apps.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312
  1. __package__ = 'archivebox.plugins_pkg.pip'
  2. import os
  3. import sys
  4. import site
  5. from pathlib import Path
  6. from typing import List, Dict, Optional
  7. from pydantic import InstanceOf, Field, model_validator, validate_call
  8. import django
  9. import django.db.backends.sqlite3.base
  10. from django.db.backends.sqlite3.base import Database as django_sqlite3 # type: ignore[import-type]
  11. from django.core.checks import Error, Tags
  12. from pydantic_pkgr import BinProvider, PipProvider, BinName, BinProviderName, ProviderLookupDict, SemVer, bin_abspath
  13. from archivebox.config import CONSTANTS, VERSION
  14. from abx.archivebox.base_plugin import BasePlugin
  15. from abx.archivebox.base_configset import BaseConfigSet
  16. from abx.archivebox.base_check import BaseCheck
  17. from abx.archivebox.base_binary import BaseBinary, BaseBinProvider, env, apt, brew
  18. from abx.archivebox.base_hook import BaseHook
  19. from ...misc.logging import hint
  20. ###################### Config ##########################
  21. class PipDependencyConfigs(BaseConfigSet):
  22. USE_PIP: bool = True
  23. PIP_BINARY: str = Field(default='pip')
  24. PIP_ARGS: Optional[List[str]] = Field(default=None)
  25. PIP_EXTRA_ARGS: List[str] = []
  26. PIP_DEFAULT_ARGS: List[str] = []
  27. PIP_CONFIG = PipDependencyConfigs()
  28. class SystemPipBinProvider(PipProvider, BaseBinProvider):
  29. name: BinProviderName = "sys_pip"
  30. INSTALLER_BIN: BinName = "pip"
  31. pip_venv: Optional[Path] = None # global pip scope
  32. def on_install(self, bin_name: str, **kwargs):
  33. # never modify system pip packages
  34. return 'refusing to install packages globally with system pip, use a venv instead'
  35. class SystemPipxBinProvider(PipProvider, BaseBinProvider):
  36. name: BinProviderName = "pipx"
  37. INSTALLER_BIN: BinName = "pipx"
  38. pip_venv: Optional[Path] = None # global pipx scope
  39. IS_INSIDE_VENV = sys.prefix != sys.base_prefix
  40. class VenvPipBinProvider(PipProvider, BaseBinProvider):
  41. name: BinProviderName = "venv_pip"
  42. INSTALLER_BIN: BinName = "pip"
  43. pip_venv: Optional[Path] = Path(sys.prefix if IS_INSIDE_VENV else os.environ.get("VIRTUAL_ENV", '/tmp/NotInsideAVenv/lib'))
  44. def setup(self):
  45. """never attempt to create a venv here, this is just used to detect if we are inside an existing one"""
  46. return None
  47. class LibPipBinProvider(PipProvider, BaseBinProvider):
  48. name: BinProviderName = "lib_pip"
  49. INSTALLER_BIN: BinName = "pip"
  50. pip_venv: Optional[Path] = CONSTANTS.LIB_PIP_DIR / 'venv'
  51. SYS_PIP_BINPROVIDER = SystemPipBinProvider()
  52. PIPX_PIP_BINPROVIDER = SystemPipxBinProvider()
  53. VENV_PIP_BINPROVIDER = VenvPipBinProvider()
  54. LIB_PIP_BINPROVIDER = LibPipBinProvider()
  55. pip = LIB_PIP_BINPROVIDER
  56. # ensure python libraries are importable from these locations (if archivebox wasnt executed from one of these then they wont already be in sys.path)
  57. assert VENV_PIP_BINPROVIDER.pip_venv is not None
  58. assert LIB_PIP_BINPROVIDER.pip_venv is not None
  59. major, minor, patch = sys.version_info[:3]
  60. site_packages_dir = f'lib/python{major}.{minor}/site-packages'
  61. LIB_SITE_PACKAGES = (LIB_PIP_BINPROVIDER.pip_venv / site_packages_dir,)
  62. VENV_SITE_PACKAGES = (VENV_PIP_BINPROVIDER.pip_venv / site_packages_dir,)
  63. USER_SITE_PACKAGES = site.getusersitepackages()
  64. SYS_SITE_PACKAGES = site.getsitepackages()
  65. ALL_SITE_PACKAGES = (
  66. *LIB_SITE_PACKAGES,
  67. *VENV_SITE_PACKAGES,
  68. *USER_SITE_PACKAGES,
  69. *SYS_SITE_PACKAGES,
  70. )
  71. for site_packages_dir in ALL_SITE_PACKAGES:
  72. if site_packages_dir not in sys.path:
  73. sys.path.append(str(site_packages_dir))
  74. class ArchiveboxBinary(BaseBinary):
