apps.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298
  1. __package__ = 'archivebox.plugins_pkg.pip'
  2. import os
  3. import sys
  4. from pathlib import Path
  5. from typing import List, Dict, Optional
  6. from pydantic import InstanceOf, Field, model_validator, validate_call
  7. import django
  8. import django.db.backends.sqlite3.base
  9. from django.db.backends.sqlite3.base import Database as django_sqlite3 # type: ignore[import-type]
  10. from django.core.checks import Error, Tags
  11. from pydantic_pkgr import BinProvider, PipProvider, BinName, BinProviderName, ProviderLookupDict, SemVer, bin_abspath
  12. from archivebox.config import CONSTANTS, VERSION
  13. from abx.archivebox.base_plugin import BasePlugin
  14. from abx.archivebox.base_configset import BaseConfigSet
  15. from abx.archivebox.base_check import BaseCheck
  16. from abx.archivebox.base_binary import BaseBinary, BaseBinProvider, env, apt, brew
  17. from abx.archivebox.base_hook import BaseHook
  18. from ...misc.logging import hint
  19. ###################### Config ##########################
  20. class PipDependencyConfigs(BaseConfigSet):
  21. USE_PIP: bool = True
  22. PIP_BINARY: str = Field(default='pip')
  23. PIP_ARGS: Optional[List[str]] = Field(default=None)
  24. PIP_EXTRA_ARGS: List[str] = []
  25. PIP_DEFAULT_ARGS: List[str] = []
  26. PIP_CONFIG = PipDependencyConfigs()
  27. class SystemPipBinProvider(PipProvider, BaseBinProvider):
  28. name: BinProviderName = "sys_pip"
  29. INSTALLER_BIN: BinName = "pip"
  30. pip_venv: Optional[Path] = None # global pip scope
  31. def on_install(self, bin_name: str, **kwargs):
  32. # never modify system pip packages
  33. return 'refusing to install packages globally with system pip, use a venv instead'
  34. class SystemPipxBinProvider(PipProvider, BaseBinProvider):
  35. name: BinProviderName = "pipx"
  36. INSTALLER_BIN: BinName = "pipx"
  37. pip_venv: Optional[Path] = None # global pipx scope
  38. IS_INSIDE_VENV = sys.prefix != sys.base_prefix
  39. class VenvPipBinProvider(PipProvider, BaseBinProvider):
  40. name: BinProviderName = "venv_pip"
  41. INSTALLER_BIN: BinName = "pip"
  42. pip_venv: Optional[Path] = Path(sys.prefix if IS_INSIDE_VENV else os.environ.get("VIRTUAL_ENV", '/tmp/NotInsideAVenv/lib'))
  43. def setup(self):
  44. """never attempt to create a venv here, this is just used to detect if we are inside an existing one"""
  45. return None
  46. class LibPipBinProvider(PipProvider, BaseBinProvider):
  47. name: BinProviderName = "lib_pip"
  48. INSTALLER_BIN: BinName = "pip"
  49. pip_venv: Optional[Path] = CONSTANTS.LIB_PIP_DIR / 'venv'
  50. SYS_PIP_BINPROVIDER = SystemPipBinProvider()
  51. PIPX_PIP_BINPROVIDER = SystemPipxBinProvider()
  52. VENV_PIP_BINPROVIDER = VenvPipBinProvider()
  53. LIB_PIP_BINPROVIDER = LibPipBinProvider()
  54. pip = LIB_PIP_BINPROVIDER
  55. # ensure python libraries are importable from these locations (if archivebox wasnt executed from one of these then they wont already be in sys.path)
  56. assert VENV_PIP_BINPROVIDER.pip_venv is not None
  57. assert LIB_PIP_BINPROVIDER.pip_venv is not None
  58. site_packages_dir = 'lib/python{}.{}/site-packages'.format(*sys.version_info[:2])
  59. if os.environ.get("VIRTUAL_ENV", None):
  60. sys.path.append(str(VENV_PIP_BINPROVIDER.pip_venv / site_packages_dir))
  61. sys.path.append(str(LIB_PIP_BINPROVIDER.pip_venv / site_packages_dir))
  62. class ArchiveboxBinary(BaseBinary):
  63. name: BinName = 'archivebox'
  64. binproviders_supported: List[InstanceOf[BinProvider]] = [VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  65. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  66. VENV_PIP_BINPROVIDER.name: {'packages': lambda: [], 'version': lambda: VERSION, 'abspath': lambda: bin_abspath('archivebox')},
