|
1
|
|
|
import os |
|
2
|
|
|
|
|
3
|
|
|
from looptools import Counter |
|
4
|
|
|
|
|
5
|
|
|
|
|
6
|
|
View Code Duplication |
class Crawler: |
|
|
|
|
|
|
7
|
|
|
|
|
8
|
|
|
def __init__(self, directory, filters, full_paths, topown, _printer): |
|
9
|
|
|
"""Sub class of DirPaths used for sequential directory parsing""" |
|
10
|
|
|
self.directory = directory |
|
11
|
|
|
self.filters = filters |
|
12
|
|
|
self.topdown = topown |
|
13
|
|
|
self._printer = _printer |
|
14
|
|
|
|
|
15
|
|
|
self.filepaths = [] |
|
16
|
|
|
|
|
17
|
|
|
if full_paths: |
|
18
|
|
|
self.add_path = self._add_filepath_absolute |
|
19
|
|
|
self._printer('Absolute paths') |
|
20
|
|
|
else: |
|
21
|
|
|
self.add_path = self._add_filepath_relative |
|
22
|
|
|
self._printer('Relative paths') |
|
23
|
|
|
|
|
24
|
|
|
def __iter__(self): |
|
25
|
|
|
return iter(self.filepaths) |
|
26
|
|
|
|
|
27
|
|
|
def __len__(self): |
|
28
|
|
|
return len(self.filepaths) |
|
29
|
|
|
|
|
30
|
|
|
def _add_filepath_relative(self, directory, fullname): |
|
31
|
|
|
self.filepaths.append(fullname) |
|
32
|
|
|
|
|
33
|
|
|
def _add_filepath_absolute(self, directory, fullname): |
|
34
|
|
|
self.filepaths.append(os.path.join(directory, fullname)) |
|
35
|
|
|
|
|
36
|
|
|
def crawler(self): |
|
37
|
|
|
if self.filters: |
|
38
|
|
|
self._printer('Filtering enabled') |
|
39
|
|
|
self.filter() |
|
40
|
|
|
else: |
|
41
|
|
|
self._printer('Filtering disabled') |
|
42
|
|
|
self.encompass() |
|
43
|
|
|
return self.filepaths |
|
44
|
|
|
|
|
45
|
|
|
def encompass(self): |
|
46
|
|
|
""" |
|
47
|
|
|
Called when parallelize is False. |
|
48
|
|
|
This function will generate the file names in a directory tree by walking the tree either top-down or |
|
49
|
|
|
bottom-up. For each directory in the tree rooted at directory top (including top itself), it yields a 3-tuple |
|
50
|
|
|
(dirpath, dirnames, filenames). |
|
51
|
|
|
""" |
|
52
|
|
|
self._printer('Standard Walk') |
|
53
|
|
|
count = Counter(length=3) |
|
54
|
|
|
for directory in self.directory: |
|
55
|
|
|
for root, directories, files in os.walk(directory, topdown=self.topdown): |
|
56
|
|
|
root = root[len(str(directory)) + 1:] |
|
57
|
|
|
self._printer(str(count.up) + ": Explored path - " + str(root), stream=True) |
|
58
|
|
|
for filename in files: |
|
59
|
|
|
fullname = os.path.join(root, filename) |
|
60
|
|
|
# Join the two strings in order to form the full filepath. |
|
61
|
|
|
self.add_path(directory, fullname) |
|
62
|
|
|
|
|
63
|
|
|
def filter(self): |
|
64
|
|
|
""" |
|
65
|
|
|
Called when parallelize is False. |
|
66
|
|
|
This function will generate the file names in a directory tree by walking the tree either top-down or |
|
67
|
|
|
bottom-up. For each directory in the tree rooted at directory top (including top itself), it yields a 3-tuple |
|
68
|
|
|
(dirpath, dirnames, filenames). |
|
69
|
|
|
""" |
|
70
|
|
|
self._printer('Standard Walk') |
|
71
|
|
|
count = Counter(length=3) |
|
72
|
|
|
for directory in self.directory: |
|
73
|
|
|
self._printer('Searching ' + directory) |
|
74
|
|
|
for root, directories, files in os.walk(directory, topdown=self.topdown): |
|
75
|
|
|
root = root[len(str(directory)) + 1:] |
|
76
|
|
|
self._printer(str(count.up) + ": Explored path - " + str(root), stream=True) |
|
77
|
|
|
if self.filters.validate(root): |
|
78
|
|
|
# Check that non-empty folders flag is on and we're at the max directory level |
|
79
|
|
|
if self.filters.non_empty_folders and self.filters.get_level(root) == self.filters.max_level: |
|
80
|
|
|
# Check that the path is not an empty folder |
|
81
|
|
|
if os.path.isdir(directory + os.sep + root): |
|
82
|
|
|
# Get paths in folder without walking directory |
|
83
|
|
|
paths = os.listdir(directory + os.sep + root) |
|
84
|
|
|
|
|
85
|
|
|
# Check that any of the paths are files and not just directories |
|
86
|
|
|
if paths and any(os.path.isfile(os.path.join(directory, p)) for p in paths): |
|
87
|
|
|
self.add_path(directory, root) |
|
88
|
|
|
|
|
89
|
|
|
else: |
|
90
|
|
|
for filename in files: |
|
91
|
|
|
fullname = os.path.join(root, filename) |
|
92
|
|
|
if self.filters.validate(fullname): |
|
93
|
|
|
# Join the two strings in order to form the full filepath. |
|
94
|
|
|
self.add_path(directory, fullname) |
|
95
|
|
|
|
sfneal /
dirutility
Loading history...