This commit is contained in:
Nick Sweeting 2024-10-25 01:06:12 -07:00
parent 4b6f08b0fe
commit 5d9a32c364
No known key found for this signature in database
178 changed files with 2982 additions and 1322 deletions

View file

@ -0,0 +1,46 @@
__package__ = 'plugins_extractor.readability'
__label__ = 'readability'
__version__ = '2024.10.14'
__author__ = 'ArchiveBox'
__homepage__ = 'https://github.com/ArchiveBox/readability-extractor'
__dependencies__ = ['npm']
import abx
@abx.hookimpl
def get_PLUGIN():
return {
'readability': {
'PACKAGE': __package__,
'LABEL': __label__,
'VERSION': __version__,
'AUTHOR': __author__,
'HOMEPAGE': __homepage__,
'DEPENDENCIES': __dependencies__,
}
}
@abx.hookimpl
def get_CONFIG():
from .config import READABILITY_CONFIG
return {
'readability': READABILITY_CONFIG
}
@abx.hookimpl
def get_BINARIES():
from .binaries import READABILITY_BINARY
return {
'readability': READABILITY_BINARY,
}
@abx.hookimpl
def get_EXTRACTORS():
from .extractors import READABILITY_EXTRACTOR
return {
'readability': READABILITY_EXTRACTOR,
}

View file

@ -0,0 +1,27 @@
__package__ = 'plugins_extractor.readability'
from typing import List
from pydantic import InstanceOf
from pydantic_pkgr import BinProvider, BinaryOverrides, BinName
from abx.archivebox.base_binary import BaseBinary, env
from plugins_pkg.npm.binproviders import SYS_NPM_BINPROVIDER, LIB_NPM_BINPROVIDER
from .config import READABILITY_CONFIG
READABILITY_PACKAGE_NAME = 'github:ArchiveBox/readability-extractor'
class ReadabilityBinary(BaseBinary):
name: BinName = READABILITY_CONFIG.READABILITY_BINARY
binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_NPM_BINPROVIDER, SYS_NPM_BINPROVIDER, env]
overrides: BinaryOverrides = {
LIB_NPM_BINPROVIDER.name: {"packages": [READABILITY_PACKAGE_NAME]},
SYS_NPM_BINPROVIDER.name: {"packages": [READABILITY_PACKAGE_NAME], "install": lambda: None}, # prevent modifying system global npm packages
}
READABILITY_BINARY = ReadabilityBinary()

View file

@ -0,0 +1,19 @@
__package__ = 'plugins_extractor.readability'
from pydantic import Field
from abx.archivebox.base_configset import BaseConfigSet
from archivebox.config.common import ARCHIVING_CONFIG
class ReadabilityConfig(BaseConfigSet):
SAVE_READABILITY: bool = Field(default=True, alias='USE_READABILITY')
READABILITY_TIMEOUT: int = Field(default=lambda: ARCHIVING_CONFIG.TIMEOUT)
READABILITY_BINARY: str = Field(default='readability-extractor')
# READABILITY_EXTRA_ARGS: List[str] = [] # readability-extractor doesn't take any extra args
READABILITY_CONFIG = ReadabilityConfig()

View file

@ -0,0 +1,20 @@
__package__ = 'plugins_extractor.readability'
from pathlib import Path
from pydantic_pkgr import BinName
from abx.archivebox.base_extractor import BaseExtractor
from .binaries import READABILITY_BINARY
class ReadabilityExtractor(BaseExtractor):
name: str = 'readability'
binary: BinName = READABILITY_BINARY.name
def get_output_path(self, snapshot) -> Path:
return Path(snapshot.link_dir) / 'readability' / 'content.html'
READABILITY_EXTRACTOR = ReadabilityExtractor()

View file

@ -0,0 +1,7 @@
[project]
name = "abx-readability-extractor"
version = "0.1.0"
description = "Add your description here"
readme = "README.md"
requires-python = ">=3.10"
dependencies = []