wip

packages/abx-plugin-git-extractor/__init__.py

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

packages/abx-plugin-git-extractor/binaries.py

@@ -0,0 +1,18 @@
+__package__ = 'plugins_extractor.git'
+
+from typing import List
+
+from pydantic import InstanceOf
+from pydantic_pkgr import BinProvider, BinName
+
+from abx.archivebox.base_binary import BaseBinary, env, apt, brew
+
+from .config import GIT_CONFIG
+
+
+
+class GitBinary(BaseBinary):
+    name: BinName = GIT_CONFIG.GIT_BINARY
+    binproviders_supported: List[InstanceOf[BinProvider]] = [apt, brew, env]
+
+GIT_BINARY = GitBinary()

packages/abx-plugin-git-extractor/config.py

@@ -0,0 +1,28 @@
+__package__ = 'plugins_extractor.git'
+
+from typing import List
+
+from pydantic import Field
+
+from abx.archivebox.base_configset import BaseConfigSet
+
+from archivebox.config.common import ARCHIVING_CONFIG
+
+
+class GitConfig(BaseConfigSet):
+
+    SAVE_GIT: bool = True
+    
+    GIT_DOMAINS: str = Field(default='github.com,bitbucket.org,gitlab.com,gist.github.com,codeberg.org,gitea.com,git.sr.ht')
+    
+    GIT_BINARY: str = Field(default='git')
+    GIT_ARGS: List[str] = [
+        '--recursive',
+    ]
+    GIT_EXTRA_ARGS: List[str] = []
+    
+    GIT_TIMEOUT: int =  Field(default=lambda: ARCHIVING_CONFIG.TIMEOUT)
+    GIT_CHECK_SSL_VALIDITY: bool = Field(default=lambda: ARCHIVING_CONFIG.CHECK_SSL_VALIDITY)
+    
+
+GIT_CONFIG = GitConfig()

packages/abx-plugin-git-extractor/extractors.py

@@ -0,0 +1,17 @@
+__package__ = 'plugins_extractor.git'
+
+from pathlib import Path
+
+from abx.archivebox.base_extractor import BaseExtractor, ExtractorName
+
+from .binaries import GIT_BINARY
+
+
+class GitExtractor(BaseExtractor):
+    name: ExtractorName = 'git'
+    binary: str = GIT_BINARY.name
+
+    def get_output_path(self, snapshot) -> Path | None:
+        return snapshot.as_link() / 'git'
+
+GIT_EXTRACTOR = GitExtractor()

packages/abx-plugin-git-extractor/pyproject.toml

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