path: root/pym/repoman/utilities.py

# repoman: Utilities
# Copyright 2007-2011 Gentoo Foundation
# Distributed under the terms of the GNU General Public License v2

"""This module contains utility functions to help repoman find ebuilds to
scan"""

from __future__ import print_function

__all__ = [
	"detect_vcs_conflicts",
	"editor_is_executable",
	"FindPackagesToScan",
	"FindPortdir",
	"FindVCS",
	"format_qa_output",
	"get_commit_message_with_editor",
	"get_commit_message_with_stdin",
	"have_profile_dir",
	"parse_metadata_use",
	"UnknownHerdsError",
	"check_metadata",
	"UpdateChangeLog"
]

import errno
import io
from itertools import chain
import logging
import pwd
import sys
import time
import textwrap
import difflib
import shutil
from tempfile import mkstemp

from portage import os
from portage import subprocess_getstatusoutput
from portage import _encodings
from portage import _unicode_decode
from portage import _unicode_encode
from portage import output
from portage.localization import _
from portage.output import red, green
from portage.process import find_binary
from portage import exception
from portage import util
normalize_path = util.normalize_path
util.initialize_logger()

if sys.hexversion >= 0x3000000:
	basestring = str

def detect_vcs_conflicts(options, vcs):
	"""Determine if the checkout has problems like cvs conflicts.
	
	If you want more vcs support here just keep adding if blocks...
	This could be better.
	
	TODO(antarus): Also this should probably not call sys.exit() as
	repoman is run on >1 packages and one failure should not cause
	subsequent packages to fail.
	
	Args:
		vcs - A string identifying the version control system in use
	Returns:
		None (calls sys.exit on fatal problems)
	"""
	retval = ("","")
	if vcs == 'cvs':
		logging.info("Performing a " + output.green("cvs -n up") + \
			" with a little magic grep to check for updates.")
		retval = subprocess_getstatusoutput("cvs -n up 2>/dev/null | " + \
			"egrep '^[^\?] .*' | " + \
			"egrep -v '^. .*/digest-[^/]+|^cvs server: .* -- ignored$'")
	if vcs == 'svn':
		logging.info("Performing a " + output.green("svn status -u") + \
			" with a little magic grep to check for updates.")
		retval = subprocess_getstatusoutput("svn status -u 2>&1 | " + \
			"egrep -v '^.  +.*/digest-[^/]+' | " + \
			"head -n-1")

	if vcs in ['cvs', 'svn']:
		mylines = retval[1].splitlines()
		myupdates = []
		for line in mylines:
			if not line:
				continue
			if line[0] not in " UPMARD": # unmodified(svn),Updates,Patches,Modified,Added,Removed/Replaced(svn),Deleted(svn)
				# Stray Manifest is fine, we will readd it anyway.
				if line[0] == '?' and line[1:].lstrip() == 'Manifest':
					continue
				logging.error(red("!!! Please fix the following issues reported " + \
					"from cvs: ")+green("(U,P,M,A,R,D are ok)"))
				logging.error(red("!!! Note: This is a pretend/no-modify pass..."))
				logging.error(retval[1])
				sys.exit(1)
			elif vcs == 'cvs' and line[0] in "UP":
				myupdates.append(line[2:])
			elif vcs == 'svn' and line[8] == '*':
				myupdates.append(line[9:].lstrip(" 1234567890"))

		if myupdates:
			logging.info(green("Fetching trivial updates..."))
			if options.pretend:
				logging.info("(" + vcs + " update " + " ".join(myupdates) + ")")
				retval = os.EX_OK
			else:
				retval = os.system(vcs + " update " + " ".join(myupdates))
			if retval != os.EX_OK:
				logging.fatal("!!! " + vcs + " exited with an error. Terminating.")
				sys.exit(retval)


def have_profile_dir(path, maxdepth=3, filename="profiles.desc"):
	""" 
	Try to figure out if 'path' has a profiles/
	dir in it by checking for the given filename.
	"""
	while path != "/" and maxdepth:
		if os.path.exists(os.path.join(path, "profiles", filename)):
			return normalize_path(path)
		path = normalize_path(path + "/..")
		maxdepth -= 1

def parse_metadata_use(xml_tree):
	"""
	Records are wrapped in XML as per GLEP 56
	returns a dict with keys constisting of USE flag names and values
	containing their respective descriptions
	"""
	uselist = {}

	usetags = xml_tree.findall("use")
	if not usetags:
		return uselist

	# It's possible to have multiple 'use' elements.
	for usetag in usetags:
		flags = usetag.findall("flag")
		if not flags:
			# DTD allows use elements containing no flag elements.
			continue

		for flag in flags:
			pkg_flag = flag.get("name")
			if pkg_flag is None:
				raise exception.ParseError("missing 'name' attribute for 'flag' tag")
			flag_restrict = flag.get("restrict")

