Add filename de-duplication to contents scanning.

[dak.git] / daklib / contents.py

diff --git a/daklib/contents.py b/daklib/contents.py
index e2e381736bc218a4b1713a10664c72520dea46e5..27ffdda6cf71879c75abd34b01ee98d3b2fe796c 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
@@ -91,7 +92,7 @@ unique_override as
         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, o.section || '/' || b.package as package
     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'''
@@ -119,7 +120,7 @@ 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, substring(o.section from position('/' in o.section) + 1) || '/' || b.package as package
+select bc.file, o.section || '/' || b.package as package
     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'''
@@ -190,7 +191,9 @@ 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():
@@ -198,6 +201,9 @@ 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):
@@ -251,7 +257,10 @@ 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()