Add support for debian-based distros

3 files changed, 131 insertions, 22 deletions

diff --git a/inspect.py b/inspect.py
index 07b8d09..d5f41d4 100755
--- a/inspect.py
+++ b/inspect.py
@@ -5,14 +5,19 @@ import os
 import re
 import sys
 
-from tools import libvslvm, lklfuse, nbdfuse, unmount, rmdir
-from tools.inspect_apps import list_applications_rpm, list_applications_windows
+from tools import libvslvm, lklfuse, nbdfuse, subfiles, unmount, rmdir
+from tools.inspect_apps import (
+    list_applications_deb,
+    list_applications_rpm,
+    list_applications_windows
+)
 from tools.inspect_os import get_linux_os_info, get_windows_os_info
 from tools.pyparted import list_partitions
 
 fmt = "{asctime}, {name}:{lineno}:{funcName}(), {levelname}, {message}"
 logging.basicConfig(level=logging.DEBUG, format=fmt, style="{")
 
+DEB = re.compile(r"^(Debian|Ubuntu|Linux\sMint|LMDE).*$")
 RPM = re.compile(r"^(CentOS|AlmaLinux|Scientific|Rocky|Oracle|openSUSE|Fedora).*$") # noqa
 WIN = re.compile(r"^(Microsoft|Windows).*$")
 
@@ -36,17 +41,15 @@ def main(vmdk_path):
             lvm_mp = libvslvm.mount(nbd, part["offset"])
             if not lvm_mp:
                 continue
-            for vol in os.listdir(lvm_mp):
+            for vol in subfiles(lvm_mp):
                 vol_path = os.path.join(lvm_mp, vol)
                 vol_parts = list_partitions(vol_path)
                 if not vol_parts:
                     continue
-                mp = lklfuse.mount(vol_path, vol_parts[0]["type"])
-                if mp:
+                if mp := lklfuse.mount(vol_path, vol_parts[0]["type"]):
                     fs_mps.append((mp, parts[0]["type"]))
         else:
-            mp = lklfuse.mount(nbd, part["type"], part["nr"])
-            if mp:
+            if mp := lklfuse.mount(nbd, part["type"], part["nr"]):
                 fs_mps.append((mp, part["type"]))
 
     os_info = {}
@@ -60,7 +63,9 @@ def main(vmdk_path):
             break
 
     for fspath, _ in fs_mps:
-        if RPM.match(os_info["name"]):
+        if DEB.match(os_info["name"]):
+            apps = list_applications_deb(fspath)
+        elif RPM.match(os_info["name"]):
             apps = list_applications_rpm(fspath)
         elif WIN.match(os_info["name"]):
             apps = list_applications_windows(fspath)
diff --git a/tools/__init__.py b/tools/__init__.py
index e25ef6f..d97a80a 100644
--- a/tools/__init__.py
+++ b/tools/__init__.py
@@ -4,7 +4,12 @@ import os
 from functools import wraps
 from subprocess import run
 
-__all__ = ["unmount", "rmdir"]
+__all__ = [
+    "unmount",
+    "rmdir",
+    "subdirs",
+    "subfiles"
+]
 
 L = logging.getLogger(__name__)
 
@@ -71,3 +76,29 @@ def rmdir(path):
         return False
 
     return True
+
+
+@log
+def subdirs(path):
+    """Yield directory names under given path using os.scandir.
+
+    See also:
+    https://docs.python.org/3/library/os.html#os.scandir
+    """
+    with os.scandir(path) as it:
+        for entry in it:
+            if not entry.name.startswith(".") and entry.is_dir():
+                yield entry.name
+
+
+@log
+def subfiles(path):
+    """Yield file names under given path using os.scandir.
+
+    See also:
+    https://docs.python.org/3/library/os.html#os.scandir
+    """
+    with os.scandir(path) as it:
+        for entry in it:
+            if not entry.name.startswith(".") and entry.is_file():
+                yield entry.name
diff --git a/tools/inspect_apps.py b/tools/inspect_apps.py
index 014625c..94a8768 100644
--- a/tools/inspect_apps.py
+++ b/tools/inspect_apps.py
@@ -2,9 +2,10 @@ import logging
 import os
 import tempfile
 