			# emulate the Element.itertext() method from python-2.7
			inner_text = []
			stack = []
			stack.append(flag)
			while stack:
				obj = stack.pop()
				if isinstance(obj, basestring):
					inner_text.append(obj)
					continue
				if isinstance(obj.text, basestring):
					inner_text.append(obj.text)
				if isinstance(obj.tail, basestring):
					stack.append(obj.tail)
				stack.extend(reversed(obj))

			if pkg_flag not in uselist:
				uselist[pkg_flag] = {}

			# (flag_restrict can be None)
			uselist[pkg_flag][flag_restrict] = " ".join("".join(inner_text).split())

	return uselist

class UnknownHerdsError(ValueError):
	def __init__(self, herd_names):
		_plural = len(herd_names) != 1
		super(UnknownHerdsError, self).__init__(
			'Unknown %s %s' % (_plural and 'herds' or 'herd',
			','.join('"%s"' % e for e in herd_names)))


def check_metadata_herds(xml_tree, herd_base):
	herd_nodes = xml_tree.findall('herd')
	unknown_herds = [name for name in
			(e.text.strip() for e in herd_nodes if e.text is not None)
			if not herd_base.known_herd(name)]

	if unknown_herds:
		raise UnknownHerdsError(unknown_herds)

def check_metadata(xml_tree, herd_base):
	if herd_base is not None:
		check_metadata_herds(xml_tree, herd_base)

def FindPackagesToScan(settings, startdir, reposplit):
	""" Try to find packages that need to be scanned
	
	Args:
		settings - portage.config instance, preferably repoman_settings
		startdir - directory that repoman was run in
		reposplit - root of the repository
	Returns:
		A list of directories to scan
	"""
	
	
	def AddPackagesInDir(path):
		""" Given a list of dirs, add any packages in it """
		ret = []
		pkgdirs = os.listdir(path)
		for d in pkgdirs:
			if d == 'CVS' or d.startswith('.'):
				continue
			p = os.path.join(path, d)

			if os.path.isdir(p):
				cat_pkg_dir = os.path.join(*p.split(os.path.sep)[-2:])
				logging.debug('adding %s to scanlist' % cat_pkg_dir)
				ret.append(cat_pkg_dir)
		return ret
	
	scanlist = []
	repolevel = len(reposplit)
	if repolevel == 1: # root of the tree, startdir = repodir
		for cat in settings.categories:
			path = os.path.join(startdir, cat)
			if not os.path.isdir(path):
				continue
			pkgdirs = os.listdir(path)
			scanlist.extend(AddPackagesInDir(path))
	elif repolevel == 2: # category level, startdir = catdir
		# we only want 1 segment of the directory, is why we use catdir instead of startdir
		catdir = reposplit[-2]
		if catdir not in settings.categories:
			logging.warn('%s is not a valid category according to profiles/categories, ' \
				'skipping checks in %s' % (catdir, catdir))
		else:
			scanlist = AddPackagesInDir(catdir)
	elif repolevel == 3: # pkgdir level, startdir = pkgdir
		catdir = reposplit[-2]
		pkgdir = reposplit[-1]
		if catdir not in settings.categories:
			logging.warn('%s is not a valid category according to profiles/categories, ' \
			'skipping checks in %s' % (catdir, catdir))
		else:
			path = os.path.join(catdir, pkgdir)
			logging.debug('adding %s to scanlist' % path)
			scanlist.append(path)
	return scanlist


def format_qa_output(formatter, stats, fails, dofull, dofail, options, qawarnings):
	"""Helper function that formats output properly
	
