apps.py 6.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174
  1. __package__ = 'archivebox.plugins_pkg.puppeteer'
  2. import os
  3. import platform
  4. from pathlib import Path
  5. from typing import List, Optional, Dict, ClassVar
  6. # Depends on other PyPI/vendor packages:
  7. from pydantic import InstanceOf, Field
  8. from pydantic_pkgr import (
  9. BinProvider,
  10. BinName,
  11. BinProviderName,
  12. BinProviderOverrides,
  13. InstallArgs,
  14. PATHStr,
  15. HostBinPath,
  16. )
  17. from archivebox.config import CONSTANTS
  18. from archivebox.config.permissions import ARCHIVEBOX_USER
  19. # Depends on other Django apps:
  20. from abx.archivebox.base_plugin import BasePlugin
  21. from abx.archivebox.base_configset import BaseConfigSet
  22. from abx.archivebox.base_binary import BaseBinary, BaseBinProvider, env
  23. # from abx.archivebox.base_extractor import BaseExtractor
  24. # from abx.archivebox.base_queue import BaseQueue
  25. from abx.archivebox.base_hook import BaseHook
  26. # Depends on Other Plugins:
  27. from plugins_pkg.npm.apps import LIB_NPM_BINPROVIDER, SYS_NPM_BINPROVIDER
  28. ###################### Config ##########################
  29. class PuppeteerConfigs(BaseConfigSet):
  30. # PUPPETEER_BINARY: str = Field(default='wget')
  31. # PUPPETEER_ARGS: Optional[List[str]] = Field(default=None)
  32. # PUPPETEER_EXTRA_ARGS: List[str] = []
  33. # PUPPETEER_DEFAULT_ARGS: List[str] = ['--timeout={TIMEOUT-10}']
  34. pass
  35. PUPPETEER_CONFIG = PuppeteerConfigs()
  36. LIB_DIR_BROWSERS = CONSTANTS.LIB_BROWSERS_DIR
  37. class PuppeteerBinary(BaseBinary):
  38. name: BinName = "puppeteer"
  39. binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_NPM_BINPROVIDER, SYS_NPM_BINPROVIDER, env]
  40. PUPPETEER_BINARY = PuppeteerBinary()
  41. class PuppeteerBinProvider(BaseBinProvider):
  42. name: BinProviderName = "puppeteer"
  43. INSTALLER_BIN: BinName = "npx"
  44. PATH: PATHStr = str(CONSTANTS.LIB_BIN_DIR)
  45. euid: Optional[int] = ARCHIVEBOX_USER
  46. puppeteer_browsers_dir: Path = LIB_DIR_BROWSERS
  47. puppeteer_install_args: List[str] = ['--yes', "@puppeteer/browsers", "install", "--path", str(LIB_DIR_BROWSERS)]
  48. packages_handler: BinProviderOverrides = Field(default={
  49. "chrome": lambda:
  50. ['chrome@stable'],
  51. }, exclude=True)
  52. _browser_abspaths: ClassVar[Dict[str, HostBinPath]] = {}
  53. def setup(self) -> None:
  54. assert SYS_NPM_BINPROVIDER.INSTALLER_BIN_ABSPATH, "NPM bin provider not initialized"
  55. if self.puppeteer_browsers_dir:
  56. self.puppeteer_browsers_dir.mkdir(parents=True, exist_ok=True)
  57. def installed_browser_bins(self, browser_name: str='*') -> List[Path]:
  58. # if on macOS, browser binary is inside a .app, otherwise it's just a plain binary
  59. if platform.system().lower() == 'darwin':
  60. # /data/lib/browsers/chrome/mac_arm-129.0.6668.58/chrome-mac-arm64/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing
  61. return sorted(self.puppeteer_browsers_dir.glob(f'{browser_name}/mac*/chrome*/Google Chrome for Testing.app/Contents/MacOS/Google Chrome for Testing'))
  62. # /data/lib/browsers/chrome/linux-131.0.6730.0/chrome-linux64/chrome
  63. return sorted(self.puppeteer_browsers_dir.glob(f"{browser_name}/linux*/chrome*/chrome"))
  64. def default_abspath_handler(self, bin_name: BinName, **context) -> Optional[HostBinPath]:
  65. assert bin_name == 'chrome', 'Only chrome is supported using the @puppeteer/browsers install method currently.'
