Merge pull request #19 from bootandy/rm_dup_input

code to remove duplicate arguments

src/display.rs

@@ -1,6 +1,7 @@
 extern crate ansi_term;
 
 use self::ansi_term::Colour::Fixed;
+use std::collections::HashSet;
 
 static UNITS: [char; 4] = ['T', 'G', 'M', 'K'];
 
@@ -8,15 +9,18 @@ pub fn draw_it(
     permissions: bool,
     short_paths: bool,
     depth: Option<u64>,
-    base_dirs: Vec<String>,
+    base_dirs: HashSet<String>,
     to_display: Vec<(String, u64)>,
 ) -> () {
     if !permissions {
         eprintln!("Did not have permissions for all directories");
     }
+    let mut found = HashSet::new();
 
-    for f in base_dirs {
+    for &(ref k, _) in to_display.iter() {
-        display_node(f, &to_display, true, short_paths, depth, "")
+        if base_dirs.contains(k) {
+            display_node(&k, &mut found, &to_display, true, short_paths, depth, "─┬")
+        }
     }
 }
 
@@ -30,65 +34,59 @@ fn get_size(nodes: &Vec<(String, u64)>, node_to_print: &String) -> Option<u64> {
 }
 
 fn display_node<S: Into<String>>(
-    node_to_print: String,
+    node_to_print: &String,
+    found: &mut HashSet<String>,
     to_display: &Vec<(String, u64)>,
     is_biggest: bool,
     short_paths: bool,
     depth: Option<u64>,
     indentation_str: S,
 ) {
+    if found.contains(node_to_print) {
+        return;
+    }
+    found.insert(node_to_print.to_string());
+
     let new_depth = match depth {
         None => None,
         Some(0) => return,
         Some(d) => Some(d - 1),
     };
-    match get_size(to_display, &node_to_print) {
+    match get_size(to_display, node_to_print) {
         None => println!("Can not find path: {}", node_to_print),
         Some(size) => {
-            let mut is = indentation_str.into();
+            let is = indentation_str.into();
             let ntp: &str = node_to_print.as_ref();
 
             print_this_node(ntp, size, is_biggest, short_paths, is.as_ref());
+            let new_indent_str = clean_indentation_string(is);
 
-            is = is.replace("└─┬", "  ");
+            let num_slashes = node_to_print.matches('/').count();
-            is = is.replace("└──", "  ");
+            let mut num_siblings = count_siblings(to_display, num_slashes, ntp);
-            is = is.replace("├──", "│ ");
-            is = is.replace("├─┬", "│ ");
-
-            let printable_node_slashes = node_to_print.matches('/').count();
-
-            let mut num_siblings = to_display.iter().fold(0, |a, b| {
-                if b.0.starts_with(ntp) && b.0.matches('/').count() == printable_node_slashes + 1 {
-                    a + 1
-                } else {
-                    a
-                }
-            });
 
             let mut is_biggest = true;
             for &(ref k, _) in to_display.iter() {
-                if k.starts_with(ntp) && k.matches('/').count() == printable_node_slashes + 1 {
+                if k.starts_with(ntp) && k.matches('/').count() == num_slashes + 1 {
                     num_siblings -= 1;
 
                     let mut has_children = false;
                     if new_depth.is_none() || new_depth.unwrap() != 1 {
                         for &(ref k2, _) in to_display.iter() {
                             let kk: &str = k.as_ref();
-                            if k2.starts_with(kk)
+                            if k2.starts_with(kk) && k2.matches('/').count() == num_slashes + 2 {
-                                && k2.matches('/').count() == printable_node_slashes + 2
-                            {
                                 has_children = true;
                             }
                         }
                     }
 
                     display_node(
-                        k.to_string(),
+                        k,
+                        found,
                         to_display,
                         is_biggest,
                         short_paths,
                         new_depth,
-                        is.to_string() + get_tree_chars(num_siblings, has_children),
+                        new_indent_str.to_string() + get_tree_chars(num_siblings, has_children),
                     );
                     is_biggest = false;
                 }
@@ -97,6 +95,26 @@ fn display_node<S: Into<String>>(
     }
 }
 
