#!/usr/bin/python import datetime import sqlite3 from wsgiref.simple_server import make_server import jinja2 from werkzeug.exceptions import HTTPException, NotFound from werkzeug.routing import Map, Rule, RequestRedirect from werkzeug.wrappers import Request, Response hash_functions = [ ("sha512", "sha512"), ("image_sha512", "image_sha512"), ("gzip_sha512", "gzip_sha512"), ("sha512", "gzip_sha512"), ("gzip_sha512", "sha512")] jinjaenv = jinja2.Environment(loader=jinja2.FileSystemLoader(".")) def format_size(size): assert isinstance(size, int) size = float(size) fmt = "%d B" if size >= 1024: size /= 1024 fmt = "%.1f KB" if size >= 1024: size /= 1024 fmt = "%.1f MB" if size >= 1024: size /= 1024 fmt = "%.1f GB" return fmt % size jinjaenv.filters["format_size"] = format_size base_template = jinjaenv.get_template("base.html") package_template = jinjaenv.from_string( """{% extends "base.html" %} {% block title %}duplication of {{ package|e }}{% endblock %} {% block header %}{% endblock %} {% block content %}

{{ package|e }}

Version: {{ version|e }}

Architecture: {{ architecture|e }}

Number of files: {{ num_files }}

Total size: {{ total_size|format_size }}

{%- if shared -%} {%- for function, sharing in shared.items() -%}

sharing with respect to {{ function|e }}

{%- for entry in sharing|sort(attribute="savable", reverse=true) -%} {%- if entry.package %}{{ entry.package|e }}{% else %}self{% endif %} compare {%- endfor -%}
packagefiles shareddata shared
{{ entry.duplicate }} ({{ (100 * entry.duplicate / num_files)|int }}%) {{ entry.savable|format_size }} ({{ (100 * entry.savable / total_size)|int }}%)
{%- endfor -%} {%- endif -%} {% endblock %}""") detail_template = jinjaenv.from_string( """{% extends "base.html" %} {% block title %}sharing between {{ details1.package|e }} and {{ details2.package|e }}{% endblock%} {% block content %}

{{ details1.package|e }} <-> {{ details2.package|e }}

{%- if shared -%} {%- for entry in shared|sort(attribute="size1", reverse=true) -%} {%- endfor -%}
{{ details1.package|e }}{{ details2.package|e }}
sizefilenamehash functionssizefilenamehash functions
{{ entry.size1|format_size }}{{ entry.filename1 }} {%- for funccomb, hashvalue in entry.functions.items() %}{{ funccomb[0]|e }} {% endfor %} {{ entry.size2|format_size }}{{ entry.filename2 }} {%- for funccomb, hashvalue in entry.functions.items() %}{{ funccomb[1]|e }} {% endfor %}
{%- endif -%} {% endblock %}""") hash_template = jinjaenv.from_string( """{% extends "base.html" %} {% block title %}information on {{ function|e }} hash {{ hashvalue|e }}{% endblock %} {% block content %}

{{ function|e }} {{ hashvalue|e }}

{%- for entry in entries -%} {%- endfor -%}
packagefilenamesizedifferent function
{{ entry.package|e }} {{ entry.filename|e }}{{ entry.size|format_size }} {% if function != entry.function %}{{ entry.function|e }}{% endif %}
{% endblock %}""") index_template = jinjaenv.from_string( """{% extends "base.html" %} {% block title %}Debian duplication detector{% endblock %} {% block header %} {% endblock %} {% block content %}

