add symlink handling

Summary:
Symlinks are treated as files with "special" mode and link target in contents.

`hg diff --git` and `git` behaviours differ wrt to replacing files with
symlinks:
 * git renders that as two operations: removing a file and adding a symlink
 * hg renders that as file modification

 I'm choosing the `hg --git` behaviour for now (as it was easier to implement)

Reviewed By: krallin

Differential Revision: D18169625

fbshipit-source-id: 9ef746c7d242f7142a5ac89924a3e9bed08cb383

lib/xdiff/src/bin/diff.rs

@@ -31,6 +31,10 @@ struct Opt {
     #[structopt(short, long)]
     move_: bool,
 
+    /// Do not follow symlinks - compare them instead (POSIX-only)
+    #[structopt(short, long)]
+    symlink: bool,
+
     /// Number of lines of unified context (default: 3)
     #[structopt(short = "U", long, default_value = "3")]
     unified: usize,
@@ -40,17 +44,30 @@ fn main() -> Result<(), std::io::Error> {
     let opt = Opt::from_args();
 
     #[cfg(target_family = "unix")]
-    fn file_mode(path: &Path) -> Result<FileType, std::io::Error> {
-        if (path.metadata()?.permissions().mode() & EXEC_BIT) > 0 {
-            Ok(FileType::Executable)
+    fn file_mode_and_contents(
+        opt: &Opt,
+        path: &Path,
+    ) -> Result<(FileType, Vec<u8>), std::io::Error> {
+        use std::ffi::OsStr;
+        use std::os::unix::ffi::OsStrExt;
+
+        if opt.symlink && path.symlink_metadata()?.file_type().is_symlink() {
+            let dest = path.read_link()?;
+            let dest: &OsStr = dest.as_ref();
+            Ok((FileType::Symlink, dest.as_bytes().to_owned()))
+        } else if (path.metadata()?.permissions().mode() & EXEC_BIT) > 0 {
+            Ok((FileType::Executable, fs::read(path)?))
         } else {
-            Ok(FileType::Regular)
+            Ok((FileType::Regular, fs::read(path)?))
         }
     }
 
     #[cfg(target_family = "windows")]
-    fn file_mode(path: &Path) -> Result<FileType, std::io::Error> {
-        Ok(FileType::Regular)
+    fn file_mode_and_contents(
+        _opt: &Opt,
+        path: &Path,
+    ) -> Result<(FileType, Vec<u8>), std::io::Error> {
+        Ok((FileType::Regular, fs::read(path)?))
     }
 
     let copy_info = match (opt.copy, opt.move_) {
@@ -62,21 +79,15 @@ fn main() -> Result<(), std::io::Error> {
 
     let a_path_str = opt.file_a.to_string_lossy();
     let a = if opt.file_a.is_file() {
-        Some(DiffFile::new(
-            a_path_str.as_bytes(),
-            fs::read(&opt.file_a)?,
-            file_mode(&opt.file_a)?,
-        ))
+        let (mode, contents) = file_mode_and_contents(&opt, &opt.file_a)?;
+        Some(DiffFile::new(a_path_str.as_bytes(), contents, mode))
     } else {
         None
     };
     let b_path_str = opt.file_b.to_string_lossy();
     let b = if opt.file_b.is_file() {
-        Some(DiffFile::new(
-            b_path_str.as_bytes(),
-            fs::read(&opt.file_b)?,
-            file_mode(&opt.file_b)?,
-        ))
+        let (mode, contents) = file_mode_and_contents(&opt, &opt.file_b)?;
+        Some(DiffFile::new(b_path_str.as_bytes(), contents, mode))
     } else {
         None
     };

lib/xdiff/src/lib.rs

@@ -491,10 +491,10 @@ where
         reduce,
     };
     fn file_type_to_mode(file_type: FileType) -> &'static [u8] {
-        if let FileType::Executable = file_type {
-            b"100755"
-        } else {
-            b"100644"
+        match file_type {
+            FileType::Executable => b"100755",
+            FileType::Symlink => b"120000",
+            FileType::Regular => b"100644",
         }
     }
     if let (None, None) = (&old_file, &new_file) {

lib/xdiff/tests/test-basic.t

@@ -136,3 +136,48 @@ Test with binary file
   diff --git a/binary_b b/binary_b
   deleted file mode 100644
   Binary file binary_b has changed
+
+Test symlinks
+  $ ln -s a link_to_a
+  $ ln -s b link_to_b
+  $ ln -s a_exec link_to_a_exec
+
+  $ xdiff -s link_to_a link_to_b
+  diff --git a/link_to_a b/link_to_b
+  --- a/link_to_a
+  +++ b/link_to_b
+  @@ -1,1 +1,1 @@
+  -a
+  \ No newline at end of file
+  +b
+  \ No newline at end of file
+
+  $ xdiff -s link_to_a link_to_a
+
+  $ xdiff -s link_to_a link_to_a_exec
+  diff --git a/link_to_a b/link_to_a_exec
+  --- a/link_to_a
+  +++ b/link_to_a_exec
+  @@ -1,1 +1,1 @@
+  -a
+  \ No newline at end of file
+  +a_exec
+  \ No newline at end of file
+
+  $ xdiff -s link_to_a a
+  diff --git a/link_to_a b/a
+  old mode 120000
+  new mode 100644
+  --- a/link_to_a
+  +++ b/a
+  @@ -1,1 +1,6 @@
+  -a
+  \ No newline at end of file
+  +a
+  +b
+  +c
+  +d
+  +e
+  +f
+
+  $ xdiff link_to_a a