4 from wsgiref.simple_server import make_server
6 from werkzeug.debug import DebuggedApplication
7 from werkzeug.exceptions import HTTPException, NotFound
8 from werkzeug.routing import Map, Rule
9 from werkzeug.wrappers import Request, Response
11 def format_size(size):
25 class Application(object):
27 self.db = sqlite3.connect("test.sqlite3")
28 self.cur = self.db.cursor()
29 self.routingmap = Map([
30 Rule("/<package>", methods=("GET",),
35 def __call__(self, request):
36 mapadapter = self.routingmap.bind_to_environ(request.environ)
38 endpoint, args = mapadapter.match()
39 assert endpoint == "package"
40 return self.show_package(args["package"])
41 except HTTPException as e:
44 def show_package(self, package):
45 self.cur.execute("SELECT version, architecture FROM content WHERE package = ? LIMIT 1;", (package,))
46 row = self.cur.fetchone()
49 version, architecture = row
50 self.cur.execute("SELECT count(filename) FROM content WHERE package = ?;", (package,))
51 num_files = self.cur.fetchone()[0]
52 self.cur.execute("SELECT sum(size) FROM content WHERE package = ?;", (package,))
53 total_size = self.cur.fetchone()[0]
54 content = "<p>Version: %s</p><p>Architecture: %s</p>" % (version, architecture)
55 content += "<p>Number of files: %d</p>" % num_files
56 content += "<p>Total size: %s</p>" % format_size(total_size)
59 self.cur.execute("SELECT a.filename, a.hash, a.size, b.package FROM content AS a JOIN content AS b ON a.hash = b.hash WHERE a.package = ? AND (a.filename != b.filename OR b.package != ?);", (package, package))
60 for afile, hashval, size, bpkg in self.cur.fetchall():
61 shared.setdefault(bpkg, dict()).setdefault(hashval, (size, set()))[1].add(afile)
64 mapping = shared.pop(package, dict())
66 duplicate = sum(len(files) for _, files in mapping.values())
67 savable = sum(size * (len(files) - 1) for size, files in mapping.values())
68 sharedstats.append(("self", duplicate, savable))
69 for pkg, mapping in shared.items():
70 pkglink = '<a href="%s">%s</a>' % (pkg, pkg)
71 duplicate = sum(len(files) for _, files in mapping.values())
72 savable = sum(size * len(files) for size, files in mapping.values())
73 sharedstats.append((pkglink, duplicate, savable))
74 sharedstats.sort(key=lambda row: row[2], reverse=True)
75 content += "<table border='1'><tr><th>package</th><th>files shared</th><th>data shared</th></tr>"
76 for pkg, duplicate, savable in sharedstats:
77 content += "<tr><td>%s</td><td>%d (%d%%)</td><td>%s (%d%%)</td></tr>" % (pkg, duplicate, 100. * duplicate / num_files, format_size(savable), 100. * savable / total_size)
80 r = Response(content_type="text/html")
81 r.data = "<html><head><title>duplication of %(package)s</title></head><body><h1>%(package)s</h1>%(content)s</body></html>" % dict(package=package, content=content)
86 app = DebuggedApplication(app, evalex=True)
87 make_server("localhost", 8800, app).serve_forever()
89 if __name__ == "__main__":