mvw-dl/mvw-dl.py

import configparser
import json
import logging
import os
import sys
import requests
import inflect
from rich.logging import RichHandler
from rich.traceback import install
import typing as t
from rich.console import Console
from type_def.mvw_json_request import MVWJSONRequest
from type_def.mvw_json_response import MVWJSONResponse

# Without width
console = Console(width=180)
p = inflect.engine()

# We use Python 3.5+ type hints; we're working with JSON objects; we're following a 2016 suggestion from
# Python's "typing" GitHub issue tracker on how to create a "JSONType" hint since such a thing does not yet
# officially exist: https://github.com/python/typing/issues/182#issuecomment-186684288
# JSONType = t.Union[str, int, float, bool, None, t.Dict[str, t.Any], t.List[t.Any]]
JSONType = t.Union[str, int, float, bool, None, t.Dict[str, t.Any], t.List[t.Any]]


# Exit codes
# 1: Config file invalid, it has no sections
# 2: Config file invalid, sections must define at least CONST.CFG_MANDATORY


class CONST(object):
    __slots__ = ()
    LOG_FORMAT = "%(message)s"
    CFG_THIS_FILE_DIRNAME = os.path.dirname(__file__)
    CFG_DEFAULT_FILENAME = "config.ini"
    CFG_DEFAULT_ABS_PATH = os.path.join(CFG_THIS_FILE_DIRNAME, CFG_DEFAULT_FILENAME)
    CFG_KNOWN_DEFAULTS = [
        {"key": "self_name", "value": "mvw-dl"},
        {"key": "tmp_base_dir", "value": os.path.join(CFG_THIS_FILE_DIRNAME, "data/tmp/%(self_name)s")},
        {"key": "state_base_dir", "value": os.path.join(CFG_THIS_FILE_DIRNAME, "data/var/lib/%(self_name)s")},
        {"key": "state_files_dir", "value": "%(state_base_dir)s/state"},
        {"key": "state_file_retention", "value": "50"},
        {"key": "state_file_name_prefix", "value": "state-"},
        {"key": "state_file_name_suffix", "value": ".log"},
        {"key": "mvw_endpoint", "value": "http://localhost:8000/api/query"},
        {"key": "title_dedup_winner", "value": "first"}
    ]
    CFG_KNOWN_SECTION = [
        {"key": "min_duration", "is_mandatory": False},
        {"key": "max_duration", "is_mandatory": False},
        {"key": "title_not_regex", "is_mandatory": False},
        {"key": "query", "is_mandatory": True},
        {"key": "dl_dir", "is_mandatory": True}
    ]
    CFG_MANDATORY = [section_cfg["key"] for section_cfg in CFG_KNOWN_SECTION if section_cfg["is_mandatory"]]


CONST = CONST()
logging.basicConfig(
    # Default for all modules in NOTSET so log everything
    level="NOTSET",
    format=CONST.LOG_FORMAT,
    datefmt="[%X]",
    handlers=[RichHandler(
        show_time=False if "SYSTEMD_EXEC_PID" in os.environ else True,
        rich_tracebacks=True
    )]
)
log = logging.getLogger("rich")
# Our own code logs with this level
log.setLevel(logging.DEBUG)
# connectionpool logs with WARNING, we don't need its verbosity
log_connectionpool = logging.getLogger("urllib3.connectionpool")
log_connectionpool.setLevel(logging.WARNING)
install(show_locals=True)


class ConfigParser(
        configparser.ConfigParser):
    """Can get options() without defaults

    Taken from https://stackoverflow.com/a/12600066.
    """

    def options(self, section, no_defaults=False, **kwargs):
        if no_defaults:
            try:
                return list(self._sections[section].keys())
            except KeyError:
                raise configparser.NoSectionError(section)
        else:
            return super().options(section)


ini_defaults = []
internal_defaults = {default["key"]: default["value"] for default in CONST.CFG_KNOWN_DEFAULTS}
config = ConfigParser(defaults=internal_defaults)
config.read(CONST.CFG_DEFAULT_FILENAME)


def print_section_header(
        header: str) -> str:
    return f"Loading config section '[{header}]' ..."


def validate_default_section(
        config_obj: configparser.ConfigParser()) -> None:
    log.debug(f"Loading config from file '{CONST.CFG_DEFAULT_ABS_PATH}' ...")
    if not config_obj.sections():
        log.error(f"No config sections found in '{CONST.CFG_DEFAULT_ABS_PATH}'. Exiting 1 ...")
        sys.exit(1)
    if config.defaults():
        log.debug(f"Symbol legend:\n"
                  f"* Global default from section '[{config_obj.default_section}]'\n"
                  f"~ Local option, doesn't exist in '[{config_obj.default_section}]'\n"
                  f"+ Local override of a value from '[{config_obj.default_section}]'\n"
                  f"= Local override, same value as in '[{config_obj.default_section}]'")
        log.debug(print_section_header(config_obj.default_section))
        for default in config_obj.defaults():
            ini_defaults.append({default: config_obj[config_obj.default_section][default]})
            log.debug(f"* {default} = {config_obj[config_obj.default_section][default]}")
    else:
        log.debug(f"No defaults defined")


def config_has_valid_section(
        config_obj: configparser.ConfigParser()) -> bool:
    has_valid_section = False
    for config_obj_section in config_obj.sections():
        if set(CONST.CFG_MANDATORY).issubset(config_obj.options(config_obj_section)):
            has_valid_section = True
            break
    return has_valid_section


def is_default(
        config_key: str) -> bool:
    return any(config_key in ini_default for ini_default in ini_defaults)


def is_same_as_default(
        config_kv_pair: dict) -> bool:
    return config_kv_pair in ini_defaults


def validate_config_sections(
        config_obj: configparser.ConfigParser()) -> None:
    for this_section in config_obj.sections():
        log.debug(print_section_header(this_section))
        if not set(CONST.CFG_MANDATORY).issubset(config_obj.options(this_section, no_defaults=True)):
            log.warning(f"Config section '[{this_section}]' does not have all mandatory options "
                        f"{CONST.CFG_MANDATORY} set, skipping section ...")
            config_obj.remove_section(this_section)
        else:
            for key in config_obj.options(this_section, no_defaults=True):
                kv_prefix = "~"
                if is_default(key):
                    kv_prefix = "+"
                    if is_same_as_default({key: config_obj[this_section][key]}):
                        kv_prefix = "="
                log.debug(f"{kv_prefix} {key} = {config_obj[this_section][key]}")


def query_string_from_file(
        filename: str) -> str:
    with open(filename, "r") as jsonfile:
        query_string = jsonfile.read()
        return query_string


def get_query_payload(
        section_name: str,
        config_obj: configparser.ConfigParser()) -> MVWJSONRequest:
    log.debug(f"Generating HTTP POST JSON payload ...")
    query = config_obj.get(section_name, "query")
    if query[0] == "@":
        query = query.split("@", 1)[1]
        query = query_string_from_file(query)
    got_query_payload = MVWJSONRequest(**json.loads(query))
    return got_query_payload


def get_json_response(
        section_name: str,
        config_obj: configparser.ConfigParser(),
        payload: MVWJSONRequest) -> MVWJSONResponse:
    log.debug(f"Downloading JSON list of Mediathek files that match search criteria")
    serialized_payload = json.dumps(payload)
    url = config_obj.get(section_name, "mvw_endpoint")
    req_header = {"Content-Type": "text/plain"}
    s = requests.Session()
    req = requests.Request("POST", url, data=serialized_payload, headers=req_header)
    prepped = req.prepare()
    newline = "\n"
    log.debug(f"Request method: {req.method}\n"
              f"URL: {req.url}\n"
              f"""{newline.join(f"Header '{header}': '{value}'" for header, value in list(req.headers.items()))}\n"""
              f"Payload: {payload}")
    with s.send(prepped) as s:
        # return json.loads(s.content)
        got_json_response = MVWJSONResponse(**json.loads(s.content))
        return got_json_response


def filter_json_by_duration(
        section_name: str,
        config_obj: configparser.ConfigParser(),
        json_obj: MVWJSONResponse) -> MVWJSONResponse:
    min_length = config_obj.getint(section_name, "min_duration")
    if min_length >= 0:
        log.debug(f"""Filtering JSON for minimum length of {min_length} {p.plural("second", min_length)}""")
        console.log(json_obj)
        #for result in json_obj["result"]["results"]:
        #    console.log(result)
        #    console.log(f"0000000000000000000000")
        #    if not result["duration"] >= min_length:
        #        pass
                # json_str.
                # console.log(f"{result}\n......................")
        # console.log(json_obj)
        # console.log(f"ssssssssss")

        # json_matches_min_length =
        pass


if __name__ == '__main__':
    validate_default_section(config)
    if config_has_valid_section(config):
        validate_config_sections(config)
    else:
        log.error(f"No valid config section found. A valid config section has at least the mandatory options "
                  f"{CONST.CFG_MANDATORY} set. Exiting 2 ...")
        sys.exit(2)

    log.debug(f"Iterating over config sections ...")
    for section in config.sections():
        log.debug(f"Processing section '[{section}]' ...")
        query_payload = get_query_payload(section, config)
        json_response = get_json_response(section, config, query_payload)
        log.debug(json_response)
        quit()
        log.debug(CONST.CFG_KNOWN_SECTION[0])
        if config.has_option(section, "min_duration") or config.has_option(section, "max_duration"):
            json_matches_duration = filter_json_by_duration(section, config, json_response)
    # console.log(json_response)