]> git.decadent.org.uk Git - dak.git/blobdiff - dak/contents.py
do all binaries, not just the first
[dak.git] / dak / contents.py
old mode 100644 (file)
new mode 100755 (executable)
index 5bb1b14..4e6ef6b
@@ -36,14 +36,16 @@ Create all the contents files
 
 import sys
 import os
-import tempfile
 import logging
-import math
 import gzip
+import threading
+import Queue
 import apt_pkg
 from daklib import utils
-from daklib.Config import Config
-from daklib.DBConn import DBConn
+from daklib.binary import Binary
+from daklib.config import Config
+from daklib.dbconn import *
+
 ################################################################################
 
 def usage (exit_code=0):
@@ -53,8 +55,8 @@ COMMANDS
     generate
         generate Contents-$arch.gz files
 
-    bootstrap
-        scan the debs in the existing pool and load contents in the the database
+    bootstrap_bin
+        scan the debs in the existing pool and load contents into the bin_contents table
 
     cruft
         remove files/paths which are no longer referenced by a binary
@@ -70,10 +72,7 @@ OPTIONS
         supress all output but errors
 
      -s, --suite={stable,testing,unstable,...}
-        only operate on a signle suite
-
-     -a, --arch={i386,amd64}
-        only operate on a signle architecture
+        only operate on a single suite
 """
     sys.exit(exit_code)
 
@@ -82,138 +81,180 @@ OPTIONS
 # where in dak.conf all of our configuration will be stowed
 
 options_prefix = "Contents"
-options_prefix = "%s::Opitons" % options_prefix
-header_prefix = "%s::Header" % options_prefix
+options_prefix = "%s::Options" % options_prefix
 
 log = logging.getLogger()
 
 ################################################################################
 
-latin1_q = """SET CLIENT_ENCODING TO 'LATIN1'"""
-
-arches_q = """PREPARE arches_q as
-              SELECT s.architecture, a.arch_string
-              FROM suite_architectures s
-              JOIN architecture a ON (s.architecture=a.id)
-                  WHERE suite = $1"""
-
-debs_q = """PREPARE debs_q 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
-              WHERE suite = $1
-                  AND arch = $2"""
-
-olddeb_q = """PREPARE olddeb_q as
-              SELECT 1 FROM content_associations
-              WHERE binary_pkg = $1
-              LIMIT 1"""
-
-contents_q = """PREPARE contents_q as
-              SELECT (p.path||'/'||n.file) AS fn,
-                      comma_separated_list(s.section||'/'||b.package)
-              FROM content_associations c
-              JOIN content_file_paths p ON (c.filepath=p.id)
-              JOIN content_file_names n ON (c.filename=n.id)
-              JOIN binaries b ON (b.id=c.binary_pkg)
-              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 (b.architecture = $1 OR b.architecture = $2)
-                  AND ba.suite = $3
-                  AND o.suite = $4
-                  AND b.type = 'deb'
-                  AND o.type = '7'
-              GROUP BY fn
-              ORDER BY fn"""
-
-udeb_contents_q = """PREPARE udeb_contents_q as
-              SELECT (p.path||'/'||n.file) as fn,
-                      comma_separated_list(s.section||'/'||b.package)
-              FROM content_associations c
-              JOIN content_file_paths p ON (c.filepath=p.id)
-              JOIN content_file_names n ON (c.filename=n.id)
-              JOIN binaries b ON (b.id=c.binary_pkg)
-              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 = $3
-                  AND b.type = 'udeb'
-                  AND o.type = '8'
-              GROUP BY fn
-              ORDER BY fn"""
-
-class Contents(object):
+class EndOfContents(object):
     """
-    Class capable of generating Contents-$arch.gz files
+    A sentry object for the end of the filename stream
+    """
+    pass
 
-    Usage GenerateContents().generateContents( ["main","contrib","non-free"] )
+class GzippedContentWriter(object):
+    """
+    An object which will write contents out to a Contents-$arch.gz
+    file on a separate thread
     """
 
