From c50cbacfd87bf8ac9923fa63f8c818dbf6b2ede6 Mon Sep 17 00:00:00 2001
From: Melody Horn <melody@boringcactus.com>
Date: Fri, 26 Mar 2021 10:52:33 -0600
Subject: become faster & more accurate on APK

---
 noRMS-apk.py | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)

diff --git a/noRMS-apk.py b/noRMS-apk.py
index a3039c3..b05456a 100644
--- a/noRMS-apk.py
+++ b/noRMS-apk.py
@@ -17,7 +17,7 @@ args = parser.parse_args()
 
 verbose = args.verbose
 
-PACKAGE = re.compile(r'^(?P<name_and_version>\S+) (?P<arch>\S+) \{(?P<origin>[^}]+)\} \((?P<licenses>[^)]+)\) \[installed\]$')
+PACKAGE = re.compile(r'^(?P<name>\S+)-[\d\w\._]+-r\d+ \S+ \{(?P<origin>[^}]+)\} \([^)]+\) \[installed\]$')
 
 # from https://directory.fsf.org/wiki/GNU:
 # `new Set([].map.call(document.querySelectorAll('table tr td:nth-child(3)'), x => new URL(x.innerText).host))`
@@ -49,25 +49,31 @@ origins = dict()
 all_packages = subprocess.check_output(['apk', 'list', '--installed'], text=True)
 for package in all_packages.splitlines():
     match = PACKAGE.match(package)
-    name_and_version, arch, origin, licenses = match['name_and_version'], match['arch'], match['origin'], match['licenses']
+    if match is None:
+        print('oops!', package)
+    name, origin = match['name'], match['origin']
     if origin not in origins:
         origins[origin] = []
-    origins[origin].append(name_and_version)
+    origins[origin].append(name)
+
+info_output = subprocess.check_output(['apk', 'info', '-wv', *origins.keys()], text=True)
+package_webpage_map = dict(x.split(': ', 1) for x in info_output.splitlines())
 
 rms_count = 0
 
-origin_keys = list(origins.keys())
-info_output = subprocess.check_output(['apk', 'info', '-w', *origin_keys], text=True)
-info_pieces = info_output.split('\n\n')
-for origin, info in zip(origin_keys, info_pieces):
-    webpage = info.splitlines()[1]
+for origin in origins:
+    webpage = package_webpage_map[origin]
     parsed_url = urllib.parse.urlparse(webpage)
     host = parsed_url.netloc
     if verbose >= 1:
         print('origin:', origin, 'host:', host, 'webpage:', webpage)
     if host in GNU_DOMAINS:
-        package_text = ', '.join(origins[origin])
-        print(f'error: {origin} ({package_text}) is GNU software (website is {webpage})')
+        subpackages = [x for x in origins[origin] if x != origin]
+        if len(subpackages) == 0:
+            package_text = ''
+        else:
+            package_text = ' (& ' + ', '.join(sorted(subpackages)) + ')'
+        print(f'error: {origin}{package_text} is GNU software (website is {webpage})')
         rms_count += 1
 
 print()
-- 
cgit v1.2.3