X-Git-Url: https://git.decadent.org.uk/gitweb/?a=blobdiff_plain;f=dak%2Fexamine_package.py;h=9448724e164f22ad4c31a4e40819a2919b76ea90;hb=c1e498c0e251f48d68996ddf05ecb39981361c86;hp=c7fd38e60acd0e17943764f6d1b491227a52e52f;hpb=42bfecd65546322bdbecda3321e99d5a5311048e;p=dak.git diff --git a/dak/examine_package.py b/dak/examine_package.py old mode 100644 new mode 100755 index c7fd38e6..4b315ded --- a/dak/examine_package.py +++ b/dak/examine_package.py @@ -1,7 +1,17 @@ #!/usr/bin/env python -# Script to automate some parts of checking NEW packages -# Copyright (C) 2000, 2001, 2002, 2003, 2006 James Troup +""" +Script to automate some parts of checking NEW packages + +Most functions are written in a functional programming style. They +return a string avoiding the side effect of directly printing the string +to stdout. Those functions can be used in multithreaded parts of dak. + +@contact: Debian FTP Master +@copyright: 2000, 2001, 2002, 2003, 2006 James Troup +@copyright: 2009 Joerg Jaspert +@license: GNU General Public License version 2 or later +""" # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by @@ -32,44 +42,44 @@ ################################################################################ -import errno, os, pg, re, sys, md5 -import apt_pkg, apt_inst -import daklib.database, daklib.utils, daklib.queue - -################################################################################ - -re_package = re.compile(r"^(.+?)_.*") -re_doc_directory = re.compile(r".*/doc/([^/]*).*") - -re_contrib = re.compile('^contrib/') -re_nonfree = re.compile('^non\-free/') - -re_arch = re.compile("Architecture: .*") -re_builddep = re.compile("Build-Depends: .*") -re_builddepind = re.compile("Build-Depends-Indep: .*") - -re_localhost = re.compile("localhost\.localdomain") -re_version = re.compile('^(.*)\((.*)\)') - -re_newlinespace = re.compile('\n') -re_spacestrip = re.compile('(\s)') - -html_escaping = {'"':'"', '&':'&', '<':'<', '>':'>'} -re_html_escaping = re.compile('|'.join(map(re.escape, html_escaping.keys()))) +# suppress some deprecation warnings in squeeze related to md5 module +import warnings +warnings.filterwarnings('ignore', \ + "the md5 module is deprecated; use hashlib instead", \ + DeprecationWarning) + +import errno +import os +import re +import sys +import md5 +import apt_pkg +import apt_inst +import shutil +import subprocess +import threading + +from daklib import utils +from daklib.config import Config +from daklib.dbconn import DBConn, get_component_by_package_suite +from daklib.gpg import SignedFile +from daklib.regexes import html_escaping, re_html_escaping, re_version, re_spacestrip, \ + re_contrib, re_nonfree, re_localhost, re_newlinespace, \ + re_package, re_doc_directory +from daklib.dak_exceptions import ChangesUnicodeError +import daklib.daksubprocess ################################################################################ Cnf = None -projectB = None - -Cnf = daklib.utils.get_conf() -projectB = pg.connect(Cnf["DB::Name"], Cnf["DB::Host"], int(Cnf["DB::Port"])) -daklib.database.init(Cnf, projectB) +Cnf = utils.get_conf() -printed_copyrights = {} +printed = threading.local() +printed.copyrights = {} +package_relations = {} #: Store relations of packages for later output # default is to not output html. -use_html = 0 +use_html = False ################################################################################ @@ -88,9 +98,6 @@ PACKAGE can be a .changes, .dsc, .deb or .udeb filename.""" ################################################################################ # probably xml.sax.saxutils would work as well -def html_escape(s): - return re_html_escaping.sub(lambda x: html_escaping.get(x.group(0)), s) - def escape_if_needed(s): if use_html: return re_html_escaping.sub(lambda x: html_escaping.get(x.group(0)), s) @@ -100,13 +107,13 @@ def escape_if_needed(s): def headline(s, level=2, bodyelement=None): if use_html: if bodyelement: - print """ + return """ %(title)s (click to toggle) - """%{"bodyelement":bodyelement,"title":html_escape(s)} + \n"""%{"bodyelement":bodyelement,"title":utils.html_escape(os.path.basename(s))} else: - print "%s" % (level, html_escape(s), level) + return "%s\n" % (level, utils.html_escape(s), level) else: - print "---- %s ----" % (s) + return "---- %s ----\n" % (s) # Colour definitions, 'end' isn't really for use @@ -114,22 +121,26 @@ ansi_colours = { 'main': "\033[36m", 'contrib': "\033[33m", 'nonfree': "\033[31m", + 'provides': "\033[35m", 'arch': "\033[32m", 'end': "\033[0m", 'bold': "\033[1m", - 'maintainer': "\033[32m"} + 'maintainer': "\033[32m", + 'distro': "\033[1m\033[41m"} html_colours = { 'main': ('',""), 'contrib': ('',""), 'nonfree': ('',""), + 'provides': ('',""), 'arch': ('',""), 'bold': ('',""), - 'maintainer': ('',"")} + 'maintainer': ('',""), + 'distro': ('',"")} def colour_output(s, colour): if use_html: - return ("%s%s%s" % (html_colours[colour][0], html_escape(s), html_colours[colour][1])) + return ("%s%s%s" % (html_colours[colour][0], utils.html_escape(s), html_colours[colour][1])) else: return ("%s%s%s" % (ansi_colours[colour], s, ansi_colours['end'])) @@ -145,7 +156,7 @@ def formatted_text(s, strip=False): if use_html: if strip: s = s.strip() - return "
%s
" % (html_escape(s)) + return "
%s
" % (utils.html_escape(s)) else: return s @@ -163,18 +174,20 @@ def format_field(k,v): def foldable_output(title, elementnameprefix, content, norow=False): d = {'elementnameprefix':elementnameprefix} + result = '' if use_html: - print """""" + result += """""" + return result ################################################################################ @@ -220,15 +233,16 @@ def split_depends (d_str) : def read_control (filename): recommends = [] + predepends = [] depends = [] section = '' maintainer = '' arch = '' - deb_file = daklib.utils.open_file(filename) + deb_file = utils.open_file(filename) try: - extracts = apt_inst.debExtractControl(deb_file) - control = apt_pkg.ParseSection(extracts) + extracts = utils.deb_extract_control(deb_file) + control = apt_pkg.TagSection(extracts) except: print formatted_text("can't parse control info") deb_file.close() @@ -238,17 +252,21 @@ def read_control (filename): control_keys = control.keys() - if control.has_key("Depends"): - depends_str = control.Find("Depends") + if "Pre-Depends" in control: + predepends_str = control["Pre-Depends"] + predepends = split_depends(predepends_str) + + if "Depends" in control: + depends_str = control["Depends"] # create list of dependancy lists depends = split_depends(depends_str) - if control.has_key("Recommends"): - recommends_str = control.Find("Recommends") + if "Recommends" in control: + recommends_str = control["Recommends"] recommends = split_depends(recommends_str) - if control.has_key("Section"): - section_str = control.Find("Section") + if "Section" in control: + section_str = control["Section"] c_match = re_contrib.search(section_str) nf_match = re_nonfree.search(section_str) @@ -261,12 +279,12 @@ def read_control (filename): else : # main section = colour_output(section_str, 'main') - if control.has_key("Architecture"): - arch_str = control.Find("Architecture") + if "Architecture" in control: + arch_str = control["Architecture"] arch = colour_output(arch_str, 'arch') - if control.has_key("Maintainer"): - maintainer = control.Find("Maintainer") + if "Maintainer" in control: + maintainer = control["Maintainer"] localhost = re_localhost.search(maintainer) if localhost: #highlight bad email @@ -274,14 +292,14 @@ def read_control (filename): else: maintainer = escape_if_needed(maintainer) - return (control, control_keys, section, depends, recommends, arch, maintainer) + return (control, control_keys, section, predepends, depends, recommends, arch, maintainer) -def read_changes_or_dsc (filename): +def read_changes_or_dsc (suite, filename, session = None): dsc = {} - dsc_file = daklib.utils.open_file(filename) + dsc_file = utils.open_file(filename) try: - dsc = daklib.utils.parse_changes(filename) + dsc = utils.parse_changes(filename, dsc_file=1) except: return formatted_text("can't parse .dsc control info") dsc_file.close() @@ -295,10 +313,13 @@ def read_changes_or_dsc (filename): for k in dsc.keys(): if k in ("build-depends","build-depends-indep"): - dsc[k] = create_depends_string(split_depends(dsc[k])) + dsc[k] = create_depends_string(suite, split_depends(dsc[k]), session) elif k == "architecture": if (dsc["architecture"] != "any"): dsc['architecture'] = colour_output(dsc["architecture"], 'arch') + elif k == "distribution": + if dsc["distribution"] not in ('unstable', 'experimental'): + dsc['distribution'] = colour_output(dsc["distribution"], 'distro') elif k in ("files","changes","description"): if use_html: dsc[k] = formatted_text(dsc[k], strip=True) @@ -312,10 +333,38 @@ def read_changes_or_dsc (filename): filecontents = '\n'.join(map(lambda x: format_field(x,dsc[x.lower()]), keysinorder))+'\n' return filecontents -def create_depends_string (depends_tree): - # just look up unstable for now. possibly pull from .changes later - suite = "unstable" +def get_provides(suite): + provides = set() + session = DBConn().session() + query = '''SELECT DISTINCT value + FROM binaries_metadata m + JOIN bin_associations b + ON b.bin = m.bin_id + WHERE key_id = ( + SELECT key_id + FROM metadata_keys + WHERE key = 'Provides' ) + AND b.suite = ( + SELECT id + FROM suite + WHERE suite_name = '%(suite)s' + OR codename = '%(suite)s')''' % \ + {'suite': suite} + for p in session.execute(query): + for e in p: + for i in e.split(','): + provides.add(i.strip()) + session.close() + return provides + +def create_depends_string (suite, depends_tree, session = None): result = "" + if suite == 'experimental': + suite_list = ['experimental','unstable'] + else: + suite_list = [suite] + + provides = set() comma_count = 1 for l in depends_tree: if (comma_count >= 2): @@ -326,18 +375,16 @@ def create_depends_string (depends_tree): result += " | " # doesn't do version lookup yet. - q = projectB.query("SELECT DISTINCT(b.package), b.version, c.name, su.suite_name FROM binaries b, files fi, location l, component c, bin_associations ba, suite su WHERE b.package='%s' AND b.file = fi.id AND fi.location = l.id AND l.component = c.id AND ba.bin=b.id AND ba.suite = su.id AND su.suite_name='%s' ORDER BY b.version desc" % (d['name'], suite)) - ql = q.getresult() - if ql: - i = ql[0] - + component = get_component_by_package_suite(d['name'], suite_list, \ + session = session) + if component is not None: adepends = d['name'] if d['version'] != '' : adepends += " (%s)" % (d['version']) - if i[2] == "contrib": + if component == "contrib": result += colour_output(adepends, "contrib") - elif i[2] == "non-free": + elif component == "non-free": result += colour_output(adepends, "nonfree") else : result += colour_output(adepends, "main") @@ -345,80 +392,169 @@ def create_depends_string (depends_tree): adepends = d['name'] if d['version'] != '' : adepends += " (%s)" % (d['version']) - result += colour_output(adepends, "bold") + if not provides: + provides = get_provides(suite) + if d['name'] in provides: + result += colour_output(adepends, "provides") + else: + result += colour_output(adepends, "bold") or_count += 1 comma_count += 1 return result -def output_deb_info(filename): - (control, control_keys, section, depends, recommends, arch, maintainer) = read_control(filename) +def output_package_relations (): + """ + Output the package relations, if there is more than one package checked in this run. + """ + + if len(package_relations) < 2: + # Only list something if we have more than one binary to compare + package_relations.clear() + return + + to_print = "" + for package in package_relations: + for relation in package_relations[package]: + to_print += "%-15s: (%s) %s\n" % (package, relation, package_relations[package][relation]) + + package_relations.clear() + return foldable_output("Package relations", "relations", to_print) + +def output_deb_info(suite, filename, packagename, session = None): + (control, control_keys, section, predepends, depends, recommends, arch, maintainer) = read_control(filename) if control == '': return formatted_text("no control