apps.py 6.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169
  1. import platform
  2. from pathlib import Path
  3. from typing import List, Optional, Dict, ClassVar
  4. from django.conf import settings
  5. # Depends on other PyPI/vendor packages:
  6. from pydantic import InstanceOf, Field
  7. from pydantic_pkgr import (
  8. BinProvider,
  9. BinName,
  10. BinProviderName,
  11. ProviderLookupDict,
  12. InstallArgs,
  13. PATHStr,
  14. HostBinPath,
  15. )
  16. import archivebox
  17. # Depends on other Django apps:
  18. from plugantic.base_plugin import BasePlugin
  19. from plugantic.base_configset import BaseConfigSet
  20. from plugantic.base_binary import BaseBinary, BaseBinProvider, env
  21. # from plugantic.base_extractor import BaseExtractor
  22. # from plugantic.base_queue import BaseQueue
  23. from plugantic.base_hook import BaseHook
  24. # Depends on Other Plugins:
  25. from plugins_pkg.npm.apps import LIB_NPM_BINPROVIDER, SYS_NPM_BINPROVIDER
  26. ###################### Config ##########################
  27. class PuppeteerConfigs(BaseConfigSet):
  28. # section: ConfigSectionName = 'DEPENDENCY_CONFIG'
  29. # PUPPETEER_BINARY: str = Field(default='wget')
  30. # PUPPETEER_ARGS: Optional[List[str]] = Field(default=None)
  31. # PUPPETEER_EXTRA_ARGS: List[str] = []
  32. # PUPPETEER_DEFAULT_ARGS: List[str] = ['--timeout={TIMEOUT-10}']
  33. pass
  34. PUPPETEER_CONFIG = PuppeteerConfigs()
  35. LIB_DIR_BROWSERS = archivebox.CONSTANTS.LIB_BROWSERS_DIR
  36. class PuppeteerBinary(BaseBinary):
  37. name: BinName = "puppeteer"
  38. binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_NPM_BINPROVIDER, SYS_NPM_BINPROVIDER, env]
  39. PUPPETEER_BINARY = PuppeteerBinary()
  40. class PuppeteerBinProvider(BaseBinProvider):
  41. name: BinProviderName = "puppeteer"
  42. INSTALLER_BIN: BinName = "npx"
  43. PATH: PATHStr = str(archivebox.CONSTANTS.LIB_BIN_DIR)
  44. puppeteer_browsers_dir: Optional[Path] = LIB_DIR_BROWSERS
  45. puppeteer_install_args: List[str] = ["@puppeteer/browsers", "install", "--path", str(LIB_DIR_BROWSERS)]
  46. packages_handler: ProviderLookupDict = Field(default={
  47. "chrome": lambda:
  48. ['chrome@stable'],
  49. }, exclude=True)
  50. _browser_abspaths: ClassVar[Dict[str, HostBinPath]] = {}
  51. def setup(self) -> None:
  52. assert SYS_NPM_BINPROVIDER.INSTALLER_BIN_ABSPATH, "NPM bin provider not initialized"
  53. if self.puppeteer_browsers_dir:
  54. self.puppeteer_browsers_dir.mkdir(parents=True, exist_ok=True)
  55. def installed_browser_bins(self, browser_name: str='*') -> List[Path]:
  56. # if on macOS, browser binary is inside a .app, otherwise it's just a plain binary
  57. if platform.system().lower() == 'darwin':
  58. # /data/lib/browsers/chrome/mac_arm-129.0.6668.58/chrome-mac-arm64/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing
  59. return sorted(self.puppeteer_browsers_dir.glob(f'{browser_name}/mac*/chrome*/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing'))
  60. # /data/lib/browsers/chrome/linux-131.0.6730.0/chrome-linux64/chrome
  61. return sorted(self.puppeteer_browsers_dir.glob(f"{browser_name}/linux*/chrome*/chrome"))
  62. def on_get_abspath(self, bin_name: BinName, **context) -> Optional[HostBinPath]:
  63. assert bin_name == 'chrome', 'Only chrome is supported using the @puppeteer/browsers install method currently.'
  64. # already loaded, return abspath from cache
  65. if bin_name in self._browser_abspaths:
  66. return self._browser_abspaths[bin_name]
  67. # first time loading, find browser in self.puppeteer_browsers_dir by searching filesystem for installed binaries
  68. matching_bins = [abspath for abspath in self.installed_browser_bins() if bin_name in str(abspath)]
  69. if matching_bins:
  70. newest_bin = matching_bins[-1] # already sorted alphabetically, last should theoretically be highest version number
  71. self._browser_abspaths[bin_name] = newest_bin
  72. return self._browser_abspaths[bin_name]
  73. return None
  74. def on_install(self, bin_name: str, packages: Optional[InstallArgs] = None, **context) -> str:
  75. """npx @puppeteer/browsers install chrome@stable"""
  76. self.setup()
  77. assert bin_name == 'chrome', 'Only chrome is supported using the @puppeteer/browsers install method currently.'
  78. if not self.INSTALLER_BIN_ABSPATH:
  79. raise Exception(
  80. f"{self.__class__.__name__} install method is not available on this host ({self.INSTALLER_BIN} not found in $PATH)"
  81. )
  82. packages = packages or self.on_get_packages(bin_name)
  83. # print(f'[*] {self.__class__.__name__}: Installing {bin_name}: {self.INSTALLER_BIN_ABSPATH} install {packages}')
  84. install_args = [*self.puppeteer_install_args]
  85. proc = self.exec(bin_name=self.INSTALLER_BIN_ABSPATH, cmd=[*install_args, *packages])
  86. if proc.returncode != 0:
  87. print(proc.stdout.strip())
  88. print(proc.stderr.strip())
  89. raise Exception(f"{self.__class__.__name__}: install got returncode {proc.returncode} while installing {packages}: {packages}")
  90. # [email protected] /data/lib/browsers/chrome/mac_arm-129.0.6668.58/chrome-mac-arm64/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing
  91. output_info = proc.stdout.strip().split('\n')[-1]
  92. browser_abspath = output_info.split(' ', 1)[-1]
  93. # browser_version = output_info.split('@', 1)[-1].split(' ', 1)[0]
  94. self._browser_abspaths[bin_name] = Path(browser_abspath)
  95. return proc.stderr.strip() + "\n" + proc.stdout.strip()
  96. PUPPETEER_BINPROVIDER = PuppeteerBinProvider()
  97. # ALTERNATIVE INSTALL METHOD using Ansible:
  98. # install_playbook = self.plugin_dir / 'install_puppeteer.yml'
  99. # chrome_bin = run_playbook(install_playbook, data_dir=archivebox.DATA_DIR, quiet=quiet).BINARIES.chrome
  100. # return self.__class__.model_validate(
  101. # {
  102. # **self.model_dump(),
  103. # "loaded_abspath": chrome_bin.symlink,
  104. # "loaded_version": chrome_bin.version,
  105. # "loaded_binprovider": env,
  106. # "binproviders_supported": self.binproviders_supported,
  107. # }
  108. # )
  109. class PuppeteerPlugin(BasePlugin):
  110. app_label: str ='puppeteer'
  111. verbose_name: str = 'Puppeteer (NPM)'
  112. hooks: List[InstanceOf[BaseHook]] = [
  113. PUPPETEER_CONFIG,
  114. PUPPETEER_BINPROVIDER,
  115. PUPPETEER_BINARY,
  116. ]
  117. PLUGIN = PuppeteerPlugin()
  118. PLUGIN.register(settings)
  119. DJANGO_APP = PLUGIN.AppConfig