4 from wsgiref.simple_server import make_server
7 from werkzeug.exceptions import HTTPException, NotFound
8 from werkzeug.routing import Map, Rule, RequestRedirect
9 from werkzeug.wrappers import Request, Response
11 jinjaenv = jinja2.Environment()
13 def format_size(size):
14 assert isinstance(size, int)
28 jinjaenv.filters["format_size"] = format_size
30 package_template = jinjaenv.from_string(
31 """<html><head><title>duplication of {{ package|e }}</title><style type="text/css">.dependency { background-color: yellow; } </style></head>
32 <body><h1>{{ package|e }}</h1>
33 <p>Version: {{ version|e }}</p>
34 <p>Architecture: {{ architecture|e }}</p>
35 <p>Number of files: {{ num_files }}</p>
36 <p>Total size: {{ total_size|format_size }}</p>
38 {%- for function, sharing in shared.items() -%}
39 <h3>sharing with respect to {{ function }}</h3>
40 <table border='1'><tr><th>package</th><th>files shared</th><th>data shared</th></tr>
41 {%- for entry in sharing|sort(attribute="savable", reverse=true) -%}
42 <tr><td{% if not entry.package or entry.package in dependencies %} class="dependency"{% endif %}>
43 {%- if entry.package %}<a href="{{ entry.package|e }}">{{ entry.package|e }}</a>{% else %}self{% endif %}
44 <a href="../compare/{{ package|e }}/{{ entry.package|default(package, true)|e }}">compare</a></td>
45 <td>{{ entry.duplicate }} ({{ (100 * entry.duplicate / num_files)|int }}%)</td>
46 <td>{{ entry.savable|format_size }} ({{ (100 * entry.savable / total_size)|int }}%)</td></tr>
54 detail_template = jinjaenv.from_string(
55 """<html><head><title>sharing between {{ details1.package|e }} and {{ details2.package|e }}</title></head>
56 <body><h1><a href="../../binary/{{ details1.package|e }}">{{ details1.package|e }}</a> <-> <a href="../../binary/{{ details2.package|e }}">{{ details2.package|e }}</a></h1>
58 <table border='1'><tr><th>size</th><th>filename in {{ details1.package|e }}</th><th>filename in {{ details2.package|e }}</th><th>hash functions</th></tr>
59 {%- for entry in shared|sort(attribute="size", reverse=true) -%}
60 <tr><td>{{ entry.size|format_size }}</td><td>{{ entry.filename1 }}</td><td>{{ entry.filename2 }}</td><td>
61 {%- for function, hashvalue in entry.functions.items() %}<a href="../../hash/{{ function|e }}/{{ hashvalue|e }}">{{ function|e }}</a> {% endfor %}</td></tr>
67 hash_template = jinjaenv.from_string(
68 """<html><head><title>information on {{ function|e }} hash {{ hashvalue|e }}</title></head>
69 <body><h1>{{ function|e }} {{ hashvalue|e }}</h1>
70 <table border='1'><tr><th>package</th><th>filename</th><th>size</th></tr>
71 {%- for entry in entries -%}
72 <tr><td><a href="../../binary/{{ entry.package|e }}">{{ entry.package|e }}</a></td>
73 <td>{{ entry.filename|e }}</td><td>{{ entry.size|format_size }}</td></tr>
78 class Application(object):
80 self.db = sqlite3.connect("test.sqlite3")
81 self.cur = self.db.cursor()
82 self.routingmap = Map([
83 Rule("/", methods=("GET",), endpoint="index"),
84 Rule("/binary/<package>", methods=("GET",), endpoint="package"),
85 Rule("/compare/<package1>/<package2>", methods=("GET",), endpoint="detail"),
86 Rule("/hash/<function>/<hashvalue>", methods=("GET",), endpoint="hash"),
90 def __call__(self, request):
91 mapadapter = self.routingmap.bind_to_environ(request.environ)
93 endpoint, args = mapadapter.match()
94 if endpoint == "package":
95 return self.show_package(args["package"])
96 elif endpoint == "detail":
97 return self.show_detail(args["package1"], args["package2"])
98 elif endpoint == "hash":
99 return self.show_hash(args["function"], args["hashvalue"])
100 elif endpoint == "index":
101 if not request.environ["PATH_INFO"]:
102 raise RequestRedirect(request.environ["SCRIPT_NAME"] + "/")
103 return Response("""<html><head><title>Debian duplication detector</title></head>
104 <body><h1>Debian duplication detector</h1>
106 <li>To inspect a particlar binary package, go to <pre>binary/<packagename></pre> Example: <a href="binary/git">binary/git</a></li>
107 <li>To inspect a combination of binary packages go to <pre>compare/<firstpackage>/<secondpackage></pre> Example: <a href="compare/git/git">compare/git/git</a></li>
