X-Git-Url: https://git.decadent.org.uk/gitweb/?a=blobdiff_plain;f=dak%2Fexamine_package.py;h=0e1b34ff23ee9b86802af1668c7d574b868e34a1;hb=4830d9be143c7645ea932b53fae095e275ad7814;hp=5080d3482b3876e6f4de6297769a83f19aa6ccd6;hpb=06b17e68fd4a76e7a12f741f26654e55bff05c79;p=dak.git diff --git a/dak/examine_package.py b/dak/examine_package.py old mode 100644 new mode 100755 index 5080d348..0e1b34ff --- 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 @@ -32,41 +42,37 @@ ################################################################################ -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 commands +import threading + +from daklib import utils +from daklib.dbconn import DBConn, get_binary_from_name_suite +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 ################################################################################ Cnf = None -projectB = None +Cnf = utils.get_conf() -Cnf = daklib.utils.get_conf() -projectB = pg.connect(Cnf["DB::Name"], Cnf["DB::Host"], int(Cnf["DB::Port"])) -daklib.database.init(Cnf, projectB) - -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 @@ -88,20 +94,22 @@ 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(html_escaping.get, s) + return re_html_escaping.sub(lambda x: html_escaping.get(x.group(0)), s) else: return s -def headline(s, level=2): +def headline(s, level=2, bodyelement=None): if use_html: - print "%s" % (level, html_escape(s), level) + if bodyelement: + return """ + %(title)s (click to toggle) + \n"""%{"bodyelement":bodyelement,"title":utils.html_escape(s)} + else: + 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 @@ -112,7 +120,8 @@ ansi_colours = { 'arch': "\033[32m", 'end': "\033[0m", 'bold': "\033[1m", - 'maintainer': "\033[32m"} + 'maintainer': "\033[32m", + 'distro': "\033[1m\033[41m"} html_colours = { 'main': ('',""), @@ -120,25 +129,59 @@ html_colours = { 'nonfree': ('',""), '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'])) -def print_escaped_text(s): +def escaped_text(s, strip=False): if use_html: - print "
%s
" % (s) + if strip: + s = s.strip() + return "
%s
" % (s) else: - print s + return s -def print_formatted_text(s): +def formatted_text(s, strip=False): if use_html: - print "
%s
" % (html_escape(s)) + if strip: + s = s.strip() + return "
%s
" % (utils.html_escape(s)) else: - print s + return s + +def output_row(s): + if use_html: + return """"""+s+"""""" + else: + return s + +def format_field(k,v): + if use_html: + return """%s:%s"""%(k,v) + else: + return "%s: %s"%(k,v) + +def foldable_output(title, elementnameprefix, content, norow=False): + d = {'elementnameprefix':elementnameprefix} + result = '' + if use_html: + result += """""" + return result ################################################################################ @@ -189,14 +232,14 @@ def read_control (filename): 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) except: - print_formatted_text("can't parse control info") - # TV-COMMENT: this will raise exceptions in two lines - control = '' + print formatted_text("can't parse control info") + deb_file.close() + raise deb_file.close() @@ -240,38 +283,52 @@ def read_control (filename): return (control, control_keys, section, depends, recommends, arch, maintainer) -def read_dsc (dsc_filename): +def read_changes_or_dsc (suite, filename, session = None): dsc = {} - dsc_file = daklib.utils.open_file(dsc_filename) + dsc_file = utils.open_file(filename) try: - dsc = daklib.utils.parse_changes(dsc_filename) + dsc = utils.parse_changes(filename, dsc_file=1) except: - print_formatted_text("can't parse control info") + return formatted_text("can't parse .dsc control info") dsc_file.close() - filecontents = escape_if_needed(strip_pgp_signature(dsc_filename)) - - if dsc.has_key("build-depends"): - builddep = split_depends(dsc["build-depends"]) - builddepstr = create_depends_string(builddep) - filecontents = re_builddep.sub("Build-Depends: "+builddepstr, filecontents) - - if dsc.has_key("build-depends-indep"): - builddepindstr = create_depends_string(split_depends(dsc["build-depends-indep"])) - filecontents = re_builddepind.sub("Build-Depends-Indep: "+builddepindstr, filecontents) + filecontents = strip_pgp_signature(filename) + keysinorder = [] + for l in filecontents.split('\n'): + m = re.match(r'([-a-zA-Z0-9]*):', l) + if m: + keysinorder.append(m.group(1)) + + for k in dsc.keys(): + if k in ("build-depends","build-depends-indep"): + 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) + else: + dsc[k] = ('\n'+'\n'.join(map(lambda x: ' '+x, dsc[k].split('\n')))).rstrip() + else: + dsc[k] = escape_if_needed(dsc[k]) - if dsc.has_key("architecture") : - if (dsc["architecture"] != "any"): - newarch = colour_output(dsc["architecture"], 'arch') - filecontents = re_arch.sub("Architecture: " + newarch, filecontents) + keysinorder = filter(lambda x: not x.lower().startswith('checksums-'), keysinorder) + 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 create_depends_string (suite, depends_tree, session = None): result = "" + if suite == 'experimental': + suite_where = "in ('experimental','unstable')" + else: + suite_where = "= '%s'" % suite + comma_count = 1 for l in depends_tree: if (comma_count >= 2): @@ -282,10 +339,9 @@ 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] + res = get_binary_from_name_suite(d['name'], suite_where, session) + if res.rowcount > 0: + i = res.fetchone() adepends = d['name'] if d['version'] != '' : @@ -306,119 +362,176 @@ def create_depends_string (depends_tree): 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, depends, recommends, arch, maintainer) = read_control(filename) + if control == '': - print_formatted_text("no control info") - else: - for key in control_keys : - output = " " + key + ": " - if key == 'Depends': - output += create_depends_string(depends) - elif