
-
-    display_node(to_display[0], &to_display, true, 1, "")
-}
-
-fn display_node<S: Into<String>>(
-    node_to_print: &Node,
-    to_display: &Vec<&Node>,
-    is_first: bool,
-    depth: u8,
-    indentation_str: S,
-) {
-    let mut is = indentation_str.into();
-    print_this_node(node_to_print, is_first, depth, is.as_ref());
-
-    is = is.replace("└──", "   ");
-    is = is.replace("├──", "│  ");
-
-    let printable_node_slashes = node_to_print.dir.name.matches("/").count();
-
-    let mut num_sibblings = to_display.iter().fold(0, |a, b| {
-        if node_to_print.children.contains(b)
-            && b.dir.name.matches("/").count() == printable_node_slashes + 1
-        {
-            a + 1
-        } else {
-            a
-        }
-    });
-
-    let mut is_biggest = true;
-    for node in to_display {
-        if node_to_print.children.contains(node) {
-            if node.dir.name.matches("/").count() == printable_node_slashes + 1 {
-                num_sibblings -= 1;
-                let tree_chars = {
-                    if num_sibblings == 0 {
-                        "└──"
-                    } else {
-                        "├──"
-                    }
-                };
-                display_node(
-                    &node,
-                    to_display,
-                    is_biggest,
-                    depth + 1,
-                    is.to_string() + tree_chars,
-                );
-                is_biggest = false;
-            }
-        }
-    }
-}
-
-fn print_this_node(node_to_print: &Node, is_biggest: bool, depth: u8, indentation_str: &str) {
-    let padded_size = format!("{:>5}", human_readable_number(node_to_print.dir.size),);
-    println!(
-        "{} {} {}",
-        if is_biggest {
-            Fixed(196).paint(padded_size)
-        } else {
-            Fixed(7).paint(padded_size)
-        },
-        indentation_str,
-        Fixed(7)
-            .on(Fixed(cmp::min(8, (depth) as u8) + 231))
-            .paint(node_to_print.dir.name.to_string())
-    );
-}
-
-fn human_readable_number(size: u64) -> (String) {
-    let units = vec!["T", "G", "M", "K"]; //make static
-
-    //return format!("{}B", size);
-
-    for (i, u) in units.iter().enumerate() {
-        let marker = 1024u64.pow((units.len() - i) as u32);
-        if size >= marker {
-            if size / marker < 10 {
-                return format!("{:.1}{}", (size as f32 / marker as f32), u);
-            } else {
-                return format!("{}{}", (size / marker), u);
-            }
-        }
-    }
-    return format!("{}B", size);
-}
-
-mod tests {
-    use super::*;
-
-    #[test]
-    fn test_human_readable_number() {
-        assert_eq!(human_readable_number(1), "1B");
-        assert_eq!(human_readable_number(956), "956B");
-        assert_eq!(human_readable_number(1004), "1004B");
-        assert_eq!(human_readable_number(1024), "1.0K");
-        assert_eq!(human_readable_number(1536), "1.5K");
-        assert_eq!(human_readable_number(1024 * 512), "512K");
-        assert_eq!(human_readable_number(1024 * 1024), "1.0M");
-        assert_eq!(human_readable_number(1024 * 1024 * 1024 - 1), "1023M");
-        assert_eq!(human_readable_number(1024 * 1024 * 1024 * 20), "20G");
-        assert_eq!(human_readable_number(1024 * 1024 * 1024 * 1024), "1.0T");
-    }
-}
+#[cfg(test)]
+mod tests;

