]> git.decadent.org.uk Git - dak.git/blobdiff - dak/contents.py
Rewrite 'dak contents bootstrap_bin' to use ContentsScanner.
[dak.git] / dak / contents.py
index be1e6d05ceef672389b364949c87ceeedb4a33ed..e292ac91a9b806701897f960bdad8d3314c00431 100755 (executable)
@@ -37,15 +37,22 @@ Create all the contents files
 import sys
 import os
 import logging
-import math
 import gzip
 import threading
+import traceback
 import Queue
 import apt_pkg
+import datetime
+import traceback
 from daklib import utils
 from daklib.binary import Binary
 from daklib.config import Config
-from daklib.dbconn import DBConn
+from daklib.dbconn import *
+from daklib.contents import ContentsScanner
+from daklib import daklog
+
+Logger = None
+
 ################################################################################
 
 def usage (exit_code=0):
@@ -55,8 +62,11 @@ COMMANDS
     generate
         generate Contents-$arch.gz files
 
+    bootstrap_bin
+        scan the debs in the existing pool and load contents into the bin_contents table
+
     bootstrap
-        scan the debs in the existing pool and load contents in the the database
+        copy data from the bin_contents table into the deb_contents / udeb_contents tables
 
     cruft
         remove files/paths which are no longer referenced by a binary
@@ -73,6 +83,9 @@ OPTIONS
 
      -s, --suite={stable,testing,unstable,...}
         only operate on a single suite
+
+     -l, --limit=NUMBER
+        optional package limit for bootstrap_bin
 """
     sys.exit(exit_code)
 
@@ -87,360 +100,500 @@ log = logging.getLogger()
 
 ################################################################################
 
-# get all the arches delivered for a given suite
-# this should probably exist somehere common
-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(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
-              WHERE suite = $1
-                  AND arch = $2"""
-
-# ask if we already have contents associated with this binary
-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(int,int) as
-                SELECT (p.path||'/'||n.file) AS fn,
-                        s.section,
-                        b.package,
-                        b.architecture
-               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 override o ON (o.package=b.package)
-               JOIN section s ON (s.id=o.section)
-               WHERE o.suite = $1 AND o.type = $2
-               AND b.type='deb'
-               ORDER BY fn"""
-
-# find me all of the contents for a given .udeb
-udeb_contents_q = """PREPARE udeb_contents_q(int,int,int) as
-              SELECT (p.path||'/'||n.file) AS fn,
-                        s.section,
-                        b.package,
-                        b.architecture
-               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 override o ON (o.package=b.package)
-               JOIN section s ON (s.id=o.section)
-               WHERE o.suite = $1 AND o.type = $2
-               AND s.id = $3
-               AND b.type='udeb'
-               ORDER BY fn"""
-
-#               FROM content_file_paths p join content_associations c 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 override o ON (o.package=b.package)
-#               JOIN section s ON (s.id=o.section)
-#               WHERE o.suite = $1 AND o.type = $2
-#               AND s.id = $3
-#               AND b.id in (SELECT ba.bin from bin_associations ba join binaries b on b.id=ba.bin where (b.architecture=$3 or b.architecture=$4)and ba.suite=$1 and b.type='udeb')
-#               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_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)"""
-
-# delete any paths we are storing which have no binary associated with them
-remove_filepath_cruft_q = """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)"""
-
 class EndOfContents(object):
     """
     A sentry object for the end of the filename stream
     """
     pass
 
