path: root/tupkg/update/tupkgupdate-lite

#!/usr/bin/python -O

import re
import os
import sys
import pacman
import getopt
import glob
import MySQLdb
import MySQLdb.connections
import ConfigParser

###########################################################
# Deal with configuration
###########################################################

conffile = '/etc/tupkgs.conf'

if not os.path.isfile(conffile):
  print "Error: cannot access config file ("+conffile+")"
  usage(argv[0])
  sys.exit(1)

config = ConfigParser.ConfigParser()
config.read(conffile)

############################################################

# Define some classes we need
class Version:
  def __init__(self):
    self.version = None
    self.file = None

class Package:
  def __init__(self):
    self.name = None
    self.category = None
    self.old = None
    self.new = None
    self.desc = None
    self.url = None
    self.depends = None
    self.sources = None

class PackageDatabase:
  def __init__(self, host, user, password, dbname):
    self.host = host
    self.user = user
    self.password = password
    self.dbname = dbname
    self.connection = MySQLdb.connect(host=host, user=user, passwd=password, db=dbname)
  def cursor(self):
    return self.connection.cursor()
  def lookup(self, packagename):
    warning("DB: Looking up package: " + packagename)
    q = self.cursor()
    q.execute("SELECT ID FROM Packages WHERE Name = '" + 
               MySQLdb.escape_string(packagename) + "'")
    if (q.rowcount != 0):
      row = q.fetchone()
      return row[0]
    return None
  def getCategoryID(self, package):
    category_id = self.lookupCategory(package.category)
    if (category_id == None):
      category_id = 1
    warning("DB: Got category ID '" + str(category_id) + "' for package '" + package.name + "'")
    return category_id
  def insert(self, package, locationId):
    warning("DB: Inserting package: " + package.name)
    global repo_dir
    q = self.cursor()
    q.execute("INSERT INTO Packages " +
      "(Name, CategoryID, Version, FSPath, LocationID, SubmittedTS, Description, URL) VALUES ('" +
      MySQLdb.escape_string(package.name) + "', " + 
      str(self.getCategoryID(package)) + ", '" +
      MySQLdb.escape_string(package.new.version) + "', '" +
      MySQLdb.escape_string(
        os.path.join(repo_dir, os.path.basename(package.new.file))) + "', " +
      str(locationId) + ", " +
      "UNIX_TIMESTAMP(), '" +
      MySQLdb.escape_string(str(package.desc)) + "', '" +
      MySQLdb.escape_string(str(package.url)) + "')")
    id = self.lookup(package.name)
    self.insertNewInfo(package, id, locationId)    
  def update(self, id, package, locationId):
    warning("DB: Updating package: " + package.name + " with id " + str(id))
    global repo_dir
    q = self.cursor()
    if (self.isdummy(package.name)):
      q.execute("UPDATE Packages SET " +
        "Version = '" + MySQLdb.escape_string(package.new.version) + "', " +
        "CategoryID = " + str(self.getCategoryID(package)) + ", " +
        "FSPath = '" + MySQLdb.escape_string(
          os.path.join(repo_dir, os.path.basename(package.new.file))) + "', " +
        "Description = '" + MySQLdb.escape_string(str(package.desc)) + "', " +
        "DummyPkg = 0, " +
	"SubmittedTS = UNIX_TIMESTAMP(), " +
        "URL = '" + MySQLdb.escape_string(str(package.url)) + "' " +
        "WHERE ID = " + str(id))
    else:
      q.execute("UPDATE Packages SET " +
        "Version = '" + MySQLdb.escape_string(package.new.version) + "', " +
        "CategoryID = " + str(self.getCategoryID(package)) + ", " +
        "FSPath = '" + MySQLdb.escape_string(
          os.path.join(repo_dir, os.path.basename(package.new.file))) + "', " +
        "Description = '" + MySQLdb.escape_string(str(package.desc)) + "', " +
        "ModifiedTS = UNIX_TIMESTAMP(), " +
        "URL = '" + MySQLdb.escape_string(str(package.url)) + "' " +
        "WHERE ID = " + str(id))
    self.insertNewInfo(package, id, locationId)
    # we must lastly check to see if this is a move of a package from
    # unsupported to community, because we'd have to reset maintainer and location
    q = self.cursor()
    q.execute("SELECT LocationID FROM Packages WHERE ID = " + str(id))
    if (q.rowcount != 0):
      row = q.fetchone()
      if (row[0] != 3):
        q = self.cursor()
	q.execute("UPDATE Packages SET LocationID = 3, MaintainerUID = null WHERE ID = " + str(id))
  def remove(self, id, locationId):
    warning("DB: Removing package with id: " + str(id))
    q = self.cursor()
    q.execute("DELETE FROM Packages WHERE " +
      "LocationID = " + str(locationId) + " AND ID = " + str(id))
  def clearOldInfo(self, id):
    warning("DB: Clearing old info for package with id : " + str(id))
    q = self.cursor()
    q.execute("DELETE FROM PackageContents WHERE PackageID = " + str(id))
    q.execute("DELETE FROM PackageDepends WHERE PackageID = " + str(id))
    q.execute("DELETE FROM PackageSources WHERE PackageID = " + str(id))
  def lookupOrDummy(self, packagename):
    retval = self.lookup(packagename)
    if (retval != None):
      return retval
    return self.createDummy(packagename)
  def lookupCategory(self, categoryname):
    warning("DB: Looking up category: " + categoryname)
    q = self.cursor()
    q.execute("SELECT ID from PackageCategories WHERE Category = '" + MySQLdb.escape_string(categoryname) + "'")
    if (q.rowcount != 0):
      row = q.fetchone()
      return row[0]
    return None
  def createDummy(self, packagename):
    warning("DB: Creating dummy package for: " + packagename)
    q = self.cursor()
    q.execute("INSERT INTO Packages " +
      "(Name, Description, LocationID, DummyPkg) " +
      "VALUES ('" +
      MySQLdb.escape_string(packagename) + "', '" +
      MySQLdb.escape_string("A dummy package") + "', 1, 1)")
    return self.lookup(packagename)
  def insertNewInfo(self, package, id, locationId):
    q = self.cursor()

