cp: fix the result of inodes are not the same when preserve links is flagged

src/uu/cp/src/cp.rs

@@ -13,7 +13,7 @@
 
 use quick_error::quick_error;
 use std::borrow::Cow;
-use std::collections::HashSet;
+use std::collections::{HashMap, HashSet};
 use std::env;
 #[cfg(not(windows))]
 use std::ffi::CString;
@@ -1134,11 +1134,13 @@
  verify_target_type(target, &target_type)?;
 
  let preserve_hard_links = options.preserve_hard_links();
+ let cli_dereference = options.cli_dereference;
 
  let mut hard_links: Vec<(String, u64)> = vec![];
 
  let mut non_fatal_errors = false;
- let mut seen_sources = HashSet::with_capacity(sources.len());
+ let mut seen_sources = HashMap::with_capacity(sources.len());
+ let mut should_hard_linked_files = HashMap::with_capacity(sources.len());
  let mut symlinked_files = HashSet::new();
 
  let progress_bar = if options.progress_bar {
@@ -1157,12 +1159,51 @@
  };
 
  for source in sources.iter() {
- if seen_sources.contains(source) {
+ let dest = construct_dest_path(source, target, &target_type, options)?;
+ if seen_sources.contains_key(source) && !should_hard_linked_files.contains_key(source) {
  // FIXME: compare sources by the actual file they point to, not their path. (e.g. dir/file == dir/../dir/file in most cases)
  show_warning!("source {} specified more than once", source.quote());
+ } else if let Some(new_source) = should_hard_linked_files.get(source) {
+ std::fs::hard_link(new_source, &dest)?;
+ should_hard_linked_files.remove(source);
+ } else if preserve_hard_links && source.is_symlink() && cli_dereference {
+ // issue 5031 case
+ // touch a && ln -s a b && mkdir c
+ // cp --preserve=links -R -H a b c
+
+ let mut original_source = source.read_link()?;
+ while original_source.is_symlink() {
+ original_source = original_source.read_link()?;
+ }
+ // unwrap it because the read_link method is successed
+ // so file_name won't be failed
+ original_source = original_source.file_name().unwrap().into();
+
+ match seen_sources.get(&original_source) {
+ Some(new_source) => {
+ std::fs::hard_link(new_source, &dest)?;
+ }
+ None => {
+ // TODO: GNU cp can deal with this case
+ // for example:
+ // touch a && ln -s a b && mkdir c
+ // cp --preserve=links -R -H b a c
+ if let Err(error) = copy_source(
+ &progress_bar,
+ source,
+ target,
+ &target_type,
+ options,
+ &mut symlinked_files,
+ ) {
+ show_error_if_needed(&error);
+ non_fatal_errors = true;
+ }
+ should_hard_linked_files.insert(original_source, dest.clone());
+ }
+ }
  } else {
  let found_hard_link = if preserve_hard_links && !source.is_dir() {
- let dest = construct_dest_path(source, target, &target_type, options)?;
  preserve_hardlinks(&mut hard_links, source, &dest)?
  } else {
  false
@@ -1180,8 +1221,8 @@
  non_fatal_errors = true;
  }
  }
- seen_sources.insert(source);
  }
+ seen_sources.insert(source, dest);
  }
  if non_fatal_errors {
  Err(Error::NotAllFilesCopied)

tests/by-util/test_cp.rs

@@ -1349,6 +1349,72 @@ fn test_cp_preserve_xattr_fails_on_android() {
  .fails();
 }
 
+#[test]
+fn test_cp_issue_5031_case_1() {
+ let (at, mut ucmd) = at_and_ucmd!();
+
+ at.touch("a");
+ at.symlink_file("a", "b");
+ at.mkdir("c");
+
+ assert!(at.file_exists("a"));
+ assert!(at.symlink_exists("b"));
+ assert!(at.dir_exists("c"));
+
+ ucmd.arg("--preserve=links")
+ .arg("-R")
+ .arg("-H")
+ .arg("a")
+ .arg("b")
+ .arg("c")
+ .succeeds();
+
+ #[cfg(not(any(target_os = "freebsd", target_os = "macos")))]
+ {
+ assert!(at.dir_exists("c"));
+ assert!(at.plus("c").join("a").exists());
+ assert!(at.plus("c").join("b").exists());
+
+ let metadata_a = std_fs::metadata(at.subdir.join("c").join("a")).unwrap();
+ let metadata_b = std_fs::metadata(at.subdir.join("c").join("b")).unwrap();
+
+ assert_eq!(metadata_a.ino(), metadata_b.ino());
+ }
+}
+
+#[test]
+fn test_cp_issue_5031_case_2() {
+ let (at, mut ucmd) = at_and_ucmd!();
+
+ at.touch("a");
+ at.symlink_file("a", "b");
+ at.mkdir("c");
+
+ assert!(at.file_exists("a"));
+ assert!(at.symlink_exists("b"));
+ assert!(at.dir_exists("c"));
+
+ ucmd.arg("--preserve=links")
+ .arg("-R")
+ .arg("-H")
+ .arg("b")
+ .arg("a")
+ .arg("c")
+ .succeeds();
+
+ #[cfg(not(any(target_os = "freebsd", target_os = "macos")))]
+ {
+ assert!(at.dir_exists("c"));
+ assert!(at.plus("c").join("a").exists());
+ assert!(at.plus("c").join("b").exists());
+
+ let metadata_a = std_fs::metadata(at.subdir.join("c").join("a")).unwrap();
+ let metadata_b = std_fs::metadata(at.subdir.join("c").join("b")).unwrap();
+
+ assert_eq!(metadata_a.ino(), metadata_b.ino());
+ }
+}
+
 #[test]
 // For now, disable the test on Windows. Symlinks aren't well support on Windows.
 // It works on Unix for now and it works locally when run from a powershell