-class GzippedContentWriter(object):
+class OneAtATime(object):
+    """
+    a one space queue which sits between multiple possible producers
+    and multiple possible consumers
+    """
+    def __init__(self):
+        self.next_in_line = None
+        self.read_lock = threading.Condition()
+        self.write_lock = threading.Condition()
+        self.die = False
+
+    def enqueue(self, next):
+        self.write_lock.acquire()
+        while self.next_in_line:
+            if self.die:
+                return
+            self.write_lock.wait()
+
+        assert( not self.next_in_line )
+        self.next_in_line = next
+        self.write_lock.release()
+        self.read_lock.acquire()
+        self.read_lock.notify()
+        self.read_lock.release()
+
+    def dequeue(self):
+        self.read_lock.acquire()
+        while not self.next_in_line:
+            if self.die:
+                return
+            self.read_lock.wait()
+
+        result = self.next_in_line
+
+        self.next_in_line = None
+        self.read_lock.release()
+        self.write_lock.acquire()
+        self.write_lock.notify()
+        self.write_lock.release()
+
+        return result
+
+
+class ContentsWorkThread(threading.Thread):
     """
-    An object which will write contents out to a Contents-$arch.gz
-    file on a separate thread
     """
+    def __init__(self, upstream, downstream):
+        threading.Thread.__init__(self)
+        self.upstream = upstream
+        self.downstream = downstream
 
-    header = None # a class object holding the header section of contents file
+    def run(self):
+        while True:
+            try:
+                contents_file = self.upstream.dequeue()
+                if isinstance(contents_file,EndOfContents):
+                    if self.downstream:
+                        self.downstream.enqueue(contents_file)
+                    break
+
+                s = datetime.datetime.now()
+                print("%s start: %s" % (self,contents_file) )
+                self._run(contents_file)
+                print("%s finished: %s in %d seconds" % (self, contents_file, (datetime.datetime.now()-s).seconds ))
+                if self.downstream:
+                    self.downstream.enqueue(contents_file)
+            except:
+                traceback.print_exc()
+
+class QueryThread(ContentsWorkThread):
+    def __init__(self, upstream, downstream):
+        ContentsWorkThread.__init__(self, upstream, downstream)
+
+    def __str__(self):
+        return "QueryThread"
+    __repr__ = __str__
+
+    def _run(self, contents_file):
+        contents_file.query()
+
+class IngestThread(ContentsWorkThread):
+    def __init__(self, upstream, downstream):
+        ContentsWorkThread.__init__(self, upstream, downstream)
+
+    def __str__(self):
+        return "IngestThread"
+    __repr__ = __str__
+
+    def _run(self, contents_file):
+        contents_file.ingest()
+
+class SortThread(ContentsWorkThread):
+    def __init__(self, upstream, downstream):
+        ContentsWorkThread.__init__(self, upstream, downstream)
+
+    def __str__(self):
+        return "SortThread"
+    __repr__ = __str__
+
+    def _run(self, contents_file):
+        contents_file.sorted_keys = sorted(contents_file.filenames.keys())
+
+class OutputThread(ContentsWorkThread):
+    def __init__(self, upstream, downstream):
+        ContentsWorkThread.__init__(self, upstream, downstream)
+
+    def __str__(self):
+        return "OutputThread"
+    __repr__ = __str__
+
+    def _run(self, contents_file):
+        contents_file.open_file()
+        for fname in contents_file.sorted_keys:
+            contents_file.filehandle.write("%s\t%s\n" % (fname,contents_file.filenames[fname]))
+        contents_file.sorted_keys = None
+        contents_file.filenames.clear()
+
+class GzipThread(ContentsWorkThread):
+    def __init__(self, upstream, downstream):
+        ContentsWorkThread.__init__(self, upstream, downstream)
+
+    def __str__(self):
+        return "GzipThread"
+    __repr__ = __str__
+
+    def _run(self, contents_file):
+        os.system("gzip -f %s" % contents_file.filename)
+
+class ContentFile(object):
+    def __init__(self,
+                 filename,
+                 suite_str,
+                 suite_id):
+
+        self.filename = filename
+        self.filenames = {}
+        self.sorted_keys = None
+        self.suite_str = suite_str
+        self.suite_id = suite_id
+        self.session = None
+        self.filehandle = None
+        self.results = None
+
+    def __str__(self):
+        return self.filename
+    __repr__ = __str__
+
+
+    def cleanup(self):
+        self.filenames = None
+        self.sortedkeys = None
+        self.filehandle.close()
+        self.session.close()
+
+    def ingest(self):
+        while True:
+            r = self.results.fetchone()
+            if not r:
+                break
+            filename, package = r
+            self.filenames[filename]=package
 