info") to_print = "" + if not package_relations.has_key(packagename): + package_relations[packagename] = {} for key in control_keys : - if key == 'Depends': - field_value = create_depends_string(depends) + if key == 'Pre-Depends': + field_value = create_depends_string(suite, predepends, session) + package_relations[packagename][key] = field_value + elif key == 'Depends': + field_value = create_depends_string(suite, depends, session) + package_relations[packagename][key] = field_value elif key == 'Recommends': - field_value = create_depends_string(recommends) + field_value = create_depends_string(suite, recommends, session) + package_relations[packagename][key] = field_value elif key == 'Section': field_value = section elif key == 'Architecture': field_value = arch elif key == 'Maintainer': field_value = maintainer + elif key == 'Homepage': + field_value = escape_if_needed(control.find(key)) + if use_html: + field_value = '%s' % \ + (field_value, field_value) elif key == 'Description': - desc = control.Find(key) - desc = re_newlinespace.sub('\n ', desc) - field_value = escape_if_needed(desc) + if use_html: + field_value = formatted_text(control.find(key), strip=True) + else: + desc = control.find(key) + desc = re_newlinespace.sub('\n ', desc) + field_value = escape_if_needed(desc) else: - field_value = escape_if_needed(control.Find(key)) + field_value = escape_if_needed(control.find(key)) to_print += " "+format_field(key,field_value)+'\n' return to_print -def do_command (command, filename, escaped=0): - o = os.popen("%s %s" % (command, filename)) - if escaped: - return escaped_text(o.read()) - else: - return formatted_text(o.read()) +def do_command (command, escaped=False): + process = daklib.daksubprocess.Popen(command, stdout=subprocess.PIPE) + o = process.stdout + try: + if escaped: + return escaped_text(o.read()) + else: + return formatted_text(o.read()) + finally: + process.wait() def do_lintian (filename): + cnf = Config() + cmd = [] + + user = cnf.get('Dinstall::UnprivUser') or None + if user is not None: + cmd.extend(['sudo', '-H', '-u', user]) + + color = 'always' if use_html: - return do_command("lintian --show-overrides --color html", filename, 1) - else: - return do_command("lintian --show-overrides --color always", filename, 1) + color = 'html' + + cmd.extend(['lintian', '--show-overrides', '--color', color, "--", filename]) + + return do_command(cmd, escaped=True) def get_copyright (deb_filename): - package = re_package.sub(r'\1', deb_filename) + global printed + + package = re_package.sub(r'\1', os.path.basename(deb_filename)) o = os.popen("dpkg-deb -c %s | egrep 'usr(/share)?/doc/[^/]*/copyright' | awk '{print $6}' | head -n 1" % (deb_filename)) - copyright = o.read()[:-1] + cright = o.read()[:-1] - if copyright == "": + if cright == "": return formatted_text("WARNING: No copyright found, please check package manually.") - doc_directory = re_doc_directory.sub(r'\1', copyright) + doc_directory = re_doc_directory.sub(r'\1', cright) if package != doc_directory: return formatted_text("WARNING: wrong doc directory (expected %s, got %s)." % (package, doc_directory)) - o = os.popen("dpkg-deb --fsys-tarfile %s | tar xvOf - %s 2>/dev/null" % (deb_filename, copyright)) - copyright = o.read() - copyrightmd5 = md5.md5(copyright).hexdigest() + o = os.popen("dpkg-deb --fsys-tarfile %s | tar xvOf - %s 2>/dev/null" % (deb_filename, cright)) + cright = o.read() + copyrightmd5 = md5.md5(cright).hexdigest() res = "" - if printed_copyrights.has_key(copyrightmd5) and printed_copyrights[copyrightmd5] != "%s (%s)" % (package, deb_filename): + if printed.copyrights.has_key(copyrightmd5) and printed.copyrights[copyrightmd5] != "%s (%s)" % (package, os.path.basename(deb_filename)): res += formatted_text( "NOTE: Copyright is the same as %s.