new vastly simplified plugin spec without pydantic

archivebox/plugins_extractor/readability/__init__.py

@@ -0,0 +1,46 @@
+__package__ = 'plugins_extractor.readability'
+__label__ = 'readability'
+__version__ = '2024.10.14'
+__author__ = 'Nick Sweeting'
+__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,
+    }

archivebox/plugins_extractor/readability/apps.py

@@ -1,86 +0,0 @@
-__package__ = 'archivebox.plugins_extractor.readability'
-
-from pathlib import Path
-from typing import List
-# from typing_extensions import Self
-
-# Depends on other PyPI/vendor packages:
-from pydantic import InstanceOf, Field
-from pydantic_pkgr import BinProvider, BinaryOverrides, BinName
-
-# Depends on other Django apps:
-from abx.archivebox.base_plugin import BasePlugin
-from abx.archivebox.base_configset import BaseConfigSet
-from abx.archivebox.base_binary import BaseBinary, env
-from abx.archivebox.base_extractor import BaseExtractor
-from abx.archivebox.base_hook import BaseHook
-
-# Depends on Other Plugins:
-from archivebox.config.common import ARCHIVING_CONFIG
-from plugins_pkg.npm.apps import SYS_NPM_BINPROVIDER, LIB_NPM_BINPROVIDER
-
-###################### 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()
-
-
-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()
-
-
-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_BINARY = ReadabilityBinary()
-READABILITY_EXTRACTOR = ReadabilityExtractor()
-
-# class ReadabilityQueue(BaseQueue):
-#     name: str = 'singlefile'
-    
-#     binaries: List[InstanceOf[BaseBinary]] = [READABILITY_BINARY]
-
-# READABILITY_QUEUE = ReadabilityQueue()
-
-class ReadabilityPlugin(BasePlugin):
-    app_label: str ='readability'
-    verbose_name: str = 'Readability'
-
-    hooks: List[InstanceOf[BaseHook]] = [
-        READABILITY_CONFIG,
-        READABILITY_BINARY,
-        READABILITY_EXTRACTOR,
-        # READABILITY_QUEUE,
-    ]
-
-
-
-PLUGIN = ReadabilityPlugin()
-# PLUGIN.register(settings)
-DJANGO_APP = PLUGIN.AppConfig

archivebox/plugins_extractor/readability/binaries.py

@@ -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()

archivebox/plugins_extractor/readability/config.py

@@ -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()

archivebox/plugins_extractor/readability/extractors.py

@@ -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()