-    def __init__(self, filename):
-        """
-        @ptype 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):
+        self.session.close()
+
+    def open_file(self):
         """
         opens a gzip stream to the contents file
         """
-        filepath = Config()["Contents::Root"] + filename
-        filedir = os.path.dirname(filepath)
+        filepath = Config()["Contents::Root"] + self.filename
+        filedir = os.path.dirname(self.filename)
         if not os.path.isdir(filedir):
             os.makedirs(filedir)
-        return gzip.open(filepath, "w")
+        self.filehandle = open(self.filename, "w")
+        self._write_header()
 
-    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
+    def _write_header(self):
+        self._get_header();
+        self.filehandle.write(ContentFile.header)
 
-            (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())
+    header=None
 
     @classmethod
-    def _getHeader(self):
+    def _get_header(self):
         """
         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 not ContentFile.header:
             if Config().has_key("Contents::Header"):
                 try:
                     h = open(os.path.join( Config()["Dir::Templates"],
                                            Config()["Contents::Header"] ), "r")
-                    GzippedContentWriter.header = h.read()
+                    ContentFile.header = h.read()
                     h.close()
                 except:
                     log.error( "error opening header file: %d\n%s" % (Config()["Contents::Header"],
                                                                       traceback.format_exc() ))
-                    GzippedContentWriter.header = None
+                    ContentFile.header = None
             else:
-                GzippedContentWriter.header = None
-
-        return GzippedContentWriter.header
-
+                ContentFile.header = None
+
+        return ContentFile.header
+
+
+class DebContentFile(ContentFile):
+    def __init__(self,
+                 filename,
+                 suite_str,
+                 suite_id,
+                 arch_str,
+                 arch_id):
+        ContentFile.__init__(self,
+                             filename,
+                             suite_str,
+                             suite_id )
+        self.arch_str = arch_str
+        self.arch_id = arch_id
+
+    def query(self):
+        self.session = DBConn().session();
+
+        self.results = self.session.execute("""SELECT filename, comma_separated_list(section || '/' || package)
+        FROM deb_contents
+        WHERE ( arch=2 or arch = :arch) AND suite = :suite
+        """, { 'arch':self.arch_id, 'suite':self.suite_id } )
+
+class UdebContentFile(ContentFile):
+    def __init__(self,
+                 filename,
+                 suite_str,
+                 suite_id,
+                 section_name,
+                 section_id):
+        ContentFile.__init__(self,
+                             filename,
+                             suite_str,
+                             suite_id )
+
+    def query(self):
+        self.session = DBConn().session();
+
+        self.results = self.session.execute("""SELECT filename, comma_separated_list(section || '/' || package)
+        FROM udeb_contents
+        WHERE suite = :suite
+        group by filename
+        """ , { 'suite': self.suite_id } )
 
 class Contents(object):
     """
     Class capable of generating Contents-$arch.gz files
-
-    Usage GenerateContents().generateContents( ["main","contrib","non-free"] )
     """
-
     def __init__(self):
         self.header = None
 
     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
         """
-        cursor = DBConn().cursor();
-        cursor.execute( "BEGIN WORK" )
-        cursor.execute( remove_pending_contents_cruft_q )
-        cursor.execute( remove_filename_cruft_q )
-        cursor.execute( remove_filepath_cruft_q )
-        cursor.execute( "COMMIT" )
+        s = DBConn().session()
 
+        # 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()
 
-    def bootstrap(self):
-        """
-        scan the existing debs in the pool to populate the contents database tables
-        """
-        pooldir = Config()[ 'Dir::Pool' ]
+        # 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()
 
