]> git.decadent.org.uk Git - dak.git/blobdiff - daklib/dbconn.py
Add class SourceContentsScanner.
[dak.git] / daklib / dbconn.py
index 1bf5a0f997c7bb84aa5821aa4f9ab3c074ec3422..014f1082e4b0216f7a74b55c261ff89e95156e6b 100755 (executable)
@@ -59,8 +59,10 @@ import sqlalchemy
 from sqlalchemy import create_engine, Table, MetaData, Column, Integer, desc, \
     Text, ForeignKey
 from sqlalchemy.orm import sessionmaker, mapper, relation, object_session, \
-    backref, MapperExtension, EXT_CONTINUE, object_mapper
+    backref, MapperExtension, EXT_CONTINUE, object_mapper, clear_mappers
 from sqlalchemy import types as sqltypes
+from sqlalchemy.orm.collections import attribute_mapped_collection
+from sqlalchemy.ext.associationproxy import association_proxy
 
 # Don't remove this, we re-export the exceptions to scripts which import us
 from sqlalchemy.exc import *
@@ -493,12 +495,14 @@ class DBBinary(ORMObject):
     def properties(self):
         return ['package', 'version', 'maintainer', 'source', 'architecture', \
             'poolfile', 'binarytype', 'fingerprint', 'install_date', \
-            'suites_count', 'binary_id', 'contents_count']
+            'suites_count', 'binary_id', 'contents_count', 'extra_sources']
 
     def not_null_constraints(self):
         return ['package', 'version', 'maintainer', 'source',  'poolfile', \
             'binarytype']
 
+    metadata = association_proxy('key', 'value')
+
     def get_component_name(self):
         return self.poolfile.location.component.component_name
 
@@ -2153,6 +2157,18 @@ __all__.append('get_sections')
 
 ################################################################################
 
+class SrcContents(ORMObject):
+    def __init__(self, file = None, source = None):
+        self.file = file
+        self.source = source
+
+    def properties(self):
+        return ['file', 'source']
+
+__all__.append('SrcContents')
+
+################################################################################
+
 class DBSource(ORMObject):
     def __init__(self, source = None, version = None, maintainer = None, \
         changedby = None, poolfile = None, install_date = None):
@@ -2172,6 +2188,27 @@ class DBSource(ORMObject):
         return ['source', 'version', 'install_date', 'maintainer', \
             'changedby', 'poolfile', 'install_date']
 
+    metadata = association_proxy('key', 'value')
+
+    def scan_contents(self):
+        '''
+        Returns a set of names for non directories. The path names are
+        normalized after converting them from either utf-8 or iso8859-1
+        encoding.
+        '''
+        fullpath = self.poolfile.fullpath
+        from daklib.contents import UnpackedSource
+        unpacked = UnpackedSource(fullpath)
+        fileset = set()
+        for name in unpacked.get_all_filenames():
+            # enforce proper utf-8 encoding
+            try:
+                name.decode('utf-8')
+            except UnicodeDecodeError:
+                name = name.decode('iso8859-1').encode('utf-8')
+            fileset.add(name)
+        return fileset
+
 __all__.append('DBSource')
 
 @session_wrapper
@@ -2465,6 +2502,16 @@ def add_deb_to_db(u, filename, session=None):
 
     bin.source_id = bin_sources[0].source_id
 
+    if entry.has_key("built-using"):
+        for srcname, version in entry["built-using"]:
+            exsources = get_sources_from_name(srcname, version, session=session)
+            if len(exsources) != 1:
+                raise NoSourceFieldError, "Unable to find source package (%s = %s) in Built-Using for %s (%s), %s, file %s, type %s, signed by %s" % \
+                                          (srcname, version, bin.package, bin.version, entry["architecture"],
+                                           filename, bin.binarytype, u.pkg.changes["fingerprint"])
+
+            bin.extra_sources.append(exsources[0])
+
     # Add and flush object so it has an ID
     session.add(bin)
 
@@ -2792,6 +2839,52 @@ __all__.append('UploadBlock')
 
