]> git.decadent.org.uk Git - dak.git/blobdiff - daklib/dbconn.py
Test and refactor get_suites_source_in().
[dak.git] / daklib / dbconn.py
index 59fe5dc3c0f6fd5a1ce8c8ca1edda8958b4064cf..6f598bb5dd5597c061395bbceff55a6a4c8a6334 100755 (executable)
@@ -46,7 +46,7 @@ from inspect import getargspec
 
 import sqlalchemy
 from sqlalchemy import create_engine, Table, MetaData, Column, Integer
-from sqlalchemy.orm import sessionmaker, mapper, relation
+from sqlalchemy.orm import sessionmaker, mapper, relation, object_session, backref
 from sqlalchemy import types as sqltypes
 
 # Don't remove this, we re-export the exceptions to scripts which import us
@@ -59,6 +59,16 @@ from config import Config
 from textutils import fix_maintainer
 from dak_exceptions import NoSourceFieldError
 
+# suppress some deprecation warnings in squeeze related to sqlalchemy
+import warnings
+warnings.filterwarnings('ignore', \
+    "The SQLAlchemy PostgreSQL dialect has been renamed from 'postgres' to 'postgresql'.*", \
+    SADeprecationWarning)
+# TODO: sqlalchemy needs some extra configuration to correctly reflect
+# the ind_deb_contents_* indexes - we ignore the warnings at the moment
+warnings.filterwarnings("ignore", 'Predicate of partial index', SAWarning)
+
+
 ################################################################################
 
 # Patch in support for the debversion field type so that it works during
@@ -193,6 +203,7 @@ def get_architecture(architecture, session=None):
 
 __all__.append('get_architecture')
 
+# TODO: should be removed because the implementation is too trivial
 @session_wrapper
 def get_architecture_suites(architecture, session=None):
     """
@@ -209,13 +220,7 @@ def get_architecture_suites(architecture, session=None):
     @return: list of Suite objects for the given name (may be empty)
     """
 
-    q = session.query(Suite)
-    q = q.join(SuiteArchitecture)
-    q = q.join(Architecture).filter_by(arch_string=architecture).order_by('suite_name')
-
-    ret = q.all()
-
-    return ret
+    return get_architecture(architecture, session).suites
 
 __all__.append('get_architecture_suites')
 
@@ -1065,8 +1070,12 @@ __all__.append('get_dscfiles')
 ################################################################################
 
 class PoolFile(object):
-    def __init__(self, *args, **kwargs):
-        pass
+    def __init__(self, filename = None, location = None, filesize = -1, \
+        md5sum = None):
+        self.filename = filename
+        self.location = location
+        self.filesize = filesize
+        self.md5sum = md5sum
 
     def __repr__(self):
         return '<PoolFile %s>' % self.filename
@@ -1075,13 +1084,16 @@ class PoolFile(object):
     def fullpath(self):
         return os.path.join(self.location.path, self.filename)
 
+    def is_valid(self, filesize = -1, md5sum = None):\
+        return self.filesize == filesize and self.md5sum == md5sum
+
 __all__.append('PoolFile')
 
 @session_wrapper
 def check_poolfile(filename, filesize, md5sum, location_id, session=None):
     """
     Returns a tuple:
-    (ValidFileFound [boolean or None], PoolFile object or None)
+    (ValidFileFound [boolean], PoolFile object or None)
 
     @type filename: string
     @param filename: the filename of the file to check against the DB
@@ -1097,34 +1109,24 @@ def check_poolfile(filename, filesize, md5sum, location_id, session=None):
 
     @rtype: tuple
     @return: Tuple of length 2.
-                 - If more than one file found with that name: (C{None},  C{None})
                  - If valid pool file found: (C{True}, C{PoolFile object})
                  - If valid pool file not found:
                      - (C{False}, C{None}) if no file found
                      - (C{False}, C{PoolFile object}) if file found with size/md5sum mismatch
     """
 
-    q = session.query(PoolFile).filter_by(filename=filename)
-    q = q.join(Location).filter_by(location_id=location_id)
+    poolfile = session.query(Location).get(location_id). \
+        files.filter_by(filename=filename).first()
+    valid = False
+    if poolfile and poolfile.is_valid(filesize = filesize, md5sum = md5sum):
+        valid = True
 
