apps.py 5.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153
  1. __package__ = 'archivebox.plugins_extractor.singlefile'
  2. from pathlib import Path
  3. from typing import List, Dict, Optional, ClassVar
  4. # from typing_extensions import Self
  5. from django.conf import settings
  6. # Depends on other PyPI/vendor packages:
  7. from pydantic import InstanceOf, Field, validate_call
  8. from pydantic_pkgr import BinProvider, BinProviderName, ProviderLookupDict, BinName, bin_abspath, ShallowBinary
  9. # Depends on other Django apps:
  10. from plugantic.base_plugin import BasePlugin
  11. from plugantic.base_configset import BaseConfigSet, ConfigSectionName
  12. from plugantic.base_binary import BaseBinary, env
  13. from plugantic.base_extractor import BaseExtractor
  14. from plugantic.base_queue import BaseQueue
  15. from plugantic.base_hook import BaseHook
  16. # Depends on Other Plugins:
  17. from plugins_sys.config.apps import ARCHIVING_CONFIG
  18. from plugins_pkg.npm.apps import SYS_NPM_BINPROVIDER, LIB_NPM_BINPROVIDER
  19. ###################### Config ##########################
  20. class SinglefileToggleConfigs(BaseConfigSet):
  21. section: ClassVar[ConfigSectionName] = 'ARCHIVE_METHOD_TOGGLES'
  22. SAVE_SINGLEFILE: bool = True
  23. class SinglefileOptionsConfigs(BaseConfigSet):
  24. section: ClassVar[ConfigSectionName] = 'ARCHIVE_METHOD_OPTIONS'
  25. SINGLEFILE_USER_AGENT: str = Field(default=lambda: ARCHIVING_CONFIG.USER_AGENT)
  26. SINGLEFILE_TIMEOUT: int = Field(default=lambda: ARCHIVING_CONFIG.TIMEOUT)
  27. SINGLEFILE_CHECK_SSL_VALIDITY: bool = Field(default=lambda: ARCHIVING_CONFIG.CHECK_SSL_VALIDITY)
  28. SINGLEFILE_COOKIES_FILE: Optional[Path] = Field(default=lambda: ARCHIVING_CONFIG.COOKIES_FILE)
  29. class SinglefileDependencyConfigs(BaseConfigSet):
  30. section: ClassVar[ConfigSectionName] = 'DEPENDENCY_CONFIG'
  31. SINGLEFILE_BINARY: str = Field(default='wget')
  32. SINGLEFILE_ARGS: Optional[List[str]] = Field(default=None)
  33. SINGLEFILE_EXTRA_ARGS: List[str] = []
  34. SINGLEFILE_DEFAULT_ARGS: List[str] = ['--timeout={TIMEOUT-10}']
  35. class SinglefileConfigs(SinglefileToggleConfigs, SinglefileOptionsConfigs, SinglefileDependencyConfigs):
  36. # section: ClassVar[ConfigSectionName] = 'ALL_CONFIGS'
  37. pass
  38. DEFAULT_GLOBAL_CONFIG = {
  39. 'CHECK_SSL_VALIDITY': False,
  40. 'SAVE_SINGLEFILE': True,
  41. 'TIMEOUT': 120,
  42. }
  43. SINGLEFILE_CONFIG = SinglefileConfigs(**DEFAULT_GLOBAL_CONFIG)
  44. SINGLEFILE_MIN_VERSION = '1.1.54'
  45. SINGLEFILE_MAX_VERSION = '1.1.60'
  46. class SinglefileBinary(BaseBinary):
  47. name: BinName = 'single-file'
  48. binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_NPM_BINPROVIDER, SYS_NPM_BINPROVIDER, env]
  49. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  50. env.name: {
  51. 'abspath': lambda:
  52. bin_abspath('single-file', PATH=env.PATH) or bin_abspath('single-file-node.js', PATH=env.PATH),
  53. },
  54. LIB_NPM_BINPROVIDER.name: {
  55. "abspath": lambda:
  56. bin_abspath("single-file", PATH=LIB_NPM_BINPROVIDER.PATH) or bin_abspath("single-file-node.js", PATH=LIB_NPM_BINPROVIDER.PATH),
  57. "packages": lambda:
  58. [f"single-file-cli@>={SINGLEFILE_MIN_VERSION} <{SINGLEFILE_MAX_VERSION}"],
  59. },
  60. SYS_NPM_BINPROVIDER.name: {
  61. "packages": lambda:
  62. [], # prevent modifying system global npm packages
  63. },
  64. }
  65. @validate_call
  66. def install(self, binprovider_name: Optional[BinProviderName]=None) -> ShallowBinary:
  67. # force install to only use lib/npm provider, we never want to modify global NPM packages
  68. return BaseBinary.install(self, binprovider_name=binprovider_name or LIB_NPM_BINPROVIDER.name)
  69. @validate_call
  70. def load_or_install(self, binprovider_name: Optional[BinProviderName] = None) -> ShallowBinary:
  71. # force install to only use lib/npm provider, we never want to modify global NPM packages
  72. try:
  73. return self.load()
  74. except Exception:
  75. return BaseBinary.install(self, binprovider_name=binprovider_name or LIB_NPM_BINPROVIDER.name)
  76. # ALTERNATIVE INSTALL METHOD using Ansible:
  77. # install_playbook = PLUGANTIC_DIR / 'ansible' / 'install_singlefile.yml'
  78. # singlefile_bin = run_playbook(install_playbook, data_dir=settings.CONFIG.OUTPUT_DIR, quiet=quiet).BINARIES.singlefile
  79. # return self.__class__.model_validate(
  80. # {
  81. # **self.model_dump(),
  82. # "loaded_abspath": singlefile_bin.abspath,
  83. # "loaded_version": singlefile_bin.version,
  84. # "loaded_binprovider": env,
  85. # "binproviders_supported": self.binproviders_supported,
  86. # }
  87. # )
  88. SINGLEFILE_BINARY = SinglefileBinary()
  89. PLUGIN_BINARIES = [SINGLEFILE_BINARY]
  90. class SinglefileExtractor(BaseExtractor):
  91. name: str = 'singlefile'
  92. binary: BinName = SINGLEFILE_BINARY.name
  93. def get_output_path(self, snapshot) -> Path:
  94. return Path(snapshot.link_dir) / 'singlefile.html'
  95. SINGLEFILE_BINARY = SinglefileBinary()
  96. SINGLEFILE_EXTRACTOR = SinglefileExtractor()
  97. class SinglefileQueue(BaseQueue):
  98. name: str = 'singlefile'
  99. binaries: List[InstanceOf[BaseBinary]] = [SINGLEFILE_BINARY]
  100. SINGLEFILE_QUEUE = SinglefileQueue()
  101. class SinglefilePlugin(BasePlugin):
  102. app_label: str ='singlefile'
  103. verbose_name: str = 'SingleFile'
  104. hooks: List[InstanceOf[BaseHook]] = [
  105. SINGLEFILE_CONFIG,
  106. SINGLEFILE_BINARY,
  107. SINGLEFILE_EXTRACTOR,
  108. SINGLEFILE_QUEUE,
  109. ]
  110. PLUGIN = SinglefilePlugin()
  111. PLUGIN.register(settings)
  112. DJANGO_APP = PLUGIN.AppConfig