  75. name: BinName = 'archivebox'
  76. binproviders_supported: List[InstanceOf[BinProvider]] = [VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  77. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  78. VENV_PIP_BINPROVIDER.name: {'packages': lambda: [], 'version': lambda: VERSION},
  79. SYS_PIP_BINPROVIDER.name: {'packages': lambda: [], 'version': lambda: VERSION},
  80. apt.name: {'packages': lambda: [], 'version': lambda: VERSION},
  81. brew.name: {'packages': lambda: [], 'version': lambda: VERSION},
  82. }
  83. @validate_call
  84. def install(self, **kwargs):
  85. return self.load() # obviously it's already installed if we are running this ;)
  86. @validate_call
  87. def load_or_install(self, **kwargs):
  88. return self.load() # obviously it's already installed if we are running this ;)
  89. ARCHIVEBOX_BINARY = ArchiveboxBinary()
  90. class PythonBinary(BaseBinary):
  91. name: BinName = 'python'
  92. binproviders_supported: List[InstanceOf[BinProvider]] = [VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  93. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  94. SYS_PIP_BINPROVIDER.name: {
  95. 'abspath': lambda: sys.executable,
  96. 'version': lambda: '{}.{}.{}'.format(*sys.version_info[:3]),
  97. },
  98. }
  99. @validate_call
  100. def install(self, **kwargs):
  101. return self.load() # obviously it's already installed if we are running this ;)
  102. @validate_call
  103. def load_or_install(self, **kwargs):
  104. return self.load() # obviously it's already installed if we are running this ;)
  105. PYTHON_BINARY = PythonBinary()
  106. LOADED_SQLITE_PATH = Path(django.db.backends.sqlite3.base.__file__)
  107. LOADED_SQLITE_VERSION = SemVer(django_sqlite3.version)
  108. LOADED_SQLITE_FROM_VENV = str(LOADED_SQLITE_PATH.absolute().resolve()).startswith(str(VENV_PIP_BINPROVIDER.pip_venv.absolute().resolve()))
  109. class SqliteBinary(BaseBinary):
  110. name: BinName = 'sqlite'
  111. binproviders_supported: List[InstanceOf[BaseBinProvider]] = Field(default=[VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER])
  112. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  113. VENV_PIP_BINPROVIDER.name: {
  114. "abspath": lambda: LOADED_SQLITE_PATH if LOADED_SQLITE_FROM_VENV else None,
  115. "version": lambda: LOADED_SQLITE_VERSION if LOADED_SQLITE_FROM_VENV else None,
  116. },
  117. SYS_PIP_BINPROVIDER.name: {
  118. "abspath": lambda: LOADED_SQLITE_PATH if not LOADED_SQLITE_FROM_VENV else None,
  119. "version": lambda: LOADED_SQLITE_VERSION if not LOADED_SQLITE_FROM_VENV else None,
  120. },
  121. }
  122. @model_validator(mode='after')
  123. def validate_json_extension_is_available(self):
  124. # Check to make sure JSON extension is available in our Sqlite3 instance
  125. try:
  126. cursor = django_sqlite3.connect(':memory:').cursor()
  127. cursor.execute('SELECT JSON(\'{"a": "b"}\')')
  128. except django_sqlite3.OperationalError as exc:
  129. print(f'[red][X] Your SQLite3 version is missing the required JSON1 extension: {exc}[/red]')
  130. hint([
  131. 'Upgrade your Python version or install the extension manually:',
  132. 'https://code.djangoproject.com/wiki/JSON1Extension'
  133. ])
  134. return self
  135. @validate_call
  136. def install(self, **kwargs):
  137. return self.load() # obviously it's already installed if we are running this ;)
  138. @validate_call
  139. def load_or_install(self, **kwargs):
  140. return self.load() # obviously it's already installed if we are running this ;)
  141. SQLITE_BINARY = SqliteBinary()
  142. LOADED_DJANGO_PATH = Path(django.__file__)
  143. LOADED_DJANGO_VERSION = SemVer(django.VERSION[:3])
  144. LOADED_DJANGO_FROM_VENV = str(LOADED_DJANGO_PATH.absolute().resolve()).startswith(str(VENV_PIP_BINPROVIDER.pip_venv.absolute().resolve()))
