2021-04-30 16:22:14 +02:00
|
|
|
from typing import Dict
|
|
|
|
|
|
|
|
from rich import print
|
|
|
|
from rich.markup import escape
|
|
|
|
|
2021-04-29 13:45:04 +02:00
|
|
|
from .config import Config
|
2021-04-30 16:22:14 +02:00
|
|
|
from .crawler import Crawler
|
2021-04-29 13:45:04 +02:00
|
|
|
from .crawlers import CRAWLERS
|
2020-04-23 20:31:32 +02:00
|
|
|
|
2020-04-23 11:44:13 +02:00
|
|
|
|
2021-04-30 16:22:14 +02:00
|
|
|
class PferdLoadException(Exception):
|
|
|
|
pass
|
|
|
|
|
|
|
|
|
2021-04-29 13:45:04 +02:00
|
|
|
class Pferd:
|
|
|
|
def __init__(self, config: Config):
|
|
|
|
self._config = config
|
2021-04-30 16:22:14 +02:00
|
|
|
self._crawlers: Dict[str, Crawler] = {}
|
|
|
|
|
|
|
|
def _load_crawlers(self) -> None:
|
|
|
|
abort = False
|
|
|
|
for name, section in self._config.crawler_sections():
|
|
|
|
print(f"[bold bright_cyan]Loading[/] crawler:{escape(name)}")
|
|
|
|
crawler_type = section.get("type")
|
|
|
|
crawler_constructor = CRAWLERS.get(crawler_type)
|
|
|
|
if crawler_constructor is None:
|
|
|
|
abort = True
|
|
|
|
if crawler_type is None:
|
|
|
|
print("[red]Error: No type")
|
|
|
|
else:
|
|
|
|
t = escape(repr(crawler_type))
|
|
|
|
print(f"[red]Error: Unknown type {t}")
|
|
|
|
continue
|
|
|
|
|
|
|
|
crawler = crawler_constructor(name, self._config, section)
|
|
|
|
self._crawlers[name] = crawler
|
|
|
|
|
|
|
|
if abort:
|
|
|
|
raise PferdLoadException()
|
2020-04-23 11:44:13 +02:00
|
|
|
|
2021-04-29 13:45:04 +02:00
|
|
|
async def run(self) -> None:
|
2021-04-30 16:22:14 +02:00
|
|
|
try:
|
|
|
|
self._load_crawlers()
|
|
|
|
except PferdLoadException:
|
|
|
|
print("[bold red]Could not initialize PFERD properly")
|
|
|
|
exit(1)
|
|
|
|
|
|
|
|
for name, crawler in self._crawlers.items():
|
|
|
|
print()
|
|
|
|
print(f"[bold bright_cyan]Running[/] crawler:{escape(name)}")
|
|
|
|
|
|
|
|
await crawler.run()
|