blendsdev_compare

#!/usr/bin/env python

##dummy script to quickly compare control/taskdesc files form current and new blends-dev

import re, sys, pprint
import argparse
from debian import deb822


#function taken from udd blends_metadata_gathener
def clean_up_packages(packages):
	# Hack: Debian Edu tasks files are using '\' at EOL which is broken
	#       in RFC 822 files, but blend-gen-control from blends-dev relies
	#       on this.  So remove this stuff here for the Moment
	pkgs = re.sub('\\\\\n\s+', '', packages)

	# Remove versions from versioned depends
	#pkgs = re.sub(' *\([ ><=\.0-9]+\) *', '', pkgs)

	#remove excluded archs
	pkgs = re.sub('\[.*\]', '', pkgs)
	# temporary strip spaces from alternatives ('|') to enable erroneous space handling as it was done before
	pkgs = re.sub('\s*\|\s*', '|', pkgs)

	# turn alternatives ('|') into real depends for this purpose
	# because we are finally interested in all alternatives
	pkgslist = pkgs.split(',')
	# Collect all dependencies in one line first,
	# create an object for each later
	pkgs_in_one_line = []
	for depl in pkgslist:
		dl = depl.strip()
		if dl != '': # avoid confusion when ',' is at end of line
			if re.search('\s', dl):
				#print >> sys.stderr , "Blend %s task %s: Syntax error '%s'" % (blend, task, dl)
				# trying to fix the syntax error after issuing error message
				dlspaces = re.sub('\s+', ',', dl).split(',')
				for dls in dlspaces:
					pkgs_in_one_line.append(dls.strip())
					#print >> sys.stderr, "Blend %s task %s: Found '%s' package inside broken syntax string - please fix task file anyway" % (blend, task, dls.strip())
			else:
				# in case we have to deal with a set of alternatives
				if re.search('\|', dl):
					#for da in dl.split('|'):
					#  deps_in_one_line.append(da)
					dl = re.sub('\|', ' | ', dl)
				pkgs_in_one_line.append(dl)

	return pkgs_in_one_line

def load_control(path_to_control):
	"""
	parses a task file and return a dictionary containing all its package headers elements
	(depends, suggests etc)
	"""
	fcontrol = open(path_to_control, 'r')
	packages = {}

	current_package = ''

	for paragraph in deb822.Sources.iter_paragraphs(fcontrol, shared_storage=False):
		if paragraph.has_key("package"):
			current_package = paragraph["package"]
			#if not current_package in packages:
			packages[current_package] = {}
			for header in ["suggests", "recommends"]:
				packages[current_package][header] = []

		#no need for this, control files are generated with -D (nodepends)
		#if paragraph.has_key("depends"):
		#	packages[current_package]["depends"] += clean_up_packages(paragraph["depends"])

		if paragraph.has_key("suggests"):
			packages[current_package]["suggests"] += clean_up_packages(paragraph["suggests"])

		if paragraph.has_key("recommends"):
			packages[current_package]["recommends"] += clean_up_packages(paragraph["recommends"])

	return packages

def load_taskdesc(path_to_taskdesc):
	"""
	parses a task file and return a dictionary containing all its package headers elements
	(depends, suggests etc)
	"""
	fcontrol = open(path_to_taskdesc, 'r')
	packages = {}

	current_task = ''

	for paragraph in deb822.Sources.iter_paragraphs(fcontrol, shared_storage=False):
		if paragraph.has_key("task"):
			print paragraph["task"]
			current_task = paragraph["task"]
			#if not current_package in packages:
			packages[current_task] = []
			
		if paragraph.has_key("packages"):
			#packages[current_task]["recommends"] += clean_up_packages(paragraph["recommends"])
			for pkg in paragraph["packages"].split(' '):
				stripped = pkg.strip()

				if stripped:
					packages[current_task].append(stripped)

	return packages

def compare_controls(controlA, controlB):

	missing_tasks = []

	for task in controlA:
		if not task in controlB:
			missing_tasks.append(task)
			continue

		print "* Comparing {0}".format(task)
		for header in ["suggests", "recommends"]:

			diffs = set(controlA[task][header]) - set(controlB[task][header])

			if diffs:
				print "  --> diff in {0}".format(header)
				pprint.pprint(list(diffs), indent=4)

	if missing_tasks:
		print "Missing tasks:"
		print "\t{0}".format(missing_tasks)

def compare_taskdescs(taskdescA, taskdescB):

	missing_tasks = []

	for task in taskdescA:
		if not task in taskdescB:
			missing_tasks.append(task)
			continue

		print "* Comparing {0}".format(task)
		diffs = set(taskdescA[task]) - set(taskdescB[task])

		if diffs:
			print "  --> diffs"
			pprint.pprint(list(diffs), indent=4)

	if missing_tasks:
		print "** Missing tasks:"
		print "\t{0}".format(missing_tasks)

def main():
	##TODO add proper epilog giving example usage
	parser = argparse.ArgumentParser(epilog="")
	
	parser.add_argument("-t", "--taskdesc", dest="taskdesc", action="store_true",
	                    help="Compare taskdesc files", default=False)
	parser.add_argument("-c", "--control", dest="control", action="store_true",
						help="Compare control files", default=False)
	parser.add_argument("-d", "--diff", dest="compare", type=str,
						help="Provide two comma separated(without spaces) paths to files(control/taskdescription) to be compared")
	#parse the command line arguments
	args = parser.parse_args()

	fileA_path, fileB_path = args.compare.split(',')

	if args.control:
		
		controlA = load_control(fileA_path)
		controlB = load_control(fileB_path)
		
		#pprint.pprint(controlA)
		#pprint.pprint(controlB)

		compare_controls(controlA, controlB)

	elif args.taskdesc:
		taskdescA = load_taskdesc(fileA_path)
		taskdescB = load_taskdesc(fileB_path)

		#pprint.pprint(taskdescA)
		#pprint.pprint(taskdescB)
		
		compare_taskdescs(taskdescA, taskdescB)

if __name__ == "__main__":
	main()