find_file_dependencies.py 3.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596
  1. # mypy: allow-untyped-defs
  2. import ast
  3. from typing import Optional
  4. from ._importlib import _resolve_name
  5. class _ExtractModuleReferences(ast.NodeVisitor):
  6. """
  7. Extract the list of global variables a block of code will read and write
  8. """
  9. @classmethod
  10. def run(cls, src: str, package: str) -> list[tuple[str, Optional[str]]]:
  11. visitor = cls(package)
  12. tree = ast.parse(src)
  13. visitor.visit(tree)
  14. return list(visitor.references.keys())
  15. def __init__(self, package):
  16. super().__init__()
  17. self.package = package
  18. self.references = {}
  19. def _absmodule(self, module_name: str, level: int) -> str:
  20. if level > 0:
  21. return _resolve_name(module_name, self.package, level)
  22. return module_name
  23. def visit_Import(self, node):
  24. for alias in node.names:
  25. self.references[(alias.name, None)] = True
  26. def visit_ImportFrom(self, node):
  27. name = self._absmodule(node.module, 0 if node.level is None else node.level)
  28. for alias in node.names:
  29. # from my_package import foo
  30. # foo may be a module, so we have to add it to the list of
  31. # potential references, if import of it fails, we will ignore it
  32. if alias.name != "*":
  33. self.references[(name, alias.name)] = True
  34. else:
  35. self.references[(name, None)] = True
  36. def _grab_node_int(self, node):
  37. return node.value
  38. def _grab_node_str(self, node):
  39. return node.value
  40. def visit_Call(self, node):
  41. # __import__ calls aren't routed to the visit_Import/From nodes
  42. if hasattr(node.func, "id") and node.func.id == "__import__":
  43. try:
  44. name = self._grab_node_str(node.args[0])
  45. fromlist: list[str] = []
  46. level = 0
  47. if len(node.args) > 3:
  48. fromlist.extend(self._grab_node_str(v) for v in node.args[3].elts)
  49. elif hasattr(node, "keywords"):
  50. for keyword in node.keywords:
  51. if keyword.arg == "fromlist":
  52. fromlist.extend(
  53. self._grab_node_str(v) for v in keyword.value.elts
  54. )
  55. if len(node.args) > 4:
  56. level = self._grab_node_int(node.args[4])
  57. elif hasattr(node, "keywords"):
  58. for keyword in node.keywords:
  59. if keyword.arg == "level":
  60. level = self._grab_node_int(keyword.value)
  61. if fromlist == []:
  62. # the top-level package (the name up till the first dot) is returned
  63. # when the fromlist argument is empty in normal import system,
  64. # we need to include top level package to match this behavior and last
  65. # level package to capture the intended dependency of user
  66. self.references[(name, None)] = True
  67. top_name = name.rsplit(".", maxsplit=1)[0]
  68. if top_name != name:
  69. top_name = self._absmodule(top_name, level)
  70. self.references[(top_name, None)] = True
  71. else:
  72. name = self._absmodule(name, level)
  73. for alias in fromlist:
  74. # fromlist args may be submodules, so we have to add the fromlist args
  75. # to the list of potential references. If import of an arg fails we
  76. # will ignore it, similar to visit_ImportFrom
  77. if alias != "*":
  78. self.references[(name, alias)] = True
  79. else:
  80. self.references[(name, None)] = True
  81. except Exception:
  82. return
  83. find_files_source_depends_on = _ExtractModuleReferences.run