From 068d47703aae803e5ae04056063a5a1540c0e2e8 Mon Sep 17 00:00:00 2001
From: Darryl Nixon <git@nixon.mozmail.com>
Date: Sun, 16 Jul 2023 12:20:57 -0700
Subject: [PATCH] ext2/3 is hard, use find as backup

---
 melamine/classes.py | 31 ++++++++++++++++++++++++++++++-
 melamine/shred.py   |  8 ++++++--
 2 files changed, 36 insertions(+), 3 deletions(-)

diff --git a/melamine/classes.py b/melamine/classes.py
index 7a1c63e..2452ca4 100644
--- a/melamine/classes.py
+++ b/melamine/classes.py
@@ -35,6 +35,7 @@ class ShredDir(AsyncObject):
         self.mount_points.add(self.mount_point)
         self.fs_handler = await mount_to_fs_handler(self.mount_point)
         self.byte_size = sum(item.byte_size for item in self.contents)
+        self.inode = path.stat().st_ino
 
     async def _get_contents(self) -> List:
         contents = []
@@ -63,13 +64,30 @@ class ShredDir(AsyncObject):
     def __hash__(self) -> int:
         return hash(self.absolute_path)
 
+    async def delete_hardlinks_by_inode(self) -> None:
+        logger.info(f"Finding and deleting hardlinks inside {self.absolute_path.name}")
+        for path in self.contents:
+            await path.delete_hardlinks_by_inode()
+
+        proc = await asyncio.create_subprocess_exec("find", str(self.mount_point), "-inum", self.get_inode(), "-delete")
+        stdout, _ = await proc.communicate()
+
+        if proc.returncode != 0:
+            err = f"Unable to delete hardlinks for {self.absolute_path.name}"
+            logger.error(err)
+            raise RuntimeError(err)
+
+        logger.info(f"Deleted hardlink for {self.absolute_path.name}")
+
 
 class ShredFile(AsyncObject):
     """Class for tracking each file to be shredded."""
 
     async def __init__(self, path: Path) -> None:
         self.absolute_path = path.resolve().absolute()
-        self.byte_size = path.stat().st_size
+        stat = path.stat()
+        self.byte_size = stat.st_size
+        self.inode = stat.st_ino
         self.mount_point = find_mount(self.absolute_path)
         self.fs_handler = await mount_to_fs_handler(self.mount_point)
         self.hardlinks = None
@@ -145,6 +163,17 @@ class ShredFile(AsyncObject):
     def __hash__(self) -> int:
         return hash(self.absolute_path)
 
+    async def delete_hardlinks_by_inode(self) -> None:
+        proc = await asyncio.create_subprocess_exec("find", str(self.mount_point), "-inum", self.get_inode(), "-delete")
+        stdout, _ = await proc.communicate()
+
+        if proc.returncode != 0:
+            err = f"Unable to delete hardlinks for {self.absolute_path.name}"
+            logger.error(err)
+            raise RuntimeError(err)
+
+        logger.info("Deleted hardlink for {self.absolute_path.name}")
+
 
 async def get_all_hardlinks(paths: Set[Union[ShredFile, ShredDir]]) -> None:
     for path in paths:
diff --git a/melamine/shred.py b/melamine/shred.py
index f658cd6..0c172ca 100644
--- a/melamine/shred.py
+++ b/melamine/shred.py
@@ -12,7 +12,6 @@ async def main(job: argparse.Namespace) -> bool:
     It is called by the CLI and builds a job queue based on the arguments passed.
     """
     new_paths = set()
-    logger.info(f"job type is {type(job)}")
 
     # Expand all directories and files, and collect mount point information
     for path in job.paths:
@@ -28,9 +27,14 @@ async def main(job: argparse.Namespace) -> bool:
         else:
             raise TypeError(f"Not a file or directory: {path}")
 
-    # Get hardlinks to subsequently unlink for all files
+    # Try to delete hardlinks based on the filesystem type
     job.paths = await get_all_hardlinks(new_paths)
 
+    # Just in case, use "find" to delete any remaining hardlinks
+    # from the mount point
+    for path in job.paths:
+        await path.delete_hardlinks_by_inode()
+
     # Shred all physical files including hardlinks
     for path in job.paths:
         if isinstance(path, ShredFile):