src/test_dir/many/hello_file

@@ -0,0 +1 @@
+hello

src/tests.rs

@@ -0,0 +1,140 @@
+extern crate ansi_term;
+extern crate tempfile;
+use self::tempfile::Builder;
+use self::tempfile::TempDir;
+use super::*;
+use display::format_string;
+use std::fs::File;
+use std::io::Write;
+use std::path::PathBuf;
+use std::process::Command;
+
+#[test]
+pub fn test_main() {
+    let r = format!(
+        "{}
+{}
+{}
+{}",
+        format_string("src/test_dir", true, " 4.0K", ""),
+        format_string("src/test_dir/many", true, " 4.0K", "└─┬",),
+        format_string("src/test_dir/many/hello_file", true, " 4.0K", "  ├──",),
+        format_string("src/test_dir/many/a_file", false, "   0B", "  └──",),
+    );
+
+    assert_cli::Assert::main_binary()
+        .with_args(&["src/test_dir"])
+        .stdout()
+        .is(r)
+        .unwrap();
+}
+
+#[test]
+pub fn test_apparent_size() {
+    let r = format!(
+        "{}",
+        format_string("src/test_dir/many/hello_file", true, "   6B", "  ├──",),
+    );
+
+    assert_cli::Assert::main_binary()
+        .with_args(&["-s", "src/test_dir"])
+        .stdout()
+        .contains(r)
+        .unwrap();
+}
+
+fn build_temp_file(dir: &TempDir) -> (PathBuf) {
+    let file_path = dir.path().join("notes.txt");
+    let mut file = File::create(&file_path).unwrap();
+    writeln!(file, "I am a temp file").unwrap();
+    file_path
+}
+
+#[test]
+pub fn test_soft_sym_link() {
+    let dir = Builder::new().tempdir().unwrap();
+    let file = build_temp_file(&dir);
+    let dir_s = dir.path().to_str().unwrap();
+    let file_path_s = file.to_str().unwrap();
+
+    let link_name = dir.path().join("the_link");
+    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 r = format!(
+        "{}
+{}
+{}",
+        format_string(dir_s, true, " 8.0K", ""),
+        format_string(file_path_s, true, " 4.0K", "├──",),
+        format_string(link_name_s, false, " 4.0K", "└──",),
+    );
+
+    assert_cli::Assert::main_binary()
+        .with_args(&[dir_s])
+        .stdout()
+        .contains(r)
+        .unwrap();
+}
+
+// Hard links are ignored as the inode is the same as the file
+#[test]
+pub fn test_hard_sym_link() {
+    let dir = Builder::new().tempdir().unwrap();
+    let file = build_temp_file(&dir);
+    let dir_s = dir.path().to_str().unwrap();
+    let file_path_s = file.to_str().unwrap();
+
+    let link_name = dir.path().join("the_link");
+    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 r = format!(
+        "{}
+{}",
+        format_string(dir_s, true, " 4.0K", ""),
+        format_string(file_path_s, true, " 4.0K", "└──")
+    );
+
+    assert_cli::Assert::main_binary()
+        .with_args(&[dir_s])
+        .stdout()
+        .contains(r)
+        .unwrap();
+}
+
+//Check we don't recurse down an infinite symlink tree
+#[test]
+pub fn test_recursive_sym_link() {
+    let dir = Builder::new().tempdir().unwrap();
+    let dir_s = dir.path().to_str().unwrap();
+
+    let link_name = dir.path().join("the_link");
+    let link_name_s = link_name.to_str().unwrap();
+
+    let c = Command::new("cd").arg(dir_s).output();
+    assert!(c.is_ok());
+    let c = Command::new("ln")
+        .arg("-s")
+        .arg(".")
+        .arg(link_name_s)
+        .output();
+    assert!(c.is_ok());
+
+    let r = format!("{}", format_string(dir_s, true, " 4.0K", ""));
+
+    assert_cli::Assert::main_binary()
+        .with_args(&[dir_s])
+        .stdout()
+        .contains(r)
+        .unwrap();
+}