-    def __init__(self):
-        self.header = None
+    header = None # a class object holding the header section of contents file
+
+    def __init__(self, filename):
+        """
+        @type filename: string
+        @param filename: the name of the file to write to
+        """
+        self.queue = Queue.Queue()
+        self.current_file = None
+        self.first_package = True
+        self.output = self.open_file(filename)
+        self.thread = threading.Thread(target=self.write_thread,
+                                       name='Contents writer')
+        self.thread.start()
+
+    def open_file(self, filename):
+        """
+        opens a gzip stream to the contents file
+        """
+        filepath = Config()["Contents::Root"] + filename
+        filedir = os.path.dirname(filepath)
+        if not os.path.isdir(filedir):
+            os.makedirs(filedir)
+        return gzip.open(filepath, "w")
+
+    def write(self, filename, section, package):
+        """
+        enqueue content to be written to the file on a separate thread
+        """
+        self.queue.put((filename,section,package))
+
+    def write_thread(self):
+        """
+        the target of a Thread which will do the actual writing
+        """
+        while True:
+            next = self.queue.get()
+            if isinstance(next, EndOfContents):
+                self.output.write('\n')
+                self.output.close()
+                break
+
+            (filename,section,package)=next
+            if next != self.current_file:
+                # this is the first file, so write the header first
+                if not self.current_file:
+                    self.output.write(self._getHeader())
+
+                self.output.write('\n%s\t' % filename)
+                self.first_package = True
+
+            self.current_file=filename
+
+            if not self.first_package:
+                self.output.write(',')
+            else:
+                self.first_package=False
+            self.output.write('%s/%s' % (section,package))
 
+    def finish(self):
+        """
+        enqueue the sentry object so that writers will know to terminate
+        """
+        self.queue.put(EndOfContents())
+
+    @classmethod
     def _getHeader(self):
-        # Internal method to return the header for Contents.gz files
-        if self.header == None:
+        """
+        Internal method to return the header for Contents.gz files
+
+        This is boilerplate which explains the contents of the file and how
+        it can be used.
+        """
+        if not GzippedContentWriter.header:
             if Config().has_key("Contents::Header"):
                 try:
                     h = open(os.path.join( Config()["Dir::Templates"],
                                            Config()["Contents::Header"] ), "r")
-                    self.header = h.read()
+                    GzippedContentWriter.header = h.read()
                     h.close()
                 except:
-                    log.error( "error openeing header file: %d\n%s" % (Config()["Contents::Header"],
-                                                                       traceback.format_exc() ))
-                    self.header = False
+                    log.error( "error opening header file: %d\n%s" % (Config()["Contents::Header"],
+                                                                      traceback.format_exc() ))
+                    GzippedContentWriter.header = None
             else:
-                self.header = False
+                GzippedContentWriter.header = None
+
+        return GzippedContentWriter.header
+
+
+class Contents(object):
+    """
+    Class capable of generating Contents-$arch.gz files
+    """
+
+    def __init__(self):
+        self.header = None
 
-        return self.header
+    def reject(self, message):
+        log.error("E: %s" % message)
 
-    # goal column for section column
-    _goal_column = 54
+    def cruft(self):
+        """
+        remove files/paths from the DB which are no longer referenced
+        by binaries and clean the temporary table
+        """
+        s = DBConn().session()
 
-    def _write_content_file(self, cursor, filename):
-        # Internal method for writing all the results to a given file
-        f = gzip.open(Config()["Dir::Root"] + filename, "w")
-        try:
-            header = self._getHeader()
+        # 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
+        s.query(PendingContentAssociation).delete()
 
-            if header:
-                f.write(header)
+        # delete any filenames we are storing which have no binary associated
+        # with them
+        cafq = s.query(ContentAssociation.filename_id).distinct()
+        cfq = s.query(ContentFilename)
+        cfq = cfq.filter(~ContentFilename.cafilename_id.in_(cafq))
+        cfq.delete()
 
-            while True:
-                contents = cursor.fetchone()
-                if not contents:
-                    return
+        # delete any paths we are storing which have no binary associated with
+        # them
+        capq = s.query(ContentAssociation.filepath_id).distinct()
+        cpq = s.query(ContentFilepath)
+        cpq = cpq.filter(~ContentFilepath.cafilepath_id.in_(capq))
+        cpq.delete()
 
-                num_tabs = max(1,
-                               int( math.ceil( (self._goal_column - len(contents[0])) / 8) ) )
-                f.write(contents[0] + ( '\t' * num_tabs ) + contents[-1] + "\n")
+        s.commit()
 
