Implement support for following symlinks

Signed-off-by: Ryan Gonzalez <[email protected]>
uutils · Jan 22, 2022 · 80c0577 · 80c0577
1 parent 50d2bde
commit 80c0577
Show file tree

Hide file tree

Showing 6 changed files with 132 additions and 84 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -24,6 +24,9 @@ filetime = "0.2"
 predicates = "2"
 serial_test = "0.5"
 
+[patch.crates-io]
+walkdir = { git = "https://github.com/refi64/walkdir", branch = "symlinks" }
+
 [[bin]]
 name = "find"
 path = "src/find/main.rs"

diff --git a/src/find/matchers/printf.rs b/src/find/matchers/printf.rs
@@ -349,11 +349,13 @@ fn get_starting_point(file_info: &walkdir::DirEntry) -> &Path {
         .unwrap()
 }
 
-fn format_non_link_file_type(file_type: fs::FileType) -> char {
+fn format_file_type(file_type: fs::FileType) -> char {
     if file_type.is_file() {
         'f'
     } else if file_type.is_dir() {
         'd'
+    } else if file_type.is_symlink() {
+        'l'
     } else {
         #[cfg(unix)]
         if file_type.is_block_device() {
@@ -377,19 +379,7 @@ fn format_directive<'entry>(
     directive: &FormatDirective,
     meta_cell: &OnceCell<fs::Metadata>,
 ) -> Result<Cow<'entry, str>, Box<dyn Error>> {
-    let meta = || {
-        meta_cell.get_or_try_init(|| {
-            if file_info.path_is_symlink() && !file_info.file_type().is_symlink() {
-                // The file_info already followed the symlink, meaning that the
-                // metadata will be for the target file, which isn't the
-                // behavior we want, so manually re-compute the metadata for the
-                // symlink itself instead.
-                file_info.path().symlink_metadata()
-            } else {
-                file_info.metadata().map_err(|e| e.into())
-            }
-        })
-    };
+    let meta = || meta_cell.get_or_try_init(|| file_info.metadata());
 
     // NOTE ON QUOTING:
     // GNU find's man page claims that several directives that print names (like
@@ -542,7 +532,7 @@ fn format_directive<'entry>(
         FormatDirective::StartingPoint => get_starting_point(file_info).to_string_lossy(),
 
         FormatDirective::SymlinkTarget => {
-            if file_info.path_is_symlink() {
+            if meta()?.is_symlink() {
                 fs::read_link(file_info.path())?
                     .to_string_lossy()
                     .into_owned()
@@ -552,10 +542,10 @@ fn format_directive<'entry>(
             }
         }
 
-        FormatDirective::Type { follow_links } => if file_info.path_is_symlink() {
-            if *follow_links {
+        FormatDirective::Type { follow_links } => {
+            if file_info.file_type().is_symlink() && *follow_links {
                 match file_info.path().metadata() {
-                    Ok(meta) => format_non_link_file_type(meta.file_type()),
+                    Ok(meta) => format_file_type(meta.file_type()),
                     Err(e) if e.kind() == std::io::ErrorKind::NotFound => 'N',
                     // The ErrorKinds corresponding to ELOOP and ENOTDIR are
                     // nightly-only:
@@ -568,13 +558,11 @@ fn format_directive<'entry>(
                     Err(_) => '?',
                 }
             } else {
-                'l'
+                format_file_type(file_info.file_type())
             }
-        } else {
-            format_non_link_file_type(file_info.file_type())
+            .to_string()
+            .into()
         }
-        .to_string()
-        .into(),
 
         #[cfg(not(unix))]
         FormatDirective::User { .. } => "0".into(),

diff --git a/src/find/mod.rs b/src/find/mod.rs
@@ -14,6 +14,7 @@ use std::time::SystemTime;
 use walkdir::WalkDir;
 
 pub struct Config {
+    follow_links: bool,
     depth_first: bool,
     min_depth: usize,
     max_depth: usize,
@@ -25,6 +26,7 @@ pub struct Config {
 impl Default for Config {
     fn default() -> Config {
         Config {
+            follow_links: false,
             depth_first: false,
             min_depth: 0,
             max_depth: usize::max_value(),
@@ -86,6 +88,16 @@ fn parse_args(args: &[&str]) -> Result<ParsedInfo, Box<dyn Error>> {
     let mut i = 0;
     let mut config = Config::default();
 