src/utils/mod.rs

@@ -0,0 +1,113 @@
+use std::collections::HashSet;
+
+use std::fs::{self, ReadDir};
+use std::io;
+
+use dust::Node;
+
+mod platform;
+use self::platform::*;
+
+pub fn get_dir_tree(filenames: &Vec<&str>, apparent_size: bool) -> (bool, Vec<Node>) {
+    let mut permissions = true;
+    let mut results = vec![];
+    for &b in filenames {
+        let mut new_name = String::from(b);
+        while new_name.chars().last() == Some('/') && new_name.len() != 1 {
+            new_name.pop();
+        }
+        let (hp, data) = examine_dir_str(&new_name, apparent_size);
+        permissions = permissions && hp;
+        results.push(data);
+    }
+    (permissions, results)
+}
+
+fn examine_dir_str(loc: &str, apparent_size: bool) -> (bool, Node) {
+    let mut inodes: HashSet<(u64, u64)> = HashSet::new();
+    let (hp, result) = examine_dir(fs::read_dir(loc), apparent_size, &mut inodes);
+
+    // This needs to be folded into the below recursive call somehow
+    let new_size = result.iter().fold(0, |a, b| a + b.size());
+    (hp, Node::new(loc, new_size, result))
+}
+
+fn examine_dir(
+    a_dir: io::Result<ReadDir>,
+    apparent_size: bool,
+    inodes: &mut HashSet<(u64, u64)>,
+) -> (bool, Vec<Node>) {
+    let mut result = vec![];
+    let mut have_permission = true;
+
+    if a_dir.is_ok() {
+        let paths = a_dir.unwrap();
+        for dd in paths {
+            match dd {
+                Ok(d) => {
+                    let file_type = d.file_type().ok();
+                    let maybe_size_and_inode = get_metadata(&d, apparent_size);
+
+                    match (file_type, maybe_size_and_inode) {
+                        (Some(file_type), Some((size, inode))) => {
+                            let s = d.path().to_string_lossy().to_string();
+                            if !apparent_size {
+                                if let Some(inode_dev_pair) = inode {
+                                    if inodes.contains(&inode_dev_pair) {
+                                        continue;
+                                    }
+                                    inodes.insert(inode_dev_pair);
+                                }
+                            }
+
+                            if d.path().is_dir() && !file_type.is_symlink() {
+                                let (hp, recursive) =
+                                    examine_dir(fs::read_dir(d.path()), apparent_size, inodes);
+                                have_permission = have_permission && hp;
+                                let new_size = recursive.iter().fold(size, |a, b| a + b.size());
+                                result.push(Node::new(s, new_size, recursive))
+                            } else {
+                                result.push(Node::new(s, size, vec![]))
+                            }
+                        }
+                        (_, None) => have_permission = false,
+                        (_, _) => (),
+                    }
+                }
+                Err(_) => (),
+            }
+        }
+    } else {
+        have_permission = false;
+    }
+    (have_permission, result)
+}
+
+// We start with a list of root directories - these must be the biggest folders
+// We then repeadedly merge in the children of the biggest directory - Each iteration
+// the next biggest directory's children are merged in.
+pub fn find_big_ones<'a>(l: &'a Vec<Node>, max_to_show: usize) -> Vec<&Node> {
+    let mut new_l: Vec<&Node> = l.iter().map(|a| a).collect();
+    new_l.sort();
+
+    for processed_pointer in 0..max_to_show {
+        if new_l.len() == processed_pointer {
+            break;
+        }
+        // Must be a list of pointers into new_l otherwise b_list will go out of scope
+        // when it is deallocated
+        let mut b_list: Vec<&Node> = new_l[processed_pointer]
+            .children()
+            .iter()
+            .map(|a| a)
+            .collect();
+        new_l.extend(b_list);
+        new_l.sort();
+    }
+
+    if new_l.len() > max_to_show {
+        new_l[0..max_to_show + 1].to_vec()
+    } else {
+        new_l
+    }
+}

src/utils/platform.rs

@@ -0,0 +1,72 @@
+use std;
+
+fn get_block_size() -> u64 {
+    // All os specific implementations of MetatdataExt seem to define a block as 512 bytes
+    // https://doc.rust-lang.org/std/os/linux/fs/trait.MetadataExt.html#tymethod.st_blocks
+    512
+}
+
+#[cfg(target_os = "linux")]
+pub fn get_metadata(
+    d: &std::fs::DirEntry,
+    use_apparent_size: bool,
+) -> Option<(u64, Option<(u64, u64)>)> {
+    use std::os::linux::fs::MetadataExt;
+    match d.metadata().ok() {
+        Some(md) => {
+            let inode = Some((md.st_ino(), md.st_dev()));
+            if use_apparent_size {
+                Some((md.len(), inode))
+            } else {
+                Some((md.st_blocks() * get_block_size(), inode))
+            }
+        }
+        None => None,
+    }
+}
+
+#[cfg(target_os = "unix")]
+pub fn get_metadata(
+    d: &std::fs::DirEntry,
+    use_apparent_size: bool,
+) -> Option<(u64, Option<(u64, u64)>)> {
+    use std::os::unix::fs::MetadataExt;
+    match d.metadata().ok() {
+        Some(md) => {
+            let inode = Some((md.ino(), md.dev()));
+            if use_apparent_size {
+                Some((md.len(), inode))
+            } else {
+                Some((md.blocks() * get_block_size(), inode))
+            }
+        }
+        None => None,
+    }
+}
+
+#[cfg(target_os = "macos")]
+pub fn get_metadata(
+    d: &std::fs::DirEntry,
+    use_apparent_size: bool,
+) -> Option<(u64, Option<(u64, u64)>)> {
+    use std::os::macos::fs::MetadataExt;
+    match d.metadata().ok() {
+        Some(md) => {
+            let inode = Some((md.st_ino(), md.st_dev()));
+            if use_apparent_size {
+                Some((md.len(), inode))
+            } else {
+                Some((md.st_blocks() * get_block_size(), inode))
+            }
+        }
+        None => None,
+    }
+}
+
+#[cfg(not(any(target_os = "linux", target_os = "unix", target_os = "macos")))]
+pub fn get_metadata(d: &std::fs::DirEntry, _apparent: bool) -> Option<(u64, Option<(u64, u64)>)> {
+    match d.metadata().ok() {
+        Some(md) => Some((md.len(), None)),
+        None => None,
+    }
+}