From 0d61346b3c45b12bfb63c39f28eec95c08971d2b Mon Sep 17 00:00:00 2001
From: Darryl Nixon <git@nixon.mozmail.com>
Date: Sun, 16 Jul 2023 17:34:02 -0700
Subject: [PATCH] mvp

---
 melamine/classes.py | 12 ++++++++++++
 melamine/fileops.py |  9 ++++++---
 melamine/shred.py   |  4 +++-
 3 files changed, 21 insertions(+), 4 deletions(-)

diff --git a/melamine/classes.py b/melamine/classes.py
index b272334..76b8496 100644
--- a/melamine/classes.py
+++ b/melamine/classes.py
@@ -53,6 +53,15 @@ class ShredDir(AsyncObject):
             elif await subpath.is_file():
                 tasks.append(ShredFile(subpath))
         return set(await asyncio.gather(*tasks))
+    
+    async def get_inodes(self, recursive: bool) -> Generator:
+        for path in self.contents:
+            if isinstance(path, ShredDir):
+                if recursive:
+                    async for sub_inode in path.get_inodes(recursive):
+                        yield sub_inode
+            elif isinstance(path, ShredFile):
+                yield path.inode
 
     def enumerate_mount_points(self) -> Generator:
         for item in self.contents:
@@ -196,6 +205,9 @@ class ShredFile(AsyncObject):
     def __hash__(self) -> int:
         return hash(self.absolute_path)
 
+    async def get_inodes(self, recursive: bool) -> Generator:
+        return self.inode
+
     async def delete_hardlinks_by_inode(self) -> None:
         proc = await asyncio.create_subprocess_exec("find", str(self.mount_point), "-inum", str(self.inode), "-delete")
         stdout, _ = await proc.communicate()
diff --git a/melamine/fileops.py b/melamine/fileops.py
index 72212a2..08c03fa 100644
--- a/melamine/fileops.py
+++ b/melamine/fileops.py
@@ -29,15 +29,18 @@ async def get_all_mounts() -> Generator:
 async def mount_bound_rglob(path: AsyncPath, mount: AsyncPath, pattern: str, ignoredirs: List[AsyncPath]) -> Generator:
     """Recursively glob a path, but stop at mount points."""
     path = await path.absolute()
+    if str(path).startswith("/home/parallels"):
+        logger.warning(f"Processing {str(path)}")
+    path_mount = await find_mount(path)
+    if path_mount != mount:
+        logger.info(f"Skipping due to mount diff: {path} (wanted {mount})")
+        return
     # Skip any ignored directories/files with a sanity check
     for ignore in ignoredirs:
         if str(path).startswith(str(ignore)) and await path.is_relative_to(ignore):
             logger.info(f"Skipping ignored subdir: {path}")
             return
     if await path.is_dir():
-        if await find_mount(path) != mount:
-            logger.info(f"Skipping differently mounted subdir: {path} (wanted {mount})")
-            return
         async for subpath in path.glob(pattern):
             async for subitem in mount_bound_rglob(subpath, mount, pattern, ignoredirs):
                 yield subitem
diff --git a/melamine/shred.py b/melamine/shred.py
index 1a0980e..b918489 100644
--- a/melamine/shred.py
+++ b/melamine/shred.py
@@ -58,7 +58,8 @@ async def main(job: argparse.Namespace) -> bool:
     logger.info("Deleting remaining hardlinks using find")
     inodes_in_mount_points = defaultdict(set)
     for path in job.paths:
-        inodes_in_mount_points[path.mount_point].add(path.inode)
+        async for inode in path.get_inodes(job.recursive):
+            inodes_in_mount_points[path.mount_point].add(inode)
 
     # We'll also limit concurrency to something reasonable since stat
     # on an entire filesystem might be a bit burdensome
@@ -81,6 +82,7 @@ async def main(job: argparse.Namespace) -> bool:
         # checking for . and .. should not be neccessary w/ rglob
         # scandir/glob/rglob doesn't play nice with FileNotFound errors,
         # so let's avoid them in dynamic fs areas
+        logger.warning(f"Scanning {mount_point} for inodes: {inodes}")
         if str(mount_point) == "/":
             logger.info("Root filesystem mount processing")
             async for item in mount_point.glob("*"):