X-Git-Url: https://git.decadent.org.uk/gitweb/?a=blobdiff_plain;f=daklib%2Futils.py;h=02278e9130f78579a8e04dea6aef8cd998a55ca8;hb=b5d21dfae245e479a1dfd261b7f1a9d9bf2e9b99;hp=16cc1ff16ed54df640d0ccbfb1da8e5879e4eef2;hpb=1271680e053c92528ade608def345cb5a88d4cae;p=dak.git diff --git a/daklib/utils.py b/daklib/utils.py index 16cc1ff1..02278e91 100755 --- a/daklib/utils.py +++ b/daklib/utils.py @@ -1,4 +1,5 @@ #!/usr/bin/env python +# vim:set et ts=4 sw=4: # Utility functions # Copyright (C) 2000, 2001, 2002, 2003, 2004, 2005, 2006 James Troup @@ -22,9 +23,10 @@ ################################################################################ import codecs, commands, email.Header, os, pwd, re, select, socket, shutil, \ - sys, tempfile, traceback + sys, tempfile, traceback, stat import apt_pkg import database +from dak_exceptions import * ################################################################################ @@ -41,48 +43,30 @@ 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.*<([^>]*)>') re_srchasver = re.compile(r"^(\S+)\s+\((\S+)\)$") re_verwithext = re.compile(r"^(\d+)(?:\.(\d+))(?:\s+\((\S+)\))?$") -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+\((\S+)\)$") default_config = "/etc/dak/dak.conf" default_apt_config = "/etc/dak/apt.conf" -################################################################################ - -class Error(Exception): - """Base class for exceptions in this module.""" - pass - -class ParseMaintError(Error): - """Exception raised for errors in parsing a maintainer field. +alias_cache = None +key_uid_email_cache = {} - Attributes: - message -- explanation of the error - """ - - def __init__(self, message): - self.args = message, - self.message = message +# (hashname, function, earliest_changes_version) +known_hashes = [("sha1", apt_pkg.sha1sum, (1, 8)), + ("sha256", apt_pkg.sha256sum, (1, 8))] ################################################################################ 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 ################################################################################ @@ -117,35 +101,15 @@ def extract_component_from_section(section): ################################################################################ -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... @@ -167,10 +131,10 @@ 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: @@ -191,7 +155,7 @@ 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' @@ -199,35 +163,240 @@ 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. + # put it in the "source-version" field instead. srcver = re_srchasver.search(changes["source"]) - if srcver: + if srcver: changes["source"] = srcver.group(1) - changes["source-version"] = srcver.group(2) + 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(): + 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: + # XXX: IS THIS THE BLOODY CASE WHEN THE FILE'S IN THE POOL!? + continue + finally: + 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(): + actual_size = os.stat(f)[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): + rejmsg.append("%s: not present in files but in checksums-%s in %s" % + (file, hashname, where)) + if not files[file]["size"] == size: + rejmsg.append("%s: size differs for files and checksums-%s entry "\ + "in %s" % (file, hashname, where)) + 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, field="files", hashname="md5sum"): @@ -235,12 +404,12 @@ def build_file_list(changes, is_a_dsc=0, field="files", hashname="md5sum"): # Make sure we have a Files: field to parse... if not changes.has_key(field): - raise no_files_exc + raise NoFilesFieldError # Make sure we recognise the format of the Files: field format = re_verwithext.search(changes.get("format", "0.0")) if not format: - raise nk_format_exc, "%s" % (changes.get("format","0.0")) + raise UnknownFormatError, "%s" % (changes.get("format","0.0")) format = format.groups() if format[1] == None: @@ -252,12 +421,12 @@ def build_file_list(changes, is_a_dsc=0, field="files", hashname="md5sum"): if is_a_dsc: if format != (1,0): - raise nk_format_exc, "%s" % (changes.get("format","0.0")) + raise UnknownFormatError, "%s" % (changes.get("format","0.0")) else: if (format < (1,5) or format > (1,8)): - raise nk_format_exc, "%s" % (changes.get("format","0.0")) - if field != "files" and format < (1,8): - raise nk_format_exc, "%s" % (changes.get("format","0.0")) + 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" @@ -273,7 +442,7 @@ def build_file_list(changes, is_a_dsc=0, field="files", hashname="md5sum"): else: (md5, size, name) = s except ValueError: - raise changes_parse_error_exc, i + raise ParseChangesError, i if section == "": section = "-" @@ -371,46 +540,46 @@ 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 += '/' + 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: @@ -424,23 +593,23 @@ 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) @@ -450,23 +619,31 @@ 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 + 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 None ################################################################################ @@ -576,7 +753,7 @@ 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 ################################################################################ @@ -652,9 +829,9 @@ 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] ################################################################################ @@ -672,7 +849,7 @@ def pp_deps (deps): ################################################################################ def get_conf(): - return Cnf + return Cnf ################################################################################ @@ -915,7 +1092,7 @@ on error.""" return "%s: tainted filename" % (filename) # Invoke gpgv on the file - status_read, status_write = os.pipe(); + 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) @@ -987,7 +1164,7 @@ used.""" return None # Build the command line - status_read, status_write = os.pipe(); + status_read, status_write = os.pipe(); cmd = "gpgv --status-fd %s %s %s %s" % ( status_write, gpg_keyring_args(keyrings), sig_filename, data_filename) @@ -1081,6 +1258,25 @@ used.""" ################################################################################ +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=""): @@ -1148,12 +1344,27 @@ 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()) ################################################################################