-    ret = None
-
-    if q.count() > 1:
-        ret = (None, None)
-    elif q.count() < 1:
-        ret = (False, None)
-    else:
-        obj = q.one()
-        if obj.md5sum != md5sum or obj.filesize != int(filesize):
-            ret = (False, obj)
-
-    if ret is None:
-        ret = (True, obj)
-
-    return ret
+    return (valid, poolfile)
 
 __all__.append('check_poolfile')
 
+# TODO: the implementation can trivially be inlined at the place where the
+# function is called
 @session_wrapper
 def get_poolfile_by_id(file_id, session=None):
     """
@@ -1137,41 +1139,10 @@ def get_poolfile_by_id(file_id, session=None):
     @return: either the PoolFile object or None
     """
 
-    q = session.query(PoolFile).filter_by(file_id=file_id)
-
-    try:
-        return q.one()
-    except NoResultFound:
-        return None
+    return session.query(PoolFile).get(file_id)
 
 __all__.append('get_poolfile_by_id')
 
-
-@session_wrapper
-def get_poolfile_by_name(filename, location_id=None, session=None):
-    """
-    Returns an array of PoolFile objects for the given filename and
-    (optionally) location_id
-
-    @type filename: string
-    @param filename: the filename of the file to check against the DB
-
-    @type location_id: int
-    @param location_id: the id of the location to look in (optional)
-
-    @rtype: array
-    @return: array of PoolFile objects
-    """
-
-    q = session.query(PoolFile).filter_by(filename=filename)
-
-    if location_id is not None:
-        q = q.join(Location).filter_by(location_id=location_id)
-
-    return q.all()
-
-__all__.append('get_poolfile_by_name')
-
 @session_wrapper
 def get_poolfile_like_name(filename, session=None):
     """
@@ -1517,8 +1488,10 @@ __all__.append('get_dbchange')
 ################################################################################
 
 class Location(object):
-    def __init__(self, *args, **kwargs):
-        pass
+    def __init__(self, path = None):
+        self.path = path
+        # the column 'type' should go away, see comment at mapper
+        self.archive_type = 'pool'
 
     def __repr__(self):
         return '<Location %s (%s)>' % (self.path, self.location_id)
@@ -1562,8 +1535,8 @@ __all__.append('get_location')
 ################################################################################
 
 class Maintainer(object):
-    def __init__(self, *args, **kwargs):
-        pass
+    def __init__(self, name = None):
+        self.name = name
 
     def __repr__(self):
         return '''<Maintainer '%s' (%s)>''' % (self.name, self.maintainer_id)
@@ -2101,8 +2074,14 @@ __all__.append('get_sections')
 ################################################################################
 
 class DBSource(object):
-    def __init__(self, *args, **kwargs):
-        pass
+    def __init__(self, source = None, version = None, maintainer = None, \
+        changedby = None, poolfile = None, install_date = None):
+        self.source = source
+        self.version = version
+        self.maintainer = maintainer
+        self.changedby = changedby
+        self.poolfile = poolfile
+        self.install_date = install_date
 
     def __repr__(self):
         return '<DBSource %s (%s)>' % (self.source, self.version)
@@ -2189,7 +2168,7 @@ def get_suites_source_in(source, session=None):
     @return: list of Suite objects for the given source
     """
 
-    return session.query(Suite).join(SrcAssociation).join(DBSource).filter_by(source=source).all()
+    return session.query(Suite).filter(Suite.sources.any(source=source)).all()
 
 __all__.append('get_suites_source_in')
 
@@ -2228,10 +2207,12 @@ def get_sources_from_name(source, version=None, dm_upload_allowed=None, session=
 
 __all__.append('get_sources_from_name')
 
+# FIXME: This function fails badly if it finds more than 1 source package and
+# its implementation is trivial enough to be inlined.
 @session_wrapper
 def get_source_in_suite(source, suite, session=None):
     """
-    Returns list of DBSource objects for a combination of C{source} and C{suite}.
+    Returns a DBSource object for a combination of C{source} and C{suite}.
 
