From c8d64e153413042f79d21c04b2b82365ae2714d4 Mon Sep 17 00:00:00 2001
From: Mürsel Türk
Date: Thu, 8 Sep 2022 22:30:54 +0200
Subject: Add support for arch-based distros

---
 inspect.py            | 11 ++++++---
 tools/__init__.py     |  2 --
 tools/inspect_apps.py | 67 +++++++++++++++++++++++++++++++++++++++++++++++++++
 tools/inspect_os.py   |  4 +++
 4 files changed, 79 insertions(+), 5 deletions(-)

diff --git a/inspect.py b/inspect.py
index d5f41d4..c73f87e 100755
--- a/inspect.py
+++ b/inspect.py
@@ -8,6 +8,7 @@ import sys
 from tools import libvslvm, lklfuse, nbdfuse, subfiles, unmount, rmdir
 from tools.inspect_apps import (
     list_applications_deb,
+    list_applications_pacman,
     list_applications_rpm,
     list_applications_windows
 )
@@ -18,6 +19,7 @@ fmt = "{asctime}, {name}:{lineno}:{funcName}(), {levelname}, {message}"
 logging.basicConfig(level=logging.DEBUG, format=fmt, style="{")
 
 DEB = re.compile(r"^(Debian|Ubuntu|Linux\sMint|LMDE).*$")
+PACMAN = re.compile(r"^(Arch|Manjaro).*$")
 RPM = re.compile(r"^(CentOS|AlmaLinux|Scientific|Rocky|Oracle|openSUSE|Fedora).*$") # noqa
 WIN = re.compile(r"^(Microsoft|Windows).*$")
 
@@ -62,12 +64,15 @@ def main(vmdk_path):
         if os_info:
             break
 
+    os_name = os_info.get("name", "")
     for fspath, _ in fs_mps:
-        if DEB.match(os_info["name"]):
+        if DEB.match(os_name):
             apps = list_applications_deb(fspath)
-        elif RPM.match(os_info["name"]):
+        elif PACMAN.match(os_name):
+            apps = list_applications_pacman(fspath)
+        elif RPM.match(os_name):
             apps = list_applications_rpm(fspath)
-        elif WIN.match(os_info["name"]):
+        elif WIN.match(os_name):
             apps = list_applications_windows(fspath)
         if apps:
             break
diff --git a/tools/__init__.py b/tools/__init__.py
index d97a80a..374823f 100644
--- a/tools/__init__.py
+++ b/tools/__init__.py
@@ -78,7 +78,6 @@ def rmdir(path):
     return True
 
 
-@log
 def subdirs(path):
     """Yield directory names under given path using os.scandir.
 
@@ -91,7 +90,6 @@ def subdirs(path):
                 yield entry.name
 
 
-@log
 def subfiles(path):
     """Yield file names under given path using os.scandir.
 
diff --git a/tools/inspect_apps.py b/tools/inspect_apps.py
index 58f0f0f..6c814d2 100644
--- a/tools/inspect_apps.py
+++ b/tools/inspect_apps.py
@@ -1,11 +1,13 @@
 import logging
 import os
+import re
 import tempfile
 
 from . import log, subdirs
 
 __all__ = [
     "list_applications_deb",
+    "list_applications_pacman",
     "list_applications_rpm",
     "list_applications_windows"
 ]
@@ -99,6 +101,71 @@ def list_applications_deb(path):
     return pkgs
 
 
+@log
+def list_applications_pacman(path):
+    """Find all packages installed on a arch-based linux distribution.
+
+    See also:
+    https://wiki.archlinux.org/title/pacman
+
+    Args:
+        path (str): Path to the mounted filesystem.
+
+    Returns:
+        List of packages. For example:
+        [{'name': 'python', 'version': '3.10.6-1'}, ...]
+    """
+    pacman_db = None
+
+    locations = [
+        "var/lib/pacman/local",
+        "lib/pacman/local"  # separated /var partition
+    ]
+
+    for location in locations:
+        db = os.path.join(path, location)
+        if os.path.exists(db):
+            pacman_db = db
+            break
+
+    # Just in case the system is using btrfs.
+    if pacman_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):
+                    pacman_db = db
+                    break
+            if pacman_db is not None:
+                break
+
+    if pacman_db is None:
+        L.debug("pacman database not found")
+        return []
+
+    pkgs = []
+    for dir in subdirs(pacman_db):
+        desc = os.path.join(pacman_db, dir, "desc")
+        kv = {}
+        with open(desc) as f:
+            sections = re.split(r"\n(?=%[A-Z]+%)", f.read())
+            for section in sections:
+                section = section.strip()
+                if lines := section.split("\n"):
+                    k = lines[0].strip("%")
+                    n = len(lines)
+                    v = "" if n < 2 else lines[1:] if n > 2 else lines[1]
+                    kv[k] = v
+        if (name := kv.get("NAME", "")) and (version := kv.get("VERSION", "")):
+            pkgs.append({
+                "name": name,
+                "version": version
+            })
+
+    return pkgs
+
+
 @log
 def list_applications_rpm(path):
     """Find all packages installed on a rpm-based linux distribution.
diff --git a/tools/inspect_os.py b/tools/inspect_os.py
index ed4ee45..9848c2e 100644
--- a/tools/inspect_os.py
+++ b/tools/inspect_os.py
@@ -72,6 +72,10 @@ def get_linux_os_info(path):
                 kv[k] = v
             name = kv.get("NAME", "")
             version = kv.get("VERSION", kv.get("VERSION_ID", ""))
+            # Arch-based distros have neither VERSION nor VERSION_ID.
+            # These are using a rolling release model.
+            if not version:
+                version = kv.get("BUILD_ID", "")
     elif "system-release" in release_files:
         # RedHat (RHEL) provides the redhat-release file. However, it does not
         # seem to be reliable for determining which operating system it is.
-- 
cgit v1.2.3-55-g7522