+    while i < args.len() {
+        if args[i] == "-L" {
+            config.follow_links = true;
+        } else {
+            break;
+        }
+
+        i += 1;
+    }
+
     while i < args.len()
         && (args[i] == "-" || !args[i].starts_with('-'))
         && args[i] != "!"
@@ -115,7 +127,10 @@ fn process_dir<'a>(
     let mut walkdir = WalkDir::new(dir)
         .contents_first(config.depth_first)
         .max_depth(config.max_depth)
-        .min_depth(config.min_depth);
+        .min_depth(config.min_depth)
+        .follow_links(config.follow_links)
+        .follow_root_link(config.follow_links)
+        .yield_link_on_error(true);
     if config.sorted_output {
         walkdir = walkdir.sort_by(|a, b| a.file_name().cmp(b.file_name()));
     }
@@ -171,6 +186,7 @@ fn print_help() {
 If no path is supplied then the current working directory is used by default.
 
 Early alpha implementation. Currently the only expressions supported are
+ -L
  -print
  -printf
  -name case-sensitive_filename_pattern

diff --git a/tests/common/test_helpers.rs b/tests/common/test_helpers.rs
@@ -93,3 +93,65 @@ pub fn get_dir_entry_for(directory: &str, filename: &str) -> DirEntry {
     }
     panic!("Couldn't find {} in {}", directory, filename);
 }
+
+pub fn create_testing_symlinks() {
+    use std::io::ErrorKind;
+
+    #[cfg(unix)]
+    use std::os::unix::fs::symlink;
+
+    #[cfg(windows)]
+    use std::os::windows::fs::{symlink_dir, symlink_file};
+
+    #[cfg(unix)]
+    {
+        if let Err(e) = symlink("abbbc", "test_data/links/link-f") {
+            if e.kind() != ErrorKind::AlreadyExists {
+                panic!("Failed to create sym link: {:?}", e);
+            }
+        }
+        if let Err(e) = symlink("subdir", "test_data/links/link-d") {
+            if e.kind() != ErrorKind::AlreadyExists {
+                panic!("Failed to create sym link: {:?}", e);
+            }
+        }
+        if let Err(e) = symlink("missing", "test_data/links/link-missing") {
+            if e.kind() != ErrorKind::AlreadyExists {
+                panic!("Failed to create sym link: {:?}", e);
+            }
+        }
+        if let Err(e) = symlink("abbbc/x", "test_data/links/link-notdir") {
+            if e.kind() != ErrorKind::AlreadyExists {
+                panic!("Failed to create sym link: {:?}", e);
+            }
+        }
+        if let Err(e) = symlink("link-loop", "test_data/links/link-loop") {
+            if e.kind() != ErrorKind::AlreadyExists {
+                panic!("Failed to create sym link: {:?}", e);
+            }
+        }
+    }
+    #[cfg(windows)]
+    {
+        if let Err(e) = symlink_file("abbbc", "test_data/links/link-f") {
+            if e.kind() != ErrorKind::AlreadyExists {
+                panic!("Failed to create sym link: {:?}", e);
+            }
+        }
+        if let Err(e) = symlink_dir("subdir", "test_data/links/link-d") {
+            if e.kind() != ErrorKind::AlreadyExists {
+                panic!("Failed to create sym link: {:?}", e);
+            }
+        }
+        if let Err(e) = symlink_file("missing", "test_data/links/link-missing") {
+            if e.kind() != ErrorKind::AlreadyExists {
+                panic!("Failed to create sym link: {:?}", e);
+            }
+        }
+        if let Err(e) = symlink_file("abbbc/x", "test_data/links/link-notdir") {
+            if e.kind() != ErrorKind::AlreadyExists {
+                panic!("Failed to create sym link: {:?}", e);
+            }
+        }
+    }
+}
diff --git a/tests/find_cmd_tests.rs b/tests/find_cmd_tests.rs
@@ -11,16 +11,10 @@
 use assert_cmd::Command;
 use predicates::prelude::*;
 use serial_test::serial;
+use std::env;
 use std::fs::File;
-use std::{env, io::ErrorKind};
 use tempfile::Builder;
 
