|
1
|
|
|
# Author: Simon Blanke |
|
2
|
|
|
# Email: [email protected] |
|
3
|
|
|
# License: MIT License |
|
4
|
|
|
|
|
5
|
|
|
|
|
6
|
|
|
import os |
|
7
|
|
|
import contextlib |
|
8
|
|
|
import pandas as pd |
|
9
|
|
|
from filelock import FileLock |
|
10
|
|
|
|
|
11
|
|
|
|
|
12
|
|
|
@contextlib.contextmanager |
|
13
|
|
|
def atomic_overwrite(filename): |
|
14
|
|
|
# from: https://stackoverflow.com/questions/42409707/pandas-to-csv-overwriting-prevent-data-loss |
|
15
|
|
|
temp = filename + "~" |
|
16
|
|
|
with open(temp, "w") as f: |
|
17
|
|
|
yield f |
|
18
|
|
|
os.rename(temp, filename) # this will only happen if no exception was raised |
|
19
|
|
|
|
|
20
|
|
|
|
|
21
|
|
|
class DataIO: |
|
22
|
|
|
def __init__(self, path, drop_duplicates): |
|
23
|
|
|
self.path = path |
|
24
|
|
|
self.replace_existing = False |
|
25
|
|
|
self.drop_duplicates = drop_duplicates |
|
26
|
|
|
|
|
27
|
|
|
if self.replace_existing: |
|
28
|
|
|
self.mode = "w" |
|
29
|
|
|
else: |
|
30
|
|
|
self.mode = "a" |
|
31
|
|
|
|
|
32
|
|
|
def _get_header(self, search_data, path): |
|
33
|
|
|
if os.path.isfile(path): |
|
34
|
|
|
if self.replace_existing: |
|
35
|
|
|
header = search_data.columns |
|
36
|
|
|
else: |
|
37
|
|
|
header = False |
|
38
|
|
|
else: |
|
39
|
|
|
header = search_data.columns |
|
40
|
|
|
return header |
|
41
|
|
|
|
|
42
|
|
|
def _save_dataframe(self, dataframe, io_wrap): |
|
43
|
|
|
if self.drop_duplicates: |
|
44
|
|
|
dataframe.drop_duplicates(subset=self.drop_duplicates, inplace=True) |
|
45
|
|
|
|
|
46
|
|
|
dataframe.to_csv(io_wrap, index=False, header=not io_wrap.tell()) |
|
47
|
|
|
|
|
48
|
|
|
def atomic_write(self, dataframe, path, replace_existing): |
|
49
|
|
|
self.replace_existing = replace_existing |
|
50
|
|
|
|
|
51
|
|
|
with atomic_overwrite(path) as io_wrap: |
|
52
|
|
|
self._save_dataframe(dataframe, io_wrap) |
|
53
|
|
|
|
|
54
|
|
|
def locked_write(self, dataframe, path): |
|
55
|
|
|
|
|
56
|
|
|
lock = FileLock(path + ".lock") |
|
57
|
|
|
with lock: |
|
58
|
|
|
with open(path, self.mode) as io_wrap: |
|
59
|
|
|
self._save_dataframe(dataframe, io_wrap) |
|
60
|
|
|
|
|
61
|
|
|
""" |
|
62
|
|
|
import fcntl |
|
63
|
|
|
|
|
64
|
|
|
with open(path, self.mode) as io_wrap: |
|
65
|
|
|
fcntl.flock(io_wrap, fcntl.LOCK_EX) |
|
66
|
|
|
self._save_dataframe(dataframe, io_wrap) |
|
67
|
|
|
fcntl.flock(io_wrap, fcntl.LOCK_UN) |
|
68
|
|
|
""" |
|
69
|
|
|
|
|
70
|
|
|
def load(self, path): |
|
71
|
|
|
if os.path.isfile(self.path) and os.path.getsize(self.path) > 0: |
|
72
|
|
|
return pd.read_csv(self.path) |
|
73
|
|
|
|
|
74
|
|
|
|
|
75
|
|
|
class DataCollector: |
|
76
|
|
|
def __init__(self, path, drop_duplicates=False): |
|
77
|
|
|
self.path = path |
|
78
|
|
|
self.drop_duplicates = drop_duplicates |
|
79
|
|
|
|
|
80
|
|
|
self.path2file = path.rsplit("/", 1)[0] + "/" |
|
81
|
|
|
self.file_name = path.rsplit("/", 1)[1] |
|
82
|
|
|
|
|
83
|
|
|
self.io = DataIO(path, drop_duplicates) |
|
84
|
|
|
|
|
85
|
|
|
def load(self): |
|
86
|
|
|
return self.io.load(self.path) |
|
87
|
|
|
|
|
88
|
|
|
def append(self, dictionary): |
|
89
|
|
|
dataframe = pd.DataFrame(dictionary, index=[0]) |
|
90
|
|
|
self.io.locked_write(dataframe, self.path) |
|
91
|
|
|
|
|
92
|
|
|
def save(self, dataframe, replace_existing=False): |
|
93
|
|
|
self.io.atomic_write(dataframe, self.path, replace_existing) |
|
94
|
|
|
|
SimonBlanke /
Hyperactive