-        finally:
-            f.close()
 
-    def cruft(self):
+    def bootstrap_bin(self):
         """
-        remove files/paths from the DB which are no longer referenced by binaries
+        scan the existing debs in the pool to populate the bin_contents table
         """
-        cursor = DBConn().cursor();
-        cursor.execute( "BEGIN WORK" )
-        cursor.execute( """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)""" );
-        cursor.execute( """DELETE FROM content_file_paths
-                           WHERE id IN (SELECT cfn.id FROM content_file_paths cfn
-                                        LEFT JOIN content_associations ca
-                                            ON ca.filepath=cfn.id
-                                        WHERE ca.id IS NULL)""" );
-        cursor.execute( "COMMIT" )
+        pooldir = Config()[ 'Dir::Pool' ]
+
+        s = DBConn().session()
+
+        for binary in s.query(DBBinary).all() ):
+            filename = binary.poolfile.filename
+             # Check for existing contents
+            existingq = s.execute( "select 1 from bin_contents where binary_id=:id", {'id':binary.binary_id} );
+            if existingq.fetchone():
+                log.debug( "already imported: %s" % (filename))
+            else:
+                # We don't have existing contents so import them
+                log.debug( "scanning: %s" % (filename) )
+
+                debfile = os.path.join(pooldir, filename)
+                if os.path.exists(debfile):
+                    Binary(debfile, self.reject).scan_package(binary.binary_id, True)
+                else:
+                    log.error("missing .deb: %s" % filename)
+
+
 
     def bootstrap(self):
         """
@@ -221,107 +262,77 @@ 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 )
-
-        suites = self._suites()
-        for suite in [i.lower() for i in suites]:
-            suite_id = DBConn().get_suite_id(suite)
-
-            arch_list = self._arches(cursor, suite_id)
-            arch_all_id = DBConn().get_architecture_id("all")
-            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:
-                    count += 1
-                    cursor.execute( "EXECUTE olddeb_q(%d)" % (deb[0] ) )
-                    old = cursor.fetchone()
-                    if old:
-                        log.debug( "already imported: %s" % deb[1] )
+        s = DBConn().session()
+
+        for suite in s.query(Suite).all():
+            for arch in get_suite_architectures(suite.suite_name, skipsrc=True, skipall=True, session=s):
+                q = s.query(BinAssociation).join(Suite)
+                q = q.join(Suite).filter_by(suite_name=suite.suite_name)
+                q = q.join(DBBinary).join(Architecture).filter_by(arch.arch_string)
+                for ba in q:
+                    filename = ba.binary.poolfile.filename
+                    # Check for existing contents
+                    existingq = s.query(ContentAssociations).filter_by(binary_pkg=ba.binary_id).limit(1)
+                    if existingq.count() > 0:
+                        log.debug( "already imported: %s" % (filename))
                     else:
-                        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] ) )
+                        # We don't have existing contents so import them
+                        log.debug( "scanning: %s" % (filename) )
+                        debfile = os.path.join(pooldir, filename)
+                        if os.path.exists(debfile):
+                            Binary(debfile, self.reject).scan_package(ba.binary_id, True)
                         else:
-                            log.error( "missing .deb: %s" % deb[1] )
+                            log.error("missing .deb: %s" % filename)
+
 
     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 )
+        session = DBConn().session()
 
-        suites = self._suites()
+        arch_all_id = get_architecture("all", session).arch_id
 
-        # Get our suites, and the architectures
-        for suite in [i.lower() for i in suites]:
-            suite_id = DBConn().get_suite_id(suite)
+        # The MORE fun part. Ok, udebs need their own contents files, udeb, and udeb-nf (not-free)
+        # This is HORRIBLY debian specific :-/
+        for dtype, section, fn_pattern in \
+              [('deb',  None,                        "dists/%s/Contents-%s.gz"),
+               ('udeb', "debian-installer",          "dists/%s/Contents-udeb-%s.gz"),
+               ('udeb', "non-free/debian-installer", "dists/%s/Contents-udeb-nf-%s.gz")]:
 
-            arch_list = self._arches(cursor, suite_id)
+            overridetype = get_override_type(dtype, session)
 
