filter by file type

8 files changed, 178 insertions, 18 deletions

diff --git a/.gitignore b/.gitignore
index 0592392..6bc848e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,3 @@
 /target
 .DS_Store
+foobar
diff --git a/src/file/mod.rs b/src/file/mod.rs
index db1ba20..3f34397 100644
--- a/src/file/mod.rs
+++ b/src/file/mod.rs
@@ -14,6 +14,9 @@ use std::{
     path::{Path, PathBuf},
 };
 
+#[cfg(unix)]
+use crate::file::unix::permissions::{file_type::FileType, SymbolicNotation};
+
 /// Concerned with querying information about a file's underlying inode.
 pub mod inode;
 use inode::{INodeError, Inode};
@@ -87,8 +90,12 @@ impl File {
     ) -> Result<Self, io::Error> {
         let path = data.path();
 
-        let (symlink_target, metadata) = if *follow {
-            (fs::read_link(path).ok(), fs::metadata(path)?)
+        let (symlink_target, metadata) = if data.file_type().is_some_and(|ft| ft.is_symlink()) {
+            if *follow {
+                (fs::read_link(path).ok(), fs::metadata(path)?)
+            } else {
+                (fs::read_link(path).ok(), fs::symlink_metadata(path)?)
+            }
         } else {
             (None, fs::symlink_metadata(path)?)
         };
@@ -179,9 +186,49 @@ impl File {
             .map(|dt| format!("{dt}"))
     }
 
+    #[cfg(unix)]
+    pub fn is_fifo(&self) -> bool {
+        self.metadata()
+            .permissions()
+            .try_mode_symbolic_notation()
+            .map_or(false, |mode| mode.file_type() == &FileType::Fifo)
+    }
+
+    #[cfg(unix)]
+    pub fn is_socket(&self) -> bool {
+        self.metadata()
+            .permissions()
+            .try_mode_symbolic_notation()
+            .map_or(false, |mode| mode.file_type() == &FileType::Socket)
+    }
+
+    #[cfg(unix)]
+    pub fn is_char_device(&self) -> bool {
+        self.metadata()
+            .permissions()
+            .try_mode_symbolic_notation()
+            .map_or(false, |mode| mode.file_type() == &FileType::CharDevice)
+    }
+
+    #[cfg(unix)]
+    pub fn is_block_device(&self) -> bool {
+        self.metadata()
+            .permissions()
+            .try_mode_symbolic_notation()
+            .map_or(false, |mode| mode.file_type() == &FileType::BlockDevice)
+    }
+
     pub fn is_dir(&self) -> bool {
         self.file_type().is_some_and(|ft| ft.is_dir())
     }
+
+    pub fn is_file(&self) -> bool {
+        self.file_type().is_some_and(|ft| ft.is_file())
+    }
+
+    pub fn is_symlink(&self) -> bool {
+        self.file_type().is_some_and(|ft| ft.is_symlink())
+    }
 }
 
 impl Deref for File {
diff --git a/src/file/tree/filter.rs b/src/file/tree/filter.rs
new file mode 100644
index 0000000..f7e361c
--- /dev/null
+++ b/src/file/tree/filter.rs
@@ -0,0 +1,61 @@
+use crate::{
+    file::{tree::Tree, File},
+    user::{
+        args::{FileType, Layout},
+        Context,
+    },
+};
+use ahash::HashSet;
+use indextree::NodeId;
+
+/// Predicate used for filtering a [`File`] based on [`FileType`].
+pub type FileTypeFilter = dyn Fn(&File) -> bool;
+
+impl Tree {
+    /// Updates the [`Tree`]'s inner [`indextree::Arena`] to only contain files of certain
+    /// file-types.
+    pub fn filter_file_type(
+        &mut self,
+        Context {
+            layout, file_type, ..
+        }: &Context,
+    ) {
+        if file_type.is_empty() {
+            return;
+        }
+
+        let mut filters = Vec::<Box<FileTypeFilter>>::new();
+
+        for ft in HashSet::from_iter(file_type) {
+            match ft {
+                FileType::Dir if matches!(layout, Layout::Tree | Layout::InvertedTree) => {
+                    filters.push(Box::new(|f| f.is_dir()))
+                },
+                FileType::Dir => filters.push(Box::new(|f| f.is_dir())),
+                FileType::File => filters.push(Box::new(|f| f.is_file())),
+                FileType::Symlink => filters.push(Box::new(|f| f.is_symlink())),
+
+                #[cfg(unix)]
+                FileType::Pipe => filters.push(Box::new(|f| f.is_fifo())),
+                #[cfg(unix)]
+                FileType::Socket => filters.push(Box::new(|f| f.is_socket())),
+                #[cfg(unix)]
+                FileType::Char => filters.push(Box::new(|f| f.is_char_device())),
+                #[cfg(unix)]
+                FileType::Block => filters.push(Box::new(|f| f.is_block_device())),
+            }
+        }
+
+        let no_match = |node_id: &NodeId| !filters.iter().any(|f| f(self.arena[*node_id].get()));
+
+        let to_remove = self
+            .root_id
+            .descendants(&self.arena)
+            .filter(no_match)
+            .collect::<Vec<_>>();
+
+        to_remove
+            .into_iter()
+            .for_each(|n| n.detach(&mut self.arena));
+    }
+}
diff --git a/src/file/tree/mod.rs b/src/file/tree/mod.rs
index c805a27..bd25e21 100644
--- a/src/file/tree/mod.rs
+++ b/src/file/tree/mod.rs
@@ -11,7 +11,10 @@ use ahash::{HashMap, HashSet};
 use indextree::{Arena, NodeId};
 use std::{ops::Deref, path::PathBuf};
 