    # first delete the old; this is never bad
    self.clearOldInfo(id)

    warning("DB: Inserting new package info for " + package.name +
            " with id " + str(id))

    # PackageSources
    for source in package.sources:
      q.execute("INSERT INTO PackageSources (PackageID, Source) " + 
        "VALUES (" + str(id) + ", '" + source + "')")
    # PackageDepends
    for dep in package.depends:
      depid = self.lookupOrDummy(dep)
      q.execute("INSERT INTO PackageDepends (PackageID, DepPkgID) " +
        "VALUES (" + str(id) + ", " + str(depid) + ")")
  def isdummy(self, packagename):
    warning("DB: Looking up package: " + packagename)
    q = self.cursor()
    q.execute("SELECT * FROM Packages WHERE Name = '" +
      MySQLdb.escape_string(packagename) + "' AND DummyPkg = 1")
    if (q.rowcount != 0):
      return True
    return False

############################################################
# Functions for walking the file trees
############################################################

def filesForRegexp(topdir, regexp):
  retval = []
  def matchfile(regexp, dirpath, namelist):
    for name in namelist:
      if (regexp.match(name)):
        retval.append(os.path.join(dirpath, name))
  os.path.walk(topdir, matchfile, regexp)
  return retval

def packagesInTree(topdir):
  return filesForRegexp(topdir, re.compile("^.*\.pkg\.tar\.gz$"))

def pkgbuildsInTree(topdir):
  return filesForRegexp(topdir, re.compile("^PKGBUILD$"))

############################################################
# Function for testing if two files are identical 
############################################################

def areFilesIdentical(file_a, file_b):
  command = "cmp '" + file_a + "' '" + file_b + "' >/dev/null"
  retval = os.system(command)
  if (retval == 0):
    return True
  return False

############################################################
# Function for fetching info from PKGBUILDs and packages
############################################################

def infoFromPackageFile(filename):
  pkg = pacman.load(filename)
  return pkg.name, pkg.version + "-" + pkg.release

def infoFromPkgbuildFile(filename):
  # first grab the category based on the file path
  pkgdirectory = os.path.dirname(filename)
  catdirectory = os.path.dirname(pkgdirectory)
  m = re.match(r".*/([^/]+)$", catdirectory)
  if (m):
    category = m.group(1)
  else:
    category = "none"

  # open and source the file
  pf_stdin, pf_stdout = os.popen2("/bin/bash", 't', 0)
  print >>pf_stdin, ". " + filename
  #print "PKGBUILD: " + filename

  # get pkgname
  print >>pf_stdin, 'echo $pkgname'
  pkgname = pf_stdout.readline().strip()
  #print "PKGBUILD: pkgname: " + pkgname