-        cursor = DBConn().cursor();
-        DBConn().prepare("debs_q",debs_q)
-        DBConn().prepare("olddeb_q",olddeb_q)
-        DBConn().prepare("arches_q",arches_q)
+        # 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()
 
-        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] ) )
-
-                count = 0
-                while True:
-                    deb = cursor.fetchone()
-                    if not deb:
-                        break
-                    count += 1
-                    cursor1 = DBConn().cursor();
-                    cursor1.execute( "EXECUTE olddeb_q(%d)" % (deb[0] ) )
-                    old = cursor1.fetchone()
-                    if old:
-                        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 ):
-                            Binary(debfile, self.reject).scan_package(deb[0],True)
-                        else:
-                            log.error("missing .deb: %s" % deb[1])
+        s.commit()
 
-    def generate(self):
+
+    def bootstrap_bin(self):
         """
-        Generate Contents-$arch.gz files for every available arch in each given suite.
+        scan the existing debs in the pool to populate the bin_contents table
         """
-        cursor = DBConn().cursor()
+        pooldir = Config()[ 'Dir::Pool' ]
 
-        DBConn().prepare("arches_q", arches_q)
-        DBConn().prepare("contents_q", contents_q)
-        DBConn().prepare("udeb_contents_q", udeb_contents_q)
+        s = DBConn().session()
 
-        debtype_id=DBConn().get_override_type_id("deb")
-        udebtype_id=DBConn().get_override_type_id("udeb")
+        for binary in s.query(DBBinary).yield_per(100):
+            print( "binary: %s" % binary.package )
+            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) )
 
-        arch_all_id = DBConn().get_architecture_id("all")
-        suites = self._suites()
+                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)
 
 
-        # Get our suites, and the architectures
-        for suite in [i.lower() for i in suites]:
-            suite_id = DBConn().get_suite_id(suite)
-            arch_list = self._arches(cursor, suite_id)
 
-            file_writers = {}
+    def bootstrap(self):
+        """
+        scan the existing debs in the pool to populate the contents database tables
+        """
+        s = DBConn().session()
+
+
+        # get a mapping of all the override types we care about (right now .deb an .udeb)
+        override_type_map = {};
+        for override_type in s.query(OverrideType).all():
+            if override_type.overridetype.endswith('deb' ):
+                override_type_map[override_type.overridetype_id] = override_type.overridetype;
+
+        for override in s.query(Override).yield_per(100):
+            if not override_type_map.has_key(override.overridetype_id):
+                #this isn't an override we care about
+                continue
+
+            binaries = s.execute("""SELECT b.id, b.architecture
+                                    FROM binaries b
+                                    JOIN bin_associations ba ON ba.bin=b.id
+                                    WHERE ba.suite=:suite
+                                    AND b.package=:package""", {'suite':override.suite_id, 'package':override.package})
+            while True:
+                binary = binaries.fetchone()
+                if not binary:
+                    break
+
+                exists = s.execute("SELECT 1 FROM %s_contents WHERE binary_id=:id limit 1" % override_type_map[override.overridetype_id], {'id':binary.id})
+
+
+                if exists.fetchone():
+                    print '.',
+                    continue
+                else:
+                    print '+',
+
+                s.execute( """INSERT INTO %s_contents (filename,section,package,binary_id,arch,suite)
+                              SELECT file, :section, :package, :binary_id, :arch, :suite
+                              FROM bin_contents
+                              WHERE binary_id=:binary_id;""" % override_type_map[override.overridetype_id],
+                           { 'section' : override.section_id,
+                             'package' : override.package,
+                             'binary_id' : binary.id,
+                             'arch' : binary.architecture,
+                             'suite' : override.suite_id } )
+                s.commit()
 
-            try:
-                for arch_id in arch_list:
-                    file_writers[arch_id[0]] = GzippedContentWriter("dists/%s/Contents-%s.gz" % (suite, arch_id[1]))
+    def generate(self):
+        """
+        Generate contents files for both deb and udeb
+        """
+        self.deb_generate()
+        self.udeb_generate()
 