\n\n" % \ - (printed_copyrights[copyrightmd5])) + (printed.copyrights[copyrightmd5])) else: - printed_copyrights[copyrightmd5] = "%s (%s)" % (package, deb_filename) - return res+formatted_text(copyright) + printed.copyrights[copyrightmd5] = "%s (%s)" % (package, os.path.basename(deb_filename)) + return res+formatted_text(cright) + +def get_readme_source (dsc_filename): + tempdir = utils.temp_dirname() + os.rmdir(tempdir) -def check_dsc (dsc_filename): - (dsc) = read_changes_or_dsc(dsc_filename) - foldable_output(dsc_filename, "dsc", dsc, norow=True) - foldable_output("lintian check for %s" % dsc_filename, "source-lintian", do_lintian(dsc_filename)) + cmd = ('dpkg-source', '--no-check', '--no-copy', '-x', dsc_filename, tempdir) + try: + daklib.daksubprocess.check_output(cmd, stderr=1) + except subprocess.CalledProcessError as e: + res = "How is education supposed to make me feel smarter? Besides, every time I learn something new, it pushes some\n old stuff out of my brain. Remember when I took that home winemaking course, and I forgot how to drive?\n" + res += "Error, couldn't extract source, WTF?\n" + res += "'dpkg-source -x' failed. return code: %s.\n\n" % (e.returncode) + res += e.output + return res + + path = os.path.join(tempdir, 'debian/README.source') + res = "" + if os.path.exists(path): + res += do_command(["cat", "--", path]) + else: + res += "No README.source in this package\n\n" -def check_deb (deb_filename): + try: + shutil.rmtree(tempdir) + except OSError as e: + if errno.errorcode[e.errno] != 'EACCES': + res += "%s: couldn't remove tmp dir %s for source tree." % (dsc_filename, tempdir) + + return res + +def check_dsc (suite, dsc_filename, session = None): + dsc = read_changes_or_dsc(suite, dsc_filename, session) + dsc_basename = os.path.basename(dsc_filename) + return foldable_output(dsc_filename, "dsc", dsc, norow=True) + \ + "\n" + \ + foldable_output("lintian check for %s" % dsc_basename, + "source-lintian", do_lintian(dsc_filename)) + \ + "\n" + \ + foldable_output("README.source for %s" % dsc_basename, + "source-readmesource", get_readme_source(dsc_filename)) + +def check_deb (suite, deb_filename, session = None): filename = os.path.basename(deb_filename) packagename = filename.split('_')[0] @@ -427,78 +563,61 @@ def check_deb (deb_filename): else: is_a_udeb = 0 - - foldable_output("control file for %s" % (filename), "binary-%s-control"%packagename, - output_deb_info(deb_filename), norow=True) + result = foldable_output("control file for %s" % (filename), "binary-%s-control"%packagename, + output_deb_info(suite, deb_filename, packagename, session), norow=True) + "\n" if is_a_udeb: - foldable_output("skipping lintian check for udeb", "binary-%s-lintian"%packagename, - "") + result += foldable_output("skipping lintian check for udeb", + "binary-%s-lintian"%packagename, "") + "\n" else: - foldable_output("lintian check for %s" % (filename), "binary-%s-lintian"%packagename, - do_lintian(deb_filename)) + result += foldable_output("lintian check for %s" % (filename), + "binary-%s-lintian"%packagename, do_lintian(deb_filename)) + "\n" - foldable_output("contents of %s" % (filename), "binary-%s-contents"%packagename, - do_command("dpkg -c", deb_filename)) + result += foldable_output("contents of %s" % (filename), "binary-%s-contents"%packagename, + do_command(["dpkg", "-c", deb_filename])) + "\n" if is_a_udeb: - foldable_output("skipping copyright for udeb", "binary-%s-copyright"%packagename, - "") + result += foldable_output("skipping copyright for udeb", + "binary-%s-copyright"%packagename, "") + "\n" else: - foldable_output("copyright of %s" % (filename), "binary-%s-copyright"%packagename, - get_copyright(deb_filename)) + result += foldable_output("copyright of %s" % (filename), + "binary-%s-copyright"%packagename, get_copyright(deb_filename)) + "\n" - foldable_output("file listing of %s" % (filename), "binary-%s-file-listing"%packagename, - do_command("ls -l", deb_filename)) + return result # Read a file, strip the signature and return the modified contents as # a string. def strip_pgp_signature (filename): - file = daklib.utils.open_file (filename) - contents = "" - inside_signature = 0 - skip_next = 0 - for line in file.readlines(): - if line[:-1] == "": - continue - if inside_signature: - continue - if skip_next: - skip_next = 0 - continue - if line.startswith("-----BEGIN PGP SIGNED MESSAGE"): - skip_next = 1 - continue - if line.startswith("-----BEGIN PGP SIGNATURE"): - inside_signature = 1 - continue - if line.startswith("-----END PGP SIGNATURE"): - inside_signature = 0 - continue - contents += line - file.close() - return contents - -def display_changes(changes_filename): - changes = read_changes_or_dsc(changes_filename) - foldable_output(changes_filename, "changes", changes, norow=True) + with utils.open_file(filename) as f: + data = f.read() + signedfile = SignedFile(data, keyrings=(), require_signature=False) + return signedfile.contents + +def display_changes(suite, changes_filename): + global printed + changes = read_changes_or_dsc(suite, changes_filename) + printed.copyrights = {} + return foldable_output(changes_filename, "changes", changes, norow=True) def check_changes (changes_filename): - display_changes(changes_filename) - - changes = daklib.utils.parse_changes (changes_filename) - files = daklib.utils.build_file_list(changes) - for file in files.keys(): - if file.endswith(".deb") or file.endswith(".udeb"): - check_deb(file) - if file.endswith(".dsc"): - check_dsc(file) + try: + changes = utils.parse_changes (changes_filename) + except ChangesUnicodeError: + utils.warn("Encoding problem with changes file %s" % (changes_filename)) + print display_changes(changes['distribution'], changes_filename) + + files = utils.build_file_list(changes) + for f in files.keys(): + if f.endswith(".deb") or f.endswith(".udeb"): + print check_deb(changes['distribution'], f) + if f.endswith(".dsc"): + print check_dsc(changes['distribution'], f) # else: => byhand def main (): - global Cnf, projectB, db_files, waste, excluded + global Cnf, db_files, waste, excluded -# Cnf = daklib.utils.get_conf() +# Cnf = utils.get_conf() Arguments = [('h',"help","Examine-Package::Options::Help"), ('H',"html-output","Examine-Package::Options::Html-Output"), @@ -507,43 +626,53 @@ def main (): if not Cnf.has_key("Examine-Package::Options::%s" % (i)): Cnf["Examine-Package::Options::%s" % (i)] = "" - args = apt_pkg.ParseCommandLine(Cnf,Arguments,sys.argv) - Options = Cnf.SubTree("Examine-Package::Options") + args = apt_pkg.parse_commandline(Cnf,Arguments,sys.argv) + Options = Cnf.subtree("Examine-Package::Options") if Options["Help"]: usage() + if Options["Html-Output"]: + global use_html + use_html = True + stdout_fd = sys.stdout - for file in args: + for f in args: try: if not Options["Html-Output"]: # Pipe output for each argument through less - less_fd = os.popen("less -R -", 'w', 0) + less_cmd = ("less", "-R", "-") + less_process = daklib.daksubprocess.Popen(less_cmd, stdin=subprocess.PIPE, bufsize=0) + less_fd = less_process.stdin # -R added to display raw control chars for colour sys.stdout = less_fd try: - if file.endswith(".changes"): - check_changes(file) - elif file.endswith(".deb") or file.endswith(".udeb"): - check_deb(file) - elif file.endswith(".dsc"): - check_dsc(file) + if f.endswith(".changes"): + check_changes(f) + elif f.endswith(".deb") or f.endswith(".udeb"): + # default to unstable when we don't have a .changes file + # perhaps this should be a command line option? + print check_deb('unstable', f) + elif f.endswith(".dsc"): + print check_dsc('unstable', f) else: - daklib.utils.fubar("Unrecognised file type: '%s'." % (file)) + utils.fubar("Unrecognised file type: '%s'." % (f)) finally: + print output_package_relations() if not Options["Html-Output"]: # Reset stdout here so future less invocations aren't FUBAR less_fd.close() + less_process.wait() sys.stdout = stdout_fd - except IOError, e: + except IOError as e: if errno.errorcode[e.errno] == 'EPIPE': - daklib.utils.warn("[examine-package] Caught EPIPE; skipping.") + utils.warn("[examine-package] Caught EPIPE; skipping.") pass else: raise except KeyboardInterrupt: - daklib.utils.warn("[examine-package] Caught C-c; skipping.") + utils.warn("[examine-package] Caught C-c; skipping.") pass #######################################################################################