+fn clean_indentation_string<S: Into<String>>(s: S) -> String {
+    let mut is = s.into();
+    is = is.replace("└─┬", "  ");
+    is = is.replace("└──", "  ");
+    is = is.replace("├──", "│ ");
+    is = is.replace("├─┬", "│ ");
+    is = is.replace("─┬", " ");
+    is
+}
+
+fn count_siblings(to_display: &Vec<(String, u64)>, num_slashes: usize, ntp: &str) -> u64 {
+    to_display.iter().fold(0, |a, b| {
+        if b.0.starts_with(ntp) && b.0.matches('/').count() == num_slashes + 1 {
+            a + 1
+        } else {
+            a
+        }
+    })
+}
+
 fn get_tree_chars(num_siblings: u64, has_children: bool) -> &'static str {
     if num_siblings == 0 {
         if has_children {

src/main.rs

@@ -5,7 +5,7 @@ extern crate walkdir;
 
 use self::display::draw_it;
 use clap::{App, AppSettings, Arg};
-use utils::{find_big_ones, get_dir_tree, sort};
+use utils::{find_big_ones, get_dir_tree, simplify_dir_names, sort};
 
 mod display;
 mod utils;
@@ -46,7 +46,7 @@ fn main() {
         .arg(Arg::with_name("inputs").multiple(true))
         .get_matches();
 
-    let filenames = {
+    let target_dirs = {
         match options.values_of("inputs") {
             None => vec!["."],
             Some(r) => r.collect(),
@@ -82,7 +82,8 @@ fn main() {
     let use_apparent_size = options.is_present("display_apparent_size");
     let use_full_path = options.is_present("display_full_paths");
 
-    let (permissions, nodes, top_level_names) = get_dir_tree(&filenames, use_apparent_size);
+    let simplified_dirs = simplify_dir_names(target_dirs);
+    let (permissions, nodes, top_level_names) = get_dir_tree(simplified_dirs, use_apparent_size);
     let sorted_data = sort(nodes);
     let biggest_ones = {
         if depth.is_none() {

src/tests.rs

@@ -28,6 +28,15 @@ pub fn test_main_long_paths() {
         .unwrap();
 }
 
+#[test]
+pub fn test_main_multi_arg() {
+    assert_cli::Assert::main_binary()
+        .with_args(&["src/test_dir/many/", "src/test_dir/", "src/test_dir"])
+        .stdout()
+        .is(main_output(true))
+        .unwrap();
+}
+
 #[cfg(target_os = "macos")]
 fn main_output(short_paths: bool) -> String {
     format!(
@@ -35,21 +44,21 @@ fn main_output(short_paths: bool) -> String {
 {}
 {}
 {}",
-        format_string("src/test_dir", true, short_paths, " 4.0K", ""),
+        format_string("src/test_dir", true, short_paths, " 4.0K", "─┬"),
-        format_string("src/test_dir/many", true, short_paths, " 4.0K", "└─┬",),
+        format_string("src/test_dir/many", true, short_paths, " 4.0K", " └─┬",),
         format_string(
             "src/test_dir/many/hello_file",
             true,
             short_paths,
             " 4.0K",
-            "  ├──",
+            "   ├──",
         ),
         format_string(
             "src/test_dir/many/a_file",
             false,
             short_paths,
             "   0B",
-            "  └──",
+            "   └──",
         ),
     )
 }
@@ -61,21 +70,21 @@ fn main_output(short_paths: bool) -> String {
 {}
 {}
 {}",
-        format_string("src/test_dir", true, short_paths, "  12K", ""),
+        format_string("src/test_dir", true, short_paths, "  12K", "─┬"),
-        format_string("src/test_dir/many", true, short_paths, " 8.0K", "└─┬",),
+        format_string("src/test_dir/many", true, short_paths, " 8.0K", " └─┬",),
         format_string(
             "src/test_dir/many/hello_file",
             true,
             short_paths,
             " 4.0K",
-            "  ├──",
+            "   ├──",
         ),
         format_string(
             "src/test_dir/many/a_file",
             false,
             short_paths,
             "   0B",
-            "  └──",
+            "   └──",
         ),
     )
 }
@@ -89,7 +98,7 @@ pub fn test_apparent_size() {
             true,
             true,
             "   6B",
-            "  ├──",
+            "   ├──",
         ),
     );
 
