From 3282d5a6550ece1544ce63d111dd36da3077bdd3 Mon Sep 17 00:00:00 2001
From: Darryl Nixon <git@nixon.mozmail.com>
Date: Sun, 16 Jul 2023 14:16:32 -0700
Subject: [PATCH] ignoredir

---
 melamine/classes.py | 22 +++++++++++++++++++---
 melamine/shred.py   | 18 +++++++++++++-----
 2 files changed, 32 insertions(+), 8 deletions(-)

diff --git a/melamine/classes.py b/melamine/classes.py
index 90fe6f6..b272334 100644
--- a/melamine/classes.py
+++ b/melamine/classes.py
@@ -60,9 +60,13 @@ class ShredDir(AsyncObject):
                 yield from item.enumerate_mount_points()
         yield self.mount_point
 
-    async def shred(self, hash: bool = False, dryrun: bool = False) -> bool:
+    async def shred(self, hash: bool = False, dryrun: bool = False, ignoredirs: Set[AsyncPath] = None) -> bool:
+        if ignoredirs is None:
+            ignoredirs = set()
         tasks = []
         for item in self.contents:
+            if any(str(item).startswith(str(path)) for path in ignoredirs):
+                continue
             tasks.append(item.shred(hash, dryrun))
         try:
             if all(await asyncio.gather(*tasks)):
@@ -105,9 +109,15 @@ class ShredFile(AsyncObject):
         self.fs_handler = await mount_to_fs_handler(self.mount_point)
         self.hardlinks = None
 
-    async def shred(self, hash: bool = False, dryrun: bool = False) -> Union[bool, bytes]:
+    async def shred(
+        self, hash: bool = False, dryrun: bool = False, ignoredirs: Set[AsyncPath] = None
+    ) -> Union[bool, bytes]:
         """Shred the file with a single file descriptor."""
+        if ignoredirs is None:
+            ignoredirs = set()
         try:
+            if any(str(path).startswith(str(path)) for path in ignoredirs):
+                return False
             logger.info(f"Shredding: {self.absolute_path}")
 
             async with aiofiles.open(self.absolute_path, "rb+") as file:
@@ -161,7 +171,13 @@ class ShredFile(AsyncObject):
             if self.hardlinks:
                 log_buf = f"[5/4] Unlinking {len(self.hardlinks)} hardlinks"
                 if not dryrun:
-                    tasks = [link.unlink() for link in self.hardlinks if link.name != "." and link.name != ".."]
+                    tasks = []
+                    for link in self.hardlinks:
+                        if link.name == "." and link.name == "..":
+                            continue
+                        if any(str(await link.absolute()).startswith(str(path)) for path in ignoredirs):
+                            return False
+                        tasks.append(link.unlink())
                     done, _ = await asyncio.wait(tasks)
                     for task in done:
                         e = task.exception()
diff --git a/melamine/shred.py b/melamine/shred.py
index 316d04c..0e06b99 100644
--- a/melamine/shred.py
+++ b/melamine/shred.py
@@ -2,6 +2,8 @@ import argparse
 import asyncio
 from collections import defaultdict
 
+from aiopath import AsyncPath
+
 from .classes import get_all_hardlinks
 from .classes import ShredDir
 from .classes import ShredFile
@@ -30,13 +32,17 @@ async def main(job: argparse.Namespace) -> bool:
     # Try to delete hardlinks based on the filesystem type
     job.paths = await get_all_hardlinks(new_paths)
 
+    tasks = [path.absolute() for path in job.ignoredir]
+    tasks.append(AsyncPath("/proc").absolute())
+    job.ignoredir = set(await asyncio.gather(*tasks))
+
     # Shred all physical files including hardlinks
     for path in job.paths:
         tasks = []
         if isinstance(path, ShredFile):
-            tasks.append(path.shred(hash=job.exhaustive, dryrun=job.dryrun))
+            tasks.append(path.shred(hash=job.exhaustive, dryrun=job.dryrun, ignoredirs=job.ignoredir))
         elif isinstance(path, ShredDir):
-            tasks.append(path.shred(hash=job.exhaustive, dryrun=job.dryrun))
+            tasks.append(path.shred(hash=job.exhaustive, dryrun=job.dryrun, ignoredirs=job.ignoredir))
         done, _ = await asyncio.wait(tasks)
         for task in done:
             e = task.exception()
@@ -54,9 +60,9 @@ async def main(job: argparse.Namespace) -> bool:
     # on an entire filesystem might be a bit burdensome
     semaphore = asyncio.Semaphore(1024)
 
-    async def check_inode_and_unlink(item):
+    async def check_inode_and_unlink(item, inodes):
         async with semaphore:
-            if await item.stat().st_ino in inodes_in_mount_points[item.mount_point]:
+            if await item.stat().st_ino in inodes:
                 log_buf = f"Deleting hardlink: {item.path}"
                 if not job.dryrun:
                     log_buf = "DRY RUN " + log_buf
@@ -67,7 +73,9 @@ async def main(job: argparse.Namespace) -> bool:
         # checking for . and .. should not be neccessary w/ rglob
         tasks = []
         async for item in mount_point.rglob("*"):
-            tasks.append(check_inode_and_unlink(item))
+            if any(str(item).startswith(path) for path in job.ignoredir):
+                continue
+            tasks.append(check_inode_and_unlink(item, inodes))
         done, _ = await asyncio.wait(tasks)
         for task in done:
             e = task.exception()