]> git.decadent.org.uk Git - dak.git/commitdiff
Merge branch 'master' into contents
authorTorsten Werner <twerner@debian.org>
Thu, 24 Mar 2011 11:40:38 +0000 (12:40 +0100)
committerTorsten Werner <twerner@debian.org>
Thu, 24 Mar 2011 11:40:38 +0000 (12:40 +0100)
daklib/contents.py
tests/dbtest_contents.py

index 2a29b2e55b5080eadc574e559edf38a22ae1615a..0fa52df2823e1950a0262c26e81ff8a41d52b197 100755 (executable)
@@ -35,16 +35,11 @@ from tempfile import mkdtemp
 
 import os.path
 
-class ContentsWriter(object):
+class BinaryContentsWriter(object):
     '''
-    ContentsWriter writes the Contents-$arch.gz files.
+    BinaryContentsWriter writes the Contents-$arch.gz files.
     '''
     def __init__(self, suite, architecture, overridetype, component = None):
-        '''
-        The constructor clones its arguments into a new session object to make
-        sure that the new ContentsWriter object can be executed in a different
-        thread.
-        '''
         self.suite = suite
         self.architecture = architecture
         self.overridetype = overridetype
@@ -193,6 +188,134 @@ select bc.file, string_agg(o.section || '/' || b.package, ',' order by b.package
         os.chmod(temp_filename, 0664)
         os.rename(temp_filename, final_filename)
 
+
+class SourceContentsWriter(object):
+    '''
+    SourceContentsWriter writes the Contents-source.gz files.
+    '''
+    def __init__(self, suite, component):
+        self.suite = suite
+        self.component = component
+        self.session = suite.session()
+
+    def query(self):
+        '''
+        Returns a query object that is doing most of the work.
+        '''
+        params = {
+            'suite_id':     self.suite.suite_id,
+            'component_id': self.component.component_id,
+        }
+
+        sql = '''
+create temp table newest_sources (
+    id integer primary key,
+    source text);
+
+create index sources_binaries_by_source on newest_sources (source);
+
+insert into newest_sources (id, source)
+    select distinct on (source) s.id, s.source from source s
+        join files f on f.id = s.file
+        join location l on l.id = f.location
+        where s.id in (select source from src_associations where suite = :suite_id)
+            and l.component = :component_id
+        order by source, version desc;
+
+select sc.file, string_agg(s.source, ',' order by s.source) as pkglist
+    from newest_sources s, src_contents sc
+    where s.id = sc.source_id group by sc.file'''
+
+        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.
+        '''
+        return "%s\t%s\n" % (filename, package_list)
+
+    def fetch(self):
+        '''
+        Yields a new line of the Contents-source.gz file in filename order.
+        '''
+        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()
+
+    def get_list(self):
+        '''
+        Returns a list of lines for the Contents-source.gz file.
+        '''
+        return [item for item in self.fetch()]
+
+    def output_filename(self):
+        '''
+        Returns the name of the output file.
+        '''
+        values = {
+            'root':      Config()['Dir::Root'],
+            'suite':     self.suite.suite_name,
+            'component': self.component.component_name
+        }
+        return "%(root)s/dists/%(suite)s/%(component)s/Contents-source.gz" % values
+
+    def write_file(self):
+        '''
+        Write the output file.
+        '''
+        command = ['gzip', '--rsyncable']
+        final_filename = self.output_filename()
+        temp_filename = final_filename + '.new'
+        output_file = open(temp_filename, 'w')
+        gzip = Popen(command, stdin = PIPE, stdout = output_file)
+        for item in self.fetch():
+            gzip.stdin.write(item)
+        gzip.stdin.close()
+        output_file.close()
+        gzip.wait()
+        os.chmod(temp_filename, 0664)
+        os.rename(temp_filename, final_filename)
+
+
+def binary_helper(suite_id, arch_id, overridetype_id, component_id = None):
+    '''
+    This function is called in a new subprocess and multiprocessing wants a top
+    level function.
+    '''
+    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 = BinaryContentsWriter(suite, architecture, overridetype, component)
+    contents_writer.write_file()
+    return log_message
+
+def source_helper(suite_id, component_id):
+    '''
+    This function is called in a new subprocess and multiprocessing wants a top
+    level function.
+    '''
+    session = DBConn().session()
+    suite = Suite.get(suite_id, session)
+    component = Component.get(component_id, session)
+    log_message = [suite.suite_name, 'source', component.component_name]
+    contents_writer = SourceContentsWriter(suite, component)
+    contents_writer.write_file()
+    return log_message
+
+class ContentsWriter(object):
+    '''
+    Loop over all suites, architectures, overridetypes, and components to write
+    all contents files.
+    '''
     @classmethod
     def log_result(class_, result):
         '''
@@ -217,42 +340,32 @@ select bc.file, string_agg(o.section || '/' || b.package, ',' order by b.package
         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
+        contrib_id = get_component('contrib', 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
+            # handle source packages
+            pool.apply_async(source_helper, (suite_id, main_id),
+                callback = class_.log_result)
+            pool.apply_async(source_helper, (suite_id, contrib_id),
+                callback = class_.log_result)
+            pool.apply_async(source_helper, (suite_id, non_free_id),
+                callback = class_.log_result)
             for architecture in suite.get_architectures(skipsrc = True, skipall = True):
                 arch_id = architecture.arch_id
                 # handle 'deb' packages
-                pool.apply_async(generate_helper, (suite_id, arch_id, deb_id), \
+                pool.apply_async(binary_helper, (suite_id, arch_id, deb_id), \
                     callback = class_.log_result)
                 # handle 'udeb' packages for 'main' and 'non-free'
-                pool.apply_async(generate_helper, (suite_id, arch_id, udeb_id, main_id), \
+                pool.apply_async(binary_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), \
+                pool.apply_async(binary_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 BinaryContentsScanner(object):
     '''
index e3128161780893ea6188165a9f91a7e5cfaa768f..0f23053c3fe3d412191d48413352f8137be2594e 100755 (executable)
@@ -3,8 +3,8 @@
 from db_test import DBDakTestCase, fixture
 