  67. SYS_PIP_BINPROVIDER.name: {'packages': lambda: [], 'version': lambda: VERSION, 'abspath': lambda: bin_abspath('archivebox')},
  68. apt.name: {'packages': lambda: [], 'version': lambda: VERSION, 'abspath': lambda: bin_abspath('archivebox')},
  69. brew.name: {'packages': lambda: [], 'version': lambda: VERSION, 'abspath': lambda: bin_abspath('archivebox')},
  70. }
  71. @validate_call
  72. def install(self, **kwargs):
  73. return self.load() # obviously it's already installed if we are running this ;)
  74. @validate_call
  75. def load_or_install(self, **kwargs):
  76. return self.load() # obviously it's already installed if we are running this ;)
  77. ARCHIVEBOX_BINARY = ArchiveboxBinary()
  78. class PythonBinary(BaseBinary):
  79. name: BinName = 'python'
  80. binproviders_supported: List[InstanceOf[BinProvider]] = [VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  81. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  82. SYS_PIP_BINPROVIDER.name: {
  83. 'abspath': lambda: sys.executable,
  84. 'version': lambda: '{}.{}.{}'.format(*sys.version_info[:3]),
  85. },
  86. }
  87. @validate_call
  88. def install(self, **kwargs):
  89. return self.load() # obviously it's already installed if we are running this ;)
  90. @validate_call
  91. def load_or_install(self, **kwargs):
  92. return self.load() # obviously it's already installed if we are running this ;)
  93. PYTHON_BINARY = PythonBinary()
  94. LOADED_SQLITE_PATH = Path(django.db.backends.sqlite3.base.__file__)
  95. LOADED_SQLITE_VERSION = SemVer(django_sqlite3.version)
  96. LOADED_SQLITE_FROM_VENV = str(LOADED_SQLITE_PATH.absolute().resolve()).startswith(str(VENV_PIP_BINPROVIDER.pip_venv.absolute().resolve()))
  97. class SqliteBinary(BaseBinary):
  98. name: BinName = 'sqlite'
  99. binproviders_supported: List[InstanceOf[BaseBinProvider]] = Field(default=[VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER])
  100. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  101. VENV_PIP_BINPROVIDER.name: {
  102. "abspath": lambda: LOADED_SQLITE_PATH if LOADED_SQLITE_FROM_VENV else None,
  103. "version": lambda: LOADED_SQLITE_VERSION if LOADED_SQLITE_FROM_VENV else None,
  104. },
  105. SYS_PIP_BINPROVIDER.name: {
  106. "abspath": lambda: LOADED_SQLITE_PATH if not LOADED_SQLITE_FROM_VENV else None,
  107. "version": lambda: LOADED_SQLITE_VERSION if not LOADED_SQLITE_FROM_VENV else None,
  108. },
  109. }
  110. @model_validator(mode='after')
  111. def validate_json_extension_is_available(self):
  112. # Check to make sure JSON extension is available in our Sqlite3 instance
  113. try:
  114. cursor = django_sqlite3.connect(':memory:').cursor()
  115. cursor.execute('SELECT JSON(\'{"a": "b"}\')')
  116. except django_sqlite3.OperationalError as exc:
  117. print(f'[red][X] Your SQLite3 version is missing the required JSON1 extension: {exc}[/red]')
  118. hint([
  119. 'Upgrade your Python version or install the extension manually:',
  120. 'https://code.djangoproject.com/wiki/JSON1Extension'
  121. ])
  122. return self
  123. @validate_call
  124. def install(self, **kwargs):
  125. return self.load() # obviously it's already installed if we are running this ;)
  126. @validate_call
  127. def load_or_install(self, **kwargs):
  128. return self.load() # obviously it's already installed if we are running this ;)
  129. SQLITE_BINARY = SqliteBinary()
  130. LOADED_DJANGO_PATH = Path(django.__file__)
  131. LOADED_DJANGO_VERSION = SemVer(django.VERSION[:3])
  132. LOADED_DJANGO_FROM_VENV = str(LOADED_DJANGO_PATH.absolute().resolve()).startswith(str(VENV_PIP_BINPROVIDER.pip_venv.absolute().resolve()))
  133. class DjangoBinary(BaseBinary):
  134. name: BinName = 'django'