-            arch_all_id = DBConn().get_architecture_id("all")
+            # For udebs, we only look in certain sections (see the for loop above)
+            if section is not None:
+                section = get_section(section, session)
 
-            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))
-                self._write_content_file(cursor, "dists/%s/Contents-%s.gz" % (suite, arch_id[1]))
+            # Get our suites
+            for suite in which_suites():
+                # Which architectures do we need to work on
+                arch_list = get_suite_architectures(suite.suite_name, skipsrc=True, skipall=True, session=session)
 
-            # 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)
-
-            # Once more, with non-free
-            section_id = DBConn().get_section_id('non-free/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-nf.gz" % suite)
+                # Set up our file writer dictionary
+                file_writers = {}
+                try:
+                    # One file writer per arch
+                    for arch in arch_list:
+                        file_writers[arch.arch_id] = GzippedContentWriter(fn_pattern % (suite, arch.arch_string))
 
+                    for r in get_suite_contents(suite, overridetype, section, session=session).fetchall():
+                        filename, section, package, arch_id = r
 
-################################################################################
+                        if arch_id == arch_all_id:
+                            # It's arch all, so all contents files get it
+                            for writer in file_writers.values():
+                                writer.write(filename, section, package)
+                        else:
+                            if file_writers.has_key(arch_id):
+                                file_writers[arch_id].write(filename, section, package)
 
-    def _suites(self):
-        # return a list of suites to operate on
-        if Config().has_key( "%s::%s" %(options_prefix,"Suite")):
-            suites = utils.split_args(Config()[ "%s::%s" %(options_prefix,"Suite")])
-        else:
-            suites = Config().SubTree("Suite").List()
-
-        return suites
-
-    def _arches(self, cursor, suite):
-        # return a list of archs to operate on
-        arch_list = [ ]
-        if Config().has_key( "%s::%s" %(options_prefix,"Arch")):
-            archs = utils.split_args(Config()[ "%s::%s" %(options_prefix,"Arch")])
-            for arch_name in archs:
-                arch_list.append((DBConn().get_architecture_id(arch_name), arch_name))
-        else:
-            cursor.execute("EXECUTE arches_q(%d)" % (suite))
-            while True:
-                r = cursor.fetchone()
-                if not r:
-                    break
-
-                if r[1] != "source" and r[1] != "all":
-                    arch_list.append((r[0], r[1]))
-
-        return arch_list
+                finally:
+                    # close all the files
+                    for writer in file_writers.values():
+                        writer.finish()
 
 ################################################################################
 
@@ -332,14 +343,21 @@ def main():
                  ('s',"suite", "%s::%s" % (options_prefix,"Suite"),"HasArg"),
                  ('q',"quiet", "%s::%s" % (options_prefix,"Quiet")),
                  ('v',"verbose", "%s::%s" % (options_prefix,"Verbose")),
-                 ('a',"arch", "%s::%s" % (options_prefix,"Arch"),"HasArg"),
                 ]
 
     commands = {'generate' : Contents.generate,
-                'bootstrap' : Contents.bootstrap,
+                'bootstrap_bin' : Contents.bootstrap_bin,
                 '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
@@ -348,19 +366,23 @@ def main():
         level=logging.DEBUG
 
 
-    logging.basicConfig( level=logging.DEBUG,
+    logging.basicConfig( level=level,
                          format='%(asctime)s %(levelname)s %(message)s',
                          stream = sys.stderr )
 
-    args = apt_pkg.ParseCommandLine(cnf.Cnf, arguments,sys.argv)
+    commands[args[0]](Contents())
 
-    if (len(args) < 1) or not commands.has_key(args[0]):
-        usage()
+def which_suites(session):
+    """
+    return a list of suites to operate on
+    """
+    if Config().has_key( "%s::%s" %(options_prefix,"Suite")):
+        suites = utils.split_args(Config()[ "%s::%s" %(options_prefix,"Suite")])
+    else:
+        suites = Config().SubTree("Suite").List()
 
-    if cnf.has_key("%s::%s" % (options_prefix,"Help")):
-        usage()
+    return [get_suite(s.lower(), session) for s in suites]
 
-    commands[args[0]](Contents())
 
 if __name__ == '__main__':
     main()