  # get pkgver
  print >>pf_stdin, 'echo $pkgver'
  pkgver = pf_stdout.readline().strip()
  #print "PKGBUILD: pkgver: " + pkgver

  # get pkgrel
  print >>pf_stdin, 'echo $pkgrel'
  pkgrel = pf_stdout.readline().strip()
  #print "PKGBUILD: pkgrel: " + pkgrel

  # get url
  print >>pf_stdin, 'echo $url'
  url = pf_stdout.readline().strip()
  #print "PKGBUILD: url: " + url

  # get desc
  print >>pf_stdin, 'echo $pkgdesc'
  pkgdesc = pf_stdout.readline().strip()
  #print "PKGBUILD: pkgdesc: " + pkgdesc

  # get source array
  print >>pf_stdin, 'echo ${source[*]}'
  source = (pf_stdout.readline().strip()).split(" ")

  # get depends array
  print >>pf_stdin, 'echo ${depends[*]}'
  depends = (pf_stdout.readline().strip()).split(" ")

  # clean up
  pf_stdin.close()
  pf_stdout.close()

  return pkgname, pkgver + "-" + pkgrel, pkgdesc, url, depends, source, category

def infoFromPkgbuildFileWorse(filename):
  # load the file with pacman library
  pkg = pacman.load(filename)
  return (pkg.name, pkg.version + "-" + pkg.release, pkg.desc,
    pkg.url, pkg.depends, pkg.source)
  
############################################################
# Functions for doing the final steps of execution
############################################################

def execute(command):
  global switches
  print(command)
  if not (switches.get("-n") == True):
    return os.system(command)
  return 0

def copyFileToRepo(filename, repodir):
  destfile = os.path.join(repodir, os.path.basename(filename))
  command = "cp --preserve=timestamps '" + filename + "' '" + destfile + "'"
  return execute(command)

def deleteFile(filename):
  command = "rm '" + filename + "'"
  return execute(command)

def runGensync(repo, pkgbuild):
#target = os.path.join(repo, os.path.basename(repo) + ".db.tar.gz")
  target = os.path.join(repo, "community.db.tar.gz")
  command = "gensync '" + pkgbuild + "' '" + target + "'"
  return execute(command)

def runUpdatesyncUpd(repo, pkgbuild):
	targetDB = os.path.join(repo, "community.db.tar.gz")
	command = "updatesync upd '" + targetDB + "' '" + pkgbuild + "' '" + repo +"'"
	return execute(command)

def runUpdatesyncDel(repo, pkgname):
	targetDB = os.path.join(repo, "community.db.tar.gz")
	command = "updatesync del '" + targetDB + "' '" + pkgname +"'"
	return execute(command)

############################################################
# Functions for error handling
############################################################

def warning(string):
  print >>sys.stderr, string

had_error = 0
def error(string):
  global had_error
  warning(string)
  had_error = 1

############################################################
# MAIN
############################################################
# The purpose of this version of tupkgupdate is to avoid
# doing as much time and machine-intensive work as the full
# tupkgupdate does, by relying on the state of the mysql
# database to be correct. Since it usually is, this is a
# good means for effecting updates more quickly, and won't
# be a processor hog.
#
# Note that this version CANNOT HANDLE deleted packages. Those
# will have to be processed less frequently. The only way to
# know a package for sure is missing is to look through all
# the PKGBUILDs which is what we're trying to avoid.
#
# ARGUMENTS
#
# tupkgupdate [-n] [--delete] [--paranoid] <repo_dir> <pkgbuild_dir> <build_dir>

# First call getopt
switch_list,args_proper = getopt.getopt(sys.argv[1:], 'n',
                          [ "delete", "paranoid" ])
switches = {}
for switch in switch_list:
  switches[switch[0]] = 1

# Then handle the remaining arguments
if (len(args_proper) < 3):
  print >>sys.stderr, "syntax: tupkgupdate [-n] [--delete] [--paranoid] <repo_dir> <pkgbuild_tree> <build_tree>"
  sys.exit(-1)

repo_dir, pkgbuild_dir, build_dir = args_proper

# Open the database so we find out now if we can't!
db = PackageDatabase(config.get('mysql', 'host'),
  config.get('mysql', 'username'),
  config.get('mysql', 'password'),
  config.get('mysql', 'db'))

# Set up the lists and tables
packages = dict()
copy = list()
delete = list()

dbremove = list()
dbmodify = list()

# PASS 1: PARSING/LOCATING/CHECKING
#  
# Go through each package file in the incoming/build tree
c_files = packagesInTree(build_dir)
for c_file in c_files:

# 1 - fetch package.name and package.new.{version,file} from package file
# a) error and skip if invalid package file
	pkgname, ver = infoFromPackageFile(c_file)
	if (pkgname == None or ver == None):
		error("Package file '" + c_file + "' is invalid!")
		continue

# b) error and skip if this is a duplicate we've already seen
	if (packages.get(pkgname)):
		error("Pkgbuild '" + a_file + "' is a duplicate!")
		continue

# c) create the package structure
	package = Package()
	package.name = pkgname

	version = Version()
	version.version = ver
	version.file = c_file

	package.new = version

# 2 - use the package name to find/parse the PKGBUILD
	ver, desc, url, depends, sources, category = findPkgbuildAndGetInfo(pkgname)

# a) if no PKGBUILD file, ignore the built file (log a warning, and skip)
	if (ver == None):
		# no way to log warnings at the moment
		continue

# b) check that PKGBUILD pkgver-pkgrel == package.new.version (or log error and skip)
	if (ver != package.new.version):
		error("For package '" + pkgname + "' the PKGBUILD ver '" + ver + "' doesn't match the binary package ver '" + package.new.version + "')
		continue;

# c) populate package.{desc,url,depends,sources,category}
  package.desc = desc
	package.url = url
	package.depends = depends
	package.sources = sources
	package.category = category

# 3 - fill in package.old.version from the mysql db ([community] only) and perform checking:
  ver = db.getCommunityVersion(package.name)

# a) if no record in mysql db, verify package not in repo (or log error and skip)
	if (ver == None):
		evil_packages = glob.glob(os.path.join(repodir, package.name + "-*.pkg.tar.gz"));
		found_evil_package = 0;
		for evil_package in evil_packages:
			pkgname, ver = infoFromPackageFile(evil_package)
			if (pkgname == package.name):
				error("No record of package '" + package.name + "' in [community] yet it has a package in '" + evil_package + "'")
				found_evil_package = 1;
				continue;
		if (found_evil_package == 1):
			continue;
	
# b) if record in mysql db, infer and verify proper package.old.file in repo (or log error and skip)
	inferred_old_filepath = os.path.join(repodir, package.name + "-" + ver + ".pkg.tar.gz")
	if (not os.path.exists(inferred_old_filepath)):
		error("The old package file '" + inferred_old_filepath + "' should exist but doesn't! Aborting.")
		continue

# 4 - If new file exists in repo already, delete it (silently) and skip
	new_filepath = os.path.join(repodir, os.path.basename(package.new.file))
  if (os.path.exists(new_filepath)):
		delete.append(package.new.file)
		continue

# 5 - If we've gotten here, we are a legitimate update or new package, so:
# a) put the package in the package list
	packages.append(package)

# b) add new file to the "copy" list
	copy.append(package.new.file)

# c) add package to "dbmodify" list
	dbmodify.append(package)

# d) if package has an old, add old file to "delete" list
	if (not package.old == None):
		delete.append(package.old.file)

## IF WE HAVE HAD ANY ERRORS AT THIS POINT, ABORT! ##
if (had_error == 1):
  error("Aborting due to errors.")
  sys.exit(-1)

# FOR NOW, ALWAYS EXIT, for safety
warning("Would have succeeded in the operation.")
warning("DBMODIFY: " + ','.join(dbmodify))
warning("COPY: " + ','.join(copy))
warning("DELETE: " + ','.join(delete))
sys.exit(-1)

# PASS 3: EXECUTION
#

# First, do all the database updates

for package in dbmodify:
  warning("DB: Package in dbmodify: " + package.name)
  id = db.lookup(package.name)
  if (id == None):
    db.insert(package, 3)
  else:
    db.update(id, package, 3)

# Copy
for file in copy:
  retval = copyFileToRepo(file, repo_dir)
  if (retval != 0):
    error("Could not copy file to repo: '" + file + "'")
    sys.exit(-1)
# Delete (second, for safety's sake)
for file in delete:
  deleteFile(file)
# Now that we've copied new files and deleted, we should delete the source
# files, if we're supposed to
if (switches.get("--delete") == True):
  for file in copy:
    deleteFile(file)

# Run updatesync where it is needed
for package in dbmodify:
	retval = runUpdatesyncUpd(repo_dir, os.path.join(pkgbuild_dir,package.category,package.name,"PKGBUILD"))
	if (retval != 0):
		error("Updatesync upd returned an error!")
		sys.exit(-1)

# vim: ft=python ts=2 sw=2 noet