 ################################################################################
 
+class MetadataKey(ORMObject):
+    def __init__(self, key = None):
+        self.key = key
+
+    def properties(self):
+        return ['key']
+
+    def not_null_constraints(self):
+        return ['key']
+
+__all__.append('MetadataKey')
+
+################################################################################
+
+class BinaryMetadata(ORMObject):
+    def __init__(self, key = None, value = None, binary = None):
+        self.key = key
+        self.value = value
+        self.binary = binary
+
+    def properties(self):
+        return ['binary', 'key', 'value']
+
+    def not_null_constraints(self):
+        return ['value']
+
+__all__.append('BinaryMetadata')
+
+################################################################################
+
+class SourceMetadata(ORMObject):
+    def __init__(self, key = None, value = None, source = None):
+        self.key = key
+        self.value = value
+        self.source = source
+
+    def properties(self):
+        return ['source', 'key', 'value']
+
+    def not_null_constraints(self):
+        return ['value']
+
+__all__.append('SourceMetadata')
+
+################################################################################
+
 class DBConn(object):
     """
     database module init.
@@ -2807,11 +2900,13 @@ class DBConn(object):
             self.__createconn()
 
     def __setuptables(self):
-        tables_with_primary = (
+        tables = (
             'architecture',
             'archive',
             'bin_associations',
+            'bin_contents',
             'binaries',
+            'binaries_metadata',
             'binary_acl',
             'binary_acl_map',
             'build_queue',
@@ -2823,37 +2918,38 @@ class DBConn(object):
             'changes_pending_binaries',
             'changes_pending_files',
             'changes_pending_source',
+            'changes_pending_files_map',
+            'changes_pending_source_files',
+            'changes_pool_files',
             'dsc_files',
+            'extra_src_references',
             'files',
             'fingerprint',
             'keyrings',
             'keyring_acl_map',
             'location',
             'maintainer',
+            'metadata_keys',
             'new_comments',
+            # TODO: the maintainer column in table override should be removed.
+            'override',
             'override_type',
             'policy_queue',
             'priority',
             'section',
             'source',
             'source_acl',
+            'source_metadata',
             'src_associations',
+            'src_contents',
             'src_format',
             'src_uploaders',
             'suite',
-            'uid',
-            'upload_blocks',
-        )
-
-        tables_no_primary = (
-            'changes_pending_files_map',
-            'changes_pending_source_files',
-            'changes_pool_files',
-            # TODO: the maintainer column in table override should be removed.
-            'override',
             'suite_architectures',
-            'suite_src_formats',
             'suite_build_queue_copy',
+            'suite_src_formats',
+            'uid',
+            'upload_blocks',
         )
 
         views = (
@@ -2880,28 +2976,11 @@ class DBConn(object):
             'suite_arch_by_name',
         )
 
-        # Sqlalchemy version 0.5 fails to reflect the SERIAL type
-        # correctly and that is why we have to use a workaround. It can
-        # be removed as soon as we switch to version 0.6.
-        for table_name in tables_with_primary:
+        for table_name in tables:
             table = Table(table_name, self.db_meta, \
-                Column('id', Integer, primary_key = True), \
                 autoload=True, useexisting=True)
             setattr(self, 'tbl_%s' % table_name, table)
 
-        for table_name in tables_no_primary:
-            table = Table(table_name, self.db_meta, autoload=True)
-            setattr(self, 'tbl_%s' % table_name, table)
-
-        # bin_contents needs special attention until the SERIAL type is
-        # correctly detected and the workaround has been removed; see comment
-        # above
-        self.tbl_bin_contents = Table('bin_contents', self.db_meta, \
-            Column('file', Text, primary_key = True),
-            Column('binary_id', Integer, ForeignKey('binaries.id'), \
-                primary_key = True),
-            autoload=True, useexisting=True)
-
         for view_name in views:
             view = Table(view_name, self.db_meta, autoload=True)
             setattr(self, 'view_%s' % view_name, view)
@@ -2942,7 +3021,11 @@ class DBConn(object):
                                  fingerprint = relation(Fingerprint),
                                  install_date = self.tbl_binaries.c.install_date,
                                  suites = relation(Suite, secondary=self.tbl_bin_associations,
-                                     backref=backref('binaries', lazy='dynamic'))),
+                                     backref=backref('binaries', lazy='dynamic')),
+                                 extra_sources = relation(DBSource, secondary=self.tbl_extra_src_references,
+                                     backref=backref('extra_binary_references', lazy='dynamic')),
+                                 key = relation(BinaryMetadata, cascade='all',
+                                     collection_class=attribute_mapped_collection('key'))),
                 extension = validator)
 
