apps.py 9.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270
  1. __package__ = 'archivebox.plugins_pkg.pip'
  2. import os
  3. import sys
  4. import inspect
  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. 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. class VenvPipBinProvider(PipProvider, BaseBinProvider):
  39. name: BinProviderName = "venv_pip"
  40. INSTALLER_BIN: BinName = "pip"
  41. pip_venv: Optional[Path] = Path(os.environ.get("VIRTUAL_ENV", None) or '/tmp/NotInsideAVenv')
  42. class LibPipBinProvider(PipProvider, BaseBinProvider):
  43. name: BinProviderName = "lib_pip"
  44. INSTALLER_BIN: BinName = "pip"
  45. pip_venv: Optional[Path] = CONSTANTS.LIB_PIP_DIR / 'venv'
  46. SYS_PIP_BINPROVIDER = SystemPipBinProvider()
  47. PIPX_PIP_BINPROVIDER = SystemPipxBinProvider()
  48. VENV_PIP_BINPROVIDER = VenvPipBinProvider()
  49. LIB_PIP_BINPROVIDER = LibPipBinProvider()
  50. pip = LIB_PIP_BINPROVIDER
  51. # ensure python libraries are importable from these locations (if archivebox wasnt executed from one of these then they wont already be in sys.path)
  52. site_packages_dir = 'lib/python{}.{}/site-packages'.format(*sys.version_info[:2])
  53. sys.path.append(str(VENV_PIP_BINPROVIDER.pip_venv / site_packages_dir))
  54. sys.path.append(str(LIB_PIP_BINPROVIDER.pip_venv / site_packages_dir))
  55. class ArchiveboxBinary(BaseBinary):
  56. name: BinName = 'archivebox'
  57. binproviders_supported: List[InstanceOf[BinProvider]] = [VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  58. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  59. VENV_PIP_BINPROVIDER.name: {'packages': lambda: [], 'version': lambda: VERSION, 'abspath': lambda: bin_abspath('archivebox')},
  60. SYS_PIP_BINPROVIDER.name: {'packages': lambda: [], 'version': lambda: VERSION, 'abspath': lambda: bin_abspath('archivebox')},
  61. apt.name: {'packages': lambda: [], 'version': lambda: VERSION, 'abspath': lambda: bin_abspath('archivebox')},
  62. brew.name: {'packages': lambda: [], 'version': lambda: VERSION, 'abspath': lambda: bin_abspath('archivebox')},
  63. }
  64. @validate_call
  65. def install(self, **kwargs):
  66. return self.load() # obviously it's already installed if we are running this ;)
  67. @validate_call
  68. def load_or_install(self, **kwargs):
  69. return self.load() # obviously it's already installed if we are running this ;)
  70. ARCHIVEBOX_BINARY = ArchiveboxBinary()
  71. class PythonBinary(BaseBinary):
  72. name: BinName = 'python'
  73. binproviders_supported: List[InstanceOf[BinProvider]] = [VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  74. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  75. SYS_PIP_BINPROVIDER.name: {
  76. 'abspath': lambda: sys.executable,
  77. 'version': lambda: '{}.{}.{}'.format(*sys.version_info[:3]),
  78. },
  79. }
  80. @validate_call
  81. def install(self, **kwargs):
  82. return self.load() # obviously it's already installed if we are running this ;)
  83. @validate_call
  84. def load_or_install(self, **kwargs):
  85. return self.load() # obviously it's already installed if we are running this ;)
  86. PYTHON_BINARY = PythonBinary()
  87. class SqliteBinary(BaseBinary):
  88. name: BinName = 'sqlite'
  89. binproviders_supported: List[InstanceOf[BaseBinProvider]] = Field(default=[VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER])
  90. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  91. VENV_PIP_BINPROVIDER.name: {
  92. "abspath": lambda: Path(inspect.getfile(django_sqlite3)),
  93. "version": lambda: SemVer(django_sqlite3.version),
  94. },
  95. SYS_PIP_BINPROVIDER.name: {
  96. "abspath": lambda: Path(inspect.getfile(django_sqlite3)),
  97. "version": lambda: SemVer(django_sqlite3.version),
  98. },
  99. }
  100. @model_validator(mode='after')
  101. def validate_json_extension_is_available(self):
  102. # Check to make sure JSON extension is available in our Sqlite3 instance
  103. try:
  104. cursor = django_sqlite3.connect(':memory:').cursor()
  105. cursor.execute('SELECT JSON(\'{"a": "b"}\')')
  106. except django_sqlite3.OperationalError as exc:
