X-Git-Url: https://git.decadent.org.uk/gitweb/?a=blobdiff_plain;f=daklib%2Futils.py;h=52b902f9ffe84e298fce9fb4c127b935b5376f6a;hb=df1bf169c5b89fa9764a326bbd7a6883a8789f6b;hp=c4c13669d92ca3a2aca761c2124e800ecf55e95a;hpb=9bc8cd8151102e782fdd27e4f383575c6fed7298;p=dak.git diff --git a/daklib/utils.py b/daklib/utils.py old mode 100644 new mode 100755 index c4c13669..52b902f9 --- a/daklib/utils.py +++ b/daklib/utils.py @@ -1,8 +1,8 @@ #!/usr/bin/env python +# vim:set et ts=4 sw=4: # Utility functions # Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005, 2006 James Troup -# $Id: utils.py,v 1.73 2005-03-18 05:24:38 troup Exp $ ################################################################################ @@ -23,9 +23,11 @@ ################################################################################ import codecs, commands, email.Header, os, pwd, re, select, socket, shutil, \ - string, sys, tempfile, traceback + sys, tempfile, traceback, stat import apt_pkg import database +import time +from dak_exceptions import * ################################################################################ @@ -42,45 +44,38 @@ re_multi_line_field = re.compile(r"^\s(.*)") re_taint_free = re.compile(r"^[-+~/\.\w]+$") re_parse_maintainer = re.compile(r"^\s*(\S.*\S)\s*\<([^\>]+)\>") +re_gpg_uid = re.compile('^uid.*<([^>]*)>') -changes_parse_error_exc = "Can't parse line in .changes file" -invalid_dsc_format_exc = "Invalid .dsc file" -nk_format_exc = "Unknown Format: in .changes file" -no_files_exc = "No Files: field in .dsc or .changes file." -cant_open_exc = "Can't open file" -unknown_hostname_exc = "Unknown hostname" -cant_overwrite_exc = "Permission denied; can't overwrite existent file." -file_exists_exc = "Destination file exists" -sendmail_failed_exc = "Sendmail invocation failed" -tried_too_hard_exc = "Tried too hard to find a free filename." +re_srchasver = re.compile(r"^(\S+)\s+\((\S+)\)$") +re_verwithext = re.compile(r"^(\d+)(?:\.(\d+))(?:\s+\((\S+)\))?$") + +re_srchasver = re.compile(r"^(\S+)\s+\((\S+)\)$") + +html_escaping = {'"':'"', '&':'&', '<':'<', '>':'>'} +re_html_escaping = re.compile('|'.join(map(re.escape, html_escaping.keys()))) default_config = "/etc/dak/dak.conf" default_apt_config = "/etc/dak/apt.conf" -################################################################################ - -class Error(Exception): - """Base class for exceptions in this module.""" - pass +alias_cache = None +key_uid_email_cache = {} -class ParseMaintError(Error): - """Exception raised for errors in parsing a maintainer field. +# (hashname, function, earliest_changes_version) +known_hashes = [("sha1", apt_pkg.sha1sum, (1, 8)), + ("sha256", apt_pkg.sha256sum, (1, 8))] - Attributes: - message -- explanation of the error - """ +################################################################################ - def __init__(self, message): - self.args = message, - self.message = message +def html_escape(s): + return re_html_escaping.sub(lambda x: html_escaping.get(x.group(0)), s) ################################################################################ def open_file(filename, mode='r'): try: - f = open(filename, mode) + f = open(filename, mode) except IOError: - raise cant_open_exc, filename + raise CantOpenError, filename return f ################################################################################ @@ -98,30 +93,11 @@ def our_raw_input(prompt=""): ################################################################################ -def str_isnum (s): - for c in s: - if c not in string.digits: - return 0 - return 1 - -################################################################################ - def extract_component_from_section(section): component = "" if section.find('/') != -1: component = section.split('/')[0] - if component.lower() == "non-us" and section.find('/') != -1: - s = component + '/' + section.split('/')[1] - if Cnf.has_key("Component::%s" % s): # Avoid e.g. non-US/libs - component = s - - if section.lower() == "non-us": - component = "non-US/main" - - # non-US prefix is case insensitive - if component.lower()[:6] == "non-us": - component = "non-US"+component[6:] # Expand default component if component == "": @@ -129,42 +105,20 @@ def extract_component_from_section(section): component = section else: component = "main" - elif component == "non-US": - component = "non-US/main" return (section, component) ################################################################################ -def parse_changes(filename, signing_rules=0): - """Parses a changes file and returns a dictionary where each field is a -key. The mandatory first argument is the filename of the .changes -file. - -signing_rules is an optional argument: - - o If signing_rules == -1, no signature is required. - o If signing_rules == 0 (the default), a signature is required. - o If signing_rules == 1, it turns on the same strict format checking - as dpkg-source. - -The rules for (signing_rules == 1)-mode are: - - o The PGP header consists of "-----BEGIN PGP SIGNED MESSAGE-----" - followed by any PGP header data and must end with a blank line. - - o The data section must end with a blank line and must be followed by - "-----BEGIN PGP SIGNATURE-----". -""" - +def parse_deb822(contents, signing_rules=0): error = "" changes = {} - changes_in = open_file(filename) - lines = changes_in.readlines() + # Split the lines in the input, keeping the linebreaks. + lines = contents.splitlines(True) - if not lines: - raise changes_parse_error_exc, "[Empty changes file]" + if len(lines) == 0: + raise ParseChangesError, "[Empty changes file]" # Reindex by line number so we can easily verify the format of # .dsc files... @@ -186,10 +140,10 @@ The rules for (signing_rules == 1)-mode are: if signing_rules == 1: index += 1 if index > num_of_lines: - raise invalid_dsc_format_exc, index + raise InvalidDscError, index line = indexed_lines[index] if not line.startswith("-----BEGIN PGP SIGNATURE"): - raise invalid_dsc_format_exc, index + raise InvalidDscError, index inside_signature = 0 break else: @@ -210,7 +164,7 @@ The rules for (signing_rules == 1)-mode are: if slf: field = slf.groups()[0].lower() changes[field] = slf.groups()[1] - first = 1 + first = 1 continue if line == " .": changes[field] += '\n' @@ -218,56 +172,305 @@ The rules for (signing_rules == 1)-mode are: mlf = re_multi_line_field.match(line) if mlf: if first == -1: - raise changes_parse_error_exc, "'%s'\n [Multi-line field continuing on from nothing?]" % (line) + raise ParseChangesError, "'%s'\n [Multi-line field continuing on from nothing?]" % (line) if first == 1 and changes[field] != "": changes[field] += '\n' first = 0 - changes[field] += mlf.groups()[0] + '\n' + changes[field] += mlf.groups()[0] + '\n' continue - error += line + error += line if signing_rules == 1 and inside_signature: - raise invalid_dsc_format_exc, index + raise InvalidDscError, index - changes_in.close() changes["filecontents"] = "".join(lines) + if changes.has_key("source"): + # Strip the source version in brackets from the source field, + # put it in the "source-version" field instead. + srcver = re_srchasver.search(changes["source"]) + if srcver: + changes["source"] = srcver.group(1) + changes["source-version"] = srcver.group(2) + if error: - raise changes_parse_error_exc, error + raise ParseChangesError, error return changes ################################################################################ +def parse_changes(filename, signing_rules=0): + """Parses a changes file and returns a dictionary where each field is a +key. The mandatory first argument is the filename of the .changes +file. + +signing_rules is an optional argument: + + o If signing_rules == -1, no signature is required. + o If signing_rules == 0 (the default), a signature is required. + o If signing_rules == 1, it turns on the same strict format checking + as dpkg-source. + +The rules for (signing_rules == 1)-mode are: + + o The PGP header consists of "-----BEGIN PGP SIGNED MESSAGE-----" + followed by any PGP header data and must end with a blank line. + + o The data section must end with a blank line and must be followed by + "-----BEGIN PGP SIGNATURE-----". +""" + + changes_in = open_file(filename) + content = changes_in.read() + changes_in.close() + return parse_deb822(content, signing_rules) + +################################################################################ + +def hash_key(hashname): + return '%ssum' % hashname + +################################################################################ + +def create_hash(where, files, hashname, hashfunc): + """create_hash extends the passed files dict with the given hash by + iterating over all files on disk and passing them to the hashing + function given.""" + + rejmsg = [] + for f in files.keys(): + try: + file_handle = open_file(f) + except CantOpenError: + rejmsg.append("Could not open file %s for checksumming" % (f)) + + files[f][hash_key(hashname)] = hashfunc(file_handle) + + file_handle.close() + return rejmsg + +################################################################################ + +def check_hash(where, files, hashname, hashfunc): + """check_hash checks the given hash in the files dict against the actual + files on disk. The hash values need to be present consistently in + all file entries. It does not modify its input in any way.""" + + rejmsg = [] + for f in files.keys(): + file_handle = None + try: + try: + file_handle = open_file(f) + + # Check for the hash entry, to not trigger a KeyError. + if not files[f].has_key(hash_key(hashname)): + rejmsg.append("%s: misses %s checksum in %s" % (f, hashname, + where)) + continue + + # Actually check the hash for correctness. + if hashfunc(file_handle) != files[f][hash_key(hashname)]: + rejmsg.append("%s: %s check failed in %s" % (f, hashname, + where)) + except CantOpenError: + # TODO: This happens when the file is in the pool. + # warn("Cannot open file %s" % f) + continue + finally: + if file_handle: + file_handle.close() + return rejmsg + +################################################################################ + +def check_size(where, files): + """check_size checks the file sizes in the passed files dict against the + files on disk.""" + + rejmsg = [] + for f in files.keys(): + try: + entry = os.stat(f) + except OSError, exc: + if exc.errno == 2: + # TODO: This happens when the file is in the pool. + continue + raise + + actual_size = entry[stat.ST_SIZE] + size = int(files[f]["size"]) + if size != actual_size: + rejmsg.append("%s: actual file size (%s) does not match size (%s) in %s" + % (f, actual_size, size, where)) + return rejmsg + +################################################################################ + +def check_hash_fields(what, manifest): + """check_hash_fields ensures that there are no checksum fields in the + given dict that we do not know about.""" + + rejmsg = [] + hashes = map(lambda x: x[0], known_hashes) + for field in manifest: + if field.startswith("checksums-"): + hashname = field.split("-",1)[1] + if hashname not in hashes: + rejmsg.append("Unsupported checksum field for %s "\ + "in %s" % (hashname, what)) + return rejmsg + +################################################################################ + +def _ensure_changes_hash(changes, format, version, files, hashname, hashfunc): + if format >= version: + # The version should contain the specified hash. + func = check_hash + + # Import hashes from the changes + rejmsg = parse_checksums(".changes", files, changes, hashname) + if len(rejmsg) > 0: + return rejmsg + else: + # We need to calculate the hash because it can't possibly + # be in the file. + func = create_hash + return func(".changes", files, hashname, hashfunc) + +# We could add the orig which might be in the pool to the files dict to +# access the checksums easily. + +def _ensure_dsc_hash(dsc, dsc_files, hashname, hashfunc): + """ensure_dsc_hashes' task is to ensure that each and every *present* hash + in the dsc is correct, i.e. identical to the changes file and if necessary + the pool. The latter task is delegated to check_hash.""" + + rejmsg = [] + if not dsc.has_key('Checksums-%s' % (hashname,)): + return rejmsg + # Import hashes from the dsc + parse_checksums(".dsc", dsc_files, dsc, hashname) + # And check it... + rejmsg.extend(check_hash(".dsc", dsc_files, hashname, hashfunc)) + return rejmsg + +################################################################################ + +def ensure_hashes(changes, dsc, files, dsc_files): + rejmsg = [] + + # Make sure we recognise the format of the Files: field in the .changes + format = changes.get("format", "0.0").split(".", 1) + if len(format) == 2: + format = int(format[0]), int(format[1]) + else: + format = int(float(format[0])), 0 + + # We