← Back
Editing: utilities.py
# This file is part of the sos project: https://github.com/sosreport/sos # # This copyrighted material is made available to anyone wishing to use, # modify, copy, or redistribute it subject to the terms and conditions of # version 2 of the GNU General Public License. # # See the LICENSE file in the source distribution for further information. import os import pwd import re import inspect from subprocess import Popen, PIPE, STDOUT import logging import fnmatch import errno import shlex import glob import tempfile import threading import time import io import mmap from contextlib import closing from collections import deque try: from packaging.version import parse as parse_version except ImportError: from pkg_resources import parse_version log = logging.getLogger('sos') # try loading magic>=0.4.20 which implements detect_from_filename method magic_mod = False try: import magic magic.detect_from_filename(__file__) magic_mod = True except (ImportError, AttributeError): from textwrap import fill msg = """\ WARNING: Failed to load 'magic' module version >= 0.4.20 which sos aims to \ use for detecting binary files. A less effective method will be used. It is \ recommended to install proper python3-magic package with the module. """ log.warning(f'\n{fill(msg, 72, replace_whitespace=False)}\n') TIMEOUT_DEFAULT = 300 __all__ = [ 'TIMEOUT_DEFAULT', 'ImporterHelper', 'SoSTimeoutError', 'TempFileUtil', 'bold', 'file_is_binary', 'fileobj', 'find', 'get_human_readable', 'grep', 'import_module', 'is_executable', 'listdir', 'parse_version', 'path_exists', 'path_isdir', 'path_isfile', 'path_islink', 'path_join', 'recursive_dict_values_by_key', 'shell_out', 'sos_get_command_output', 'tac_logs', 'tail', ] def format_version_to_pep440(ver): """ Convert the version into a PEP440 compliant version scheme.""" public_version_re = re.compile( r"^([0-9][0-9.]*(?:(?:a|b|rc|.post|.dev)[0-9]+)*)\+?" ) try: _, public, local = public_version_re.split(ver, maxsplit=1) if not local: return ver sanitized_local = re.sub("[+~]+", ".", local).strip("-") pep440_version = f"{public}+{sanitized_local}" return pep440_version except Exception as err: log.debug(f"Unable to format {ver} to pep440 format: {err}") return ver def sos_parse_version(ver, pep440=True): """ Converts the version to PEP440 format before parsing """ if pep440: ver_pep440 = format_version_to_pep440(ver) return parse_version(ver_pep440) return parse_version(ver) def tail(filename, number_of_bytes): """Returns the last number_of_bytes of filename""" with open(filename, "rb") as f: if os.stat(filename).st_size > number_of_bytes: f.seek(-number_of_bytes, 2) return f.read() def fileobj(path_or_file, mode='r'): """Returns a file-like object that can be used as a context manager""" if isinstance(path_or_file, str): try: return open(path_or_file, mode, encoding='utf-8') except IOError: log.debug(f"fileobj: {path_or_file} could not be opened") return closing(io.StringIO()) else: return closing(path_or_file) def convert_bytes(num_bytes): """Converts a number of bytes to a shorter, more human friendly format""" sizes = {'T': 1 << 40, 'G': 1 << 30, 'M': 1 << 20, 'K': 1 << 10} for symbol, size in sizes.items(): if num_bytes >= size: return f"{float(num_bytes) / size:.1f}{symbol}" return f"{num_bytes}" def file_is_certificate(fname): """Helper to determine if a given file is a certificate or not. This is especially helpful for `sos clean`, which cannot obfuscate certificate files content and instead, by default, will keep them as is. :param fname: The full path of the file to check :type fname: ``str`` :returns: The type of the certificate or ``None`` :rtype: ``string`` or ``None`` """ if fname[-4:] in [".csr", ".crt", ".pem"]: with open(fname, 'r', encoding='utf-8') as f: content = f.read() if re.search(r'-----BEGIN CERTIFICATE-----', content): return "certificate" if re.search(r'-----BEGIN [A-Z]+ PRIVATE KEY-----', content): return "certificatekey" return None return None def file_is_binary(fname): """Helper to determine if a given file contains