@@ -140,9 +149,9 @@ fn soft_sym_link_output(dir: &str, file_path: &str, link_name: &str) -> String {
         "{}
 {}
 {}",
-        format_string(dir, true, true, " 8.0K", ""),
+        format_string(dir, true, true, " 8.0K", "─┬"),
-        format_string(file_path, true, true, " 4.0K", "├──",),
+        format_string(file_path, true, true, " 4.0K", " ├──",),
-        format_string(link_name, false, true, " 4.0K", "└──",),
+        format_string(link_name, false, true, " 4.0K", " └──",),
     )
 }
 
@@ -152,9 +161,9 @@ fn soft_sym_link_output(dir: &str, file_path: &str, link_name: &str) -> String {
         "{}
 {}
 {}",
-        format_string(dir, true, true, " 8.0K", ""),
+        format_string(dir, true, true, " 8.0K", "─┬"),
-        format_string(file_path, true, true, " 4.0K", "├──",),
+        format_string(file_path, true, true, " 4.0K", " ├──",),
-        format_string(link_name, false, true, "   0B", "└──",),
+        format_string(link_name, false, true, "   0B", " └──",),
     )
 }
 
@@ -200,14 +209,14 @@ fn hard_link_output(dir_s: &str, file_path_s: &str, link_name_s: &str) -> (Strin
     let r = format!(
         "{}
 {}",
-        format_string(dir_s, true, true, " 4.0K", ""),
+        format_string(dir_s, true, true, " 4.0K", "─┬"),
-        format_string(file_path_s, true, true, " 4.0K", "└──")
+        format_string(file_path_s, true, true, " 4.0K", " └──")
     );
     let r2 = format!(
         "{}
 {}",
-        format_string(dir_s, true, true, " 4.0K", ""),
+        format_string(dir_s, true, true, " 4.0K", "─┬"),
-        format_string(link_name_s, true, true, " 4.0K", "└──")
+        format_string(link_name_s, true, true, " 4.0K", " └──")
     );
     (r, r2)
 }
@@ -217,14 +226,14 @@ fn hard_link_output(dir_s: &str, file_path_s: &str, link_name_s: &str) -> (Strin
     let r = format!(
         "{}
 {}",
-        format_string(dir_s, true, true, " 8.0K", ""),
+        format_string(dir_s, true, true, " 8.0K", "─┬"),
-        format_string(file_path_s, true, true, " 4.0K", "└──")
+        format_string(file_path_s, true, true, " 4.0K", " └──")
     );
     let r2 = format!(
         "{}
 {}",
-        format_string(dir_s, true, true, " 8.0K", ""),
+        format_string(dir_s, true, true, " 8.0K", "─┬"),
-        format_string(link_name_s, true, true, " 4.0K", "└──")
+        format_string(link_name_s, true, true, " 4.0K", " └──")
     );
     (r, r2)
 }
@@ -257,8 +266,8 @@ fn recursive_sym_link_output(dir: &str, link_name: &str) -> String {
     format!(
         "{}
 {}",
-        format_string(dir, true, true, " 4.0K", ""),
+        format_string(dir, true, true, " 4.0K", "─┬"),
-        format_string(link_name, true, true, " 4.0K", "└──",),
+        format_string(link_name, true, true, " 4.0K", " └──",),
     )
 }
 #[cfg(target_os = "linux")]
@@ -266,9 +275,8 @@ fn recursive_sym_link_output(dir: &str, link_name: &str) -> String {
     format!(
         "{}
 {}",
-        format_string(dir, true, true, " 4.0K", ""),
+        format_string(dir, true, true, " 4.0K", "─┬"),
-        format_string(link_name, true, true, "   0B", "└──",),
+        format_string(link_name, true, true, "   0B", " └──",),
     )
 }
 
