Contents: use the new string_agg() function of Pg9.

[dak.git] / daklib / contents.py

diff --git a/daklib/contents.py b/daklib/contents.py
index 1148758c704e87b8e09695b5cee947fafae958b7..5bf94d9a252da7c7bdd5a6c10d70b408294b5dab 100755 (executable)
--- a/daklib/contents.py
+++ b/daklib/contents.py
@@ -31,6 +31,7 @@ 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
 
 import os.path
@@ -55,12 +56,16 @@ class ContentsWriter(object):
         '''
         Returns a query object that is doing most of the work.
         '''
+        overridesuite = self.suite
+        if self.suite.overridesuite is not None:
+            overridesuite = get_suite(self.suite.overridesuite, self.session)
         params = {
-            'suite':    self.suite.suite_id,
-            'arch_all': get_architecture('all', self.session).arch_id,
-            'arch':     self.architecture.arch_id,
-            'type_id':  self.overridetype.overridetype_id,
-            'type':     self.overridetype.overridetype,
+            'suite':         self.suite.suite_id,
+            'overridesuite': overridesuite.suite_id,
+            'arch_all':      get_architecture('all', self.session).arch_id,
+            'arch':          self.architecture.arch_id,
+            'type_id':       self.overridetype.overridetype_id,
+            'type':          self.overridetype.overridetype,
         }
 
         if self.component is not None:
@@ -84,13 +89,13 @@ with
 unique_override as
     (select o.package, s.section
         from override o, section s
-        where o.suite = :suite and o.type = :type_id and o.section = s.id and
+        where o.suite = :overridesuite and o.type = :type_id and o.section = s.id and
         o.component = :component)
 
-select bc.file, substring(o.section from position('/' in o.section) + 1) || '/' || 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 = '''
@@ -112,39 +117,29 @@ with
 unique_override as
     (select distinct on (o.package, s.section) o.package, s.section
         from override o, section s
-        where o.suite = :suite and o.type = :type_id and o.section = s.id
+        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, substring(o.section from position('/' in o.section) + 1) || '/' || 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()
 
@@ -186,7 +181,9 @@ select bc.file, substring(o.section from position('/' in o.section) + 1) || '/'
         Write the output file.
         '''
         command = ['gzip', '--rsyncable']
-        output_file = open(self.output_filename(), 'w')
+        final_filename = self.output_filename()
+        temp_filename = final_filename + '.new'
+        output_file = open(temp_filename, 'w')
         gzip = Popen(command, stdin = PIPE, stdout = output_file)
         gzip.stdin.write(self.get_header())
         for item in self.fetch():
@@ -194,6 +191,9 @@ select bc.file, substring(o.section from position('/' in o.section) + 1) || '/'
         gzip.stdin.close()
         output_file.close()
         gzip.wait()
+        os.remove(final_filename)
+        os.rename(temp_filename, final_filename)
+        os.chmod(final_filename, 0664)
 
     @classmethod
     def write_all(class_, suite_names = [], force = False):
@@ -247,7 +247,10 @@ class ContentsScanner(object):
         '''
         session = DBConn().session()
         binary = session.query(DBBinary).get(self.binary_id)
-        for filename in binary.scan_contents():
+        fileset = set(binary.scan_contents())
+        if len(fileset) == 0:
+            fileset.add('EMPTY_PACKAGE')
+        for filename in fileset:
             binary.contents.append(BinContents(file = filename))
         session.commit()
         session.close()