X-Git-Url: https://git.decadent.org.uk/gitweb/?a=blobdiff_plain;f=daklib%2Fbinary.py;h=5290410254214be9e105fc41819d346e35769827;hb=f09b6c88380bd29be8455d3734d3fb68408905ff;hp=fb1cade91daa4fa4ba20e980bf62ca68d2104267;hpb=f3db7ee80798580c7e5a0e1f48e008bc22a8bdfc;p=dak.git diff --git a/daklib/binary.py b/daklib/binary.py index fb1cade9..52904102 100755 --- a/daklib/binary.py +++ b/daklib/binary.py @@ -24,7 +24,23 @@ Functions related debian binary packages ################################################################################ +# are we going the xorg way? +# a dak without a dak.conf? +# automatically detect the wrong settings at runtime? +# yes! +# well, we'll probably always need dak.conf (how do you get the database setting +# but removing most of the config into the database seems sane +# mhy: dont spoil the fun +# mhy: and i know how. we nmap localhost and check all open ports +# maybe one answers to sql +# we will discover projectb via avahi +# you're both sick +# really fucking sick + +################################################################################ + import os +import sys import shutil import tempfile import tarfile @@ -33,12 +49,33 @@ import traceback import atexit from debian_bundle import deb822 from dbconn import DBConn +from config import Config +import logging +import utils class Binary(object): - def __init__(self, filename): + def __init__(self, filename, reject=None): + """ + @ptype filename: string + @param filename: path of a .deb + + @ptype reject: function + @param reject: a function to log reject messages to + """ self.filename = filename self.tmpdir = None self.chunks = None + self.wrapped_reject = reject + + def reject(self, message): + """ + if we were given a reject function, send the reject message, + otherwise send it to stderr. + """ + if self.wrapped_reject: + self.wrapped_reject(message) + else: + print >> sys.stderr, message def __del__(self): """ @@ -51,20 +88,19 @@ class Binary(object): we need to remove the temporary directory, if we created one """ if self.tmpdir and os.path.exists(self.tmpdir): - self.tmpdir = None shutil.rmtree(self.tmpdir) + self.tmpdir = None def __scan_ar(self): # get a list of the ar contents if not self.chunks: cmd = "ar t %s" % (self.filename) - (result, output) = commands.getstatusoutput(cmd) if result != 0: rejected = True - reject("%s: 'ar t' invocation failed." % (self.filename)) - reject(utils.prefix_multi_line_string(output, " [ar output:] "), "") + self.reject("%s: 'ar t' invocation failed." % (self.filename)) + self.reject(utils.prefix_multi_line_string(output, " [ar output:] "), "") self.chunks = output.split('\n') @@ -74,15 +110,15 @@ class Binary(object): # a temporary directory if not self.tmpdir: - tmpdir = tempfile.mkdtemp() + tmpdir = utils.temp_dirname() cwd = os.getcwd() try: os.chdir( tmpdir ) cmd = "ar x %s %s %s" % (os.path.join(cwd,self.filename), self.chunks[1], self.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:] "), "") + self.reject("%s: '%s' invocation failed." % (self.filename, cmd)) + self.reject(utils.prefix_multi_line_string(output, " [ar output:] ")) else: self.tmpdir = tmpdir atexit.register( self._cleanup ) @@ -102,20 +138,20 @@ class Binary(object): rejected = not self.chunks if len(self.chunks) != 3: rejected = True - reject("%s: found %d chunks, expected 3." % (self.filename, len(self.chunks))) + self.reject("%s: found %d chunks, expected 3." % (self.filename, len(self.chunks))) if self.chunks[0] != "debian-binary": rejected = True - reject("%s: first chunk is '%s', expected 'debian-binary'." % (self.filename, self.chunks[0])) + self.reject("%s: first chunk is '%s', expected 'debian-binary'." % (self.filename, self.chunks[0])) if