  135. binproviders_supported: List[InstanceOf[BaseBinProvider]] = Field(default=[VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER])
  136. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  137. VENV_PIP_BINPROVIDER.name: {
  138. "abspath": lambda: LOADED_DJANGO_PATH if LOADED_DJANGO_FROM_VENV else None,
  139. "version": lambda: LOADED_DJANGO_VERSION if LOADED_DJANGO_FROM_VENV else None,
  140. },
  141. SYS_PIP_BINPROVIDER.name: {
  142. "abspath": lambda: LOADED_DJANGO_PATH if not LOADED_DJANGO_FROM_VENV else None,
  143. "version": lambda: LOADED_DJANGO_VERSION if not LOADED_DJANGO_FROM_VENV else None,
  144. },
  145. }
  146. @validate_call
  147. def install(self, **kwargs):
  148. return self.load() # obviously it's already installed if we are running this ;)
  149. @validate_call
  150. def load_or_install(self, **kwargs):
  151. return self.load() # obviously it's already installed if we are running this ;)
  152. DJANGO_BINARY = DjangoBinary()
  153. class PipBinary(BaseBinary):
  154. name: BinName = "pip"
  155. binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_PIP_BINPROVIDER, VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  156. @validate_call
  157. def install(self, **kwargs):
  158. return self.load() # obviously it's already installed if we are running this ;)
  159. @validate_call
  160. def load_or_install(self, **kwargs):
  161. return self.load() # obviously it's already installed if we are running this ;)
  162. PIP_BINARY = PipBinary()
  163. class PipxBinary(BaseBinary):
  164. name: BinName = "pipx"
  165. binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_PIP_BINPROVIDER, VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  166. PIPX_BINARY = PipxBinary()
  167. class CheckUserIsNotRoot(BaseCheck):
  168. label: str = 'CheckUserIsNotRoot'
  169. tag: str = Tags.database
  170. @staticmethod
  171. def check(settings, logger) -> List[Warning]:
  172. errors = []
  173. if getattr(settings, "USER", None) == 'root' or getattr(settings, "PUID", None) == 0:
  174. errors.append(
  175. Error(
  176. "Cannot run as root!",
  177. id="core.S001",
  178. hint=f'Run ArchiveBox as a non-root user with a UID greater than 500. (currently running as UID {os.getuid()}).',
  179. )
  180. )
  181. # logger.debug('[√] UID is not root')
  182. return errors
  183. class CheckPipEnvironment(BaseCheck):
  184. label: str = "CheckPipEnvironment"
  185. tag: str = Tags.database
  186. @staticmethod
  187. def check(settings, logger) -> List[Warning]:
  188. # soft errors: check that lib/pip virtualenv is setup properly
  189. errors = []
  190. LIB_PIP_BINPROVIDER.setup()
  191. if not LIB_PIP_BINPROVIDER.is_valid:
  192. errors.append(
  193. Error(
  194. f"Failed to setup {LIB_PIP_BINPROVIDER.pip_venv} virtualenv for runtime dependencies!",
  195. id="pip.P001",
  196. hint="Make sure the data dir is writable and make sure python3-pip and python3-venv are installed & available on the host.",
  197. )
  198. )
  199. # logger.debug("[√] CheckPipEnvironment: data/lib/pip virtualenv is setup properly")
  200. return errors
  201. USER_IS_NOT_ROOT_CHECK = CheckUserIsNotRoot()
  202. PIP_ENVIRONMENT_CHECK = CheckPipEnvironment()
  203. class PipPlugin(BasePlugin):
  204. app_label: str = 'pip'
  205. verbose_name: str = 'PIP'
  206. hooks: List[InstanceOf[BaseHook]] = [
  207. PIP_CONFIG,
  208. SYS_PIP_BINPROVIDER,
  209. PIPX_PIP_BINPROVIDER,
  210. VENV_PIP_BINPROVIDER,
  211. LIB_PIP_BINPROVIDER,
  212. PIP_BINARY,
  213. PIPX_BINARY,
  214. ARCHIVEBOX_BINARY,
  215. PYTHON_BINARY,
  216. SQLITE_BINARY,
  217. DJANGO_BINARY,
  218. USER_IS_NOT_ROOT_CHECK,
  219. PIP_ENVIRONMENT_CHECK,
  220. ]
  221. PLUGIN = PipPlugin()
  222. # PLUGIN.register(settings)
  223. DJANGO_APP = PLUGIN.AppConfig