Add terrible forge scan

pkgcrap/cli.py

@@ -3,11 +3,15 @@
 
 from sys import argv
 from pkgcrap.list_update import main as update
+from pkgcrap.list_update import maintlist as maintlist
 from pkgcrap.outdated_check import main as outdated
+from pkgcrap.forge_scan import main as forgescan
 
 options = {
     'update': update,
     'outdated': outdated,
+    'maintlist': maintlist,
+    'forgescan': forgescan,
 }
 
 def main():

pkgcrap/forge_scan.py

@@ -0,0 +1,45 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+from pkgcrap.util import conf_file_path
+import pkgcrap.parse as parse
+from urllib.parse import urlparse
+
+def main(args):
+    repos = parse.repos()
+    repos.load()
+    checked = 0
+    failed = 0
+    forges = {}
+    for repo in repos.repos.values():
+        repo.load()
+        print('Scanning', repo.name)
+        for cat in repo.categories.values():
+            cat.load()
+            for pkg in cat.packages.values():
+                pkg.load()
+                if len(pkg.ebuilds) == 0 or 'EGIT_REPO_URI' not in pkg.ebuilds[0].vars:
+                    failed += 1
+                    continue
+                eb = pkg.ebuilds[0]
+                repo = pkg.ebuilds[0].vars['EGIT_REPO_URI'].replace('${PN}', pkg.name)
+                if 'HOMEPAGE' in pkg.ebuilds[0].vars:
+                    repo = repo.replace('${HOMEPAGE}', eb.vars['HOMEPAGE'])
+                if 'EGO_PN' in pkg.ebuilds[0].vars:
+                    repo = repo.replace('${EGO_PN}', eb.vars['EGO_PN'])
+                if 'MY_REPO_URI' in pkg.ebuilds[0].vars:
+                    repo = repo.replace('${MY_REPO_URI}', eb.vars['MY_REPO_URI'])
+                forge = urlparse(repo).netloc
+                if forge == '' or forge.startswith('${'):
+                    failed += 1
+                    continue
+                if forge not in forges:
+                    forges[forge] = 0
+                forges[forge] += 1
+                checked += 1
+    print('Found git URI in '+str(checked)+' packages')
+    print('Failed to find git URI in '+str(failed)+' packages')
+    for forge in dict(reversed(sorted(forges.items(), key=lambda item: item[1]))):
+        c = forges[forge]
+        p = float(c)/checked*100
+        print(forge+': '+str(c)+f' ({p:.2f}%)')

pkgcrap/list_update.py

@@ -4,13 +4,14 @@
 from pkgcrap.util import conf_file_path
 import pkgcrap.parse as parse
 
-def packages_scan(maintainer):
+def packages_scan(maintainer, verbose=False):
     repos = parse.repos()
     repos.load()
     packages = []
     for repo in repos.repos.values():
         repo.load()
-        print('Scanning', repo.name)
+        if verbose:
+            print('Scanning', repo.name)
         for cat in repo.categories.values():
             cat.load()
             for pkg in cat.packages.values():
@@ -22,11 +23,19 @@ def packages_scan(maintainer):
                     pass
     return packages
 
+def maintlist(args):
+    if len(args) < 1:
+        print('Specify the maintainer\'s email address')
+        return
+    packages = packages_scan(args[0])
+    for pkg in packages:
+        print(pkg.full_name)
+
 def main(args):
     if len(args) < 1:
         print('Specify your maintainer email address')
         return
-    packages = packages_scan(args[0])
+    packages = packages_scan(args[0], verbose=True)
     with open(conf_file_path('maintained.txt'), 'w') as fp:
         fp.write("\n".join([pkg.full_name for pkg in packages]))
     print('Updated maintained package list')