	Args:
		formatter - a subclass of Formatter
		stats - a dict of qa status items
		fails - a dict of qa status failures
		dofull - boolean to print full results or a summary
		dofail - boolean to decide if failure was hard or soft
	
	Returns:
		None (modifies formatter)
	"""
	full = options.mode == 'full'
	# we only want key value pairs where value > 0 
	for category, number in \
		filter(lambda myitem: myitem[1] > 0, iter(stats.items())):
		formatter.add_literal_data(_unicode_decode("  " + category.ljust(30)))
		if category in qawarnings:
			formatter.push_style("WARN")
		else:
			formatter.push_style("BAD")
		formatter.add_literal_data(_unicode_decode(str(number)))
		formatter.pop_style()
		formatter.add_line_break()
		if not dofull:
			if not full and dofail and category in qawarnings:
				# warnings are considered noise when there are failures
				continue
			fails_list = fails[category]
			if not full and len(fails_list) > 12:
				fails_list = fails_list[:12]
			for failure in fails_list:
				formatter.add_literal_data(_unicode_decode("   " + failure))
				formatter.add_line_break()


def editor_is_executable(editor):
	"""
	Given an EDITOR string, validate that it refers to
	an executable. This uses shlex_split() to split the
	first component and do a PATH lookup if necessary.

	@param editor: An EDITOR value from the environment.
	@type: string
	@rtype: bool
	@returns: True if an executable is found, False otherwise.
	"""
	editor_split = util.shlex_split(editor)
	if not editor_split:
		return False
	filename = editor_split[0]
	if not os.path.isabs(filename):
		return find_binary(filename) is not None
	return os.access(filename, os.X_OK) and os.path.isfile(filename)


def get_commit_message_with_editor(editor, message=None):
	"""
	Execute editor with a temporary file as it's argument
	and return the file content afterwards.

	@param editor: An EDITOR value from the environment
	@type: string
	@param message: An iterable of lines to show in the editor.
	@type: iterable
	@rtype: string or None
	@returns: A string on success or None if an error occurs.
	"""
	fd, filename = mkstemp()
	try:
		os.write(fd, _unicode_encode(_(
			"\n# Please enter the commit message " + \
			"for your changes.\n# (Comment lines starting " + \
			"with '#' will not be included)\n"),
			encoding=_encodings['content'], errors='backslashreplace'))
		if message:
			os.write(fd, b"#\n")
			for line in message:
				os.write(fd, _unicode_encode("#" + line,
					encoding=_encodings['content'], errors='backslashreplace'))
		os.close(fd)
		retval = os.system(editor + " '%s'" % filename)
		if not (os.WIFEXITED(retval) and os.WEXITSTATUS(retval) == os.EX_OK):
			return None
		try:
			mylines = io.open(_unicode_encode(filename,
				encoding=_encodings['fs'], errors='strict'),
				mode='r', encoding=_encodings['content'], errors='replace'
				).readlines()
		except OSError as e:
			if e.errno != errno.ENOENT:
				raise
			del e
			return None
		return "".join(line for line in mylines if not line.startswith("#"))
	finally:
		try:
			os.unlink(filename)
		except OSError:
			pass


def get_commit_message_with_stdin():
	"""
	Read a commit message from the user and return it.

	@rtype: string or None
	@returns: A string on success or None if an error occurs.
	"""
	print("Please enter a commit message. Use Ctrl-d to finish or Ctrl-c to abort.")
	commitmessage = []
	while True:
		commitmessage.append(sys.stdin.readline())
		if not commitmessage[-1]:
			break
	commitmessage = "".join(commitmessage)
	return commitmessage


def FindPortdir(settings):
	""" Try to figure out what repo we are in and whether we are in a regular
	tree or an overlay.
	
	Basic logic is:
	
	1. Determine what directory we are in (supports symlinks).
	2. Build a list of directories from / to our current location
	3. Iterate over PORTDIR_OVERLAY, if we find a match, search for a profiles directory
		 in the overlay.  If it has one, make it portdir, otherwise make it portdir_overlay.
	4. If we didn't find an overlay in PORTDIR_OVERLAY, see if we are in PORTDIR; if so, set
		 portdir_overlay to PORTDIR.  If we aren't in PORTDIR, see if PWD has a profiles dir, if
		 so, set portdir_overlay and portdir to PWD, else make them False.
	5. If we haven't found portdir_overlay yet, it means the user is doing something odd, report
		 an error.
	6. If we haven't found a portdir yet, set portdir to PORTDIR.
	
