path: root/config.py

#!/usr/bin/env python3

# © Copyright 2021-2022, Scott Gasch

"""Global configuration driven by commandline arguments, environment variables
and saved configuration files.  This works across several modules.

Usage:

    In your file.py::

        import config

        parser = config.add_commandline_args(
            "Module",
            "Args related to module doing the thing.",
        )
        parser.add_argument(
            "--module_do_the_thing",
            type=bool,
            default=True,
            help="Should the module do the thing?"
        )

    In your main.py::

        import config

        def main() -> None:
            parser = config.add_commandline_args(
                "Main",
                "A program that does the thing.",
            )
            parser.add_argument(
                "--dry_run",
                type=bool,
                default=False,
                help="Should we really do the thing?"
            )
            config.parse()   # Very important, this must be invoked!

    If you set this up and remember to invoke config.parse(), all commandline
    arguments will play nicely together.  This is done automatically for you
    if you're using the :meth:`bootstrap.initialize` decorator on
    your program's entry point.  See :meth:`python_modules.bootstrap.initialize`
    for more details.::

        import bootstrap

        @bootstrap.initialize
        def main():
            whatever

        if __name__ == '__main__':
            main()

    Either way, you'll get this behavior from the commandline::

        % main.py -h
        usage: main.py [-h]
                       [--module_do_the_thing MODULE_DO_THE_THING]
                       [--dry_run DRY_RUN]

        Module:
          Args related to module doing the thing.

          --module_do_the_thing MODULE_DO_THE_THING
                       Should the module do the thing?

        Main:
          A program that does the thing

          --dry_run
                       Should we really do the thing?

    Arguments themselves should be accessed via
    :code:`config.config['arg_name']`.  e.g.::

        if not config.config['dry_run']:
            module.do_the_thing()

"""

import argparse
import logging
import os
import pprint
import re
import sys
from typing import Any, Dict, List, Optional

from kazoo.client import KazooClient

import scott_secrets

# This module is commonly used by others in here and should avoid
# taking any unnecessary dependencies back on them.

# Defer logging messages until later when logging has been initialized.
SAVED_MESSAGES: List[str] = []

# Make a copy of the original program arguments.
PROGRAM_NAME: str = os.path.basename(sys.argv[0])
ORIG_ARGV: List[str] = sys.argv.copy()

# A zookeeper client that is lazily created so as to not incur the
# latency of connecting to zookeeper for programs that are not reading
# or writing their config data into zookeeper.
ZK: Optional[KazooClient] = None


class OptionalRawFormatter(argparse.HelpFormatter):
    """This formatter has the same bahavior as the normal argparse text
    formatter except when the help text of an argument begins with
    "RAW|".  In that case, the line breaks are preserved and the text
    is not wrapped.

    Use this, for example, when you need the helptext of an argument
    to have its spacing preserved exactly, e.g.::

        args.add_argument(
            '--mode',
            type=str,
            default='PLAY',
            choices=['CHEAT', 'AUTOPLAY', 'SELFTEST', 'PRECOMPUTE', 'PLAY'],
            metavar='MODE',
            help='''RAW|Our mode of operation.  One of:

                PLAY = play wordle with me!  Pick a random solution or
                       specify a solution with --template.

               CHEAT = given a --template and, optionally, --letters_in_word
                       and/or --letters_to_avoid, return the best guess word;

            AUTOPLAY = given a complete word in --template, guess it step
                       by step showing work;

            SELFTEST = autoplay every possible solution keeping track of
                       wins/losses and average number of guesses;

          PRECOMPUTE = populate hash table with optimal guesses.
            ''',
        )

    """

    def _split_lines(self, text, width):
        if text.startswith('RAW|'):
            return text[4:].splitlines()
        return argparse.HelpFormatter._split_lines(self, text, width)


# A global parser that we will collect arguments into.
ARGS = argparse.ArgumentParser(
    description=None,
    formatter_class=OptionalRawFormatter,
    fromfile_prefix_chars="@",
    epilog=f'{PROGRAM_NAME} uses config.py ({__file__}) for global, cross-module configuration setup and parsing.',
    # I don't fully understand why but when loaded by sphinx sometimes
    # the same module is loaded many times causing any arguments it
    # registers via module-level code to be redefined.  Work around
    # this iff the program is 'sphinx-build'
    conflict_handler='resolve' if PROGRAM_NAME == 'sphinx-build' else 'error',
)

# Keep track of if we've been called and prevent being called more
# than once.
CONFIG_PARSE_CALLED = False


# A global configuration dictionary that will contain parsed arguments.
# It is also this variable that modules use to access parsed arguments.
# This is the data that is most interesting to our callers; it will hold
# the configuration result.
config: Dict[str, Any] = {}

# It would be really nice if this shit worked from interactive python


def add_commandline_args(title: str, description: str = "") -> argparse._ArgumentGroup:
    """Create a new context for arguments and return a handle.

    Args:
        title: A title for your module's commandline arguments group.
        description: A helpful description of your module.