need to deal with the original changes blob, as the fields we need + # might not be in the changes dict serialised into the .dak anymore. + orig_changes = parse_deb822(changes['filecontents']) + + # Copy the checksums over to the current changes dict. This will keep + # the existing modifications to it intact. + for field in orig_changes: + if field.startswith('checksums-'): + changes[field] = orig_changes[field] + + # Check for unsupported hashes + rejmsg.extend(check_hash_fields(".changes", changes)) + rejmsg.extend(check_hash_fields(".dsc", dsc)) + + # We have to calculate the hash if we have an earlier changes version than + # the hash appears in rather than require it exist in the changes file + for hashname, hashfunc, version in known_hashes: + rejmsg.extend(_ensure_changes_hash(changes, format, version, files, + hashname, hashfunc)) + if "source" in changes["architecture"]: + rejmsg.extend(_ensure_dsc_hash(dsc, dsc_files, hashname, + hashfunc)) + + return rejmsg + +def parse_checksums(where, files, manifest, hashname): + rejmsg = [] + field = 'checksums-%s' % hashname + if not field in manifest: + return rejmsg + input = manifest[field] + for line in input.split('\n'): + if not line: + break + hash, size, file = line.strip().split(' ') + if not files.has_key(file): + # TODO: check for the file's entry in the original files dict, not + # the one modified by (auto)byhand and other weird stuff + # rejmsg.append("%s: not present in files but in checksums-%s in %s" % + # (file, hashname, where)) + continue + if not files[file]["size"] == size: + rejmsg.append("%s: size differs for files and checksums-%s entry "\ + "in %s" % (file, hashname, where)) + continue + files[file][hash_key(hashname)] = hash + for f in files.keys(): + if not files[f].has_key(hash_key(hashname)): + rejmsg.append("%s: no entry in checksums-%s in %s" % (file, + hashname, where)) + return rejmsg + +################################################################################ + # Dropped support for 1.4 and ``buggy dchanges 3.4'' (?!) compared to di.pl -def build_file_list(changes, is_a_dsc=0): +def build_file_list(changes, is_a_dsc=0, field="files", hashname="md5sum"): files = {} # Make sure we have a Files: field to parse... - if not changes.has_key("files"): - raise no_files_exc + if not changes.has_key(field): + raise NoFilesFieldError # Make sure we recognise the format of the Files: field - format = changes.get("format", "") - if format != "": - format = float(format) - if not is_a_dsc and (format < 1.5 or format > 2.0): - raise nk_format_exc, format + format = re_verwithext.search(changes.get("format", "0.0")) + if not format: + raise UnknownFormatError, "%s" % (changes.get("format","0.0")) + + format = format.groups() + if format[1] == None: + format = int(float(format[0])), 0, format[2] + else: + format = int(format[0]), int(format[1]), format[2] + if format[2] == None: + format = format[:2] + + if is_a_dsc: + # format = (1,0) are the only formats we currently accept, + # format = (0,0) are missing format headers of which we still + # have some in the archive. + if format != (1,0) and format != (0,0): + raise UnknownFormatError, "%s" % (changes.get("format","0.0")) + else: + if (format < (1,5) or format > (1,8)): + raise UnknownFormatError, "%s" % (changes.get("format","0.0")) + if field != "files" and format < (1,8): + raise UnknownFormatError, "%s" % (changes.get("format","0.0")) + + includes_section = (not is_a_dsc) and field == "files" # Parse each entry/line: - for i in changes["files"].split('\n'): + for i in changes[field].split('\n'): if not i: break s = i.split() section = priority = "" try: - if is_a_dsc: - (md5, size, name) = s - else: + if includes_section: (md5, size, section, priority, name) = s + else: + (md5, size, name) = s except ValueError: - raise changes_parse_error_exc, i + raise ParseChangesError, i if section == "": section = "-" @@ -276,8 +479,9 @@ def build_file_list(changes, is_a_dsc=0): (section, component) = extract_component_from_section(section) - files[name] = Dict(md5sum=md5, size=size, section=section, + files[name] = Dict(size=size, section=section, priority=priority, component=component) + files[name][hashname] = md5 return files @@ -364,48 +568,46 @@ switched to 