       - B{source} - source package name, eg. I{mailfilter}, I{bbdb}, I{glibc}
       - B{suite} - a suite name, eg. I{unstable}
@@ -2247,12 +2228,9 @@ def get_source_in_suite(source, suite, session=None):
 
     """
 
-    q = session.query(SrcAssociation)
-    q = q.join('source').filter_by(source=source)
-    q = q.join('suite').filter_by(suite_name=suite)
-
+    q = get_suite(suite, session).get_sources(source)
     try:
-        return q.one().source
+        return q.one()
     except NoResultFound:
         return None
 
@@ -2501,6 +2479,8 @@ SUITE_FIELDS = [ ('SuiteName', 'suite_name'),
                  ('CopyChanges', 'copychanges'),
                  ('OverrideSuite', 'overridesuite')]
 
+# Why the heck don't we have any UNIQUE constraints in table suite?
+# TODO: Add UNIQUE constraints for appropriate columns.
 class Suite(object):
     def __init__(self, suite_name = None, version = None):
         self.suite_name = suite_name
@@ -2530,38 +2510,49 @@ class Suite(object):
 
         return "\n".join(ret)
 
-__all__.append('Suite')
+    def get_architectures(self, skipsrc=False, skipall=False):
+        """
+        Returns list of Architecture objects
 
-@session_wrapper
-def get_suite_architecture(suite, architecture, session=None):
-    """
-    Returns a SuiteArchitecture object given C{suite} and ${arch} or None if it
-    doesn't exist
+        @type skipsrc: boolean
+        @param skipsrc: Whether to skip returning the 'source' architecture entry
+        (Default False)
 
-    @type suite: str
-    @param suite: Suite name to search for
+        @type skipall: boolean
+        @param skipall: Whether to skip returning the 'all' architecture entry
+        (Default False)
 
-    @type architecture: str
-    @param architecture: Architecture name to search for
+        @rtype: list
+        @return: list of Architecture objects for the given name (may be empty)
+        """
 
-    @type session: Session
-    @param session: Optional SQL session object (a temporary one will be
-    generated if not supplied)
+        q = object_session(self).query(Architecture). \
+            filter(Architecture.suites.contains(self))
+        if skipsrc:
+            q = q.filter(Architecture.arch_string != 'source')
+        if skipall:
+            q = q.filter(Architecture.arch_string != 'all')
+        return q.order_by(Architecture.arch_string).all()
 
-    @rtype: SuiteArchitecture
-    @return: the SuiteArchitecture object or None
-    """
+    def get_sources(self, source):
+        """
+        Returns a query object representing DBSource that is part of C{suite}.
 
-    q = session.query(SuiteArchitecture)
-    q = q.join(Architecture).filter_by(arch_string=architecture)
-    q = q.join(Suite).filter_by(suite_name=suite)
+          - B{source} - source package name, eg. I{mailfilter}, I{bbdb}, I{glibc}
 
-    try:
-        return q.one()
-    except NoResultFound:
-        return None
+        @type source: string
+        @param source: source package name
 
-__all__.append('get_suite_architecture')
+        @rtype: sqlalchemy.orm.query.Query
+        @return: a query of DBSource
+
+        """
+
+        session = object_session(self)
+        return session.query(DBSource).filter_by(source = source). \
+            filter(DBSource.suites.contains(self))
+
+__all__.append('Suite')
 
 @session_wrapper
 def get_suite(suite, session=None):
@@ -2590,15 +2581,7 @@ __all__.append('get_suite')
 
 ################################################################################
 
-class SuiteArchitecture(object):
-    def __init__(self, *args, **kwargs):
-        pass
-
-    def __repr__(self):
-        return '<SuiteArchitecture (%s, %s)>' % (self.suite_id, self.arch_id)
-
-__all__.append('SuiteArchitecture')
-
+# TODO: should be removed because the implementation is too trivial
 @session_wrapper
 def get_suite_architectures(suite, skipsrc=False, skipall=False, session=None):
     """
@@ -2623,19 +2606,7 @@ def get_suite_architectures(suite, skipsrc=False, skipall=False, session=None):
     @return: list of Architecture objects for the given name (may be empty)
     """
 
-    q = session.query(Architecture)
-    q = q.join(SuiteArchitecture)
-    q = q.join(Suite).filter_by(suite_name=suite)
-
-    if skipsrc:
-        q = q.filter(Architecture.arch_string != 'source')
-
-    if skipall:
-        q = q.filter(Architecture.arch_string != 'all')
-
-    q = q.order_by('arch_string')
-
-    return q.all()
+    return get_suite(suite, session).get_architectures(skipsrc, skipall)
 