binary content or not. This is especially helpful for `sos clean`, which cannot obfuscate binary data and instead, by default, will remove binary files. :param fname: The full path of the file to check binaryness of :type fname: ``str`` :returns: True if binary, else False :rtype: ``bool`` """ if magic_mod: try: _ftup = magic.detect_from_filename(fname) _mimes = ['text/', 'inode/'] return ( _ftup.encoding == 'binary' and not any(_ftup.mime_type.startswith(_mt) for _mt in _mimes) ) except Exception: pass # if for some reason the above check fails or magic>=0.4.20 is not present, # fail over to checking the very first byte of the file content with open(fname, 'tr', encoding='utf-8') as tfile: try: # when opened as above (tr), reading binary content will raise # an exception tfile.read(1) return False except UnicodeDecodeError: return True def find(file_pattern, top_dir, max_depth=None, path_pattern=None): """Generator function to find files recursively. Usage:: for filename in find("*.properties", "/var/log/foobar"): print filename """ if max_depth: base_depth = os.path.dirname(top_dir).count(os.path.sep) max_depth += base_depth for path, dirlist, filelist in os.walk(top_dir): if max_depth and path.count(os.path.sep) >= max_depth: del dirlist[:] if path_pattern and not fnmatch.fnmatch(path, path_pattern): continue for name in fnmatch.filter(filelist, file_pattern): yield os.path.join(path, name) def grep(pattern, *files_or_paths): """Returns lines matched in fnames, where fnames can either be pathnames to files to grep through or open file objects to grep through line by line""" matches = [] for fop in files_or_paths: with fileobj(fop) as fo: matches.extend((line for line in fo if re.match(pattern, line))) return matches def is_executable(command, sysroot=None): """Returns if a command matches an executable on the PATH""" paths = os.environ.get("PATH", "").split(os.path.pathsep) candidates = [command] + [os.path.join(p, command) for p in paths] if sysroot: candidates += [ os.path.join(sysroot, c.lstrip('/')) for c in candidates ] return any(os.access(path, os.X_OK) for path in candidates) def scrub_url_credential(url: str): """ Replace username:password@ with ********@ in proxy URL if present """ from urllib.parse import urlparse, urlunparse try: parsed_url = urlparse(url) if parsed_url.username or parsed_url.password: netloc = "********@" if parsed_url.hostname: netloc += parsed_url.hostname if parsed_url.port: netloc += f":{parsed_url.port}" return urlunparse(( parsed_url.scheme, netloc, parsed_url.path, parsed_url.params, parsed_url.query, parsed_url.fragment )) return url except Exception: # pylint: disable=broad-except return url def sos_get_command_output(command, timeout=TIMEOUT_DEFAULT, stderr=False, chroot=None, chdir=None, env=None, foreground=False, binary=False, sizelimit=None, poller=None, to_file=False, tac=False, runas=None): # pylint: disable=too-many-locals,too-many-branches """Execute a command and return a dictionary of status and output, optionally changing root or current working directory before executing command. """ # Change root or cwd for child only. Exceptions in the prexec_fn # closure are caught in the parent (chroot and chdir are bound from # the enclosing scope). def _child_prep_fn(): if chroot and chroot != '/': os.chroot(chroot) if runas: os.setgid(pwd.getpwnam(runas).pw_gid) os.setuid(pwd.getpwnam(runas).pw_uid) os.chdir(pwd.getpwnam(runas).pw_dir) if chdir: os.chdir(chdir) def _check_poller(proc): if poller() or proc.poll() == 124: proc.terminate() raise SoSTimeoutError time.sleep(0.01) if runas: try: pwd_user = pwd.getpwnam(runas) except KeyError: # no such user return {'status': 127, 'output': "", 'truncated': ''} env.update({ 'HOME': pwd_user.pw_dir, 'LOGNAME': runas, 'PWD': pwd_user.pw_dir, 'USER': runas }) cmd_env = os.environ.copy() # ensure consistent locale for collected command output cmd_env['LC_ALL'] = 'C.UTF-8' # optionally add an environment change for the command if env: for key, value in env.items(): if value: cmd_env[key] = value else: cmd_env.pop(key, None) # use /usr/bin/timeout to implement a timeout if timeout and is_executable("timeout"): command = (f"timeout {'--foreground' if foreground else ''} {timeout}s" f" {command}") args = shlex.split(command) # Expand arguments that are wildcard root paths. expanded_args = [] for arg in args: if arg.startswith("/") and "*" in arg: expanded_arg = glob.glob(arg) if expanded_arg: expanded_args.extend(expanded_arg) else: expanded_args.append(arg) else: expanded_args.append(arg) if to_file: if sizelimit: # going to use HeadReader _output = PIPE elif tac: # no limit but we need an intermediate file _output = tempfile.TemporaryFile(dir=os.path.dirname(to_file)) else: # pylint: disable=consider-using-with _output = open(to_file, 'wb') else: _output = PIPE try: with Popen(expanded_args, shell=False, stdout=_output, stderr=STDOUT if stderr else PIPE, bufsize=-1, env=cmd_env, close_fds=True, preexec_fn=_child_prep_fn) as p: if to_file: if sizelimit: if tac: _output = tempfile.TemporaryFile( dir=os.path.dirname(to_file) ) else: # pylint: disable=consider-using-with _output = open(to_file, 'wb') reader = HeadReader(p.stdout, _output, sizelimit, binary) else: reader = FakeReader(p, binary) else: reader = TailReader(p.stdout, sizelimit, binary) if poller: while reader.running: _check_poller(p) else: try: # override timeout=0 to timeout=None, as Popen will treat # the former as a literal 0-second timeout p.wait(timeout if timeout else None) except Exception: p.terminate() if to_file: if tac: with open(to_file, 'wb') as f_dst: tac_logs(_output, f_dst, True) # until we separate timeouts from the `timeout` command # handle per-cmd timeouts via Plugin status checks reader.running = False return {'status': 124, 'output': reader.get_contents(), 'truncated': reader.is_full} # wait for Popen to set the returncode while p.poll() is None: pass if to_file and tac: with open(to_file, 'wb') as f_dst: tac_logs(_output, f_dst, reader.is_full or p.returncode != 0) if p.returncode in (126, 127): stdout = b"" else: stdout = reader.get_contents() return { 'status': p.returncode, 'output': stdout, 'truncated': reader.is_full } except OSError as e: if e.errno == errno.ENOENT: return {'status': 127, 'output': "", 'truncated': ''} raise e finally: if hasattr(_output, 'close'): _output.close() def tac_logs(f_src, f_dst, drop_last_log=False): """Python implementation of the tac utility with support for multiline logs (starting with space). It is intended to reverse the output of 'journalctl --reverse'. """ NEWLINE_B = b'\n' NEWLINE_I = 10 SPACE_I = 32 # make sure all python/libc buffers are flushed # else fstat()/mmap() might see partial data f_src.flush() if os.fstat(f_src.fileno()).st_size == 0: return with mmap.mmap(f_src.fileno(), 0, access=mmap.ACCESS_READ) as mm: sep1 = sep2 = mm.size()-1 if mm[sep2] != NEWLINE_I: drop_last_log = True while sep2 >= 0: sep1 = mm.rfind(NEWLINE_B, 0, sep1) # multiline logs have a first line not starting with space # followed by lines starting with spaces # line 5 # line 4 # multiline 4 # line 3 if mm[sep1+1] == SPACE_I: # first line starts with a space # (this should not happen) if sep1 == -1: break # go find the previous NEWLINE continue # When we truncate or timeout, the last log # might be a partial multiline log if drop_last_log: drop_last_log = False else: # write the (multi)line log ending with the NEWLINE f_dst.write(mm[sep1+1:sep2+1]) sep2 = sep1 def import_module(module_fqname, superclasses=None): """Imports the module module_fqname and returns a list of defined classes from that module. If superclasses is defined then the classes returned will be subclasses of the specified superclass or superclasses. If superclasses is plural it must be a tuple of classes.""" module_name = module_fqname.rpartition(".")[-1] try: module = __import__(module_fqname, globals(), locals(), [module_name]) except ImportError as e: print(f'Error while trying to load module {module_fqname}: ' f' {e.__class__.__name__}') raise e modules = [class_ for cname, class_ in inspect.getmembers(module, inspect.isclass) if class_.