'email (name)' format.""" # sendmail wrapper, takes _either_ a message string or a file as arguments def send_mail (message, filename=""): - # If we've been passed a string dump it into a temporary file - if message: - filename = tempfile.mktemp() - fd = os.open(filename, os.O_RDWR|os.O_CREAT|os.O_EXCL, 0700) - os.write (fd, message) - os.close (fd) - - # Invoke sendmail - (result, output) = commands.getstatusoutput("%s < %s" % (Cnf["Dinstall::SendmailCommand"], filename)) - if (result != 0): - raise sendmail_failed_exc, output - - # Clean up any temporary files - if message: - os.unlink (filename) + # If we've been passed a string dump it into a temporary file + if message: + filename = tempfile.mktemp() + fd = os.open(filename, os.O_RDWR|os.O_CREAT|os.O_EXCL, 0700) + os.write (fd, message) + os.close (fd) + + # Invoke sendmail + (result, output) = commands.getstatusoutput("%s < %s" % (Cnf["Dinstall::SendmailCommand"], filename)) + if (result != 0): + raise SendmailFailedError, output + + # Clean up any temporary files + if message: + os.unlink (filename) ################################################################################ def poolify (source, component): if component: - component += '/' - # FIXME: this is nasty - component = component.lower().replace("non-us/", "non-US/") + component += '/' if source[:3] == "lib": - return component + source[:4] + '/' + source + '/' + return component + source[:4] + '/' + source + '/' else: - return component + source[:1] + '/' + source + '/' + return component + source[:1] + '/' + source + '/' ################################################################################ def move (src, dest, overwrite = 0, perms = 0664): if os.path.exists(dest) and os.path.isdir(dest): - dest_dir = dest + dest_dir = dest else: - dest_dir = os.path.dirname(dest) + dest_dir = os.path.dirname(dest) if not os.path.exists(dest_dir): - umask = os.umask(00000) - os.makedirs(dest_dir, 02775) - os.umask(umask) + umask = os.umask(00000) + os.makedirs(dest_dir, 02775) + os.umask(umask) #print "Moving %s to %s..." % (src, dest) if os.path.exists(dest) and os.path.isdir(dest): - dest += '/' + os.path.basename(src) + dest += '/' + os.path.basename(src) # Don't overwrite unless forced to if os.path.exists(dest): if not overwrite: @@ -419,23 +621,23 @@ def move (src, dest, overwrite = 0, perms = 0664): def copy (src, dest, overwrite = 0, perms = 0664): if os.path.exists(dest) and os.path.isdir(dest): - dest_dir = dest + dest_dir = dest else: - dest_dir = os.path.dirname(dest) + dest_dir = os.path.dirname(dest) if not os.path.exists(dest_dir): - umask = os.umask(00000) - os.makedirs(dest_dir, 02775) - os.umask(umask) + umask = os.umask(00000) + os.makedirs(dest_dir, 02775) + os.umask(umask) #print "Copying %s to %s..." % (src, dest) if os.path.exists(dest) and os.path.isdir(dest): - dest += '/' + os.path.basename(src) + dest += '/' + os.path.basename(src) # Don't overwrite unless forced to if os.path.exists(dest): if not overwrite: - raise file_exists_exc + raise FileExistsError else: if not os.access(dest, os.W_OK): - raise cant_overwrite_exc + raise CantOverwriteError shutil.copy2(src, dest) os.chmod(dest, perms) @@ -445,23 +647,31 @@ def where_am_i (): res = socket.gethostbyaddr(socket.gethostname()) database_hostname = Cnf.get("Config::" + res[0] + "::DatabaseHostname") if database_hostname: - return database_hostname + return database_hostname else: return res[0] def which_conf_file (): res = socket.gethostbyaddr(socket.gethostname()) if Cnf.get("Config::" + res[0] + "::DakConfig"): - return Cnf["Config::" + res[0] + "::DakConfig"] + return Cnf["Config::" + res[0] + "::DakConfig"] else: - return default_config + return default_config def which_apt_conf_file (): res = socket.gethostbyaddr(socket.gethostname()) if Cnf.get("Config::" + res[0] + "::AptConfig"): - return Cnf["Config::" + res[0] + "::AptConfig"] + return Cnf["Config::" + res[0] + "::AptConfig"] + else: + return default_apt_config + +def which_alias_file(): + hostname = socket.gethostbyaddr(socket.gethostname())[0] + aliasfn = '/var/lib/misc/'+hostname+'/forward-alias' + if os.path.exists(aliasfn): + return aliasfn else: - return default_apt_config + return None ################################################################################ @@ -571,7 +781,7 @@ def find_next_free (dest, too_many=100): dest = orig_dest + '.' + repr(extra) extra += 1 if extra >= too_many: - raise tried_too_hard_exc + raise NoFreeFilenameError return dest ################################################################################ @@ -617,7 +827,7 @@ argument: orig_filename = filename if filename.endswith(".dak"): - filename = filename[:-6]+".changes" + filename = filename[:-4]+".changes" if not filename.endswith(".changes"): error = "invalid file type; not a changes file" @@ -647,9 +857,9 @@ def real_arch(arch): ################################################################################ def join_with_commas_and(list): - if len(list) == 0: return "nothing" - if len(list) == 1: return list[0] - return ", ".join(list[:-1]) + " and " + list[-1] + if len(list) == 0: return "nothing" + if len(list) == 1: return list[0] + return ", ".join(list[:-1]) + " and " + list[-1] ################################################################################ @@ -667,7 +877,7 @@ def pp_deps (deps): ################################################################################ def get_conf(): - return Cnf + return Cnf ################################################################################ @@ -683,7 +893,7 @@ def parse_args(Options): else: suite_ids_list.append(suite_id) if suite_ids_list: - con_suites = "AND su.id IN (%s)" % ", ".join(map(str, suite_ids_list)) + con_suites = "AND su.id IN (%s)" % ", ".join([ str(i) for i in suite_ids_list ]) else: fubar("No valid suite given.") else: @@ -699,7 +909,7 @@ def parse_args(Options): else: component_ids_list.append(component_id) if component_ids_list: - con_components = "AND c.id IN (%s)" % ", ".join(map(str, component_ids_list)) + con_components = "AND c.id IN (%s)" % ", ".join([ str(i) for i in component_ids_list ]) else: fubar("No valid component given.") else: @@ -720,7 +930,7 @@ def parse_args(Options): else: arch_ids_list.append(architecture_id) if arch_ids_list: - con_architectures = "AND a.id IN (%s)" % ", ".join(map(str, arch_ids_list)) + con_architectures = "AND a.id IN (%s)" % ", ".join([ str(i) for i in arch_ids_list ]) else: if not check_source: fubar("No valid architecture given.") @@ -865,10 +1075,88 @@ def gpgv_get_status_output(cmd, status_read, status_write): return output, status, exit_status -############################################################ +################################################################################ + +def process_gpgv_output(status): + # Process the status-fd output + keywords = {} + internal_error = "" + for line in status.split('\n'): + line = line.strip() + if line == "": + continue + split = line.split() + if len(split) < 2: + internal_error += "gpgv status line is malformed (< 2 atoms) ['%s'].\n" % (line) + continue + (gnupg, keyword) = split[:2] + if gnupg != "[GNUPG:]": + internal_error += "gpgv status line is malformed (incorrect prefix '%s').\n" % (gnupg) + continue + args = split[2:] + if keywords.has_key(keyword) and keyword not in [ "NODATA", "SIGEXPIRED", "KEYEXPIRED" ]: + internal_error += "found duplicate status token ('%s').\n" % (keyword) + continue + else: + keywords[keyword] = args + + return (keywords, internal_error) + +################################################################################ + +def retrieve_key (filename, keyserver=None, keyring=None): + """Retrieve the key that signed 'filename' from 'keyserver' and +add it to 'keyring'. Returns nothing on success, or an error message +on error.""" + + # Defaults for keyserver and keyring + if not keyserver: + keyserver = Cnf["Dinstall::KeyServer"] + if not keyring: + keyring = Cnf.ValueList("Dinstall::GPGKeyring")[0] + + # Ensure the filename contains no shell meta-characters or other badness + if not re_taint_free.match(filename): + return "%s: tainted filename" % (filename) + + # Invoke gpgv on the file + status_read, status_write = os.pipe(); + cmd = "gpgv --status-fd %s --keyring /dev/null %s" % (status_write, filename) + (_, status, _) = gpgv_get_status_output(cmd, status_read, status_write) + + # Process the status-fd output + (keywords, internal_error) = process_gpgv_output(status) + if internal_error: + return