Update herd and active developer plugins

2 files changed, 95 insertions, 10 deletions

diff --git a/utils/qa_check_developer.py b/utils/qa_check_developer.py
new file mode 100644
index 0000000..4d353f3
--- /dev/null
+++ b/utils/qa_check_developer.py
@@ -0,0 +1,79 @@
+#! /usr/bin/env python
+import os, sys
+import urllib2
+
+from datetime import datetime
+from lxml.html import fromstring
+from optparse import OptionParser
+
+path = os.path.join(os.path.dirname(__file__), os.path.pardir)
+sys.path.insert(0, path)
+del path
+
+from grumpy import app
+from grumpy.models import db, Developer, PkgIssue, Setting
+
+DEVS_URL = 'http://www.gentoo.org/proj/en/devrel/roll-call/userinfo.xml'
+PLUGIN_NAME='qa::valid_developer'
+
+def gc_collect(timestamp):
+    """Remove old QA issues from database returning number of rows deleted."""
+    db.session.expire_all()
+    print ("DEBUG: Deleted %d old issues." % PkgIssue.query \
+                        .filter_by(plugin=PLUGIN_NAME) \
+                        .filter(PkgIssue.created_on < timestamp).delete(False))
+
+def insert_issues(invalid):
+    """Insert QA issues into db."""
+    if 'maintainer-needed@gentoo.org' in invalid:
+        h = Developer.query.filter_by(email='maintainer-needed@gentoo.org').first()
+        for pkg in h.packages:
+            pkg.qaissues.append(PkgIssue(pkg, PLUGIN_NAME, 'maintainer-needed'))
+        invalid.remove('maintainer-needed@gentoo.org')
+    for dev in invalid:
+        d = Developer.query.filter_by(email=dev).first()
+        for pkg in d.packages:
+            pkg.qaissues.append(PkgIssue(pkg, PLUGIN_NAME, 'unknown-dev', \
+                      'User %s is not as active Gentoo developer.' % dev))
+    db.session.commit()
+
+def download_and_parse_devs():
+    """Return list of herd names from 'userinfo.xml'"""
+    data = urllib2.urlopen(DEVS_URL).read()
+    devs = []
+    # Select <table class='ntable'> from page and iterate over rows
+    for tr in fromstring(data).cssselect('.ntable')[0].getchildren():
+        # Next we have tr > td > b
+        devs.append('%s@gentoo.org' % tr.getchildren()[0].getchildren()[0].text)
+    # Skip first (this is table header
+    del devs[0]
+    return devs
+
+if __name__ == '__main__':
+    parser = OptionParser(usage="usage: %prog [options] CONFFILE")
+    (opts, args) = parser.parse_args()
+    if len(args) != 1:
+        parser.error("provide path to configuration file as first argument")
+        sys.exit(1)
+    # Fetch data and parse it
+    devs = download_and_parse_devs()
+    b0rks = []
+    timestamp = datetime.now()
+    with app.test_request_context():
+        # Load configuration
+        app.config.from_pyfile(args[0])
+        # Fetch list of developers from db
+        dbdevs = [d.email for d in Developer.query.filter( \
+                    Developer.email.like('%@gentoo.org')).all()]
+        for dev in dbdevs:
+            if dev in devs:
+                devs.remove(dev)
+            else:
+                b0rks.append(dev)
+        insert_issues(b0rks)
+        # Clean up issues < timestamp
+        gc_collect(timestamp)
+        # Update settings and add info about last run..
+        Setting.query.filter_by(name=PLUGIN_NAME).delete(False)
+        db.session.add(Setting(PLUGIN_NAME, str(timestamp)))
+        db.session.commit()
diff --git a/utils/qa_check_herd.py b/utils/qa_check_herd.py
index 21c5ead..7088d5a 100644
--- a/utils/qa_check_herd.py
+++ b/utils/qa_check_herd.py
@@ -1,8 +1,10 @@
 #! /usr/bin/env python
 import os, sys
+import urllib2
 
 from datetime import datetime
-from lxml.etree import parse
+from lxml.etree import fromstring
+from optparse import OptionParser
 
 path = os.path.join(os.path.dirname(__file__), os.path.pardir)
 sys.path.insert(0, path)
@@ -11,6 +13,7 @@ del path
 from grumpy import app
 from grumpy.models import db, Herd, PkgIssue, Setting
 
+HERDS_URL = 'http://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo/xml/htdocs/proj/en/metastructure/herds/herds.xml'
 PLUGIN_NAME='qa::valid_herd'
 
 def gc_collect(timestamp):
@@ -45,26 +48,29 @@ def insert_issues(invalid):
                       'Herd %s is not listed in official herd list.' % herd))
     db.session.commit()
 
-def parse_herds_xml(file):
+def download_and_parse_herds():
     """Return list of herd names from 'herds.xml'"""
+    data = urllib2.urlopen(HERDS_URL)
     herds = []
-    if not os.path.isfile(file):
-        print ("File '%s' does not exist" % file)
-        raise RuntimeError
-    for child in parse(file).getroot().getchildren():
+    for child in fromstring(data.read()).getchildren():
         for value in child.getchildren():
             if value.tag == 'name':
                 herds.append(value.text)
     return herds
 
 if __name__ == '__main__':
-    # TODO: Download latest herds file
-    # Parse list of herds from file
-    herds = parse_herds_xml('herds.xml')
+    parser = OptionParser(usage="usage: %prog [options] CONFFILE")
+    (opts, args) = parser.parse_args()
+    if len(args) != 1:
+        parser.error("provide path to configuration file as first argument")
+        sys.exit(1)
+    # Fetch data and parse it
+    herds = download_and_parse_herds()
     b0rks = []
     timestamp = datetime.now()
-    # Setup database for application
     with app.test_request_context():
+        # Load configuration
+        app.config.from_pyfile(args[0])
         # Fetch list of herds from db
         for herd in Herd.query.all():
             if herd.name in herds: