2020-04-20 17:15:47 +02:00
|
|
|
"""
|
|
|
|
A few utility bobs and bits.
|
|
|
|
"""
|
|
|
|
|
2020-04-20 03:54:47 +02:00
|
|
|
import logging
|
2020-04-24 20:39:30 +02:00
|
|
|
import re
|
2020-04-20 03:54:47 +02:00
|
|
|
from pathlib import Path, PurePath
|
2020-04-24 13:50:27 +02:00
|
|
|
from typing import Optional, Tuple, Union
|
2018-11-24 09:27:33 +01:00
|
|
|
|
2020-04-20 18:38:18 +02:00
|
|
|
import bs4
|
2020-04-20 03:54:47 +02:00
|
|
|
import requests
|
|
|
|
from colorama import Fore, Style
|
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
PathLike = Union[PurePath, str, Tuple[str, ...]]
|
2020-04-20 03:54:47 +02:00
|
|
|
|
2020-04-20 17:15:47 +02:00
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
def to_path(pathlike: PathLike) -> Path:
|
|
|
|
if isinstance(pathlike, tuple):
|
|
|
|
return Path(*pathlike)
|
|
|
|
return Path(pathlike)
|
2018-11-27 09:52:27 +01:00
|
|
|
|
2020-04-20 14:29:28 +02:00
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
Regex = Union[str, re.Pattern]
|
|
|
|
|
2020-04-20 17:15:47 +02:00
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
def to_pattern(regex: Regex) -> re.Pattern:
|
|
|
|
if isinstance(regex, re.Pattern):
|
|
|
|
return regex
|
|
|
|
return re.compile(regex)
|
2018-11-27 09:52:27 +01:00
|
|
|
|
2020-04-20 14:29:28 +02:00
|
|
|
|
2020-04-20 18:38:18 +02:00
|
|
|
def soupify(response: requests.Response) -> bs4.BeautifulSoup:
|
2020-04-20 19:27:26 +02:00
|
|
|
"""
|
|
|
|
Wrap a requests response in a bs4 object.
|
|
|
|
"""
|
|
|
|
|
2020-04-20 18:38:18 +02:00
|
|
|
return bs4.BeautifulSoup(response.text, "html.parser")
|
|
|
|
|
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
def stream_to_path(response: requests.Response, target: Path, chunk_size: int = 1024 ** 2) -> None:
|
2020-04-20 17:15:47 +02:00
|
|
|
"""
|
|
|
|
Download a requests response content to a file by streaming it. This
|
|
|
|
function avoids excessive memory usage when downloading large files. The
|
|
|
|
chunk_size is in bytes.
|
|
|
|
"""
|
|
|
|
|
2020-04-20 19:27:26 +02:00
|
|
|
with response:
|
2020-04-24 20:39:30 +02:00
|
|
|
with open(target, 'wb') as file_descriptor:
|
2020-04-20 19:27:26 +02:00
|
|
|
for chunk in response.iter_content(chunk_size=chunk_size):
|
|
|
|
file_descriptor.write(chunk)
|
2018-11-24 09:27:33 +01:00
|
|
|
|
2020-04-20 14:29:28 +02:00
|
|
|
|
|
|
|
def prompt_yes_no(question: str, default: Optional[bool] = None) -> bool:
|
2020-04-20 17:15:47 +02:00
|
|
|
"""
|
|
|
|
Prompts the user a yes/no question and returns their choice.
|
|
|
|
"""
|
|
|
|
|
2020-04-20 14:29:28 +02:00
|
|
|
if default is True:
|
|
|
|
prompt = "[Y/n]"
|
|
|
|
elif default is False:
|
|
|
|
prompt = "[y/N]"
|
|
|
|
else:
|
|
|
|
prompt = "[y/n]"
|
|
|
|
|
|
|
|
text = f"{question} {prompt} "
|
2020-04-20 17:15:47 +02:00
|
|
|
wrong_reply = "Please reply with 'yes'/'y' or 'no'/'n'."
|
2020-04-20 14:29:28 +02:00
|
|
|
|
|
|
|
while True:
|
|
|
|
response = input(text).strip().lower()
|
|
|
|
if response in {"yes", "ye", "y"}:
|
|
|
|
return True
|
2020-04-20 17:15:47 +02:00
|
|
|
if response in {"no", "n"}:
|
2020-04-20 14:29:28 +02:00
|
|
|
return False
|
2020-04-20 17:15:47 +02:00
|
|
|
if response == "" and default is not None:
|
|
|
|
return default
|
|
|
|
print(wrong_reply)
|
2020-04-20 14:29:28 +02:00
|
|
|
|
|
|
|
|
2019-06-07 13:26:23 +02:00
|
|
|
class PrettyLogger:
|
2020-04-20 17:15:47 +02:00
|
|
|
"""
|
|
|
|
A logger that prints some specially formatted log messages in color.
|
|
|
|
"""
|
2019-06-07 13:26:23 +02:00
|
|
|
|
2020-04-20 03:54:47 +02:00
|
|
|
def __init__(self, logger: logging.Logger) -> None:
|
2019-06-07 13:26:23 +02:00
|
|
|
self.logger = logger
|
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
@staticmethod
|
|
|
|
def _format_path(path: PathLike) -> str:
|
|
|
|
return repr(str(to_path(path)))
|
|
|
|
|
|
|
|
def modified_file(self, path: PathLike) -> None:
|
2020-04-20 17:15:47 +02:00
|
|
|
"""
|
|
|
|
An existing file has changed.
|
|
|
|
"""
|
|
|
|
|
2020-04-20 14:29:28 +02:00
|
|
|
self.logger.info(
|
2020-04-24 20:39:30 +02:00
|
|
|
f"{Fore.MAGENTA}{Style.BRIGHT}Modified {self._format_path(path)}.{Style.RESET_ALL}"
|
2020-04-23 12:33:38 +02:00
|
|
|
)
|
2019-06-07 13:26:23 +02:00
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
def new_file(self, path: PathLike) -> None:
|
2020-04-20 17:15:47 +02:00
|
|
|
"""
|
|
|
|
A new file has been downloaded.
|
|
|
|
"""
|
|
|
|
|
2020-04-20 14:29:28 +02:00
|
|
|
self.logger.info(
|
2020-04-24 20:39:30 +02:00
|
|
|
f"{Fore.GREEN}{Style.BRIGHT}Created {self._format_path(path)}.{Style.RESET_ALL}"
|
2020-04-24 20:24:44 +02:00
|
|
|
)
|
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
def ignored_file(self, path: PathLike, reason: str) -> None:
|
2020-04-24 20:24:44 +02:00
|
|
|
"""
|
|
|
|
File was not downloaded or modified.
|
|
|
|
"""
|
|
|
|
|
|
|
|
self.logger.info(
|
2020-04-24 20:39:30 +02:00
|
|
|
f"{Style.DIM}Ignored {self._format_path(path)} "
|
2020-04-24 20:24:44 +02:00
|
|
|
f"({Style.NORMAL}{reason}{Style.DIM}).{Style.RESET_ALL}"
|
|
|
|
)
|
2019-06-07 13:26:23 +02:00
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
def searching(self, path: PathLike) -> None:
|
2020-04-20 17:15:47 +02:00
|
|
|
"""
|
2020-04-24 20:24:44 +02:00
|
|
|
A crawler searches a particular object.
|
2020-04-20 17:15:47 +02:00
|
|
|
"""
|
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
self.logger.info(f"Searching {self._format_path(path)}")
|
2020-04-23 12:33:38 +02:00
|
|
|
|
2020-04-24 20:39:30 +02:00
|
|
|
def not_searching(self, path: PathLike, reason: str) -> None:
|
2020-04-23 12:33:38 +02:00
|
|
|
"""
|
2020-04-24 20:24:44 +02:00
|
|
|
A crawler does not search a particular object.
|
2020-04-23 12:33:38 +02:00
|
|
|
"""
|
|
|
|
|
|
|
|
self.logger.info(
|
2020-04-24 20:39:30 +02:00
|
|
|
f"{Style.DIM}Not searching {self._format_path(path)} "
|
2020-04-24 20:24:44 +02:00
|
|
|
f"({Style.NORMAL}{reason}{Style.DIM}).{Style.RESET_ALL}"
|
2020-04-23 12:33:38 +02:00
|
|
|
)
|
2019-06-07 13:26:23 +02:00
|
|
|
|
2020-04-20 17:15:47 +02:00
|
|
|
def starting_synchronizer(
|
|
|
|
self,
|
2020-04-24 20:39:30 +02:00
|
|
|
target_directory: PathLike,
|
2020-04-20 17:15:47 +02:00
|
|
|
synchronizer_name: str,
|
|
|
|
subject: Optional[str] = None,
|
|
|
|
) -> None:
|
|
|
|
"""
|
|
|
|
A special message marking that a synchronizer has been started.
|
|
|
|
"""
|
|
|
|
|
2019-06-07 13:26:23 +02:00
|
|
|
subject_str = f"{subject} " if subject else ""
|
|
|
|
self.logger.info("")
|
|
|
|
self.logger.info((
|
2020-04-24 20:39:30 +02:00
|
|
|
f"{Fore.CYAN}{Style.BRIGHT}Synchronizing "
|
|
|
|
f"{subject_str}to {self._format_path(target_directory)} "
|
|
|
|
f"using the {synchronizer_name} synchronizer.{Style.RESET_ALL}"
|
2019-06-07 13:26:23 +02:00
|
|
|
))
|