internal_error + + if not keywords.has_key("NO_PUBKEY"): + return "didn't find expected NO_PUBKEY in gpgv status-fd output" + + fingerprint = keywords["NO_PUBKEY"][0] + # XXX - gpg sucks. You can't use --secret-keyring=/dev/null as + # it'll try to create a lockfile in /dev. A better solution might + # be a tempfile or something. + cmd = "gpg --no-default-keyring --secret-keyring=%s --no-options" \ + % (Cnf["Dinstall::SigningKeyring"]) + cmd += " --keyring %s --keyserver %s --recv-key %s" \ + % (keyring, keyserver, fingerprint) + (result, output) = commands.getstatusoutput(cmd) + if (result != 0): + return "'%s' failed with exit code %s" % (cmd, result) + + return "" + +################################################################################ + +def gpg_keyring_args(keyrings=None): + if not keyrings: + keyrings = Cnf.ValueList("Dinstall::GPGKeyring") + + return " ".join(["--keyring %s" % x for x in keyrings]) +################################################################################ -def check_signature (sig_filename, reject, data_filename="", keyrings=None): +def check_signature (sig_filename, reject, data_filename="", keyrings=None, autofetch=None): """Check the signature of a file and return the fingerprint if the signature is valid or 'None' if it's not. The first argument is the filename whose signature should be checked. The second argument is a @@ -878,8 +1166,9 @@ the second is an optional prefix string. It's possible for reject() to be called more than once during an invocation of check_signature(). The third argument is optional and is the name of the files the detached signature applies to. The fourth argument is optional and is -a *list* of keyrings to use. -""" +a *list* of keyrings to use. 'autofetch' can either be None, True or +False. If None, the default behaviour specified in the config will be +used.""" # Ensure the filename contains no shell meta-characters or other badness if not re_taint_free.match(sig_filename): @@ -891,38 +1180,27 @@ a *list* of keyrings to use. return None if not keyrings: - keyrings = (Cnf["Dinstall::PGPKeyring"], Cnf["Dinstall::GPGKeyring"]) + keyrings = Cnf.ValueList("Dinstall::GPGKeyring") + + # Autofetch the signing key if that's enabled + if autofetch == None: + autofetch = Cnf.get("Dinstall::KeyAutoFetch") + if autofetch: + error_msg = retrieve_key(sig_filename) + if error_msg: + reject(error_msg) + return None # Build the command line - status_read, status_write = os.pipe(); - cmd = "gpgv --status-fd %s" % (status_write) - for keyring in keyrings: - cmd += " --keyring %s" % (keyring) - cmd += " %s %s" % (sig_filename, data_filename) + status_read, status_write = os.pipe(); + cmd = "gpgv --status-fd %s %s %s %s" % ( + status_write, gpg_keyring_args(keyrings), sig_filename, data_filename) + # Invoke gpgv on the file (output, status, exit_status) = gpgv_get_status_output(cmd, status_read, status_write) # Process the status-fd output - keywords = {} - bad = internal_error = "" - for line in status.split('\n'): - line = line.strip() - if line == "": - continue - split = line.split() - if len(split) < 2: - internal_error += "gpgv status line is malformed (< 2 atoms) ['%s'].\n" % (line) - continue - (gnupg, keyword) = split[:2] - if gnupg != "[GNUPG:]": - internal_error += "gpgv status line is malformed (incorrect prefix '%s').\n" % (gnupg) - continue - args = split[2:] - if keywords.has_key(keyword) and (keyword != "NODATA" and keyword != "SIGEXPIRED"): - internal_error += "found duplicate status token ('%s').\n" % (keyword) - continue - else: - keywords[keyword] = args + (keywords, internal_error) = process_gpgv_output(status) # If we failed to parse the status-fd output, let's just whine and bail now if internal_error: @@ -931,10 +1209,8 @@ a *list* of keyrings to use. reject("Please report the above errors to the Archive maintainers by replying to this mail.", "") return None + bad = "" # Now check for obviously bad things in the processed output - if keywords.has_key("SIGEXPIRED"): - reject("The key used to sign %s has expired." % (sig_filename)) - bad = 1 if keywords.has_key("KEYREVOKED"): reject("The key used to sign %s has been revoked." % (sig_filename)) bad = 1 @@ -956,6 +1232,23 @@ a *list* of keyrings to use. if