108 <li>To discover package shipping a particular file go to <pre>hash/sha512/<hashvalue></pre> Example: <a href="hash/sha512/ed94df7781793f06f9426a600c1bde86397afc7b35cb3aa11b60214bd31e35ad893b53a04a2cf4676154982d7c204c4aa165d6ccdaac0170031364a05dbab3bc">hash/sha512/ed94df7781793f06f9426a600c1bde86397afc7b35cb3aa11b60214bd31e35ad893b53a04a2cf4676154982d7c204c4aa165d6ccdaac0170031364a05dbab3bc</a></li>
109 <ul></body></html>""",
110 content_type="text/html")
112 except HTTPException as e:
115 def get_details(self, package):
116 self.cur.execute("SELECT version, architecture FROM package WHERE package = ?;",
118 row = self.cur.fetchone()
121 version, architecture = row
122 details = dict(package=package,
124 architecture=architecture)
125 self.cur.execute("SELECT count(filename), sum(size) FROM content WHERE package = ?;",
127 num_files, total_size = self.cur.fetchone()
128 details.update(dict(num_files=num_files, total_size=total_size))
131 def get_dependencies(self, package):
132 self.cur.execute("SELECT required FROM dependency WHERE package = ?;",
134 return set(row[0] for row in self.cur.fetchall())
136 def show_package(self, package):
137 params = self.get_details(package)
138 params["dependencies"] = self.get_dependencies(package)
141 self.cur.execute("SELECT a.filename, a.function, a.hash, a.size, b.package FROM content AS a JOIN content AS b ON a.function = b.function AND a.hash = b.hash WHERE a.package = ? AND (a.filename != b.filename OR b.package != ?);",
143 for afile, function, hashval, size, bpkg in self.cur.fetchall():
144 pkgdict = shared.setdefault(function, dict())
145 hashdict = pkgdict.setdefault(bpkg, dict())
146 fileset = hashdict.setdefault(hashval, (size, set()))[1]
150 for function, sharing in shared.items():
151 sharedstats[function] = list()
152 mapping = sharing.pop(package, dict())
154 duplicate = sum(len(files) for _, files in mapping.values())
155 savable = sum(size * (len(files) - 1) for size, files in mapping.values())
156 sharedstats[function].append(dict(package=None, duplicate=duplicate, savable=savable))
157 for pkg, mapping in sharing.items():
158 duplicate = sum(len(files) for _, files in mapping.values())
159 savable = sum(size * len(files) for size, files in mapping.values())
160 sharedstats[function].append(dict(package=pkg, duplicate=duplicate, savable=savable))
162 params["shared"] = sharedstats
163 return Response(package_template.render(**params).encode("utf8"),
164 content_type="text/html")
166 def show_detail(self, package1, package2):
167 if package1 == package2:
168 details1 = details2 = self.get_details(package1)
170 self.cur.execute("SELECT a.filename, b.filename, a.size, a.function, a.hash FROM content AS a JOIN content AS b ON a.function = b.function AND a.hash = b.hash WHERE a.package = ? AND b.package = ? AND a.filename != b.filename;",
171 (package1, package1))
173 details1 = self.get_details(package1)
174 details2 = self.get_details(package2)
176 self.cur.execute("SELECT a.filename, b.filename, a.size, a.function, a.hash FROM content AS a JOIN content AS b ON a.function = b.function AND a.hash = b.hash WHERE a.package = ? AND b.package = ?;",
177 (package1, package2))
180 for filename1, filename2, size, function, hashvalue in self.cur.fetchall():
181 shared.setdefault((filename1, filename2, size), dict())[function] = hashvalue
182 shared = [dict(filename1=filename1, filename2=filename2, size=size,
184 for (filename1, filename2, size), functions in shared.items()]
189 return Response(detail_template.render(**params).encode("utf8"),
190 content_type="text/html")
192 def show_hash(self, function, hashvalue):
193 self.cur.execute("SELECT package, filename, size FROM content WHERE function = ? AND hash = ?;",
194 (function, hashvalue))
195 entries = [dict(package=package, filename=filename, size=size)
196 for package, filename, size in self.cur.fetchall()]
199 params = dict(function=function, hashvalue=hashvalue, entries=entries)
200 return Response(hash_template.render(**params).encode("utf8"),
201 content_type="text/html")
205 #app = DebuggedApplication(app, evalex=True)
206 make_server("0.0.0.0", 8800, app).serve_forever()
208 if __name__ == "__main__":