-#[cfg(unix)]
-use std::os::unix::fs::symlink;
-
-#[cfg(windows)]
-use std::os::windows::fs::{symlink_dir, symlink_file};
-
 use common::test_helpers::*;
 
 mod common;
@@ -126,60 +120,46 @@ fn delete_on_dot_dir() {
     assert!(temp_dir.path().exists(), "temp dir should still exist");
 }
 
+#[serial(working_dir)]
+#[test]
+fn follow_links() {
+    create_testing_symlinks();
+
+    // By default, we should not follow the root if it's a link.
+    Command::cargo_bin("find")
+        .expect("found binary")
+        .args(&[
+            &fix_up_slashes("./test_data/links/link-d"),
+            "-printf",
+            "%p %y\n",
+        ])
+        .assert()
+        .success()
+        .stderr(predicate::str::is_empty())
+        .stdout(predicate::str::diff(fix_up_slashes(
+            "./test_data/links/link-d l\n",
+        )));
+
+    Command::cargo_bin("find")
+        .expect("found binary")
+        .args(&[
+            "-L",
+            &fix_up_slashes("./test_data/links/link-d"),
+            "-printf",
+            "%p %y\n",
+        ])
+        .assert()
+        .success()
+        .stderr(predicate::str::is_empty())
+        .stdout(predicate::str::diff(fix_up_slashes(
+            "./test_data/links/link-d d\n./test_data/links/link-d/test f\n",
+        )));
+}
+
 #[serial(working_dir)]
 #[test]
 fn find_printf() {
-    #[cfg(unix)]
-    {
-        if let Err(e) = symlink("abbbc", "test_data/links/link-f") {
-            if e.kind() != ErrorKind::AlreadyExists {
-                panic!("Failed to create sym link: {:?}", e);
-            }
-        }
-        if let Err(e) = symlink("subdir", "test_data/links/link-d") {
-            if e.kind() != ErrorKind::AlreadyExists {
-                panic!("Failed to create sym link: {:?}", e);
-            }
-        }
-        if let Err(e) = symlink("missing", "test_data/links/link-missing") {
-            if e.kind() != ErrorKind::AlreadyExists {
-                panic!("Failed to create sym link: {:?}", e);
-            }
-        }
-        if let Err(e) = symlink("abbbc/x", "test_data/links/link-notdir") {
-            if e.kind() != ErrorKind::AlreadyExists {
-                panic!("Failed to create sym link: {:?}", e);
-            }
-        }
-        if let Err(e) = symlink("link-loop", "test_data/links/link-loop") {
-            if e.kind() != ErrorKind::AlreadyExists {
-                panic!("Failed to create sym link: {:?}", e);
-            }
-        }
-    }
-    #[cfg(windows)]
-    {
-        if let Err(e) = symlink_file("abbbc", "test_data/links/link-f") {
-            if e.kind() != ErrorKind::AlreadyExists {
-                panic!("Failed to create sym link: {:?}", e);
-            }
-        }
-        if let Err(e) = symlink_dir("subdir", "test_data/links/link-d") {
-            if e.kind() != ErrorKind::AlreadyExists {
-                panic!("Failed to create sym link: {:?}", e);
-            }
-        }
-        if let Err(e) = symlink_file("missing", "test_data/links/link-missing") {
-            if e.kind() != ErrorKind::AlreadyExists {
-                panic!("Failed to create sym link: {:?}", e);
-            }
-        }
-        if let Err(e) = symlink_file("abbbc/x", "test_data/links/link-notdir") {
-            if e.kind() != ErrorKind::AlreadyExists {
-                panic!("Failed to create sym link: {:?}", e);
-            }
-        }
-    }
+    create_testing_symlinks();
 
     Command::cargo_bin("find")
         .expect("found binary")