         mapper(BinaryACL, self.tbl_binary_acl,
@@ -3112,7 +3195,9 @@ class DBConn(object):
                                                      primaryjoin=(self.tbl_source.c.id==self.tbl_dsc_files.c.source)),
                                  suites = relation(Suite, secondary=self.tbl_src_associations,
                                      backref=backref('sources', lazy='dynamic')),
-                                 srcuploaders = relation(SrcUploader)),
+                                 srcuploaders = relation(SrcUploader),
+                                 key = relation(SourceMetadata, cascade='all',
+                                     collection_class=attribute_mapped_collection('key'))),
                extension = validator)
 
         mapper(SourceACL, self.tbl_source_acl,
@@ -3160,20 +3245,49 @@ class DBConn(object):
                     backref=backref('contents', lazy='dynamic', cascade='all')),
                 file = self.tbl_bin_contents.c.file))
 
+        mapper(SrcContents, self.tbl_src_contents,
+            properties = dict(
+                source = relation(DBSource,
+                    backref=backref('contents', lazy='dynamic', cascade='all')),
+                file = self.tbl_src_contents.c.file))
+
+        mapper(MetadataKey, self.tbl_metadata_keys,
+            properties = dict(
+                key_id = self.tbl_metadata_keys.c.key_id,
+                key = self.tbl_metadata_keys.c.key))
+
+        mapper(BinaryMetadata, self.tbl_binaries_metadata,
+            properties = dict(
+                binary_id = self.tbl_binaries_metadata.c.bin_id,
+                binary = relation(DBBinary),
+                key_id = self.tbl_binaries_metadata.c.key_id,
+                key = relation(MetadataKey),
+                value = self.tbl_binaries_metadata.c.value))
+
+        mapper(SourceMetadata, self.tbl_source_metadata,
+            properties = dict(
+                source_id = self.tbl_source_metadata.c.src_id,
+                source = relation(DBSource),
+                key_id = self.tbl_source_metadata.c.key_id,
+                key = relation(MetadataKey),
+                value = self.tbl_source_metadata.c.value))
+
     ## Connection functions
     def __createconn(self):
         from config import Config
         cnf = Config()
-        if cnf["DB::Host"]:
+        if cnf.has_key("DB::Service"):
+            connstr = "postgresql://service=%s" % cnf["DB::Service"]
+        elif cnf.has_key("DB::Host"):
             # TCP/IP
-            connstr = "postgres://%s" % cnf["DB::Host"]
-            if cnf["DB::Port"] and cnf["DB::Port"] != "-1":
+            connstr = "postgresql://%s" % cnf["DB::Host"]
+            if cnf.has_key("DB::Port") and cnf["DB::Port"] != "-1":
                 connstr += ":%s" % cnf["DB::Port"]
             connstr += "/%s" % cnf["DB::Name"]
         else:
             # Unix Socket
-            connstr = "postgres:///%s" % cnf["DB::Name"]
-            if cnf["DB::Port"] and cnf["DB::Port"] != "-1":
+            connstr = "postgresql:///%s" % cnf["DB::Name"]
+            if cnf.has_key("DB::Port") and cnf["DB::Port"] != "-1":
                 connstr += "?port=%s" % cnf["DB::Port"]
 
         engine_args = { 'echo': self.debug }
@@ -3185,6 +3299,20 @@ class DBConn(object):
             cnf['DB::Unicode'] == 'false':
             engine_args['use_native_unicode'] = False
 
+        # Monkey patch a new dialect in in order to support service= syntax
+        import sqlalchemy.dialects.postgresql
+        from sqlalchemy.dialects.postgresql.psycopg2 import PGDialect_psycopg2
+        class PGDialect_psycopg2_dak(PGDialect_psycopg2):
+            def create_connect_args(self, url):
+                if str(url).startswith('postgresql://service='):
+                    # Eww
+                    servicename = str(url)[21:]
+                    return (['service=%s' % servicename], {})
+                else:
+                    return PGDialect_psycopg2.create_connect_args(self, url)
+
+        sqlalchemy.dialects.postgresql.base.dialect = PGDialect_psycopg2_dak
+
         self.db_pg   = create_engine(connstr, **engine_args)
         self.db_meta = MetaData()
         self.db_meta.bind = self.db_pg
@@ -3194,8 +3322,13 @@ class DBConn(object):
 
         self.__setuptables()
         self.__setupmappers()
+        self.pid = os.getpid()
 
     def session(self):
+        # reinitialize DBConn in new processes
+        if self.pid != os.getpid():
+            clear_mappers()
+            self.__createconn()
         return self.db_smaker()
 
 __all__.append('DBConn')