__module__ == module_fqname] if superclasses: modules = [m for m in modules if issubclass(m, superclasses)] return modules def shell_out(cmd, timeout=30, chroot=None, runat=None): """Shell out to an external command and return the output or the empty string in case of error. """ return sos_get_command_output(cmd, timeout=timeout, chroot=chroot, chdir=runat)['output'] def get_human_readable(size, precision=2): # Credit to Pavan Gupta https://stackoverflow.com/questions/5194057/ suffixes = ['B', 'KiB', 'MiB', 'GiB', 'TiB'] suffixindex = 0 while size > 1024 and suffixindex < 4: suffixindex += 1 size = size/1024.0 return f"{size:.{precision}f}{suffixes[suffixindex]}" def _os_wrapper(path, sysroot, method, module=os.path): if sysroot and sysroot != os.sep: if not path.startswith(sysroot): path = os.path.join(sysroot, path.lstrip('/')) _meth = getattr(module, method) return _meth(path) def path_exists(path, sysroot): if '*' in path: return _os_wrapper(path, sysroot, 'glob', module=glob) return _os_wrapper(path, sysroot, 'exists') def path_isdir(path, sysroot): return _os_wrapper(path, sysroot, 'isdir') def path_isfile(path, sysroot): return _os_wrapper(path, sysroot, 'isfile') def path_islink(path, sysroot): return _os_wrapper(path, sysroot, 'islink') def listdir(path, sysroot): return _os_wrapper(path, sysroot, 'listdir', os) def path_join(path, *p, sysroot=os.sep): if sysroot and not path.startswith(sysroot): path = os.path.join(sysroot, path.lstrip(os.sep)) return os.path.join(path, *p) def bold(text): """Helper to make text bold in console output, without pulling in dependencies to the project unneccessarily. :param text: The text to make bold :type text: ``str`` :returns: The text wrapped in the ASCII codes to display as bold :rtype: ``str`` """ return '\033[1m' + text + '\033[0m' def recursive_dict_values_by_key(dobj, keys=[]): """Recursively compile all elements of a potentially nested dict by a set of keys. If a given key is a dict within ``dobj``, then _all_ elements within that dict, regardless of child keys, will be returned. For example, if a Plugin searches the devices dict for the 'storage' key, then all storage devices under the that dict (e.g. block, fibre, etc...) will be returned. However, if the Plugin specifies 'block' via ``keys``, then only the block devices within the devices['storage'] dict will be returned. Any elements passed here that are _not_ keys within the dict or any nested dicts will also be returned. :param dobj: The 'top-level' dict to intially search by :type dobj: ``dict`` :param keys: Which keys to compile elements from within ``dobj``. If no keys are given, all nested elements are returned :param keys: ``list`` of ``str`` :returns: All elements within the dict and any nested dicts :rtype: ``list`` """ _items = [] _filt = [] _items.extend(keys) if isinstance(dobj, dict): for k, v in dobj.items(): _filt.append(k) # get everything below this key, including nested dicts if not keys or k in keys: _items.extend(recursive_dict_values_by_key(v)) # recurse into this dict only for dict keys that match what # we're looking for elif isinstance(v, dict): try: # this will return a nested list, extract it _items.extend( recursive_dict_values_by_key( v[key] for key in keys if key in v )[0] ) except IndexError: # none of the keys given exist in the nested dict pass _filt.extend(v.keys()) else: _items.extend(dobj) return [d for d in _items if d not in _filt] class FakeReader(): """Used when we are writing directly to disk without sizelimits, this allows us to keep more simplified flows for executing, monitoring, and collecting command output. """ def __init__(self, process, binary): self.process = process self.binary = binary @property def is_full(self): return False def get_contents(self): return '' if not self.binary else b'' @property def running(self): return self.process.poll() is None class HeadReader(threading.Thread): """Used to 'head' the command output (f_src) to a given size without deadlocking sos. Takes a sizelimit value in MB. """ COPY_BUFSIZE = 1024*1024 def __init__(self, f_src, f_dst, sizelimit, binary): super().