key == 'Recommends': - output += create_depends_string(recommends) - elif key == 'Section': - output += section - elif key == 'Architecture': - output += arch - elif key == 'Maintainer': - output += maintainer - elif key == 'Description': + 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(suite, depends, session) + package_relations[packagename][key] = field_value + elif key == '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 == 'Description': + if use_html: + field_value = formatted_text(control.Find(key), strip=True) + else: desc = control.Find(key) desc = re_newlinespace.sub('\n ', desc) - output += escape_if_needed(desc) - else: - output += escape_if_needed(control.Find(key)) - to_print += output + '\n' - print_escaped_text(to_print) + field_value = escape_if_needed(desc) + else: + 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: - print_escaped_text(o.read()) + return escaped_text(o.read()) else: - print_formatted_text(o.read()) + return formatted_text(o.read()) def do_lintian (filename): if use_html: - do_command("lintian --show-overrides --color html", filename, 1) + return do_command("lintian --show-overrides --color html", filename, 1) else: - do_command("lintian --show-overrides --color always", filename, 1) + return do_command("lintian --show-overrides --color always", filename, 1) + +def get_copyright (deb_filename): + global printed -def print_copyright (deb_filename): package = re_package.sub(r'\1', 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 == "": - print_formatted_text("WARNING: No copyright found, please check package manually.") - return + 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: - print_formatted_text("WARNING: wrong doc directory (expected %s, got %s)." % (package, doc_directory)) - return + 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() - if printed_copyrights.has_key(copyrightmd5) and printed_copyrights[copyrightmd5] != "%s (%s)" % (package, deb_filename): - print_formatted_text( "NOTE: Copyright is the same as %s.\n" % \ - (printed_copyrights[copyrightmd5])) + res = "" + if printed.copyrights.has_key(copyrightmd5) and printed.copyrights[copyrightmd5] != "%s (%s)" % (package, deb_filename): + res += formatted_text( "NOTE: Copyright is the same as %s.\n\n" % \ + (printed.copyrights[copyrightmd5])) else: - printed_copyrights[copyrightmd5] = "%s (%s)" % (package, deb_filename) - - print_formatted_text(copyright) - -def check_dsc (dsc_filename): - headline(".dsc file for %s" % (dsc_filename)) - (dsc) = read_dsc(dsc_filename) - print_escaped_text(dsc) - headline("lintian check for %s" % (dsc_filename)) - do_lintian(dsc_filename) + printed.copyrights[copyrightmd5] = "%s (%s)" % (package, deb_filename) + return res+formatted_text(cright) + +def get_readme_source (dsc_filename): + tempdir = utils.temp_dirname() + os.rmdir(tempdir) + + cmd = "dpkg-source --no-check --no-copy -x %s %s" % (dsc_filename, tempdir) + (result, output) = commands.getstatusoutput(cmd) + if (result != 0): + 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" % (result) + res += 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, 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) + return foldable_output(dsc_filename, "dsc", dsc, norow=True) + \ + "\n" + \ + foldable_output("lintian check for %s" % dsc_filename, + "source-lintian", do_lintian(dsc_filename)) + \ + "\n" + \ + foldable_output("README.source for %s" % dsc_filename, + "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] if filename.endswith(".udeb"): is_a_udeb = 1 else: is_a_udeb = 0 - headline("control file for %s" % (filename)) - #do_command ("dpkg -I", deb_filename) - output_deb_info(deb_filename) + 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: - headline("skipping lintian check for udeb") - print + result += foldable_output("skipping lintian check for udeb", + "binary-%s-lintian"%packagename, "") + "\n" else: - headline("lintian check for %s" % (filename)) - do_lintian(deb_filename) + result += foldable_output("lintian check for %s" % (filename), + "binary-%s-lintian"%packagename, do_lintian(deb_filename)) + "\n" - headline("contents of %s" % (filename)) - 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: - headline("skipping copyright for udeb") + result += foldable_output("skipping copyright for udeb", + "binary-%s-copyright"%packagename, "") + "\n" else: - headline("copyright of %s" % (filename)) - print_copyright(deb_filename) + result += foldable_output("copyright of %s" % (filename), + "binary-%s-copyright"%packagename, get_copyright(deb_filename)) + "\n" - headline("file listing of %s" % (filename)) - do_command ("ls -l", deb_filename) + result += 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) + inputfile = utils.open_file (filename) contents = "" inside_signature = 0 skip_next = 0 - for line in file.readlines(): + for line in inputfile.readlines(): if line[:-1] == "": continue if inside_signature: @@ -436,30 +549,34 @@ def strip_pgp_signature (filename): inside_signature = 0 continue contents += line - file.close() + inputfile.close() return contents -# Display the .changes [without the signature] -def display_changes (changes_filename): - headline(".changes file for %s" % (changes_filename)) - print_formatted_text(strip_pgp_signature(changes_filename)) +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"), @@ -474,9 +591,13 @@ def main (): if Options["Help"]: usage() + if Options["Html-Output"]: + global use_html + use_html = 1 + 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 @@ -484,27 +605,30 @@ def main (): # -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() sys.stdout = stdout_fd except IOError, 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 #######################################################################################