    Returns:
        An argparse._ArgumentGroup to be populated by the caller.
    """
    return ARGS.add_argument_group(title, description)


group = add_commandline_args(
    f'Global Config ({__file__})',
    'Args that control the global config itself; how meta!',
)
group.add_argument(
    '--config_loadfile',
    metavar='FILENAME',
    default=None,
    help='Config file (populated via --config_savefile) from which to read args in lieu or in addition to commandline.',
)
group.add_argument(
    '--config_dump',
    default=False,
    action='store_true',
    help='Display the global configuration (possibly derived from multiple sources) on STDERR at program startup.',
)
group.add_argument(
    '--config_savefile',
    type=str,
    metavar='FILENAME',
    default=None,
    help='Populate config file compatible with --config_loadfile to save global config for later use.',
)
group.add_argument(
    '--config_rejects_unrecognized_arguments',
    default=False,
    action='store_true',
    help=(
        'If present, config will raise an exception if it doesn\'t recognize an argument.  The '
        + 'default behavior is to ignore this so as to allow interoperability with programs that '
        + 'want to use their own argparse calls to parse their own, separate commandline args.'
    ),
)


def overwrite_argparse_epilog(msg: str) -> None:
    """Allows your code to override the default epilog created by
    argparse.

    Args:
        msg: The epilog message to substitute for the default.
    """
    ARGS.epilog = msg


def is_flag_already_in_argv(var: str) -> bool:
    """Returns true if a particular flag is passed on the commandline?

    Args:
        var: The flag to search for.
    """
    for _ in sys.argv:
        if var in _:
            return True
    return False


def _reorder_arg_action_groups_before_help(entry_module: Optional[str]):
    """Internal.  Used to reorder the arguments before dumping out a
    generated help string such that the main program's arguments come
    last.

    """
    reordered_action_groups = []
    for grp in ARGS._action_groups:
        if entry_module is not None and entry_module in grp.title:  # type: ignore
            reordered_action_groups.append(grp)
        elif PROGRAM_NAME in group.title:  # type: ignore
            reordered_action_groups.append(grp)
        else:
            reordered_action_groups.insert(0, grp)
    return reordered_action_groups


def print_usage() -> None:
    """Prints the normal help usage message out."""
    ARGS.print_help()


def usage() -> str:
    """
    Returns:
        program usage help text as a string.
    """
    return ARGS.format_usage()


def _augment_sys_argv_from_environment_variables():
    """Internal.  Look at the system environment for variables that match
    arg names.  This is done via some munging such that:

    :code:`--argument_to_match`

    ...is matched by:

    :code:`ARGUMENT_TO_MATCH`

    This allows programmers to set args via shell environment variables
    in lieu of passing them on the cmdline.

    """

    usage_message = usage()
    optional = False
    var = ''
    for x in usage_message.split():
        if x[0] == '[':
            optional = True
        if optional:
            var += f'{x} '
            if x[-1] == ']':
                optional = False
                var = var.strip()
                var = var.strip('[')
                var = var.strip(']')
                chunks = var.split()
                if len(chunks) > 1:
                    var = var.split()[0]

                # Environment vars the same as flag names without
                # the initial -'s and in UPPERCASE.
                env = var.strip('-').upper()
                if env in os.environ:
                    if not is_flag_already_in_argv(var):
                        value = os.environ[env]
                        SAVED_MESSAGES.append(f'Initialized from environment: {var} = {value}')
                        from string_utils import to_bool

                        if len(chunks) == 1 and to_bool(value):
                            sys.argv.append(var)
                        elif len(chunks) > 1:
                            sys.argv.append(var)
                            sys.argv.append(value)
                var = ''
                env = ''


def _augment_sys_argv_from_loadfile():
    """Internal.  Augment with arguments persisted in a saved file."""

    global ZK
    loadfile = None
    saw_other_args = False
    grab_next_arg = False
    for arg in sys.argv[1:]:
        if 'config_loadfile' in arg:
            pieces = arg.split('=')
            if len(pieces) > 1:
                loadfile = pieces[1]
            else:
                grab_next_arg = True
        elif grab_next_arg:
            loadfile = arg
        else:
            saw_other_args = True

    if loadfile is not None:
        zkpath = None
        if loadfile[:3] == 'zk:':
            try:
                if ZK is None:
                    ZK = KazooClient(
                        hosts=scott_secrets.ZOOKEEPER_NODES,
                        use_ssl=True,
                        verify_certs=False,
                        keyfile=scott_secrets.ZOOKEEPER_CLIENT_CERT,
                        keyfile_password=scott_secrets.ZOOKEEPER_CLIENT_PASS,
                        certfile=scott_secrets.ZOOKEEPER_CLIENT_CERT,
                    )
                    ZK.start()
                zkpath = loadfile[3:]
                if not zkpath.startswith('/config/'):
                    zkpath = '/config/' + zkpath
                    zkpath = re.sub(r'//+', '/', zkpath)
                if not ZK.exists(zkpath):
                    raise Exception(
                        f'ERROR: --config_loadfile argument must be a file, {loadfile} not found (in zookeeper)'
                    )
            except Exception as e:
                raise Exception(
                    f'ERROR: Error talking with zookeeper while looking for {loadfile}'
                ) from e
        elif not os.path.exists(loadfile):
            raise Exception(
                f'ERROR: --config_loadfile argument must be a file, {loadfile} not found.'
            )

        if saw_other_args:
            msg = f'Augmenting commandline arguments with those from {loadfile}.'
        else:
            msg = f'Reading commandline arguments from {loadfile}.'
        print(msg, file=sys.stderr)
        SAVED_MESSAGES.append(msg)

        newargs = []
        if zkpath:
            try:
                assert ZK
                contents = ZK.get(zkpath)[0]
                contents = contents.decode()
                newargs = [
                    arg.strip('\n') for arg in contents.split('\n') if 'config_savefile' not in arg
                ]
            except Exception as e:
                raise Exception(f'Error reading {zkpath} from zookeeper.') from e
            SAVED_MESSAGES.append(f'Loaded config from zookeeper from {zkpath}')
        else:
            with open(loadfile, 'r') as rf:
                newargs = rf.readlines()
            newargs = [arg.strip('\n') for arg in newargs if 'config_savefile' not in arg]
        sys.argv += newargs


def parse(entry_module: Optional[str]) -> Dict[str, Any]:
    """Main program should call this early in main().  Note that the
    :code:`bootstrap.initialize` wrapper takes care of this automatically.
    This should only be called once per program invocation.

    """
    global CONFIG_PARSE_CALLED
    if CONFIG_PARSE_CALLED:
        return config
    global ZK

    # If we're about to do the usage message dump, put the main
    # module's argument group last in the list (if possible) so that
    # when the user passes -h or --help, it will be visible on the
    # screen w/o scrolling.
    for arg in sys.argv:
        if arg in ('--help', '-h'):
            if entry_module is not None:
                entry_module = os.path.basename(entry_module)
            ARGS._action_groups = _reorder_arg_action_groups_before_help(entry_module)

    # Examine the environment for variables that match known flags.
    # For a flag called --example_flag the corresponding environment
    # variable would be called EXAMPLE_FLAG.  If found, hackily add
    # these into sys.argv to be parsed.
    _augment_sys_argv_from_environment_variables()

    # Look for loadfile and read/parse it if present.  This also
    # works by jamming these values onto sys.argv.
    _augment_sys_argv_from_loadfile()

    # Parse (possibly augmented, possibly completely overwritten)
    # commandline args with argparse normally and populate config.
    known, unknown = ARGS.parse_known_args()
    config.update(vars(known))

    # Reconstruct the argv with unrecognized flags for the benefit of
    # future argument parsers.  For example, unittest_main in python
    # has some of its own flags.  If we didn't recognize it, maybe
    # someone else will.
    if len(unknown) > 0:
        if config['config_rejects_unrecognized_arguments']:
            raise Exception(
                f'Encountered unrecognized config argument(s) {unknown} with --config_rejects_unrecognized_arguments enabled; halting.'
            )
        SAVED_MESSAGES.append(f'Config encountered unrecognized commandline arguments: {unknown}')
    sys.argv = sys.argv[:1] + unknown

    # Check for savefile and populate it if requested.
    savefile = config['config_savefile']
    if savefile and len(savefile) > 0:
        data = '\n'.join(ORIG_ARGV[1:])
        if savefile[:3] == 'zk:':
            zkpath = savefile[3:]
            if not zkpath.startswith('/config/'):
                zkpath = '/config/' + zkpath
                zkpath = re.sub(r'//+', '/', zkpath)
            try:
                if not ZK:
                    ZK = KazooClient(
                        hosts=scott_secrets.ZOOKEEPER_NODES,
                        use_ssl=True,
                        verify_certs=False,
                        keyfile=scott_secrets.ZOOKEEPER_CLIENT_CERT,
                        keyfile_password=scott_secrets.ZOOKEEPER_CLIENT_PASS,
                        certfile=scott_secrets.ZOOKEEPER_CLIENT_CERT,
                    )
                    ZK.start()
                if not ZK.exists(zkpath):
                    ZK.create(zkpath, data.encode())
                else:
                    ZK.set(zkpath, data.encode())
            except Exception as e:
                raise Exception(f'Failed to create zookeeper path {zkpath}') from e
            SAVED_MESSAGES.append(f'Saved config to zookeeper in {zkpath}')
        else:
            with open(savefile, 'w') as wf:
                wf.write(data)

    # Also dump the config on stderr if requested.
    if config['config_dump']:
        dump_config()

    CONFIG_PARSE_CALLED = True
    return config


def has_been_parsed() -> bool:
    """Returns True iff the global config has already been parsed"""
    return CONFIG_PARSE_CALLED


def dump_config():
    """Print the current config to stdout."""
    print("Global Configuration:", file=sys.stderr)
    pprint.pprint(config, stream=sys.stderr)
    print()


def late_logging():
    """Log messages saved earlier now that logging has been initialized."""
    logger = logging.getLogger(__name__)
    logger.debug('Original commandline was: %s', ORIG_ARGV)
    for _ in SAVED_MESSAGES:
        logger.debug(_)