-/// Parallel disk reading
+/// Concerned with filtering via file-type, globbing, and regular expressions.
+mod filter;
+
+/// Parallel disk reading.
 mod traverse;
 
 /// Representation of the file-tree that is traversed starting from the root directory whose index
diff --git a/src/file/tree/traverse.rs b/src/file/tree/traverse.rs
index 4fd82ec..3ebbd54 100644
--- a/src/file/tree/traverse.rs
+++ b/src/file/tree/traverse.rs
@@ -121,7 +121,7 @@ mod walker {
         let walker = builder
             .follow_links(ctx.follow)
             .git_ignore(ctx.gitignore)
-            .git_global(ctx.gitignore)
+            .git_global(ctx.global_gitignore)
             .threads(ctx.threads)
             .hidden(ctx.no_hidden)
             .same_file_system(ctx.same_fs);
diff --git a/src/main.rs b/src/main.rs
index e3aa91f..402443c 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -56,6 +56,10 @@ fn run() -> Result<()> {
         file_tree.prune();
     }
 
+    if !ctx.file_type.is_empty() {
+        file_tree.filter_file_type(&ctx)
+    }
+
     let output = render::output(&file_tree, &ctx)?;
 
     let mut stdout = stdout().lock();
diff --git a/src/user/args.rs b/src/user/args.rs
index 06990c0..7469fc3 100644
--- a/src/user/args.rs
+++ b/src/user/args.rs
@@ -131,10 +131,12 @@ pub enum Sort {
 /// Whether to sort directory entries relative either to their siblings or all directory entries
 #[derive(Copy, Clone, Debug, ValueEnum, PartialEq, Eq, PartialOrd, Ord, Default)]
 pub enum SortType {
-    /// Sort directory entries relative to their siblings
+    /// Sort directory entries relative to their siblings. Only applies when using 'tree' or
+    /// 'inverted-tree' layout.
     #[default]
     Tree,
-    /// Sort directory entries relative to all directory entries
+    /// Sort directory entries relative to all directory entries. Only applies when using 'flat'
+    /// layout.
     Flat,
 }
 
@@ -151,3 +153,33 @@ pub enum DirOrder {
     /// Sort directories below files
     Last,
 }
+
+/// File-types found in both Unix and Windows.
+#[derive(Copy, Clone, Debug, ValueEnum, PartialEq, Eq, PartialOrd, Ord, Default, Hash)]
+pub enum FileType {
+    /// A regular file
+    #[default]
+    File,
+
+    /// A directory
+    Dir,
+
+    /// A symlink
+    Symlink,
+
+    /// A FIFO
+    #[cfg(unix)]
+    Pipe,
+
+    /// A socket
+    #[cfg(unix)]
+    Socket,
+
+    /// A character device
+    #[cfg(unix)]
+    Char,
+
+    /// A block device
+    #[cfg(unix)]
+    Block,
+}
diff --git a/src/user/mod.rs b/src/user/mod.rs
index 86938ad..1a3dd30 100644
--- a/src/user/mod.rs
+++ b/src/user/mod.rs
@@ -30,18 +30,30 @@ pub struct Context {
     #[arg(long)]
     pub no_git: bool,
 
-    /// Ignore files in .gitignore
-    #[arg(short = 'i', long)]
-    pub gitignore: bool,
-
     /// Report byte size in either binary or SI units
     #[arg(short, long, value_enum, default_value_t)]
     pub byte_units: args::BytePresentation,
 
+    /// Sort directories before or after all other file types
+    #[arg(short, long, value_enum, default_value_t)]
+    pub dir_order: args::DirOrder,
+
+    /// Filter for specified file types
+    #[arg(short = 'F', long, value_enum)]
+    pub file_type: Vec<args::FileType>,
+
     /// Follow symlinks
     #[arg(short = 'f', long)]
     pub follow: bool,
 
+    /// Ignore files that match rules in all .gitignore files encountered during traversal
+    #[arg(short = 'i', long)]
+    pub gitignore: bool,
+
+    /// Ignore files that match rules in the global .gitignore file
+    #[arg(long)]
+    pub global_gitignore: bool,
+
     /// Show extended metadata and attributes
     #[cfg(unix)]
     #[arg(short, long)]
@@ -85,8 +97,12 @@ pub struct Context {
     #[arg(short, long, value_enum, default_value_t)]
     pub metric: args::Metric,
 
-    /// Omit empty directories from the output
+    /// Regular expression (or glob if '--glob' or '--iglob' is used) used to match files
     #[arg(short, long)]
+    pub pattern: Option<String>,
+
+    /// Omit empty directories from the output
+    #[arg(short = 'P', long)]
     pub prune: bool,
 
     /// Field whereby to sort entries
@@ -97,9 +113,9 @@ pub struct Context {
     #[arg(long, value_enum, default_value_t)]
     pub sort_type: args::SortType,
 
-    /// Sort directories before or after all other file types
-    #[arg(short, long, value_enum, default_value_t)]
-    pub dir_order: args::DirOrder,
+    /// Don't compute disk-usage and omit file size from output
+    #[arg(short = 'S', long)]
+    pub suppress_size: bool,
 
     /// Which kind of layout to use when rendering the output
     #[arg(short = 'y', long, value_enum, default_value_t)]
@@ -113,10 +129,6 @@ pub struct Context {
     #[arg(short = 'x', long = "one-file-system")]
     pub same_fs: bool,
 
-    /// Don't compute disk-usage and omit file size from output
-    #[arg(long)]
-    pub suppress_size: bool,
-
     /// Prints logs at the end of the output
     #[arg(short = 'v', long = "verbose")]
     pub verbose: bool,