]> git.decadent.org.uk Git - dak.git/blobdiff - dak/contents.py
fix things I broke yesterday with the contents merge
[dak.git] / dak / contents.py
old mode 100644 (file)
new mode 100755 (executable)
index c7d47bd..ebeb2bc
@@ -36,12 +36,12 @@ Create all the contents files
 
 import sys
 import os
-import tempfile
 import logging
 import math
 import gzip
 import apt_pkg
 from daklib import utils
+from daklib.binary import Binary
 from daklib.config import Config
 from daklib.dbconn import DBConn
 ################################################################################
@@ -88,19 +88,16 @@ log = logging.getLogger()
 
 ################################################################################
 
-# we unfortunately still have broken stuff in headers
-latin1_q = """SET CLIENT_ENCODING TO 'LATIN1'"""
-
 # get all the arches delivered for a given suite
 # this should probably exist somehere common
-arches_q = """PREPARE arches_q as
+arches_q = """PREPARE arches_q(int) as
               SELECT s.architecture, a.arch_string
               FROM suite_architectures s
               JOIN architecture a ON (s.architecture=a.id)
                   WHERE suite = $1"""
 
 # find me the .deb for a given binary id
-debs_q = """PREPARE debs_q as
+debs_q = """PREPARE debs_q(int, int) as
               SELECT b.id, f.filename FROM bin_assoc_by_arch baa
               JOIN binaries b ON baa.bin=b.id
               JOIN files f ON b.file=f.id
@@ -108,13 +105,13 @@ debs_q = """PREPARE debs_q as
                   AND arch = $2"""
 
 # ask if we already have contents associated with this binary
-olddeb_q = """PREPARE olddeb_q as
+olddeb_q = """PREPARE olddeb_q(int) as
               SELECT 1 FROM content_associations
               WHERE binary_pkg = $1
               LIMIT 1"""
 
 # find me all of the contents for a given .deb
-contents_q = """PREPARE contents_q as
+contents_q = """PREPARE contents_q(int,int,int,int) as
               SELECT (p.path||'/'||n.file) AS fn,
                       comma_separated_list(s.section||'/'||b.package)
               FROM content_associations c
@@ -126,14 +123,13 @@ contents_q = """PREPARE contents_q as
               JOIN section s ON (s.id=o.section)
               WHERE (b.architecture = $1 OR b.architecture = $2)
                   AND ba.suite = $3
-                  AND o.suite = $4
+                  AND o.suite = $3
                   AND b.type = 'deb'
-                  AND o.type = '7'
+                  AND o.type = $4
               GROUP BY fn
               ORDER BY fn"""
 
-# find me all of the contents for a given .udeb
-udeb_contents_q = """PREPARE udeb_contents_q as
+udeb_contents_q = """PREPARE udeb_contents_q(int,int,int,int,int) as
               SELECT (p.path||'/'||n.file) as fn,
                       comma_separated_list(s.section||'/'||b.package)
               FROM content_associations c
@@ -143,26 +139,28 @@ udeb_contents_q = """PREPARE udeb_contents_q as
               JOIN bin_associations ba ON (b.id=ba.bin)
               JOIN override o ON (o.package=b.package)
               JOIN section s ON (s.id=o.section)
-              WHERE s.id = $1
-                  AND ba.suite = $2
-                  AND o.suite = $3
+              WHERE (b.architecture = $1 OR b.architecture = $2)
+                  AND s.id = $3
+                  AND ba.suite = $4
+                  AND o.suite = $4
                   AND b.type = 'udeb'
-                  AND o.type = '8'
+                  AND o.type = $5
               GROUP BY fn
               ORDER BY fn"""
 
+
 # clear out all of the temporarily stored content associations
 # this should be run only after p-a has run.  after a p-a
 # run we should have either accepted or rejected every package
 # so there should no longer be anything in the queue
-remove_temp_contents_cruft_q = """DELETE FROM temp_content_associations"""
+remove_pending_contents_cruft_q = """DELETE FROM pending_content_associations"""
 
