Coverage for skema/program_analysis/module_locate.py: 100%

1import argparse 

2import os 

3import pkg_resources 

4import requests 

5import importlib 

6import sys 

7import re 

8import subprocess 

9import tempfile 

10from pathlib import Path 

11 

12from skema.gromet.fn import TypedValue, ImportSourceType, GrometFNModuleDependencyReference 

13 

14IMPORT_PATTERN = re.compile(r'^\s*(from\s+[^\s]+\s+import\s+[^\s,]+(?:\s*,\s*[^\s,]+)*|import\s+[^\s,]+(?:\s*,\s*[^\s,]+)*)', re.MULTILINE) 

15 

16def identify_source_type(source: str): 

17 if not source: 

18 return "Unknown" 

19 elif source == "https://github.com/python/cpython": 

20 return "Compiled" 

21 elif "github" in source: 

22 return "Repository" 

23 elif source.startswith("http"): 

24 return "Url" 

25 return "Local" 

26 

27 

28def extract_imports(source: str): 

29 output_references = [] 

30 

31 import_statements = IMPORT_PATTERN.findall(source) 

32 modules = set(tuple([statement.split()[1] for statement in import_statements])) 

33 

34 for module in modules: 

35 source_value = module_locate(module) 

36 source_type = identify_source_type(source_value) 

37 output_references.append( 

38 GrometFNModuleDependencyReference( 

39 name=module, 

40 source_reference=TypedValue( 

41 type=source_type, 

42 value=source_value 

43 ))) 

44 

45 return output_references 

46 

47def module_locate(module_name: str) -> str: 

48 """ 

49 Locates the source of a Python module specified by the import statement. 

50 If the module is built-in or installed, it returns the file path. 

51 If the module is on PyPI with a GitHub link, it returns the GitHub URL. 

52 For PyPI modules, it also attempts to return the tarball URL for the current version. 

53 

54 :param module_name: The name of the module or submodule as a string. 

55 :return: The module's file path, GitHub URL, or tarball URL. 

56 """ 

57 

58 # Check if module is compiled into Python 

59 if module_name in sys.builtin_module_names: 

60 return "https://github.com/python/cpython" 

61 

62 # Attempt to find the module in the local environment 

63 try: 

64 module_obj = importlib.import_module(module_name) 

65 module_file = getattr(module_obj, '__file__', None) 

66 if module_file: 

67 module_path = Path(module_file) 

68 # Check if it's a package 

69 if module_path.name == "__init__.py": 

70 return str(module_path.parent) 

71 return str(module_path) 

72 except ImportError: 

73 pass # If module is not found locally, proceed to check on PyPI 

74 

75 # Fetch module info from PyPI 

76 try: 

77 pypi_url = f"https://pypi.org/pypi/{module_name}/json" 

78 response = requests.get(pypi_url) 

79 data = response.json() 

80 

81 project_urls = data.get('info', {}).get('project_urls', {}) 

82 github_url = project_urls.get('Source', '') or project_urls.get('Homepage', '') 

83 if 'github.com' in github_url: 

84 return github_url 

85 

86 # Get the tarball URL for the current version 

87 version = data['info']['version'] 

88 releases = data['releases'].get(version, []) 

89 for release in releases: 

90 if release['filename'].endswith('.tar.gz'): 

91 return release['url'] 

92 except Exception as e: 

93 # Handle errors related to network issues or JSON decoding 

94 print(f"Error fetching module information from PyPI: {e}") 

95 

96 return None 

97 

98 

99""" 

100# Basic tests  

101print(module_locate("import os")) 

102print(module_locate("import requests")) 

103print(module_locate("import xml.etree")) 

104print(module_locate("import minimal")) 

105 

106# PyDice tests 

107print(module_locate("import numpy as np")) 

108print(module_locate("import time")) 

109print(module_locate("from numba import njit,guvectorize,float64")) 

110print(module_locate("import scipy.optimize as opt")) 

111print(module_locate("from matplotlib import pyplot as plt")) 

112""" 

113 

114 

115 

116