  107. print(f'[red][X] Your SQLite3 version is missing the required JSON1 extension: {exc}[/red]')
  108. hint([
  109. 'Upgrade your Python version or install the extension manually:',
  110. 'https://code.djangoproject.com/wiki/JSON1Extension'
  111. ])
  112. return self
  113. @validate_call
  114. def install(self, **kwargs):
  115. return self.load() # obviously it's already installed if we are running this ;)
  116. @validate_call
  117. def load_or_install(self, **kwargs):
  118. return self.load() # obviously it's already installed if we are running this ;)
  119. SQLITE_BINARY = SqliteBinary()
  120. class DjangoBinary(BaseBinary):
  121. name: BinName = 'django'
  122. binproviders_supported: List[InstanceOf[BaseBinProvider]] = Field(default=[VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER])
  123. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  124. VENV_PIP_BINPROVIDER.name: {
  125. "abspath": lambda: inspect.getfile(django),
  126. "version": lambda: django.VERSION[:3],
  127. },
  128. SYS_PIP_BINPROVIDER.name: {
  129. "abspath": lambda: inspect.getfile(django),
  130. "version": lambda: django.VERSION[:3],
  131. },
  132. }
  133. @validate_call
  134. def install(self, **kwargs):
  135. return self.load() # obviously it's already installed if we are running this ;)
  136. @validate_call
  137. def load_or_install(self, **kwargs):
  138. return self.load() # obviously it's already installed if we are running this ;)
  139. DJANGO_BINARY = DjangoBinary()
  140. class PipBinary(BaseBinary):
  141. name: BinName = "pip"
  142. binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_PIP_BINPROVIDER, VENV_PIP_BINPROVIDER, SYS_PIP_BINPROVIDER, apt, brew, env]
  143. @validate_call
  144. def install(self, **kwargs):
  145. return self.load() # obviously it's already installed if we are running this ;)
  146. @validate_call
  147. def load_or_install(self, **kwargs):
  148. return self.load() # obviously it's already installed if we are running this ;)
  149. PIP_BINARY = PipBinary()
  150. class CheckUserIsNotRoot(BaseCheck):
  151. label: str = 'CheckUserIsNotRoot'
  152. tag: str = Tags.database
  153. @staticmethod
  154. def check(settings, logger) -> List[Warning]:
  155. errors = []
  156. if getattr(settings, "USER", None) == 'root' or getattr(settings, "PUID", None) == 0:
  157. errors.append(
  158. Error(
  159. "Cannot run as root!",
  160. id="core.S001",
  161. hint=f'Run ArchiveBox as a non-root user with a UID greater than 500. (currently running as UID {os.getuid()}).',
  162. )
  163. )
  164. # logger.debug('[√] UID is not root')
  165. return errors
  166. class CheckPipEnvironment(BaseCheck):
  167. label: str = "CheckPipEnvironment"
  168. tag: str = Tags.database
  169. @staticmethod
  170. def check(settings, logger) -> List[Warning]:
  171. # soft errors: check that lib/pip virtualenv is setup properly
  172. errors = []
  173. LIB_PIP_BINPROVIDER.setup()
  174. if not LIB_PIP_BINPROVIDER.is_valid:
  175. errors.append(
  176. Error(
  177. f"Failed to setup {LIB_PIP_BINPROVIDER.pip_venv} virtualenv for runtime dependencies!",
  178. id="pip.P001",
  179. hint="Make sure the data dir is writable and make sure python3-pip and python3-venv are installed & available on the host.",
  180. )
  181. )
  182. # logger.debug("[√] CheckPipEnvironment: data/lib/pip virtualenv is setup properly")
  183. return errors
  184. USER_IS_NOT_ROOT_CHECK = CheckUserIsNotRoot()
  185. PIP_ENVIRONMENT_CHECK = CheckPipEnvironment()
  186. class PipPlugin(BasePlugin):
  187. app_label: str = 'pip'
  188. verbose_name: str = 'PIP'
  189. hooks: List[InstanceOf[BaseHook]] = [
  190. PIP_CONFIG,
  191. SYS_PIP_BINPROVIDER,
  192. PIPX_PIP_BINPROVIDER,
  193. VENV_PIP_BINPROVIDER,
  194. LIB_PIP_BINPROVIDER,
  195. PIP_BINARY,
  196. ARCHIVEBOX_BINARY,
  197. PYTHON_BINARY,
  198. SQLITE_BINARY,
  199. DJANGO_BINARY,
  200. USER_IS_NOT_ROOT_CHECK,
  201. PIP_ENVIRONMENT_CHECK,
  202. ]
  203. PLUGIN = PipPlugin()
  204. # PLUGIN.register(settings)
  205. DJANGO_APP = PLUGIN.AppConfig