 # delete any filenames we are storing which have no binary associated with them
 remove_filename_cruft_q = """DELETE FROM content_file_names
                              WHERE id IN (SELECT cfn.id FROM content_file_names cfn
                                           LEFT JOIN content_associations ca
                                             ON ca.filename=cfn.id
-                                          WHERE ca.id IS NULL)""" );
+                                          WHERE ca.id IS NULL)"""
 
 # delete any paths we are storing which have no binary associated with them
 remove_filepath_cruft_q = """DELETE FROM content_file_paths
@@ -180,6 +178,9 @@ class Contents(object):
     def __init__(self):
         self.header = None
 
+    def reject(self, message):
+        log.error("E: %s" % message)
+
     def _getHeader(self):
         """
         Internal method to return the header for Contents.gz files
@@ -193,14 +194,12 @@ class Contents(object):
                     h = open(os.path.join( Config()["Dir::Templates"],
                                            Config()["Contents::Header"] ), "r")
                     self.header = h.read()
-                    print( "header: %s" % self.header )
                     h.close()
                 except:
                     log.error( "error opening header file: %d\n%s" % (Config()["Contents::Header"],
                                                                       traceback.format_exc() ))
                     self.header = False
             else:
-                print( "no header" )
                 self.header = False
 
         return self.header
@@ -213,7 +212,11 @@ class Contents(object):
         Internal method for writing all the results to a given file.
         The cursor should have a result set generated from a query already.
         """
-        f = gzip.open(Config()["Dir::Root"] + filename, "w")
+        filepath = Config()["Contents::Root"] + filename
+        filedir = os.path.dirname(filepath)
+        if not os.path.isdir(filedir):
+            os.makedirs(filedir)
+        f = gzip.open(filepath, "w")
         try:
             header = self._getHeader()
 
@@ -226,7 +229,7 @@ class Contents(object):
                     return
 
                 num_tabs = max(1,
-                               int( math.ceil( (self._goal_column - len(contents[0])) / 8) ) )
+                               int(math.ceil((self._goal_column - len(contents[0])-1) / 8)))
                 f.write(contents[0] + ( '\t' * num_tabs ) + contents[-1] + "\n")
 
         finally:
@@ -239,7 +242,7 @@ class Contents(object):
         """
         cursor = DBConn().cursor();
         cursor.execute( "BEGIN WORK" )
-        cursor.execute( remove_temp_contents_cruft_q )
+        cursor.execute( remove_pending_contents_cruft_q )
         cursor.execute( remove_filename_cruft_q )
         cursor.execute( remove_filepath_cruft_q )
         cursor.execute( "COMMIT" )
@@ -252,10 +255,9 @@ class Contents(object):
         pooldir = Config()[ 'Dir::Pool' ]
 
         cursor = DBConn().cursor();
-        cursor.execute( latin1_q )
-        cursor.execute( debs_q )
-        cursor.execute( olddeb_q )
-        cursor.execute( arches_q )
+        DBConn().prepare("debs_q",debs_q)
+        DBConn().prepare("olddeb_q",olddeb_q)
+        DBConn().prepare("arches_q",arches_q)
 
         suites = self._suites()
         for suite in [i.lower() for i in suites]:
@@ -266,32 +268,37 @@ class Contents(object):
             for arch_id in arch_list:
                 cursor.execute( "EXECUTE debs_q(%d, %d)" % ( suite_id, arch_id[0] ) )
 
-                debs = cursor.fetchall()
                 count = 0
-                for deb in debs:
+                while True:
+                    deb = cursor.fetchone()
+                    if not deb:
+                        break
                     count += 1
-                    cursor.execute( "EXECUTE olddeb_q(%d)" % (deb[0] ) )
-                    old = cursor.fetchone()
+                    cursor1 = DBConn().cursor();
+                    cursor1.execute( "EXECUTE olddeb_q(%d)" % (deb[0] ) )
+                    old = cursor1.fetchone()
                     if old:
-                        log.debug( "already imported: %s" % deb[1] )
+                        log.debug( "already imported: %s" % (deb[1]) )
                     else:
+                        log.debug( "scanning: %s" % (deb[1]) )
                         debfile = os.path.join( pooldir, deb[1] )
                         if os.path.exists( debfile ):
-                            contents = utils.generate_contents_information( debfile )
-                            DBConn().insert_content_paths(deb[0], contents)
-                            log.info( "imported (%d/%d): %s" % (count,len(debs),deb[1] ) )
+                            Binary(debfile, self.reject).scan_package(deb[0],True)
                         else:
-                            log.error( "missing .deb: %s" % deb[1] )
+                            log.error("missing .deb: %s" % deb[1])
 
     def generate(self):
         """
-        Generate Contents-$arch.gz files for every aviailable arch in each given suite.
+        Generate Contents-$arch.gz files for every available arch in each given suite.
         """
         cursor = DBConn().cursor();
 