	Args:
		settings - portage.config instance, preferably repoman_settings
	Returns:
		list(portdir, portdir_overlay, location)
	"""

	portdir = None
	portdir_overlay = None
	location = os.getcwd()
	pwd = os.environ.get('PWD', '')
	if pwd and pwd != location and os.path.realpath(pwd) == location:
		# getcwd() returns the canonical path but that makes it hard for repoman to
		# orient itself if the user has symlinks in their portage tree structure.
		# We use os.environ["PWD"], if available, to get the non-canonical path of
		# the current working directory (from the shell).
		location = pwd

	location = normalize_path(location)

	path_ids = {}
	p = location
	s = None
	while True:
		s = os.stat(p)
		path_ids[(s.st_dev, s.st_ino)] = p
		if p == "/":
			break
		p = os.path.dirname(p)
	if location[-1] != "/":
		location += "/"

	for overlay in settings["PORTDIR_OVERLAY"].split():
		overlay = os.path.realpath(overlay)
		try:
			s = os.stat(overlay)
		except OSError:
			continue
		overlay = path_ids.get((s.st_dev, s.st_ino))
		if overlay is None:
			continue
		if overlay[-1] != "/":
			overlay += "/"
		if True:
			portdir_overlay = overlay
			subdir = location[len(overlay):]
			if subdir and subdir[-1] != "/":
				subdir += "/"
			if have_profile_dir(location, subdir.count("/")):
				portdir = portdir_overlay
			break

	# Couldn't match location with anything from PORTDIR_OVERLAY,
	# so fall back to have_profile_dir() checks alone. Assume that
	# an overlay will contain at least a "repo_name" file while a
	# master repo (portdir) will contain at least a "profiles.desc"
	# file.
	if not portdir_overlay:
		portdir_overlay = have_profile_dir(location, filename="repo_name")
		if portdir_overlay:
			subdir = location[len(portdir_overlay):]
			if subdir and subdir[-1] != os.sep:
				subdir += os.sep
			if have_profile_dir(location, subdir.count(os.sep)):
				portdir = portdir_overlay

	if not portdir_overlay:
		if (settings["PORTDIR"] + os.path.sep).startswith(location):
			portdir_overlay = settings["PORTDIR"]
		else:
			portdir_overlay = have_profile_dir(location)
		portdir = portdir_overlay
	
	if not portdir_overlay:
		msg = 'Repoman is unable to determine PORTDIR or PORTDIR_OVERLAY' + \
			' from the current working directory'
		logging.critical(msg)
		return (None, None, None)

	if not portdir:
		portdir = settings["PORTDIR"]

	if not portdir_overlay.endswith('/'):
		portdir_overlay += '/'
	
	if not portdir.endswith('/'):
		portdir += '/'

	return [normalize_path(x) for x in (portdir, portdir_overlay, location)]

def FindVCS():
	""" Try to figure out in what VCS' working tree we are. """

	outvcs = []

	def seek(depth = None):
		""" Seek for VCSes that have a top-level data directory only. """
		retvcs = []
		pathprep = ''

		while depth is None or depth > 0:
			if os.path.isdir(os.path.join(pathprep, '.git')):
				retvcs.append('git')
			if os.path.isdir(os.path.join(pathprep, '.bzr')):
				retvcs.append('bzr')
			if os.path.isdir(os.path.join(pathprep, '.hg')):
				retvcs.append('hg')
			if os.path.isdir(os.path.join(pathprep, '.svn')):  # >=1.7
				retvcs.append('svn')

			if retvcs:
				break
			pathprep = os.path.join(pathprep, '..')
			if os.path.realpath(pathprep).strip('/') == '':
				break
			if depth is not None:
				depth = depth - 1

		return retvcs

	# Level zero VCS-es.
	if os.path.isdir('CVS'):
		outvcs.append('cvs')
	if os.path.isdir('.svn'):  # <1.7
		outvcs.append('svn')