keywords.has_key("NODATA"): reject("no signature found in %s." % (sig_filename)) bad = 1 + if keywords.has_key("EXPKEYSIG"): + args = keywords["EXPKEYSIG"] + if len(args) >= 1: + key = args[0] + reject("Signature made by expired key 0x%s" % (key)) + bad = 1 + if keywords.has_key("KEYEXPIRED") and not keywords.has_key("GOODSIG"): + args = keywords["KEYEXPIRED"] + expiredate="" + if len(args) >= 1: + timestamp = args[0] + if timestamp.count("T") == 0: + expiredate = time.strftime("%Y-%m-%d", time.gmtime(timestamp)) + else: + expiredate = timestamp + reject("The key used to sign %s has expired on %s" % (sig_filename, expiredate)) + bad = 1 if bad: return None @@ -990,7 +1283,7 @@ a *list* of keyrings to use. # Finally ensure there's not something we don't recognise known_keywords = Dict(VALIDSIG="",SIG_ID="",GOODSIG="",BADSIG="",ERRSIG="", SIGEXPIRED="",KEYREVOKED="",NO_PUBKEY="",BADARMOR="", - NODATA="") + NODATA="",NOTATION_DATA="",NOTATION_NAME="",KEYEXPIRED="") for keyword in keywords.keys(): if not known_keywords.has_key(keyword): @@ -1004,6 +1297,25 @@ a *list* of keyrings to use. ################################################################################ +def gpg_get_key_addresses(fingerprint): + """retreive email addresses from gpg key uids for a given fingerprint""" + addresses = key_uid_email_cache.get(fingerprint) + if addresses != None: + return addresses + addresses = set() + cmd = "gpg --no-default-keyring %s --fingerprint %s" \ + % (gpg_keyring_args(), fingerprint) + (result, output) = commands.getstatusoutput(cmd) + if result == 0: + for l in output.split('\n'): + m = re_gpg_uid.match(l) + if m: + addresses.add(m.group(1)) + key_uid_email_cache[fingerprint] = addresses + return addresses + +################################################################################ + # Inspired(tm) by http://www.zopelabs.com/cookbook/1022242603 def wrap(paragraph, max_length, prefix=""): @@ -1071,12 +1383,76 @@ If 'dotprefix' is non-null, the filename will be prefixed with a '.'.""" ################################################################################ +# checks if the user part of the email is listed in the alias file + +def is_email_alias(email): + global alias_cache + if alias_cache == None: + aliasfn = which_alias_file() + alias_cache = set() + if aliasfn: + for l in open(aliasfn): + alias_cache.add(l.split(':')[0]) + uid = email.split('@')[0] + return uid in alias_cache + +################################################################################ + apt_pkg.init() Cnf = apt_pkg.newConfiguration() apt_pkg.ReadConfigFileISC(Cnf,default_config) if which_conf_file() != default_config: - apt_pkg.ReadConfigFileISC(Cnf,which_conf_file()) + apt_pkg.ReadConfigFileISC(Cnf,which_conf_file()) ################################################################################ + +def generate_contents_information(filename): + """ + Generate a list of flies contained in a .deb + + @type filename: string + @param filename: the path to a .deb + + @rtype: list + @return: a list of files in the data.tar.* portion of the .deb + """ + cmd = "ar t %s" % (filename) + (result, output) = commands.getstatusoutput(cmd) + if result != 0: + reject("%s: 'ar t' invocation failed." % (filename)) + reject(utils.prefix_multi_line_string(output, " [ar output:] "), "") + + # Ugh ... this is ugly ... Code ripped from process_unchecked.py + chunks = output.split('\n') + + contents = [] + try: + cmd = "ar x %s %s" % (filename, chunks[2]) + (result, output) = commands.getstatusoutput(cmd) + if result != 0: + reject("%s: '%s' invocation failed." % (filename, cmd)) + reject(utils.prefix_multi_line_string(output, " [ar output:] "), "") + + # Got deb tarballs, now lets go through and determine what bits + # and pieces the deb had ... + if chunks[2] == "data.tar.gz": + data = tarfile.open("data.tar.gz", "r:gz") + elif data_tar == "data.tar.bz2": + data = tarfile.open("data.tar.bz2", "r:bz2") + else: + os.remove(chunks[2]) + reject("couldn't find data.tar.*") + + for tarinfo in data: + if not tarinfo.isdir(): + contents.append(tarinfo.name[2:]) + + finally: + if os.path.exists( chunks[2] ): + os.remove( chunks[2] ) + + return contents + +###############################################################################