 __all__.append('get_suite_architectures')
 
@@ -2874,8 +2845,10 @@ class DBConn(object):
 
     def __setupmappers(self):
         mapper(Architecture, self.tbl_architecture,
-               properties = dict(arch_id = self.tbl_architecture.c.id,
-                                 suites = relation(Suite, secondary=self.tbl_suite_architectures, backref='architectures')))
+           properties = dict(arch_id = self.tbl_architecture.c.id,
+               suites = relation(Suite, secondary=self.tbl_suite_architectures,
+                   order_by='suite_name',
+                   backref=backref('architectures', order_by='arch_string'))))
 
         mapper(Archive, self.tbl_archive,
                properties = dict(archive_id = self.tbl_archive.c.id,
@@ -2964,7 +2937,11 @@ class DBConn(object):
                properties = dict(file_id = self.tbl_files.c.id,
                                  filesize = self.tbl_files.c.size,
                                  location_id = self.tbl_files.c.location,
-                                 location = relation(Location)))
+                                 location = relation(Location,
+                                     # using lazy='dynamic' in the back
+                                     # reference because we have A LOT of
+                                     # files in one location
+                                     backref=backref('files', lazy='dynamic'))))
 
         mapper(Fingerprint, self.tbl_fingerprint,
                properties = dict(fingerprint_id = self.tbl_fingerprint.c.id,
@@ -3037,10 +3014,16 @@ class DBConn(object):
                                  component = relation(Component),
                                  archive_id = self.tbl_location.c.archive,
                                  archive = relation(Archive),
+                                 # FIXME: the 'type' column is old cruft and
+                                 # should be removed in the future.
                                  archive_type = self.tbl_location.c.type))
 
         mapper(Maintainer, self.tbl_maintainer,
-               properties = dict(maintainer_id = self.tbl_maintainer.c.id))
+               properties = dict(maintainer_id = self.tbl_maintainer.c.id,
+                   maintains_sources = relation(DBSource, backref='maintainer',
+                       primaryjoin=(self.tbl_maintainer.c.id==self.tbl_source.c.maintainer)),
+                   changed_sources = relation(DBSource, backref='changedby',
+                       primaryjoin=(self.tbl_maintainer.c.id==self.tbl_source.c.changedby))))
 
         mapper(NewComment, self.tbl_new_comments,
                properties = dict(comment_id = self.tbl_new_comments.c.id))
@@ -3076,19 +3059,15 @@ class DBConn(object):
                properties = dict(source_id = self.tbl_source.c.id,
                                  version = self.tbl_source.c.version,
                                  maintainer_id = self.tbl_source.c.maintainer,
-                                 maintainer = relation(Maintainer,
-                                                       primaryjoin=(self.tbl_source.c.maintainer==self.tbl_maintainer.c.id)),
                                  poolfile_id = self.tbl_source.c.file,
-                                 poolfile = relation(PoolFile),
+                                 poolfile = relation(PoolFile, backref=backref('source', uselist = False)),
                                  fingerprint_id = self.tbl_source.c.sig_fpr,
                                  fingerprint = relation(Fingerprint),
                                  changedby_id = self.tbl_source.c.changedby,
-                                 changedby = relation(Maintainer,
-                                                      primaryjoin=(self.tbl_source.c.changedby==self.tbl_maintainer.c.id)),
                                  srcfiles = relation(DSCFile,
                                                      primaryjoin=(self.tbl_source.c.id==self.tbl_dsc_files.c.source)),
-                                 srcassociations = relation(SrcAssociation,
-                                                            primaryjoin=(self.tbl_source.c.id==self.tbl_src_associations.c.source)),
+                                 suites = relation(Suite, secondary=self.tbl_src_associations,
+                                     backref='sources'),
                                  srcuploaders = relation(SrcUploader)))
 
         mapper(SourceACL, self.tbl_source_acl,
@@ -3119,12 +3098,6 @@ class DBConn(object):
                                  policy_queue = relation(PolicyQueue),
                                  copy_queues = relation(BuildQueue, secondary=self.tbl_suite_build_queue_copy)))
 
-        mapper(SuiteArchitecture, self.tbl_suite_architectures,
-               properties = dict(suite_id = self.tbl_suite_architectures.c.suite,
-                                 suite = relation(Suite, backref='suitearchitectures'),
-                                 arch_id = self.tbl_suite_architectures.c.architecture,
-                                 architecture = relation(Architecture)))
-
         mapper(SuiteSrcFormat, self.tbl_suite_src_formats,
                properties = dict(suite_id = self.tbl_suite_src_formats.c.suite,
                                  suite = relation(Suite, backref='suitesrcformats'),