apps.py 4.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138
  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 SinglefileConfig(BaseConfigSet):
  21. section: ClassVar[ConfigSectionName] = 'ARCHIVING_CONFIG'
  22. SAVE_SINGLEFILE: bool = True
  23. SINGLEFILE_USER_AGENT: str = Field(default=lambda: ARCHIVING_CONFIG.USER_AGENT)
  24. SINGLEFILE_TIMEOUT: int = Field(default=lambda: ARCHIVING_CONFIG.TIMEOUT)
  25. SINGLEFILE_CHECK_SSL_VALIDITY: bool = Field(default=lambda: ARCHIVING_CONFIG.CHECK_SSL_VALIDITY)
  26. SINGLEFILE_COOKIES_FILE: Optional[Path] = Field(default=lambda: ARCHIVING_CONFIG.COOKIES_FILE)
  27. SINGLEFILE_BINARY: str = Field(default='single-file')
  28. SINGLEFILE_EXTRA_ARGS: List[str] = []
  29. SINGLEFILE_CONFIG = SinglefileConfig()
  30. SINGLEFILE_MIN_VERSION = '1.1.54'
  31. SINGLEFILE_MAX_VERSION = '1.1.60'
  32. class SinglefileBinary(BaseBinary):
  33. name: BinName = SINGLEFILE_CONFIG.SINGLEFILE_BINARY
  34. binproviders_supported: List[InstanceOf[BinProvider]] = [LIB_NPM_BINPROVIDER, SYS_NPM_BINPROVIDER, env]
  35. provider_overrides: Dict[BinProviderName, ProviderLookupDict] = {
  36. env.name: {
  37. 'abspath': lambda:
  38. bin_abspath(SINGLEFILE_CONFIG.SINGLEFILE_BINARY, PATH=env.PATH)
  39. or bin_abspath('single-file', PATH=env.PATH)
  40. or bin_abspath('single-file-node.js', PATH=env.PATH),
  41. },
  42. LIB_NPM_BINPROVIDER.name: {
  43. "abspath": lambda:
  44. bin_abspath(SINGLEFILE_CONFIG.SINGLEFILE_BINARY, PATH=env.PATH)
  45. or bin_abspath("single-file", PATH=LIB_NPM_BINPROVIDER.PATH)
  46. or bin_abspath("single-file-node.js", PATH=LIB_NPM_BINPROVIDER.PATH),
  47. "packages": lambda:
  48. [f"single-file-cli@>={SINGLEFILE_MIN_VERSION} <{SINGLEFILE_MAX_VERSION}"],
  49. },
  50. SYS_NPM_BINPROVIDER.name: {
  51. "packages": lambda:
  52. [], # prevent modifying system global npm packages
  53. },
  54. }
  55. @validate_call
  56. def install(self, binprovider_name: Optional[BinProviderName]=None) -> ShallowBinary:
  57. # force install to only use lib/npm provider, we never want to modify global NPM packages
  58. return BaseBinary.install(self, binprovider_name=binprovider_name or LIB_NPM_BINPROVIDER.name)
  59. @validate_call
  60. def load_or_install(self, binprovider_name: Optional[BinProviderName] = None) -> ShallowBinary:
  61. # force install to only use lib/npm provider, we never want to modify global NPM packages
  62. try:
  63. return self.load()
  64. except Exception:
  65. return BaseBinary.install(self, binprovider_name=binprovider_name or LIB_NPM_BINPROVIDER.name)
  66. # ALTERNATIVE INSTALL METHOD using Ansible:
  67. # install_playbook = PLUGANTIC_DIR / 'ansible' / 'install_singlefile.yml'
  68. # singlefile_bin = run_playbook(install_playbook, data_dir=settings.CONFIG.OUTPUT_DIR, quiet=quiet).BINARIES.singlefile
  69. # return self.__class__.model_validate(
  70. # {
  71. # **self.model_dump(),
  72. # "loaded_abspath": singlefile_bin.abspath,
  73. # "loaded_version": singlefile_bin.version,
  74. # "loaded_binprovider": env,
  75. # "binproviders_supported": self.binproviders_supported,
  76. # }
  77. # )
  78. SINGLEFILE_BINARY = SinglefileBinary()
  79. PLUGIN_BINARIES = [SINGLEFILE_BINARY]
  80. class SinglefileExtractor(BaseExtractor):
  81. name: str = 'singlefile'
  82. binary: BinName = SINGLEFILE_BINARY.name
  83. def get_output_path(self, snapshot) -> Path:
  84. return Path(snapshot.link_dir) / 'singlefile.html'
  85. SINGLEFILE_BINARY = SinglefileBinary()
  86. SINGLEFILE_EXTRACTOR = SinglefileExtractor()
  87. class SinglefileQueue(BaseQueue):
  88. name: str = 'singlefile'
  89. binaries: List[InstanceOf[BaseBinary]] = [SINGLEFILE_BINARY]
  90. SINGLEFILE_QUEUE = SinglefileQueue()
  91. class SinglefilePlugin(BasePlugin):
  92. app_label: str ='singlefile'
  93. verbose_name: str = 'SingleFile'
  94. hooks: List[InstanceOf[BaseHook]] = [
  95. SINGLEFILE_CONFIG,
  96. SINGLEFILE_BINARY,
  97. SINGLEFILE_EXTRACTOR,
  98. SINGLEFILE_QUEUE,
  99. ]
  100. PLUGIN = SinglefilePlugin()
  101. # PLUGIN.register(settings)
  102. DJANGO_APP = PLUGIN.AppConfig