-                cursor.execute("EXECUTE contents_q(%d,%d);" % (suite_id, debtype_id))
+    def deb_generate(self):
+        """
+        Generate Contents-$arch.gz files for every available arch in each given suite.
+        """
+        session = DBConn().session()
+        debtype_id = get_override_type("deb", session)
+        suites = self._suites()
 
-                while True:
-                    r = cursor.fetchone()
-                    if not r:
-                        break
+        inputtoquery = OneAtATime()
+        querytoingest = OneAtATime()
+        ingesttosort = OneAtATime()
+        sorttooutput = OneAtATime()
+        outputtogzip = OneAtATime()
 
-                    filename, section, package, arch = r
+        qt = QueryThread(inputtoquery,querytoingest)
+        it = IngestThread(querytoingest,ingesttosort)
+        st = SortThread(ingesttosort,sorttooutput)
+        ot = OutputThread(sorttooutput,outputtogzip)
+        gt = GzipThread(outputtogzip, None)
 
-                    if arch == arch_all_id:
-                        ## its arch all, so all contents files get it
-                        for writer in file_writers.values():
-                            writer.write(filename, section, package)
+        qt.start()
+        it.start()
+        st.start()
+        ot.start()
+        gt.start()
 
-                    else:
-                        file_writers[arch].write(filename, section, package)
+        # Get our suites, and the architectures
+        for suite in [i.lower() for i in suites]:
+            suite_id = get_suite(suite, session).suite_id
+            print( "got suite_id: %s for suite: %s" % (suite_id, suite ) )
+            arch_list = self._arches(suite_id, session)
 
-            finally:
-                # close all the files
-                for writer in file_writers.values():
-                    writer.finish()
+            for (arch_id,arch_str) in arch_list:
+                print( "suite: %s, arch: %s time: %s" %(suite_id, arch_id, datetime.datetime.now().isoformat()) )
 
+                filename = "dists/%s/Contents-%s" % (suite, arch_str)
+                cf = DebContentFile(filename, suite, suite_id, arch_str, arch_id)
+                inputtoquery.enqueue( cf )
 
-            # The MORE fun part. Ok, udebs need their own contents files, udeb, and udeb-nf (not-free)
-            # This is HORRIBLY debian specific :-/
-        for section, fn_pattern in [("debian-installer","dists/%s/Contents-udeb-%s.gz"),
-                                    ("non-free/debian-installer", "dists/%s/Contents-udeb-nf-%s.gz")]:
+        inputtoquery.enqueue( EndOfContents() )
+        gt.join()
 
-            section_id = DBConn().get_section_id(section) # all udebs should be here)
-            if section_id != -1:
+    def udeb_generate(self):
+        """
+        Generate Contents-$arch.gz files for every available arch in each given suite.
+        """
+        session = DBConn().session()
+        udebtype_id=DBConn().get_override_type_id("udeb")
+        suites = self._suites()
 
-                # Get our suites, and the architectures
-                for suite in [i.lower() for i in suites]:
-                    suite_id = DBConn().get_suite_id(suite)
-                    arch_list = self._arches(cursor, suite_id)
+        inputtoquery = OneAtATime()
+        querytoingest = OneAtATime()
+        ingesttosort = OneAtATime()
+        sorttooutput = OneAtATime()
+        outputtogzip = OneAtATime()
 
-                    file_writers = {}
+        qt = QueryThread(inputtoquery,querytoingest)
+        it = IngestThread(querytoingest,ingesttosort)
+        st = SortThread(ingesttosort,sorttooutput)
+        ot = OutputThread(sorttooutput,outputtogzip)
+        gt = GzipThread(outputtogzip, None)
 
-                    try:
-                        for arch_id in arch_list:
-                            file_writers[arch_id[0]] = GzippedContentWriter(fn_pattern % (suite, arch_id[1]))
+        qt.start()
+        it.start()
+        st.start()
+        ot.start()
+        gt.start()
 
