def test_parse_deps():
for (arg, exp) in (('a, b < 1.0 c = 5:2.0-3 d',
[('a', 0, (None, None, None)),
('b', rpm.RPMSENSE_LESS, (None, '1.0', None)),
('c', rpm.RPMSENSE_EQUAL, ('5', '2.0', '3')),
('d', 0, (None, None, None))]), ):
assert Pkg.parse_deps(arg) == exp
def test_parse_deps():
for (arg, exp) in (("a, b < 1.0 c = 5:2.0-3 d",
[("a", 0, (None, None, None)),
("b", rpm.RPMSENSE_LESS, (None, "1.0", None)),
("c", rpm.RPMSENSE_EQUAL, ("5", "2.0", "3")),
("d", 0, (None, None, None))]), ):
assert Pkg.parse_deps(arg) == exp
def check_spec(self, pkg, spec_file, spec_lines=None):
self._spec_file = spec_file
spec_only = isinstance(pkg, Pkg.FakePkg)
if not spec_lines:
spec_lines = Pkg.readlines(spec_file)
patches = {}
applied_patches = []
applied_patches_ifarch = []
patches_auto_applied = False
source_dir = False
buildroot = False
configure_linenum = None
configure_cmdline = ''
mklibname = False
is_lib_pkg = False
if_depth = 0
ifarch_depth = -1
current_section = 'package'
buildroot_clean = {'clean': False, 'install': False}
depscript_override = False
depgen_disabled = False
patch_fuzz_override = False
indent_spaces = 0
indent_tabs = 0
section = {}
# None == main package
current_package = None
package_noarch = {}
if self._spec_file:
if not Pkg.is_utf8(self._spec_file):
self.output.add_info('E', pkg, 'non-utf8-spec-file',
self._spec_name or self._spec_file)
# gather info from spec lines
pkg.current_linenum = 0
nbsp = UNICODE_NBSP
for line in spec_lines:
pkg.current_linenum += 1
char = line.find(nbsp)
if char != -1:
self.output.add_info(
'W', pkg, 'non-break-space',
'line %s, char %d' % (pkg.current_linenum, char))
section_marker = False
for sec, regex in section_regexs.items():
res = regex.search(line)
if res:
current_section = sec
section_marker = True
section[sec] = section.get(sec, 0) + 1
if sec in ('package', 'files'):
rest = filelist_regex.sub('', line[res.end() - 1:])
res = pkgname_regex.search(rest)
if res:
current_package = res.group(1)
else:
current_package = None
break
if section_marker:
if not is_lib_pkg and lib_package_regex.search(line):
is_lib_pkg = True
continue
if (current_section in Pkg.RPM_SCRIPTLETS + ('prep', 'build')
and contains_buildroot(line)):
self.output.add_info('W', pkg, 'rpm-buildroot-usage',
'%' + current_section, line[:-1].strip())
if make_check_regex.search(line) and current_section not in \
('check', 'changelog', 'package', 'description'):
self.output.add_info('W', pkg,
'make-check-outside-check-section',
line[:-1])
if current_section in buildroot_clean and \
not buildroot_clean[current_section] and \
contains_buildroot(line) and rm_regex.search(line):
buildroot_clean[current_section] = True
if ifarch_regex.search(line):
if_depth = if_depth + 1
ifarch_depth = if_depth
if if_regex.search(line):
if_depth = if_depth + 1
if setup_regex.match(line):
if not setup_q_regex.search(line):
# Don't warn if there's a -T without -a or -b
if setup_t_regex.search(line):
if setup_ab_regex.search(line):
self.output.add_info('W', pkg, 'setup-not-quiet')
else:
self.output.add_info('W', pkg, 'setup-not-quiet')
if current_section != 'prep':
self.output.add_info('W', pkg, 'setup-not-in-prep')
elif autopatch_regex.search(line):
patches_auto_applied = True
if current_section != 'prep':
self.output.add_info('W', pkg, '%autopatch-not-in-prep')
else:
res = autosetup_regex.search(line)
if res:
if not autosetup_n_regex.search(res.group(1)):
