|
1
|
|
|
import fnmatch |
|
2
|
|
|
import operator |
|
3
|
|
|
import pathlib |
|
4
|
|
|
from collections.abc import Iterable |
|
5
|
|
|
|
|
6
|
|
|
from magika import Magika |
|
7
|
|
|
|
|
8
|
|
|
import graphinate |
|
9
|
|
|
|
|
10
|
|
|
|
|
11
|
|
|
def load_ignore_patterns(ignore_files: Iterable[str]) -> set[str]: |
|
12
|
|
|
patterns = set() |
|
13
|
|
|
for ignore_file in ignore_files: |
|
14
|
|
|
if pathlib.Path(ignore_file).exists(): |
|
15
|
|
|
with open(ignore_file) as file: |
|
16
|
|
|
patterns.update(line.strip() for line in file if line.strip() and not line.startswith('#')) |
|
17
|
|
|
|
|
18
|
|
|
expand_patterns = {f"**/*{p}" if p.startswith('/') else f"**/*/{p}" for p in patterns} |
|
19
|
|
|
patterns.update(expand_patterns) |
|
20
|
|
|
return patterns |
|
21
|
|
|
|
|
22
|
|
|
|
|
23
|
|
|
def is_ignored(path: pathlib.Path, patterns: Iterable[str]) -> bool: |
|
24
|
|
|
return any(fnmatch.fnmatch(path.as_posix(), pattern) for pattern in patterns) |
|
25
|
|
|
|
|
26
|
|
|
|
|
27
|
|
|
def create_filesystem_graph_model( |
|
28
|
|
|
input_folder: str = '.', |
|
29
|
|
|
ignore_files: Iterable[str] = ('.ignore', '.gitignore', '.dockerignore') |
|
30
|
|
|
): |
|
31
|
|
|
""" |
|
32
|
|
|
Create a graph model of the file system structure. |
|
33
|
|
|
|
|
34
|
|
|
Args: |
|
35
|
|
|
input_folder (str): The folder to start the traversal from. Defaults to the current folder. |
|
36
|
|
|
ignore_files (Iterable): A list of files containing ignore patterns. |
|
37
|
|
|
Defaults to ['.ignore', '.gitignore', '.dockerignore']. |
|
38
|
|
|
|
|
39
|
|
|
Returns: |
|
40
|
|
|
GraphModel: A graph model representing the file system structure. |
|
41
|
|
|
""" |
|
42
|
|
|
graph_model = graphinate.model(name="File System Graph") |
|
43
|
|
|
magika = Magika() |
|
44
|
|
|
|
|
45
|
|
|
root_folder = pathlib.Path(input_folder) |
|
46
|
|
|
ignore_patterns = load_ignore_patterns(ignore_files) |
|
47
|
|
|
|
|
48
|
|
|
def file_type(path: pathlib.Path) -> str: |
|
49
|
|
|
if path.is_file(): |
|
50
|
|
|
return magika.identify_path(path).output.ct_label |
|
51
|
|
|
elif path.is_dir(): |
|
52
|
|
|
return 'folder' |
|
53
|
|
|
else: |
|
54
|
|
|
return 'other' |
|
55
|
|
|
|
|
56
|
|
|
as_posix = operator.methodcaller('as_posix') |
|
57
|
|
|
|
|
58
|
|
|
@graph_model.node(file_type, key=as_posix, value=as_posix) |
|
59
|
|
|
def file_node(): |
|
60
|
|
|
yield root_folder |
|
61
|
|
|
for path in root_folder.rglob('*'): |
|
62
|
|
|
if not is_ignored(path, ignore_patterns, ): |
|
63
|
|
|
yield path |
|
64
|
|
|
|
|
65
|
|
|
@graph_model.edge() |
|
66
|
|
|
def contains(): |
|
67
|
|
|
for path in root_folder.rglob('*'): |
|
68
|
|
|
if not is_ignored(path, ignore_patterns): |
|
69
|
|
|
yield { |
|
70
|
|
|
'source': path.parent.as_posix(), |
|
71
|
|
|
'target': path.as_posix() |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
return graph_model |
|
75
|
|
|
|
|
76
|
|
|
|
|
77
|
|
|
if __name__ == '__main__': |
|
78
|
|
|
input_folder = '..' # Default to the current folder |
|
79
|
|
|
ignore_files = ['.ignore', '.gitignore'] # Example list of ignore files |
|
80
|
|
|
filesystem_model = create_filesystem_graph_model(input_folder, ignore_files) |
|
81
|
|
|
|
|
82
|
|
|
schema = graphinate.builders.GraphQLBuilder(filesystem_model).build() |
|
83
|
|
|
graphinate.graphql.server(schema) |
|
84
|
|
|
|
erivlis /
graphinate