-                        cursor.execute("EXECUTE udeb_contents_q(%d,%d,%d)" % (suite_id, udebtype_id, section_id))
 
-                        while True:
-                            r = cursor.fetchone()
-                            if not r:
-                                break
+    def generate(self):
+        """
+        Generate Contents-$arch.gz files for every available arch in each given suite.
+        """
+        session = DBConn().session()
 
-                            filename, section, package, arch = r
+        arch_all_id = get_architecture("all", session).arch_id
 
-                            if not file_writers.has_key( arch ):
-                                continue
+        # 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")]:
 
-                            if arch == arch_all_id:
-                                ## its arch all, so all contents files get it
-                                for writer in file_writers.values():
-                                    writer.write(filename, section, package)
+            overridetype = get_override_type(dtype, session)
 
-                            else:
-                                file_writers[arch].write(filename, section, package)
-                    finally:
-                        # close all the files
-                        for writer in file_writers.values():
-                            writer.finish()
+            # For udebs, we only look in certain sections (see the for loop above)
+            if section is not None:
+                section = get_section(section, session)
 
+            # Get our suites
+            for suite in which_suites(session):
+                # Which architectures do we need to work on
+                arch_list = get_suite_architectures(suite.suite_name, skipsrc=True, skipall=True, session=session)
 
+                # 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)
 
+                finally:
+                    # close all the files
+                    for writer in file_writers.values():
+                        writer.finish()
     def _suites(self):
         """
         return a list of suites to operate on
@@ -452,14 +605,20 @@ class Contents(object):
 
         return suites
 
-    def _arches(self, cursor, suite):
+    def _arches(self, suite, session):
         """
         return a list of archs to operate on
         """
         arch_list = []
-        cursor.execute("EXECUTE arches_q(%d)" % (suite))
+        arches = session.execute(
+            """SELECT s.architecture, a.arch_string
+            FROM suite_architectures s
+            JOIN architecture a ON (s.architecture=a.id)
+            WHERE suite = :suite_id""",
+            {'suite_id':suite } )
+
         while True:
-            r = cursor.fetchone()
+            r = arches.fetchone()
             if not r:
                 break
 
@@ -468,19 +627,27 @@ class Contents(object):
 
         return arch_list
 
-################################################################################
 
+def scan_all(limit):
+    result = ContentsScanner.scan_all(limit)
+    processed = '%(processed)d packages processed' % result
+    remaining = '%(remaining)d packages remaining' % result
+    Logger([processed, remaining])
+
+################################################################################
 
 def main():
     cnf = Config()
 
     arguments = [('h',"help", "%s::%s" % (options_prefix,"Help")),
                  ('s',"suite", "%s::%s" % (options_prefix,"Suite"),"HasArg"),
+                 ('l',"limit", "%s::%s" % (options_prefix,"Limit"),"HasArg"),
                  ('q',"quiet", "%s::%s" % (options_prefix,"Quiet")),
                  ('v',"verbose", "%s::%s" % (options_prefix,"Verbose")),
                 ]
 
     commands = {'generate' : Contents.generate,
+                'bootstrap_bin' : Contents.bootstrap_bin,
                 'bootstrap' : Contents.bootstrap,
                 'cruft' : Contents.cruft,
                 }
@@ -505,7 +672,29 @@ def main():
                          format='%(asctime)s %(levelname)s %(message)s',
                          stream = sys.stderr )
 
+    global Logger
+    Logger = daklog.Loggor(cnf.Conf, 'contents')
+
+    limit = None
+    if cnf.has_key("%s::%s" % (options_prefix,"Limit")):
+        limit = cnf["%s::%s" % (options_prefix,"Limit")]
+    if args[0] == 'bootstrap_bin':
+        scan_all(limit)
+        return
+
     commands[args[0]](Contents())
 
+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()
+
+    return [get_suite(s.lower(), session) for s in suites]
+
+
 if __name__ == '__main__':
     main()