  145. class DjangoBinary(BaseBinary):
  146. name: BinName = 'django'
  147. binproviders_supported: List[InstanceOf[BaseBinProvider]] = Field(default=[VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER])
  148. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  149. VENV_PIP_BINPROVIDER.name: {
  150. "abspath": lambda: LOADED_DJANGO_PATH if LOADED_DJANGO_FROM_VENV else None,
  151. "version": lambda: LOADED_DJANGO_VERSION if LOADED_DJANGO_FROM_VENV else None,
  152. },
  153. SYS_PIP_BINPROVIDER.name: {
  154. "abspath": lambda: LOADED_DJANGO_PATH if not LOADED_DJANGO_FROM_VENV else None,
  155. "version": lambda: LOADED_DJANGO_VERSION if not LOADED_DJANGO_FROM_VENV else None,
  156. },
  157. }
  158. @validate_call
  159. def install(self, **kwargs):
  160. return self.load() # obviously it's already installed if we are running this ;)
  161. @validate_call
  162. def load_or_install(self, **kwargs):
  163. return self.load() # obviously it's already installed if we are running this ;)
  164. DJANGO_BINARY = DjangoBinary()
  165. class PipBinary(BaseBinary):
  166. name: BinName = "pip"
  167. binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_PIP_BINPROVIDER, VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  168. @validate_call
  169. def install(self, **kwargs):
  170. return self.load() # obviously it's already installed if we are running this ;)
  171. @validate_call
  172. def load_or_install(self, **kwargs):
  173. return self.load() # obviously it's already installed if we are running this ;)
  174. PIP_BINARY = PipBinary()
  175. class PipxBinary(BaseBinary):
  176. name: BinName = "pipx"
  177. binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_PIP_BINPROVIDER, VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  178. PIPX_BINARY = PipxBinary()
  179. class CheckUserIsNotRoot(BaseCheck):
  180. label: str = 'CheckUserIsNotRoot'
  181. tag: str = Tags.database
  182. @staticmethod
  183. def check(settings, logger) -> List[Warning]:
  184. errors = []
  185. if getattr(settings, "USER", None) == 'root' or getattr(settings, "PUID", None) == 0:
  186. errors.append(
  187. Error(
  188. "Cannot run as root!",
  189. id="core.S001",
  190. hint=f'Run ArchiveBox as a non-root user with a UID greater than 500. (currently running as UID {os.getuid()}).',
  191. )
  192. )
  193. # logger.debug('[√] UID is not root')
  194. return errors
  195. class CheckPipEnvironment(BaseCheck):
  196. label: str = "CheckPipEnvironment"
  197. tag: str = Tags.database
  198. @staticmethod
  199. def check(settings, logger) -> List[Warning]:
  200. # soft errors: check that lib/pip virtualenv is setup properly
  201. errors = []
  202. LIB_PIP_BINPROVIDER.setup()
  203. if not LIB_PIP_BINPROVIDER.is_valid:
  204. errors.append(
  205. Error(
  206. f"Failed to setup {LIB_PIP_BINPROVIDER.pip_venv} virtualenv for runtime dependencies!",
  207. id="pip.P001",
  208. hint="Make sure the data dir is writable and make sure python3-pip and python3-venv are installed & available on the host.",
  209. )
  210. )
  211. # logger.debug("[√] CheckPipEnvironment: data/lib/pip virtualenv is setup properly")
  212. return errors
  213. USER_IS_NOT_ROOT_CHECK = CheckUserIsNotRoot()
  214. PIP_ENVIRONMENT_CHECK = CheckPipEnvironment()
  215. class PipPlugin(BasePlugin):
  216. app_label: str = 'pip'
  217. verbose_name: str = 'PIP'
  218. hooks: List[InstanceOf[BaseHook]] = [
  219. PIP_CONFIG,
  220. SYS_PIP_BINPROVIDER,
  221. PIPX_PIP_BINPROVIDER,
  222. VENV_PIP_BINPROVIDER,
  223. LIB_PIP_BINPROVIDER,
  224. PIP_BINARY,
  225. PIPX_BINARY,
  226. ARCHIVEBOX_BINARY,
  227. PYTHON_BINARY,
  228. SQLITE_BINARY,
  229. DJANGO_BINARY,
  230. USER_IS_NOT_ROOT_CHECK,
  231. PIP_ENVIRONMENT_CHECK,
  232. ]
  233. PLUGIN = PipPlugin()
  234. # PLUGIN.register(settings)
  235. DJANGO_APP = PLUGIN.AppConfig