Commit 3b5a487de871166efef86bab60586cbd7d0ff82e

Authored by Quillot Mathias
1 parent 9a2c6b4d02
Exists in master

Just tried to change float to np.float64 to have larger float

Showing 1 changed file with 1 additions and 1 deletions Inline Diff

1 ''' 1 '''
2 Data management input/output 2 Data management input/output
3 ''' 3 '''
4 4
5 # Import packages and modules 5 # Import packages and modules
6 import numpy as np 6 import numpy as np
7 import sys 7 import sys
8 8
9 # Defining some types 9 # Defining some types
10 from typing import List, Dict 10 from typing import List, Dict
11 KeyToList = Dict[str, List[str]] 11 KeyToList = Dict[str, List[str]]
12 KeyToLabels = Dict[str, List[str]] 12 KeyToLabels = Dict[str, List[str]]
13 KeyToIntLabels = Dict[str, List[int]] 13 KeyToIntLabels = Dict[str, List[int]]
14 KeyToFeatures = Dict[str, List[float]] 14 KeyToFeatures = Dict[str, List[float]]
15 15
16 16
17 def read_lst(file_path: str) -> KeyToList: 17 def read_lst(file_path: str) -> KeyToList:
18 ''' 18 '''
19 Read lst file with this structure: 19 Read lst file with this structure:
20 [id_1] 20 [id_1]
21 [id_2] 21 [id_2]
22 ... 22 ...
23 [id_n] 23 [id_n]
24 24
25 Return a list of ids. 25 Return a list of ids.
26 ''' 26 '''
27 lst = [] 27 lst = []
28 with open(file_path, "r") as f: 28 with open(file_path, "r") as f:
29 for line in f: 29 for line in f:
30 lst.append(line.replace("\n", "")) 30 lst.append(line.replace("\n", ""))
31 return lst 31 return lst
32 32
33 33
34 def read_id_values(file_path: str, value_type=str): 34 def read_id_values(file_path: str, value_type=str):
35 ''' 35 '''
36 Read file where each line is an id with its corresponding values: 36 Read file where each line is an id with its corresponding values:
37 [id_1] [value_1_1] [value_1_2] ... [value_1_k] 37 [id_1] [value_1_1] [value_1_2] ... [value_1_k]
38 [id_2] [value_2_1] [value_2_2] ... [value_2_k] 38 [id_2] [value_2_1] [value_2_2] ... [value_2_k]
39 ... 39 ...
40 [id_n] [value_n_1] [value_n_2] ... [value_n_k] 40 [id_n] [value_n_1] [value_n_2] ... [value_n_k]
41 41
42 where values are value_type type. 42 where values are value_type type.
43 43
44 Used in many reader functions with specific value_type. 44 Used in many reader functions with specific value_type.
45 Return a dictionary with id as key and values as associated values. 45 Return a dictionary with id as key and values as associated values.
46 ''' 46 '''
47 id_values = {} 47 id_values = {}
48 with open(file_path, "r") as f: 48 with open(file_path, "r") as f:
49 for line in f: 49 for line in f:
50 splited = line.replace("\n", "").split(" ") 50 splited = line.replace("\n", "").split(" ")
51 id_values[splited[0]] = np.asarray(splited[1:], dtype=value_type) 51 id_values[splited[0]] = np.asarray(splited[1:], dtype=value_type)
52 return id_values 52 return id_values
53 53
54 54
55 def read_features(file_path: str) -> KeyToFeatures: 55 def read_features(file_path: str) -> KeyToFeatures:
56 ''' 56 '''
57 Read features files with the following structure: 57 Read features files with the following structure:
58 [id_1] [value_1_1] [value_1_2] ... [value_1_k] 58 [id_1] [value_1_1] [value_1_2] ... [value_1_k]
59 [id_2] [value_2_1] [value_2_2] ... [value_2_k] 59 [id_2] [value_2_1] [value_2_2] ... [value_2_k]
60 ... 60 ...
61 [id_n] [value_n_1] [value_n_2] ... [value_n_k] 61 [id_n] [value_n_1] [value_n_2] ... [value_n_k]
62 62
63 where values are float 63 where values are float
64 64
65 Returns a dictionary with id as key and a list of values as associated values 65 Returns a dictionary with id as key and a list of values as associated values
66 ''' 66 '''
67 return read_id_values(file_path, float) 67 return read_id_values(file_path, np.float64)
68 68
69 69
70 def read_labels(file_path: str) -> KeyToLabels: 70 def read_labels(file_path: str) -> KeyToLabels:
71 ''' 71 '''
72 Read features files with the following structure : 72 Read features files with the following structure :
73 [id_1] [value_1_1] [value_1_2] ... [value_1_k] 73 [id_1] [value_1_1] [value_1_2] ... [value_1_k]
74 [id_2] [value_2_1] [value_2_2] ... [value_2_k] 74 [id_2] [value_2_1] [value_2_2] ... [value_2_k]
75 ... 75 ...
76 [id_n] [value_n_1] [value_n_2] ... [value_n_k] 76 [id_n] [value_n_1] [value_n_2] ... [value_n_k]
77 77
78 where values are int 78 where values are int
79 ''' 79 '''
80 return read_id_values(file_path, str) 80 return read_id_values(file_path, str)
81 81
82 82
83 def read_labels_integer(file_path: str) -> KeyToIntLabels: 83 def read_labels_integer(file_path: str) -> KeyToIntLabels:
84 ''' 84 '''
85 Read features files with the following structure : 85 Read features files with the following structure :
86 [id_1] [value_1_1] [value_1_2] ... [value_1_k] 86 [id_1] [value_1_1] [value_1_2] ... [value_1_k]
87 [id_2] [value_2_1] [value_2_2] ... [value_2_k] 87 [id_2] [value_2_1] [value_2_2] ... [value_2_k]
88 ... 88 ...
89 [id_n] [value_n_1] [value_n_2] ... [value_n_k] 89 [id_n] [value_n_1] [value_n_2] ... [value_n_k]
90 90
91 where values are int 91 where values are int
92 ''' 92 '''
93 return read_id_values(file_path, int) 93 return read_id_values(file_path, int)
94 94
95 95
96 def write_line(id_, values=[], out=sys.stdout): 96 def write_line(id_, values=[], out=sys.stdout):
97 """ 97 """
98 Write a line in list, labels or features files. 98 Write a line in list, labels or features files.
99 If you want to write a list, specify an empty 99 If you want to write a list, specify an empty
100 array for *values*. 100 array for *values*.
101 101
102 Args: 102 Args:
103 id_ (str): id in string. 103 id_ (str): id in string.
104 values (list, optional): list of values to write, features or labels. Defaults to []. 104 values (list, optional): list of values to write, features or labels. Defaults to [].
105 out (_io.TextIOWrapper, optional): . Defaults to sys.stdout. 105 out (_io.TextIOWrapper, optional): . Defaults to sys.stdout.
106 """ 106 """
107 if len(values) == 0: 107 if len(values) == 0:
108 out.write(str(id_) + "\n") 108 out.write(str(id_) + "\n")
109 else: 109 else:
110 out.write(str(id_) + " " + " ".join(values) + "\n") 110 out.write(str(id_) + " " + " ".join(values) + "\n")