From 1038168df096e46817c38db8b76999b969a4f91a Mon Sep 17 00:00:00 2001
From: Ludwig Nussel <ludwig.nussel@suse.de>
Date: Wed, 6 Aug 2014 14:51:10 +0200
Subject: [PATCH] enhance factory news

---
 factory-package-news.py | 92 +++++++++++++++++++++++++++++++++++++----
 1 file changed, 85 insertions(+), 7 deletions(-)

diff --git a/factory-package-news.py b/factory-package-news.py
index 6ed96fef..6f8e835b 100755
--- a/factory-package-news.py
+++ b/factory-package-news.py
@@ -28,6 +28,8 @@ import rpm
 import pickle
 import cmdln
 
+data_version = 3
+
 try:
     from xml.etree import cElementTree as ET
 except ImportError:
@@ -60,6 +62,7 @@ class ChangeLogger(cmdln.Cmdln):
     def readChangeLogs(self, args):
 
         pkgdata = dict()
+        changelogs = dict()
 
         for path in args:
             for root, dirs, files in os.walk(path):
@@ -68,16 +71,18 @@ class ChangeLogger(cmdln.Cmdln):
                     #print h.sprintf("[* %{CHANGELOGTIME:day} %{CHANGELOGNAME}\n%{CHANGELOGTEXT}\n\n]")
                     #print h['changelogname']
                     evr = dict()
-                    for tag in ['name', 'version', 'release']:
+                    for tag in ['name', 'version', 'release', 'sourcerpm']:
                         evr[tag] = h[tag]
-                    if h['sourcerpm'] in pkgdata:
-                        pkgdata[h['sourcerpm']]['packages'].append(evr)
+                    pkgdata[h['name']] = evr
+
+                    if h['sourcerpm'] in changelogs:
+                        changelogs[h['sourcerpm']]['packages'].append(h['name'])
                     else:
-                        data = { 'packages': [ evr ] }
+                        data = { 'packages': [ h['name'] ] }
                         for tag in ['changelogtime', 'changelogtext']:
                             data[tag] = h[tag]
-                        pkgdata[h['sourcerpm']] = data
-        return pkgdata
+                        changelogs[h['sourcerpm']] = data
+        return pkgdata, changelogs
 
     @cmdln.option("--snapshot", action="store", type='string', help="snapshot number")
     @cmdln.option("--dir", action="store", type='string', dest='dir', help="data directory")
@@ -96,7 +101,7 @@ class ChangeLogger(cmdln.Cmdln):
             raise Exception("missing snapshot option")
 
         f = open(os.path.join(opts.dir, opts.snapshot), 'wb')
-        pickle.dump([1, self.readChangeLogs(dirs)], f)
+        pickle.dump([data_version, self.readChangeLogs(dirs)], f)
 
     def do_dump(self, subcmd, opts, *dirs):
         """${cmd_name}: pprint the package changelog information
@@ -106,6 +111,79 @@ class ChangeLogger(cmdln.Cmdln):
         """
         pprint(self.readChangeLogs(dirs))
 
+    def _get_packages_grouped(self, pkgs, names):
+        group = dict()
+        for pkg in names:
+            if not pkgs[pkg]['sourcerpm'] in group:
+                group[pkgs[pkg]['sourcerpm']] = [pkg]
+            else:
+                group[pkgs[pkg]['sourcerpm']].append(pkg)
+        return group
+
+
+    @cmdln.option("--dir", action="store", type='string', dest='dir', help="data directory")
+    def do_diff(self, subcmd, opts, version1, version2):
+        """${cmd_name}: diff two snapshots
+
+        ${cmd_usage}
+        ${cmd_option_list}
+        """
+        if not opts.dir:
+            raise Exception("need --dir option")
+        if not os.path.isdir(opts.dir):
+            raise Exception("%s must be a directory"%opts.dir)
+
+        f = open(os.path.join(opts.dir, version1), 'rb')
+        (v, (v1pkgs, v1changelogs)) = pickle.load(f)
+        if v != data_version:
+            raise Exception("not matching version %s in %s"%(v, version1))
+        f = open(os.path.join(opts.dir, version2), 'rb')
+        (v, (v2pkgs, v2changelogs)) = pickle.load(f)
+        if v != data_version:
+            raise Exception("not matching version %s in %s"%(v, version2))
+
+        p1 = set(v1pkgs.keys())
+        p2 = set(v2pkgs.keys())
+
+        print "Changed packages:"
+        group = self._get_packages_grouped(v2pkgs, p1&p2)
+#        pprint(p1&p2)
+#        pprint(group)
+#        print "  "+"\n  ".join(["\n   * ".join(sorted(group[s])) for s in sorted(group.keys()) ])
+        for srpm in sorted(group.keys()):
+            srpm1 = v1pkgs[group[srpm][0]]['sourcerpm']
+            #print group[srpm], srpm, srpm1
+            if srpm1 == srpm:
+                continue # source package unchanged
+            try:
+                t1 = v1changelogs[srpm1]['changelogtime'][0]
+            except IndexError:
+                print >>sys.stderr, srpm1, "doesn't have a changelog"
+                continue
+            #print t1
+            if t1 == v2changelogs[srpm]['changelogtime'][0]:
+                continue # no new changelog entry, probably just rebuilt
+            pkgs = sorted(group[srpm])
+            print "\n==== %s ===="%pkgs[0]
+            if v1pkgs[pkgs[0]]['version'] != v2pkgs[pkgs[0]]['version']:
+                print "Version update (%s -> %s)"%(v1pkgs[pkgs[0]]['version'], v2pkgs[pkgs[0]]['version'])
+            if len(pkgs) > 1:
+                print "Subpackages:", " ".join(pkgs[1:])
+            print
+            for (i2, t2) in enumerate(v2changelogs[srpm]['changelogtime']):
+                if t2 == t1:
+                    break
+                #print "+++ ",t2
+                print v2changelogs[srpm]['changelogtext'][i2]
+
+        print "Removed packages:"
+        group = self._get_packages_grouped(v1pkgs, p1-p2)
+        print "  "+"\n  ".join(["\n   > ".join(sorted(group[s])) for s in sorted(group.keys()) ])
+        print "\n"
+        print "Added packages:"
+        group = self._get_packages_grouped(v2pkgs, p2-p1)
+        print "  "+"\n  ".join(["\n   > ".join(sorted(group[s])) for s in sorted(group.keys()) ])
+
     def get_optparser(self):
         parser = cmdln.CmdlnOptionParser(self)
         parser.add_option("--dry", action="store_true", help="dry run")