self.chunks[1] != "control.tar.gz": rejected = True - reject("%s: second chunk is '%s', expected 'control.tar.gz'." % (self.filename, self.chunks[1])) + self.reject("%s: second chunk is '%s', expected 'control.tar.gz'." % (self.filename, self.chunks[1])) if self.chunks[2] not in [ "data.tar.bz2", "data.tar.gz" ]: rejected = True - reject("%s: third chunk is '%s', expected 'data.tar.gz' or 'data.tar.bz2'." % (self.filename, self.chunks[2])) + self.reject("%s: third chunk is '%s', expected 'data.tar.gz' or 'data.tar.bz2'." % (self.filename, self.chunks[2])) return not rejected - def scan_package(self): + def scan_package(self, bootstrap_id=0): """ Unpack the .deb, do sanity checking, and gather info from it. @@ -123,37 +159,84 @@ class Binary(object): the hopefully near future, it should also include gathering info from the control file. + @ptype bootstrap_id: int + @param bootstrap_id: the id of the binary these packages + should be associated or zero meaning we are not bootstrapping + so insert into a temporary table + @return True if the deb is valid and contents were imported """ rejected = not self.valid_deb() self.__unpack() + result = False + + cwd = os.getcwd() if not rejected and self.tmpdir: - cwd = os.getcwd() try: os.chdir(self.tmpdir) if self.chunks[1] == "control.tar.gz": control = tarfile.open(os.path.join(self.tmpdir, "control.tar.gz" ), "r:gz") - - pkg = deb822.Packages.iter_paragraphs( control.extractfile('./control') ).next() - + control.extract('./control', self.tmpdir ) if self.chunks[2] == "data.tar.gz": data = tarfile.open(os.path.join(self.tmpdir, "data.tar.gz"), "r:gz") elif self.chunks[2] == "data.tar.bz2": data = tarfile.open(os.path.join(self.tmpdir, "data.tar.bz2" ), "r:bz2") - return DBConn().insert_content_paths(pkg, [ tarinfo.name for tarinfo in data if tarinfo.isdir()]) + if bootstrap_id: + result = DBConn().insert_content_paths(bootstrap_id, [tarinfo.name for tarinfo in data if not tarinfo.isdir()]) + else: + pkgs = deb822.Packages.iter_paragraphs(file(os.path.join(self.tmpdir,'control'))) + pkg = pkgs.next() + result = DBConn().insert_pending_content_paths(pkg, [tarinfo.name for tarinfo in data if not tarinfo.isdir()]) except: traceback.print_exc() - return False + os.chdir(cwd) + return result - finally: - os.chdir( cwd ) + def check_utf8_package(self, package): + """ + Unpack the .deb, do sanity checking, and gather info from it. + Currently information gathering consists of getting the contents list. In + the hopefully near future, it should also include gathering info from the + control file. + + @ptype bootstrap_id: int + @param bootstrap_id: the id of the binary these packages + should be associated or zero meaning we are not bootstrapping + so insert into a temporary table + + @return True if the deb is valid and contents were imported + """ + rejected = not self.valid_deb() + self.__unpack() + + if not rejected and self.tmpdir: + cwd = os.getcwd() + try: + os.chdir(self.tmpdir) + if self.chunks[1] == "control.tar.gz": + control = tarfile.open(os.path.join(self.tmpdir, "control.tar.gz" ), "r:gz") + control.extract('control', self.tmpdir ) + if self.chunks[2] == "data.tar.gz": + data = tarfile.open(os.path.join(self.tmpdir, "data.tar.gz"), "r:gz") + elif self.chunks[2] == "data.tar.bz2": + data = tarfile.open(os.path.join(self.tmpdir, "data.tar.bz2" ), "r:bz2") + + for tarinfo in data: + try: + unicode( tarinfo.name ) + except: + print >> sys.stderr, "E: %s has non-unicode filename: %s" % (package,tarinfo.name) + except: + traceback.print_exc() + result = False + os.chdir(cwd) if __name__ == "__main__": Binary( "/srv/ftp.debian.org/queue/accepted/halevt_0.1.3-2_amd64.deb" ).scan_package()