Debian duplication detector

{% endblock %}""") def fetchiter(cursor): rows = cursor.fetchmany() while rows: for row in rows: yield row rows = cursor.fetchmany() def encode_and_buffer(iterator): buff = b"" for elem in iterator: buff += elem.encode("utf8") if len(buff) >= 2048: yield buff buff = b"" if buff: yield buff def html_response(unicode_iterator, max_age=24 * 60 * 60): resp = Response(encode_and_buffer(unicode_iterator), mimetype="text/html") resp.cache_control.max_age = max_age resp.expires = datetime.datetime.now() + datetime.timedelta(seconds=max_age) return resp class Application(object): def __init__(self, db): self.db = db self.routingmap = Map([ Rule("/", methods=("GET",), endpoint="index"), Rule("/binary/", methods=("GET",), endpoint="package"), Rule("/compare//", methods=("GET",), endpoint="detail"), Rule("/hash//", methods=("GET",), endpoint="hash"), ]) @Request.application def __call__(self, request): mapadapter = self.routingmap.bind_to_environ(request.environ) try: endpoint, args = mapadapter.match() if endpoint == "package": return self.show_package(args["package"]) elif endpoint == "detail": return self.show_detail(args["package1"], args["package2"]) elif endpoint == "hash": return self.show_hash(args["function"], args["hashvalue"]) elif endpoint == "index": if not request.environ["PATH_INFO"]: raise RequestRedirect(request.environ["SCRIPT_NAME"] + "/") return html_response(index_template.stream()) raise NotFound() except HTTPException as e: return e def get_details(self, package): cur = self.db.cursor() cur.execute("SELECT version, architecture FROM package WHERE package = ?;", (package,)) row = cur.fetchone() if not row: raise NotFound() version, architecture = row details = dict(package=package, version=version, architecture=architecture) cur.execute("SELECT count(filename), sum(size) FROM content WHERE package = ?;", (package,)) num_files, total_size = cur.fetchone() details.update(dict(num_files=num_files, total_size=total_size)) return details def get_dependencies(self, package): cur = self.db.cursor() cur.execute("SELECT required FROM dependency WHERE package = ?;", (package,)) return set(row[0] for row in fetchiter(cur)) def show_package(self, package): cur = self.db.cursor() params = self.get_details(package) params["dependencies"] = self.get_dependencies(package) sharedstats = {} for func1, func2 in hash_functions: 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.function = ? AND b.function = ? AND (a.filename != b.filename OR b.package != ?);", (package, func1, func2, package)) sharing = dict() for afile, hashval, size, bpkg in fetchiter(cur): hashdict = sharing.setdefault(bpkg, dict()) fileset = hashdict.setdefault(hashval, (size, set()))[1] fileset.add(afile) if sharing: curstats = list() if func1 == func2: sharedstats[func1] = curstats else: sharedstats["%s -> %s" % (func1, func2)] = curstats mapping = sharing.pop(package, dict()) if mapping: duplicate = sum(len(files) for _, files in mapping.values()) savable = sum(size * (len(files) - 1) for size, files in mapping.values()) curstats.append(dict(package=None, duplicate=duplicate, savable=savable)) for pkg, mapping in sharing.items(): duplicate = sum(len(files) for _, files in mapping.values()) savable = sum(size * len(files) for size, files in mapping.values()) curstats.append(dict(package=pkg, duplicate=duplicate, savable=savable)) params["shared"] = sharedstats return html_response(package_template.render(params)) def show_detail(self, package1, package2): cur = self.db.cursor() if package1 == package2: details1 = details2 = self.get_details(package1) cur.execute("SELECT a.filename, a.size, a.function, b.filename, b.size, b.function, a.hash FROM content AS a JOIN content AS b ON a.hash = b.hash WHERE a.package = ? AND b.package = ? AND a.filename != b.filename;", (package1, package1)) else: details1 = self.get_details(package1) details2 = self.get_details(package2) cur.execute("SELECT a.filename, a.size, a.function, b.filename, b.size, b.function, a.hash FROM content AS a JOIN content AS b ON a.hash = b.hash WHERE a.package = ? AND b.package = ?;", (package1, package2)) shared = dict() for filename1, size1, func1, filename2, size2, func2, hashvalue in fetchiter(cur): funccomb = (func1, func2) if funccomb not in hash_functions: continue funcdict = shared.setdefault((filename1, filename2), (size1, size2, dict()))[2] funcdict[(func1, func2)] = hashvalue shared = [dict(filename1=filename1, filename2=filename2, size1=size1, size2=size2, functions=functions) for (filename1, filename2), (size1, size2, functions) in shared.items()] params = dict( details1=details1, details2=details2, shared=shared) return html_response(detail_template.render(params)) def show_hash(self, function, hashvalue): cur = self.db.cursor() cur.execute("SELECT package, filename, size, function FROM content WHERE hash = ?;", (hashvalue,)) entries = [dict(package=package, filename=filename, size=size, function=otherfunc) for package, filename, size, otherfunc in fetchiter(cur) if (function, otherfunc) in hash_functions] if not entries: raise NotFound() params = dict(function=function, hashvalue=hashvalue, entries=entries) return html_response(hash_template.render(params)) def main(): app = Application(sqlite3.connect("test.sqlite3")) #app = DebuggedApplication(app, evalex=True) make_server("0.0.0.0", 8800, app).serve_forever() if __name__ == "__main__": main()