-        cursor.execute( arches_q )
-        cursor.execute( contents_q )
-        cursor.execute( udeb_contents_q )
+        DBConn().prepare( "arches_q", arches_q )
+        DBConn().prepare( "contents_q", contents_q )
+        DBConn().prepare( "udeb_contents_q", udeb_contents_q )
+
+        debtype_id=DBConn().get_override_type_id("deb")
+        udebtype_id=DBConn().get_override_type_id("udeb")
 
         suites = self._suites()
 
@@ -303,23 +310,20 @@ class Contents(object):
             arch_all_id = DBConn().get_architecture_id("all")
 
             for arch_id in arch_list:
-                cursor.execute( "EXECUTE contents_q(%d,%d,%d,%d)" % (arch_id[0], arch_all_id, suite_id, suite_id))
+                cursor.execute("EXECUTE contents_q(%d,%d,%d,%d)" % (arch_id[0], arch_all_id, suite_id, debtype_id))
                 self._write_content_file(cursor, "dists/%s/Contents-%s.gz" % (suite, arch_id[1]))
 
             # The MORE fun part. Ok, udebs need their own contents files, udeb, and udeb-nf (not-free)
             # This is HORRIBLY debian specific :-/
-            # First off, udeb
-            section_id = DBConn().get_section_id('debian-installer') # all udebs should be here)
-            if section_id != -1:
-                cursor.execute("EXECUTE udeb_contents_q(%d,%d,%d)" % (section_id, suite_id, suite_id))
-                self._write_content_file(cursor, "dists/%s/Contents-udeb.gz" % suite)
+            for section, fn_pattern in [("debian-installer","dists/%s/Contents-udeb-%s.gz"),
+                                           ("non-free/debian-installer", "dists/%s/Contents-udeb-nf-%s.gz")]:
 
-            # Once more, with non-free
-            section_id = DBConn().get_section_id('non-free/debian-installer') # all udebs should be here)
+                for arch_id in arch_list:
+                    section_id = DBConn().get_section_id(section) # all udebs should be here)
+                    if section_id != -1:
+                        cursor.execute("EXECUTE udeb_contents_q(%d,%d,%d,%d,%d)" % (arch_id[0], arch_all_id, section_id, suite_id, udebtype_id))
 
-            if section_id != -1:
-                cursor.execute("EXECUTE udeb_contents_q(%d,%d,%d)" % (section_id, suite_id, suite_id))
-                self._write_content_file(cursor, "dists/%s/Contents-udeb-nf.gz" % suite)
+                        self._write_content_file(cursor, fn_pattern % (suite, arch_id[1]))
 
 
 ################################################################################
@@ -373,6 +377,14 @@ def main():
                 'cruft' : Contents.cruft,
                 }
 
+    args = apt_pkg.ParseCommandLine(cnf.Cnf, arguments,sys.argv)
+
+    if (len(args) < 1) or not commands.has_key(args[0]):
+        usage()
+
+    if cnf.has_key("%s::%s" % (options_prefix,"Help")):
+        usage()
+
     level=logging.INFO
     if cnf.has_key("%s::%s" % (options_prefix,"Quiet")):
         level=logging.ERROR
@@ -385,14 +397,6 @@ def main():
                          format='%(asctime)s %(levelname)s %(message)s',
                          stream = sys.stderr )
 
-    args = apt_pkg.ParseCommandLine(cnf.Cnf, arguments,sys.argv)
-
-    if (len(args) < 1) or not commands.has_key(args[0]):
-        usage()
-
-    if cnf.has_key("%s::%s" % (options_prefix,"Help")):
-        usage()
-
     commands[args[0]](Contents())
 
 if __name__ == '__main__':