-// TODO: add test for bad path

src/utils/mod.rs

@@ -7,25 +7,43 @@ use walkdir::WalkDir;
 mod platform;
 use self::platform::*;
 
+pub fn simplify_dir_names(filenames: Vec<&str>) -> HashSet<String> {
+    let mut top_level_names: HashSet<String> = HashSet::new();
+
+    for t in filenames {
+        let top_level_name = strip_end_slashes(t);
+        let mut can_add = true;
+        let mut to_remove: Vec<String> = Vec::new();
+
+        for tt in top_level_names.iter() {
+            let temp = tt.to_string();
+            if top_level_name.starts_with(&temp) {
+                can_add = false;
+            } else if tt.starts_with(&top_level_name) {
+                to_remove.push(temp);
+            }
+        }
+        for tr in to_remove {
+            top_level_names.remove(&tr);
+        }
+        if can_add {
+            top_level_names.insert(top_level_name);
+        }
+    }
+
+    top_level_names
+}
+
 pub fn get_dir_tree(
-    filenames: &Vec<&str>,
+    top_level_names: HashSet<String>,
     apparent_size: bool,
-) -> (bool, HashMap<String, u64>, Vec<String>) {
+) -> (bool, HashMap<String, u64>, HashSet<String>) {
     let mut permissions = 0;
     let mut inodes: HashSet<(u64, u64)> = HashSet::new();
     let mut data: HashMap<String, u64> = HashMap::new();
-    let mut top_level_names = Vec::new();
 
-    for b in filenames {
+    for b in top_level_names.iter() {
-        let top_level_name = strip_end_slashes(b);
+        examine_dir(&b, apparent_size, &mut inodes, &mut data, &mut permissions);
-        examine_dir(
-            &top_level_name,
-            apparent_size,
-            &mut inodes,
-            &mut data,
-            &mut permissions,
-        );
-        top_level_names.push(top_level_name);
     }
     (permissions == 0, data, top_level_names)
 }
@@ -100,3 +118,38 @@ pub fn find_big_ones<'a>(new_l: Vec<(String, u64)>, max_to_show: usize) -> Vec<(
         new_l
     }
 }
+
+mod tests {
+    #[allow(unused_imports)]
+    use super::*;
+
+    #[test]
+    fn test_simplify_dir() {
+        let mut correct = HashSet::new();
+        correct.insert("a".to_string());
+        assert!(simplify_dir_names(vec!["a"]) == correct);
+    }
+
+    #[test]
+    fn test_simplify_dir_rm_subdir() {
+        let mut correct = HashSet::new();
+        correct.insert("a/b".to_string());
+        assert!(simplify_dir_names(vec!["a/b", "a/b/c", "a/b/d/f"]) == correct);
+    }
+
+    #[test]
+    fn test_simplify_dir_duplicates() {
+        let mut correct = HashSet::new();
+        correct.insert("a/b".to_string());
+        correct.insert("c".to_string());
+        assert!(simplify_dir_names(vec!["a/b", "a/b//", "c", "c/"]) == correct);
+    }
+    #[test]
+    fn test_simplify_dir_rm_subdir_and_not_substrings() {
+        let mut correct = HashSet::new();
+        correct.insert("a/b".to_string());
+        correct.insert("c/a/b".to_string());
+        correct.insert("b".to_string());
+        assert!(simplify_dir_names(vec!["a/b", "c/a/b/", "b"]) == correct);
+    }
+}