patches_auto_applied = True
if current_section != 'prep':
self.output.add_info('W', pkg,
'%autosetup-not-in-prep')
if endif_regex.search(line):
if ifarch_depth == if_depth:
ifarch_depth = -1
if_depth = if_depth - 1
res = applied_patch_regex.search(line)
if res:
pnum = res.group(1) or 0
for tmp in applied_patch_p_regex.findall(line) or [pnum]:
pnum = int(tmp)
applied_patches.append(pnum)
if ifarch_depth > 0:
applied_patches_ifarch.append(pnum)
else:
res = applied_patch_pipe_regex.search(line)
if res:
pnum = int(res.group(1))
applied_patches.append(pnum)
if ifarch_depth > 0:
applied_patches_ifarch.append(pnum)
else:
res = applied_patch_i_regex.search(line)
if res:
pnum = int(res.group(1))
applied_patches.append(pnum)
if ifarch_depth > 0:
applied_patches_ifarch.append(pnum)
if not res and not source_dir:
res = source_dir_regex.search(line)
if res:
source_dir = True
self.output.add_info('E', pkg, 'use-of-RPM_SOURCE_DIR')
if configure_linenum:
if configure_cmdline[-1] == '\\':
configure_cmdline = configure_cmdline[:-1] + line.strip()
else:
res = configure_libdir_spec_regex.search(configure_cmdline)
if not res:
# Hack to get the correct (start of ./configure) line
# number displayed:
real_linenum = pkg.current_linenum
pkg.current_linenum = configure_linenum
self.output.add_info('W', pkg,
'configure-without-libdir-spec')
pkg.current_linenum = real_linenum
elif res.group(1):
res = re.match(hardcoded_library_paths, res.group(1))
if res:
self.output.add_info('E', pkg,
'hardcoded-library-path',
res.group(1),
'in configure options')
configure_linenum = None
hashPos = line.find('#')
if current_section != 'changelog':
cfgPos = line.find('./configure')
if cfgPos != -1 and (hashPos == -1 or hashPos > cfgPos):
# store line where it started
configure_linenum = pkg.current_linenum
configure_cmdline = line.strip()
res = hardcoded_library_path_regex.search(line)
if current_section != 'changelog' and res and not \
(biarch_package_regex.match(pkg.name) or
self.hardcoded_lib_path_exceptions_regex.search(
res.group(1).lstrip())):
self.output.add_info('E', pkg, 'hardcoded-library-path', 'in',
res.group(1).lstrip())
if '%mklibname' in line:
mklibname = True
if current_section == 'package':
# Would be cleaner to get sources and patches from the
# specfile parsed in Python (see below), but we want to
# catch %ifarch'd etc ones as well, and also catch these when
# the specfile is not parseable.
res = patch_regex.search(line)
if res:
pnum = int(res.group(1) or 0)
patches[pnum] = res.group(2)
res = obsolete_tags_regex.search(line)
if res:
self.output.add_info('W', pkg, 'obsolete-tag',
res.group(1))
res = buildroot_regex.search(line)
if res:
buildroot = True
if res.group(1).startswith('/'):
self.output.add_info(
'W', pkg, 'hardcoded-path-in-buildroot-tag',
res.group(1))
res = buildarch_regex.search(line)
if res:
if res.group(1) != 'noarch':
self.output.add_info(
'E', pkg, 'buildarch-instead-of-exclusivearch-tag',
res.group(1))
else:
package_noarch[current_package] = True
res = packager_regex.search(line)
if res:
self.output.add_info('W', pkg, 'hardcoded-packager-tag',
res.group(1))
res = prefix_regex.search(line)
if res:
if not res.group(1).startswith('%'):
self.output.add_info('W', pkg, 'hardcoded-prefix-tag',
res.group(1))
res = prereq_regex.search(line)
if res:
self.output.add_info('E', pkg, 'prereq-use', res.group(2))
res = buildprereq_regex.search(line)
if res:
self.output.add_info('E', pkg, 'buildprereq-use',