	# If we already found one of 'level zeros', just take a quick look
	# at the current directory. Otherwise, seek parents till we get
	# something or reach root.
	if outvcs:
		outvcs.extend(seek(1))
	else:
		outvcs = seek()

	return outvcs

def UpdateChangeLog(pkgdir, category, package, new, removed, changed, msg, pretend):
	""" Write an entry to an existing ChangeLog, or create a new one. """

	# figure out who to write as
	if 'GENTOO_COMMITTER_NAME' in os.environ and \
			'GENTOO_COMMITTER_EMAIL' in os.environ:
		user = '%s <%s>' % (os.environ['GENTOO_COMMITTER_NAME'], \
				os.environ['GENTOO_COMMITTER_EMAIL'])
	elif 'GENTOO_AUTHOR_NAME' in os.environ and \
			'GENTOO_AUTHOR_EMAIL' in os.environ:
		user = '%s <%s>' % (os.environ['GENTOO_AUTHOR_NAME'], \
				os.environ['GENTOO_AUTHOR_EMAIL'])
	elif 'ECHANGELOG_USER' in os.environ:
		user = os.environ['ECHANGELOG_USER']
	else:
		(login, _, _, _, gecos, _, _) = pwd.getpwuid(os.getuid())
		gecos = gecos.split(',')[0]  # bug #80011
		user = '%s <%s@gentoo.org>' % (gecos, login)

	if '<root@' in user:
		err = 'Please set ECHANGELOG_USER or run as non-root'
		logging.critical(err)
		return None

	cl_path = os.path.join(pkgdir, 'ChangeLog')
	f, clnew_path = mkstemp()

	# create an empty ChangeLog.new with correct header first
	try:
		f = os.fdopen(f, 'w+')
		f.write('# ChangeLog for %s/%s\n' % (category, package))
		year = time.strftime('%Y')
		f.write('# Copyright 1999-%s Gentoo Foundation; Distributed under the GPL v2\n' % year)
		f.write('# $Header: $\n')
		f.write('\n')

		# write new ChangeLog entry
		date = time.strftime('%d %b %Y')
		newebuild = False
		for fn in new:
			if not fn.endswith('.ebuild'):
				continue
			ebuild = fn.split(os.sep)[-1][0:-7] 
			f.write('*%s (%s)\n' % (ebuild, date))
			newebuild = True
		if newebuild:
			f.write('\n')
		new = ['+' + elem for elem in new if elem not in ['ChangeLog', 'Manifest']]
		removed = ['-' + elem for elem in removed]
		changed = [elem for elem in changed if elem not in ['ChangeLog', 'Manifest']]
		mesg = '%s; %s %s:' % (date, user, \
				', '.join(chain(new,removed,changed)))
		for line in textwrap.wrap(mesg, 80, \
				initial_indent='  ', subsequent_indent='  ', \
				break_on_hyphens=False):
			f.write('%s\n' % line)
		for line in textwrap.wrap(msg, 80, \
				initial_indent='  ', subsequent_indent='  '):
			f.write('%s\n' % line)

		# append stuff from old ChangeLog
		cl_lines = []
		if os.path.exists(cl_path):
			c = open(cl_path, 'r')
			cl_lines = c.readlines()
			for index, line in enumerate(cl_lines):
				# skip the headers
				if line.startswith('#'):
					# normalise to $Header: $ to avoid pointless diff line
					if line.startswith('# $Header:'):
						cl_lines[index] = '# $Header: $\n'
					continue
				f.write(line)
			c.close()

		# show diff (do we want to keep on doing this, or only when
		# pretend?)
		f.seek(0)
		clnew_lines = f.readlines()
		for line in difflib.unified_diff(cl_lines, clnew_lines, \
				fromfile=cl_path, tofile=cl_path + '.new', n=0):
			print(line.rstrip())
		print()

		f.close()

		if pretend:
			# remove what we've done
			os.remove(clnew_path)
		else:
			# rename ChangeLog.new to ChangeLog
			shutil.move(clnew_path, cl_path)

		if cl_lines == []:
			return True
		else:
			return False
	except IOError as e:
		err = 'Repoman is unable to create/write to Changelog.new file: %s' % (e,)
		logging.critical(err)
		# try to remove if possible
		try:
			os.remove(clnew_path)
		except OSError:
			pass
		return None