debsize.py

import subprocess
import collections
import io
import sys
import os
import shlex

__doc__ = """
Tool for printing dependency tree for Debian systems including size of package. Might help with
stripping down dependencies.

Usage:
    DPKG=ssh someone@somewhere dpkg python debsize.py > somewhere.dot
    dot -Tsvg somewhere.dot > somewhere.svg
"""

DPKG=["dpkg", ]

dpkg_v = os.getenv("DPKG")
if dpkg_v:
    DPKG=shlex.split(dpkg_v)
    del dpkg_v

def kp(string):
    ret = list()
    kill=False
    for ch in string:
        if ch == '(':
            kill=True
        if not kill:
            ret.append(ch)
        if ch == ')' and kill:
            kill=False
    return "".join(ret).split('|')[0]

def parse_depends(depends):
    return [kp(item).strip() for item in depends.split(',')]

DpkgRes = collections.namedtuple("DpkgRes", "package, depends, suggests, size")
def parse_dpkg_status(f):
    package = ""
    depends = []
    suggests = []
    size = 0
    for line in f:
        if line.startswith("Package:"):
            package = line.split(':')[1].strip()
        elif line.startswith("Installed-Size:"):
            size = int(line.split(':')[1].strip())
        elif line.startswith("Depends:"):
            depends.extend(parse_depends(line.split(':')[1]))
        elif line.startswith("Pre-Depends:"):
            depends.extend(parse_depends(line.split(':')[1]))
        elif line.startswith("Suggests:"):
            suggests.extend(parse_depends(line.split(':')[1]))
        elif line.startswith("Recommends:"):
            suggests.extend(parse_depends(line.split(':')[1]))

    return DpkgRes(package, depends, suggests, size)

def parse_dpkg_list(f):
    ret = set()
    for line in f:
        if not line.startswith("ii"):
            continue
        name = line.split()[1]
        if ':' in name:
            name = name.split(':')[0]
        ret.add(name)
    return ret

def dpkg_status(name):
    assert name
    global DPKG
    return io.BytesIO(subprocess.check_output(DPKG + ["-s", name,]))

def dpkg_list():
    global DPKG
    return io.BytesIO(subprocess.check_output(DPKG + ["-l", ]))

def self_test():
    with open("dpkg.info", "r") as f:
        print(parse_dpkg_status(f))

    with open("dpkg.list", "r") as f:
        print(parse_dpkg_list(f))

# main
print("digraph debian {");
packages = parse_dpkg_list(dpkg_list())
for package in packages:
    status = parse_dpkg_status(dpkg_status(package))

    print('"%(package)s" [label="%(package)s (%(size)d)"];' % dict(package=package, size=status.size))
    for dep in status.depends:
        print('"%(package)s" -> "%(dep)s";' % dict(package=package, dep=dep))
    sys.stdout.flush()
    for dep in (dep for dep in status.suggests if dep in packages):
        print('"%(package)s" -> "%(dep)s" [style=dotted];' % dict(package=package, dep=dep))
    sys.stdout.flush()
    #TODO suggests
print("}")