-from . import log
+from . import log, subdirs
 
 __all__ = [
+    "list_applications_deb",
     "list_applications_rpm",
     "list_applications_windows"
 ]
@@ -31,6 +32,78 @@ except ModuleNotFoundError:
 
 
 @log
+def list_applications_deb(path):
+    """Find all packages installed on a debian-based linux distribution.
+
+    See also:
+    https://man7.org/linux/man-pages/man1/dpkg.1.html
+
+    Args:
+        path (str): Path to the mounted filesystem.
+
+    Returns:
+        List of packages. For example:
+        [{'name': 'adduser', 'version': '3.118'}, ...]
+    """
+    dpkg_db = None
+
+    locations = [
+        "var/lib/dpkg/status",
+        "lib/dpkg/status"  # separated /var partition
+    ]
+
+    for location in locations:
+        db = os.path.join(path, location)
+        if os.path.exists(db):
+            dpkg_db = db
+            break
+
+    # Debian uses subvol=@rootfs for root filesystem for btrfs.
+    # Therefore, under Debian 11.* looks like this: /@rootfs/var/lib/dpkg
+    if dpkg_db is None:
+        for dir in subdirs(path):
+            new_path = os.path.join(path, dir)
+            for location in locations:
+                db = os.path.join(new_path, location)
+                if os.path.exists(db):
+                    dpkg_db = db
+                    break
+            if dpkg_db is not None:
+                break
+
+    if dpkg_db is None:
+        L.debug("dpkg database not found")
+        return []
+
+    pkgs = []
+    with open(dpkg_db) as f:
+        name = version = ""
+        installed = False
+        count = 0
+        for line in f:
+            if count >= 10:
+                break
+            line = line.strip()
+            if not line:
+                if name and version and installed:
+                    pkgs.append({
+                        "name": name,
+                        "version": version
+                    })
+                    count += 1
+                name = version = ""
+                installed = False
+            elif line.startswith("Package:"):
+                name = line[9:]
+            elif line.startswith("Status:"):
+                installed = "installed" in line[8:].split()
+            elif line.startswith("Version:"):
+                version = line[9:]
+
+    return pkgs
+
+
+@log
 def list_applications_rpm(path):
     """Find all packages installed on a rpm-based linux distribution.
 
@@ -75,16 +148,16 @@ def list_applications_rpm(path):
         L.error("failed to open RPM database: %r", e)
         return []
 
-    ret = []
+    pkgs = []
     for h in dbMatch:
-        ret.append({
+        pkgs.append({
             "name": h["name"],
             "version": h["version"]
         })
 
     rpm.delMacro("_dbpath")
 
-    return ret
+    return pkgs
 
 
 @log
@@ -121,7 +194,7 @@ def list_applications_windows(path):
         L.error("failed to open registry file %s: %r", software, e)
         return []
 
-    ret = []
+    apps = []
 
     # native applications
     hive_path = "Microsoft\\Windows\\CurrentVersion\\Uninstall"
@@ -129,9 +202,9 @@ def list_applications_windows(path):
         key = registry.open(hive_path)
     except Exception as e:
         L.error("%s not found in %s: %r", hive_path, software, e)
-        return ret
+        return apps
     if apps_native := _list_applications_windows_from_key(key):
-        ret.extend(apps_native)
+        apps.extend(apps_native)
 
     # 32-bit applications running on WOW64 emulator
     # see also: http://support.microsoft.com/kb/896459
@@ -140,11 +213,11 @@ def list_applications_windows(path):
         key = registry.open(hive_path)
     except Exception as e:
         L.error("%s not found in %s: %r", hive_path, software, e)
-        return ret
+        return apps
     if apps_emulator := _list_applications_windows_from_key(key):
-        ret.extend(apps_emulator)
+        apps.extend(apps_emulator)
 
-    return ret
+    return apps
 
 
 @log
@@ -160,7 +233,7 @@ def _list_applications_windows_from_key(key):
     Returns:
         List of applications.
     """
-    ret = []
+    apps = []
     for k in key.subkeys():
         # name = k.name()
         # name does not say much, so take the display name
@@ -172,9 +245,9 @@ def _list_applications_windows_from_key(key):
                 version = v.value()
         # ignore applications with no display name
         if name and version:
-            ret.append({
+            apps.append({
                 "name": name,
                 "version": version
             })
 
-    return ret
+    return apps