res.group(1))
if scriptlet_requires_regex.search(line):
self.output.add_info(
'E', pkg, 'broken-syntax-in-scriptlet-requires',
line.strip())
res = requires_regex.search(line)
if res:
reqs = Pkg.parse_deps(res.group(1))
for req in unversioned(reqs):
if compop_regex.search(req):
self.output.add_info(
'W', pkg, 'comparison-operator-in-deptoken',
req)
res = provides_regex.search(line)
if res:
provs = Pkg.parse_deps(res.group(1))
for prov in unversioned(provs):
if not prov.startswith('/'):
self.output.add_info(
'W', pkg, 'unversioned-explicit-provides',
prov)
if compop_regex.search(prov):
self.output.add_info(
'W', pkg, 'comparison-operator-in-deptoken',
prov)
res = obsoletes_regex.search(line)
if res:
obses = Pkg.parse_deps(res.group(1))
for obs in unversioned(obses):
if not obs.startswith('/'):
self.output.add_info(
'W', pkg, 'unversioned-explicit-obsoletes',
obs)
if compop_regex.search(obs):
self.output.add_info(
'W', pkg, 'comparison-operator-in-deptoken',
obs)
res = conflicts_regex.search(line)
if res:
confs = Pkg.parse_deps(res.group(1))
for conf in unversioned(confs):
if compop_regex.search(conf):
self.output.add_info(
'W', pkg, 'comparison-operator-in-deptoken',
conf)
if current_section == 'changelog':
for match in macro_regex.findall(line):
res = re.match('%+', match)
if len(res.group(0)) % 2:
self.output.add_info('W', pkg, 'macro-in-%changelog',
match)
else:
if not depscript_override:
depscript_override = \
depscript_override_regex.search(line) is not None
if not depgen_disabled:
depgen_disabled = \
depgen_disable_regex.search(line) is not None
if not patch_fuzz_override:
patch_fuzz_override = \
patch_fuzz_override_regex.search(line) is not None
if current_section == 'files':
# TODO: check scriptlets for these too?
if package_noarch.get(current_package) or \
(current_package not in package_noarch and
package_noarch.get(None)):
res = libdir_regex.search(line)
if res:
pkgname = current_package
if pkgname is None:
pkgname = '(main package)'
self.output.add_info('W', pkg,
'libdir-macro-in-noarch-package',
pkgname, line.rstrip())
if not indent_tabs and '\t' in line:
indent_tabs = pkg.current_linenum
if not indent_spaces and indent_spaces_regex.search(line):
indent_spaces = pkg.current_linenum
# Check if egrep or fgrep is used
if current_section not in \
('package', 'changelog', 'description', 'files'):
greps = deprecated_grep_regex.findall(line)
if greps:
self.output.add_info('W', pkg, 'deprecated-grep', greps)
# If not checking spec file only, we're checking one inside a
# SRPM -> skip this check to avoid duplicate warnings (#167)
if spec_only and self.valid_groups and \
line.lower().startswith('group:'):
group = line[6:].strip()
if group not in self.valid_groups:
self.output.add_info('W', pkg, 'non-standard-group', group)
# Test if there are macros in comments
if hashPos != -1 and \
(hashPos == 0 or line[hashPos - 1] in (' ', '\t')):
for match in macro_regex.findall(line[hashPos + 1:]):
res = re.match('%+', match)
if len(res.group(0)) % 2:
self.output.add_info('W', pkg, 'macro-in-comment',
match)
# Last line read is not useful after this point
pkg.current_linenum = None
for sect in (x for x in buildroot_clean if not buildroot_clean[x]):
self.output.add_info('W', pkg, 'no-cleaning-of-buildroot',
'%' + sect)
if not buildroot:
self.output.add_info('W', pkg, 'no-buildroot-tag')
for sec in ('prep', 'build', 'install', 'clean'):