  66. # already loaded, return abspath from cache
  67. if bin_name in self._browser_abspaths:
  68. return self._browser_abspaths[bin_name]
  69. # first time loading, find browser in self.puppeteer_browsers_dir by searching filesystem for installed binaries
  70. matching_bins = [abspath for abspath in self.installed_browser_bins() if bin_name in str(abspath)]
  71. if matching_bins:
  72. newest_bin = matching_bins[-1] # already sorted alphabetically, last should theoretically be highest version number
  73. self._browser_abspaths[bin_name] = newest_bin
  74. return self._browser_abspaths[bin_name]
  75. return None
  76. def default_install_handler(self, bin_name: str, packages: Optional[InstallArgs] = None, **context) -> str:
  77. """npx @puppeteer/browsers install chrome@stable"""
  78. self.setup()
  79. assert bin_name == 'chrome', 'Only chrome is supported using the @puppeteer/browsers install method currently.'
  80. if not self.INSTALLER_BIN_ABSPATH:
  81. raise Exception(
  82. f"{self.__class__.__name__} install method is not available on this host ({self.INSTALLER_BIN} not found in $PATH)"
  83. )
  84. packages = packages or self.get_packages(bin_name)
  85. assert packages, f"No packages specified for installation of {bin_name}"
  86. # print(f'[*] {self.__class__.__name__}: Installing {bin_name}: {self.INSTALLER_BIN_ABSPATH} install {packages}')
  87. install_args = [*self.puppeteer_install_args]
  88. proc = self.exec(bin_name=self.INSTALLER_BIN_ABSPATH, cmd=[*install_args, *packages])
  89. if proc.returncode != 0:
  90. print(proc.stdout.strip())
  91. print(proc.stderr.strip())
  92. raise Exception(f"{self.__class__.__name__}: install got returncode {proc.returncode} while installing {packages}: {packages}")
  93. # to proceed? (y) [email protected] /tmp/test3/lib/x86_64-linux/browsers/chrome/linux-129.0.6668.91/chrome-linux64/chrome
  94. # [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
  95. relpath = proc.stdout.strip().split(str(self.puppeteer_browsers_dir))[-1].split('\n', 1)[0]
  96. abspath = self.puppeteer_browsers_dir / relpath
  97. if os.path.isfile(abspath) and os.access(abspath, os.X_OK):
  98. self._browser_abspaths[bin_name] = abspath
  99. return abspath
  100. return proc.stderr.strip() + "\n" + proc.stdout.strip()
  101. PUPPETEER_BINPROVIDER = PuppeteerBinProvider()
  102. # ALTERNATIVE INSTALL METHOD using Ansible:
  103. # install_playbook = self.plugin_dir / 'install_puppeteer.yml'
  104. # chrome_bin = run_playbook(install_playbook, data_dir=DATA_DIR, quiet=quiet).BINARIES.chrome
  105. # return self.__class__.model_validate(
  106. # {
  107. # **self.model_dump(),
  108. # "loaded_abspath": chrome_bin.symlink,
  109. # "loaded_version": chrome_bin.version,
  110. # "loaded_binprovider": env,
  111. # "binproviders_supported": self.binproviders_supported,
  112. # }
  113. # )
  114. class PuppeteerPlugin(BasePlugin):
  115. app_label: str ='puppeteer'
  116. verbose_name: str = 'Puppeteer (NPM)'
  117. hooks: List[InstanceOf[BaseHook]] = [
  118. PUPPETEER_CONFIG,
  119. PUPPETEER_BINPROVIDER,
  120. PUPPETEER_BINARY,
  121. ]
  122. PLUGIN = PuppeteerPlugin()
  123. # PLUGIN.register(settings)
  124. DJANGO_APP = PLUGIN.AppConfig