X-Git-Url: https://git.decadent.org.uk/gitweb/?a=blobdiff_plain;f=daklib%2Fcontents.py;h=a158e8fcd4bb7d2ed27e8547b79b03d1cbd84b8a;hb=b8690d6b1256a4a95198b874d9eb1aab62a825fb;hp=27ffdda6cf71879c75abd34b01ee98d3b2fe796c;hpb=55c9457424113af281f88f786042c35dff410d2b;p=dak.git diff --git a/daklib/contents.py b/daklib/contents.py index 27ffdda6..a158e8fc 100755 --- a/daklib/contents.py +++ b/daklib/contents.py @@ -27,12 +27,11 @@ Helper code for contents generation. from daklib.dbconn import * from daklib.config import Config -from daklib.threadpool import ThreadPool -from multiprocessing import Pool -from sqlalchemy import desc, or_ -from sqlalchemy.exc import IntegrityError -from subprocess import Popen, PIPE, call +from multiprocessing import Pool +from shutil import rmtree +from subprocess import Popen, PIPE, check_call +from tempfile import mkdtemp import os.path @@ -92,10 +91,10 @@ unique_override as where o.suite = :overridesuite and o.type = :type_id and o.section = s.id and o.component = :component) -select bc.file, o.section || '/' || b.package as package +select bc.file, string_agg(o.section || '/' || b.package, ',' order by b.package) as pkglist from newest_binaries b, bin_contents bc, unique_override o where b.id = bc.binary_id and o.package = b.package - order by bc.file, b.package''' + group by bc.file''' else: sql = ''' @@ -120,36 +119,26 @@ unique_override as where o.suite = :overridesuite and o.type = :type_id and o.section = s.id order by o.package, s.section, o.modified desc) -select bc.file, o.section || '/' || b.package as package +select bc.file, string_agg(o.section || '/' || b.package, ',' order by b.package) as pkglist from newest_binaries b, bin_contents bc, unique_override o where b.id = bc.binary_id and o.package = b.package - order by bc.file, b.package''' + group by bc.file''' - return self.session.query("file", "package").from_statement(sql). \ + return self.session.query("file", "pkglist").from_statement(sql). \ params(params) def formatline(self, filename, package_list): ''' Returns a formatted string for the filename argument. ''' - package_list = ','.join(package_list) return "%-55s %s\n" % (filename, package_list) def fetch(self): ''' Yields a new line of the Contents-$arch.gz file in filename order. ''' - last_filename = None - package_list = [] - for filename, package in self.query().yield_per(100): - if filename != last_filename: - if last_filename is not None: - yield self.formatline(last_filename, package_list) - last_filename = filename - package_list = [] - package_list.append(package) - if last_filename is not None: - yield self.formatline(last_filename, package_list) + for filename, package_list in self.query().yield_per(100): + yield self.formatline(filename, package_list) # end transaction to return connection to pool self.session.rollback() @@ -201,53 +190,85 @@ select bc.file, o.section || '/' || b.package as package gzip.stdin.close() output_file.close() gzip.wait() - os.remove(final_filename) + try: + os.remove(final_filename) + except: + pass os.rename(temp_filename, final_filename) os.chmod(final_filename, 0664) @classmethod - def write_all(class_, suite_names = [], force = False): + def log_result(class_, result): + ''' + Writes a result message to the logfile. + ''' + class_.logger.log(result) + + @classmethod + def write_all(class_, logger, suite_names = [], force = False): ''' Writes all Contents files for suites in list suite_names which defaults to all 'touchable' suites if not specified explicitely. Untouchable suites will be included if the force argument is set to True. ''' + class_.logger = logger session = DBConn().session() suite_query = session.query(Suite) if len(suite_names) > 0: suite_query = suite_query.filter(Suite.suite_name.in_(suite_names)) if not force: suite_query = suite_query.filter_by(untouchable = False) + deb_id = get_override_type('deb', session).overridetype_id + udeb_id = get_override_type('udeb', session).overridetype_id + main_id = get_component('main', session).component_id + non_free_id = get_component('non-free', session).component_id pool = Pool() for suite in suite_query: + suite_id = suite.suite_id for architecture in suite.get_architectures(skipsrc = True, skipall = True): + arch_id = architecture.arch_id # handle 'deb' packages - command = ['dak', 'contents', '-s', suite.suite_name, \ - 'generate_helper', architecture.arch_string, 'deb'] - pool.apply_async(call, (command, )) + pool.apply_async(generate_helper, (suite_id, arch_id, deb_id), \ + callback = class_.log_result) # handle 'udeb' packages for 'main' and 'non-free' - command = ['dak', 'contents', '-s', suite.suite_name, \ - 'generate_helper', architecture.arch_string, 'udeb', 'main'] - pool.apply_async(call, (command, )) - command = ['dak', 'contents', '-s', suite.suite_name, \ - 'generate_helper', architecture.arch_string, 'udeb', 'non-free'] - pool.apply_async(call, (command, )) + pool.apply_async(generate_helper, (suite_id, arch_id, udeb_id, main_id), \ + callback = class_.log_result) + pool.apply_async(generate_helper, (suite_id, arch_id, udeb_id, non_free_id), \ + callback = class_.log_result) pool.close() pool.join() session.close() +def generate_helper(suite_id, arch_id, overridetype_id, component_id = None): + ''' + This function is called in a new subprocess. + ''' + session = DBConn().session() + suite = Suite.get(suite_id, session) + architecture = Architecture.get(arch_id, session) + overridetype = OverrideType.get(overridetype_id, session) + log_message = [suite.suite_name, architecture.arch_string, overridetype.overridetype] + if component_id is None: + component = None + else: + component = Component.get(component_id, session) + log_message.append(component.component_name) + contents_writer = ContentsWriter(suite, architecture, overridetype, component) + contents_writer.write_file() + return log_message + class ContentsScanner(object): ''' ContentsScanner provides a threadsafe method scan() to scan the contents of a DBBinary object. ''' - def __init__(self, binary): + def __init__(self, binary_id): ''' - The argument binary is the actual DBBinary object that should be - scanned. + The argument binary_id is the id of the DBBinary object that + should be scanned. ''' - self.binary_id = binary.binary_id + self.binary_id = binary_id def scan(self, dummy_arg = None): ''' @@ -279,10 +300,79 @@ class ContentsScanner(object): if limit is not None: query = query.limit(limit) processed = query.count() - threadpool = ThreadPool() + pool = Pool() for binary in query.yield_per(100): - threadpool.queueTask(ContentsScanner(binary).scan) - threadpool.joinAll() + pool.apply_async(scan_helper, (binary.binary_id, )) + pool.close() + pool.join() remaining = remaining() session.close() return { 'processed': processed, 'remaining': remaining } + +def scan_helper(binary_id): + ''' + This function runs in a subprocess. + ''' + scanner = ContentsScanner(binary_id) + scanner.scan() + + +class UnpackedSource(object): + ''' + UnpackedSource extracts a source package into a temporary location and + gives you some convinient function for accessing it. + ''' + def __init__(self, dscfilename): + ''' + The dscfilename is a name of a DSC file that will be extracted. + ''' + self.root_directory = os.path.join(mkdtemp(), 'root') + command = ('dpkg-source', '--no-copy', '--no-check', '-x', dscfilename, + self.root_directory) + # dpkg-source does not have a --quiet option + devnull = open(os.devnull, 'w') + check_call(command, stdout = devnull, stderr = devnull) + devnull.close() + + def get_root_directory(self): + ''' + Returns the name of the package's root directory which is the directory + where the debian subdirectory is located. + ''' + return self.root_directory + + def get_changelog_file(self): + ''' + Returns a file object for debian/changelog or None if no such file exists. + ''' + changelog_name = os.path.join(self.root_directory, 'debian', 'changelog') + try: + return open(changelog_name) + except IOError: + return None + + def get_all_filenames(self): + ''' + Returns an iterator over all filenames. The filenames will be relative + to the root directory. + ''' + skip = len(self.root_directory) + 1 + for root, _, files in os.walk(self.root_directory): + for name in files: + yield os.path.join(root[skip:], name) + + def cleanup(self): + ''' + Removes all temporary files. + ''' + if self.root_directory is None: + return + parent_directory = os.path.dirname(self.root_directory) + rmtree(parent_directory) + self.root_directory = None + + def __del__(self): + ''' + Enforce cleanup. + ''' + self.cleanup()