 from daklib.dbconn import *
-from daklib.contents import ContentsWriter, BinaryContentsScanner, \
-    UnpackedSource, SourceContentsScanner
+from daklib.contents import BinaryContentsWriter, BinaryContentsScanner, \
+    UnpackedSource, SourceContentsScanner, SourceContentsWriter
 
 from os.path import normpath
 from sqlalchemy.exc import FlushError, IntegrityError
@@ -131,9 +131,9 @@ class ContentsTestCase(DBDakTestCase):
         self.assertEqual(self.override['hello_sid_main_udeb'], \
             self.otype['udeb'].overrides.one())
 
-    def test_contentswriter(self):
+    def test_binarycontentswriter(self):
         '''
-        Test the ContentsWriter class.
+        Test the BinaryContentsWriter class.
         '''
         self.setup_suites()
         self.setup_architectures()
@@ -142,7 +142,7 @@ class ContentsTestCase(DBDakTestCase):
         self.setup_overrides()
         self.binary['hello_2.2-1_i386'].contents.append(BinContents(file = '/usr/bin/hello'))
         self.session.commit()
-        cw = ContentsWriter(self.suite['squeeze'], self.arch['i386'], self.otype['deb'])
+        cw = BinaryContentsWriter(self.suite['squeeze'], self.arch['i386'], self.otype['deb'])
         self.assertEqual(['/usr/bin/hello                                          python/hello\n'], \
             cw.get_list())
         # test formatline and sort order
@@ -151,7 +151,7 @@ class ContentsTestCase(DBDakTestCase):
         # test output_filename
         self.assertEqual('tests/fixtures/ftp/dists/squeeze/Contents-i386.gz', \
             normpath(cw.output_filename()))
-        cw = ContentsWriter(self.suite['squeeze'], self.arch['i386'], \
+        cw = BinaryContentsWriter(self.suite['squeeze'], self.arch['i386'], \
             self.otype['udeb'], self.comp['main'])
         self.assertEqual('tests/fixtures/ftp/dists/squeeze/main/Contents-i386.gz', \
             normpath(cw.output_filename()))
@@ -201,6 +201,22 @@ class ContentsTestCase(DBDakTestCase):
         SourceContentsScanner(source.source_id).scan()
         self.assertTrue(source.contents.count() > 0)
 
+    def test_sourcecontentswriter(self):
+        '''
+        Test the SourceContentsWriter class.
+        '''
+        self.setup_sources()
+        self.session.flush()
+        # remove newer package from sid because it disturbs the test
+        self.source['hello_2.2-2'].suites = []
+        self.session.commit()
+        source = self.source['hello_2.2-1']
+        SourceContentsScanner(source.source_id).scan()
+        cw = SourceContentsWriter(source.suites[0], source.poolfile.location.component)
+        result = cw.get_list()
+        self.assertEqual(8, len(result))
+        self.assertTrue('debian/changelog\thello\n' in result)
+
     def classes_to_clean(self):
         return [Override, Suite, BinContents, DBBinary, DBSource, Architecture, Section, \
             OverrideType, Maintainer, Component, Priority, PoolFile]