if not section.get(sec):
self.output.add_info('W', pkg, 'no-%%%s-section' % sec)
for sec in ('changelog', ):
# prep, build, install, clean, check prevented by rpmbuild 4.4
if section.get(sec, 0) > 1:
self.output.add_info('W', pkg,
'more-than-one-%%%s-section' % sec)
if is_lib_pkg and not mklibname:
self.output.add_info('E', pkg, 'lib-package-without-%mklibname')
if depscript_override and not depgen_disabled:
self.output.add_info('W', pkg,
'depscript-without-disabling-depgen')
if patch_fuzz_override:
self.output.add_info('W', pkg, 'patch-fuzz-is-changed')
if indent_spaces and indent_tabs:
pkg.current_linenum = max(indent_spaces, indent_tabs)
self.output.add_info(
'W', pkg, 'mixed-use-of-spaces-and-tabs',
'(spaces: line %d, tab: line %d)' %
(indent_spaces, indent_tabs))
pkg.current_linenum = None
# process gathered info
if not patches_auto_applied:
for pnum, pfile in patches.items():
if pnum in applied_patches_ifarch:
self.output.add_info('W', pkg, '%ifarch-applied-patch',
'Patch%d:' % pnum, pfile)
if pnum not in applied_patches:
self.output.add_info('W', pkg, 'patch-not-applied',
'Patch%d:' % pnum, pfile)
# Rest of the checks require a real spec file
if not self._spec_file:
return
# We'd like to parse the specfile only once using python bindings,
# but it seems errors from rpmlib get logged to stderr and we can't
# capture and print them nicely, so we do it once each way :P
out = Pkg.getstatusoutput(
('rpm', '-q', '--qf=', '-D', '_sourcedir %s' % pkg.dirName(),
'--specfile', self._spec_file))
parse_error = False
for line in out[1].splitlines():
# No such file or dir hack: https://bugzilla.redhat.com/487855
if 'No such file or directory' not in line:
parse_error = True
self.output.add_info('E', pkg, 'specfile-error', line)
if not parse_error:
# grab sources and patches from parsed spec object to get
# them with macros expanded for URL checking
spec_obj = None
rpm.addMacro('_sourcedir', pkg.dirName())
try:
ts = rpm.TransactionSet()
spec_obj = ts.parseSpec(self._spec_file)
except (ValueError, rpm.error):
# errors logged above already
pass
rpm.delMacro('_sourcedir')
if spec_obj:
try:
# rpm < 4.8.0
sources = spec_obj.sources()
except TypeError:
# rpm >= 4.8.0
sources = spec_obj.sources
for src in sources:
(url, num, flags) = src
(scheme, netloc) = urlparse(url)[0:2]
if flags & 1: # rpmspec.h, rpm.org ticket #123
srctype = 'Source'
else:
srctype = 'Patch'
tag = '%s%s' % (srctype, num)
if scheme and netloc:
info = self.check_url(pkg, tag, url)
if not info or not hasattr(pkg, 'files'):
continue
clen = info.get('Content-Length')
if clen is not None:
clen = int(clen)
cmd5 = info.get('Content-MD5')
if cmd5 is not None:
cmd5 = cmd5.lower()
if clen is not None or cmd5 is not None:
# Not using path from urlparse results to match how
# rpm itself parses the basename.
pkgfile = pkg.files().get(url.split('/')[-1])
if pkgfile:
if clen is not None and pkgfile.size != clen:
self.output.add_info(
'W', pkg, 'file-size-mismatch',
'%s = %s, %s = %s' %
(pkgfile.name, pkgfile.size, url,
clen))
# pkgfile.md5 could be some other digest than
# MD5, treat as MD5 only if it's 32 chars long
if cmd5 and len(pkgfile.md5) == 32 \
and pkgfile.md5 != cmd5:
self.output.add_info(
'W', pkg, 'file-md5-mismatch',
'%s = %s, %s = %s' %
(pkgfile.name, pkgfile.md5, url, cmd5))
elif srctype == 'Source' and tarball_regex.search(url):
self.output.add_info('W', pkg, 'invalid-url',
'%s:' % tag, url)