__init__() self.f_src = f_src self.f_dst = f_dst self.remaining = sizelimit * 1048576 # convert to bytes self.binary = binary self.running = True self.start() def run(self): """Reads from the f_src (Popen stdout pipe) until we reach sizelimit. once done, close f_src to signal the program that we are done. """ while self.remaining > 0: buf = self.f_src.read(min(self.remaining, self.COPY_BUFSIZE)) if not buf: break self.f_dst.write(buf) self.remaining -= len(buf) self.f_src.close() self.running = False def get_contents(self): return '' if not self.binary else b'' @property def is_full(self): return self.remaining <= 0 class TailReader(threading.Thread): """Used to tail the command output to a given size without deadlocking sos. Takes a sizelimit value in MB, and will compile stdout from Popen into a string that is limited to the given sizelimit. """ def __init__(self, channel, sizelimit, binary): super().__init__() self.chan = channel self.binary = binary self.chunksize = 2048 self.slots = None if sizelimit: sizelimit = sizelimit * 1048576 # convert to bytes self.slots = int(sizelimit / self.chunksize) self.deque = deque(maxlen=self.slots) self.running = True self.start() def run(self): """Reads from the channel (pipe) that is the output pipe for a called Popen. As we are reading from the pipe, the output is added to a deque. After the size of the deque exceeds the sizelimit earlier (older) entries are removed. This means the returned output is chunksize-sensitive, but is not really byte-sensitive. """ try: while True: line = self.chan.read(self.chunksize) if not line: # Pipe can remain open after output has completed break self.deque.append(line) except (ValueError, IOError): # pipe has closed, meaning command output is done pass self.running = False def get_contents(self): """Returns the contents of the deque as a string""" # block until command completes or timesout (separate from the plugin # hitting a timeout) while self.running: time.sleep(0.01) if not self.binary: return ''.join(ln.decode('utf-8', 'ignore') for ln in self.deque) return b''.join(ln for ln in self.deque) @property def is_full(self): """Checks if the deque is full, implying that output was truncated""" if not self.slots: return False return len(self.deque) == self.slots class ImporterHelper: """Provides a list of modules that can be imported in a package. Importable modules are located along the module __path__ list and modules are files that end in .py. """ def __init__(self, package): """package is a package module import my.package.module helper = ImporterHelper(my.package.module)""" self.package = package def _plugin_name(self, path): "Returns the plugin module name given the path" base = os.path.basename(path) name, _ = os.path.splitext(base) return name def _get_plugins_from_list(self, list_): plugins = [self._plugin_name(plugin) for plugin in list_ if "__init__" not in plugin and plugin.endswith(".py")] plugins.sort() return plugins def _find_plugins_in_dir(self, path): if os.path.exists(path): py_files = list(find("*.py", path)) pnames = self._get_plugins_from_list(py_files) if pnames: return pnames return [] def get_modules(self): """Returns the list of importable modules in the configured python package. """ plugins = [] for path in self.package.__path__: if os.path.isdir(path): plugins.extend(self._find_plugins_in_dir(path)) return plugins class TempFileUtil(): def __init__(self, tmp_dir): self.tmp_dir = tmp_dir self.files = [] def new(self): fd, fname = tempfile.mkstemp(dir=self.tmp_dir) # avoid TOCTOU race by using os.fdopen() fobj = os.fdopen(fd, 'w+') self.files.append((fname, fobj)) return fobj def clean(self): for fname, f in self.files: try: f.flush() f.close() except Exception: # file already closed or potentially already removed, ignore pass try: os.unlink(fname) except Exception: # if the above failed, this is also likely to fail, ignore pass self.files = [] class SoSTimeoutError(OSError): pass # vim: set et ts=4 sw=4 :
Save File
Cancel