|
1
|
|
|
#!/usr/bin/env python3 |
|
2
|
|
|
# -*- coding: utf-8 -*- |
|
3
|
|
|
""" |
|
4
|
|
|
Created on Tue Jul 24 15:51:05 2018 |
|
5
|
|
|
|
|
6
|
|
|
@author: Paolo Cozzi <[email protected]> |
|
7
|
|
|
""" |
|
8
|
|
|
|
|
9
|
|
|
import magic |
|
10
|
|
|
import tempfile |
|
11
|
|
|
|
|
12
|
|
|
from django import forms |
|
13
|
|
|
|
|
14
|
|
|
from common.forms import RequestFormMixin |
|
15
|
|
|
from common.constants import CRB_ANIM_TYPE, TEMPLATE_TYPE |
|
16
|
|
|
from image_app.models import Submission |
|
17
|
|
|
from crbanim.helpers import CRBAnimReader |
|
18
|
|
|
from excel.helpers import ExcelTemplateReader |
|
19
|
|
|
|
|
20
|
|
|
|
|
21
|
|
|
class SubmissionFormMixin(): |
|
22
|
|
|
def clean(self): |
|
23
|
|
|
# I can call this method without providing a 'uploaded file' |
|
24
|
|
|
# (for instance, when omitting uploaded file) |
|
25
|
|
|
if "uploaded_file" in self.cleaned_data: |
|
26
|
|
|
# avoid file type for excel types (is not an text file) |
|
27
|
|
|
if ("datasource_type" in self.cleaned_data and |
|
28
|
|
|
self.cleaned_data["datasource_type"] != TEMPLATE_TYPE): |
|
29
|
|
|
self.check_file_encoding() |
|
30
|
|
|
|
|
31
|
|
|
# check crbanim files only if provided |
|
32
|
|
|
if ("datasource_type" in self.cleaned_data and |
|
33
|
|
|
self.cleaned_data["datasource_type"] == CRB_ANIM_TYPE): |
|
34
|
|
|
self.check_crbanim_columns() |
|
35
|
|
|
|
|
36
|
|
|
# check template files only if provided |
|
37
|
|
|
if ("datasource_type" in self.cleaned_data and |
|
38
|
|
|
self.cleaned_data["datasource_type"] == TEMPLATE_TYPE): |
|
39
|
|
|
self.check_template_file() |
|
40
|
|
|
|
|
41
|
|
|
def check_file_encoding(self): |
|
42
|
|
|
uploaded_file = self.cleaned_data['uploaded_file'] |
|
43
|
|
|
|
|
44
|
|
|
# read one chunk of such file |
|
45
|
|
|
chunk = next(uploaded_file.chunks()) |
|
46
|
|
|
magic_line = magic.from_buffer(chunk) |
|
47
|
|
|
file_type = magic_line.split(",")[0] |
|
48
|
|
|
|
|
49
|
|
|
if "UTF-8" not in file_type and "ASCII" not in file_type: |
|
50
|
|
|
# create message and add error |
|
51
|
|
|
msg = ( |
|
52
|
|
|
"Error: file not in UTF-8 nor ASCII format: " |
|
53
|
|
|
"format was %s" % file_type) |
|
54
|
|
|
|
|
55
|
|
|
# raising an exception: |
|
56
|
|
|
raise forms.ValidationError(msg, code='invalid') |
|
57
|
|
|
|
|
58
|
|
|
def check_crbanim_columns(self): |
|
59
|
|
|
"""Check if a CRBanim file has mandatory columns""" |
|
60
|
|
|
|
|
61
|
|
|
uploaded_file = self.cleaned_data['uploaded_file'] |
|
62
|
|
|
|
|
63
|
|
|
# read one chunk of such file |
|
64
|
|
|
chunk = next(uploaded_file.chunks()) |
|
65
|
|
|
|
|
66
|
|
|
# now determine if CRBanim file is valid. chunk is in binary format |
|
67
|
|
|
# neet to convert to a string, fortunately I've already check that |
|
68
|
|
|
# file is in UTF-8 |
|
69
|
|
|
check, not_found = CRBAnimReader.is_valid(chunk.decode("utf-8")) |
|
70
|
|
|
|
|
71
|
|
|
if check is False: |
|
72
|
|
|
msg = "Error: file lacks of CRBanim mandatory columns: %s" % ( |
|
73
|
|
|
not_found) |
|
74
|
|
|
|
|
75
|
|
|
# raising an exception: |
|
76
|
|
|
raise forms.ValidationError(msg, code='invalid') |
|
77
|
|
|
|
|
78
|
|
|
def check_template_file(self): |
|
79
|
|
|
"""Check if template file has columns and sheets""" |
|
80
|
|
|
|
|
81
|
|
|
uploaded_file = self.cleaned_data['uploaded_file'] |
|
82
|
|
|
|
|
83
|
|
|
chunk = next(uploaded_file.chunks()) |
|
84
|
|
|
magic_line = magic.from_buffer(chunk) |
|
85
|
|
|
|
|
86
|
|
|
if 'Microsoft' not in magic_line: |
|
87
|
|
|
msg = "The file you provided is not a Template file" |
|
88
|
|
|
raise forms.ValidationError(msg, code='invalid') |
|
89
|
|
|
|
|
90
|
|
|
# xlrd can manage only files. Write a temporary file |
|
91
|
|
|
with tempfile.NamedTemporaryFile(delete=True) as tmpfile: |
|
92
|
|
|
for chunk in uploaded_file.chunks(): |
|
93
|
|
|
tmpfile.write(chunk) |
|
94
|
|
|
|
|
95
|
|
|
# open the file with proper model |
|
96
|
|
|
reader = ExcelTemplateReader() |
|
97
|
|
|
reader.read_file(tmpfile.name) |
|
98
|
|
|
|
|
99
|
|
|
# check that template has at least breed, animal, sample sheets |
|
100
|
|
|
check, not_found = reader.check_sheets() |
|
101
|
|
|
|
|
102
|
|
|
if check is False: |
|
103
|
|
|
msg = "Error: file lacks of Template mandatory sheets: %s" % ( |
|
104
|
|
|
not_found) |
|
105
|
|
|
|
|
106
|
|
|
# raising an exception: |
|
107
|
|
|
raise forms.ValidationError(msg, code='invalid') |
|
108
|
|
|
|
|
109
|
|
|
# check that template has at least breed, animal, sample sheets |
|
110
|
|
|
check, not_found = reader.check_columns() |
|
111
|
|
|
|
|
112
|
|
|
if check is False: |
|
113
|
|
|
msg = "Error: file lacks of Template mandatory columns: %s" % ( |
|
114
|
|
|
not_found) |
|
115
|
|
|
|
|
116
|
|
|
# raising an exception: |
|
117
|
|
|
raise forms.ValidationError(msg, code='invalid') |
|
118
|
|
|
|
|
119
|
|
|
|
|
120
|
|
|
class SubmissionForm(SubmissionFormMixin, RequestFormMixin, forms.ModelForm): |
|
121
|
|
|
class Meta: |
|
122
|
|
|
model = Submission |
|
123
|
|
|
fields = ( |
|
124
|
|
|
'title', |
|
125
|
|
|
'description', |
|
126
|
|
|
'gene_bank_name', |
|
127
|
|
|
'gene_bank_country', |
|
128
|
|
|
'organization', |
|
129
|
|
|
'datasource_type', |
|
130
|
|
|
'datasource_version', |
|
131
|
|
|
'uploaded_file' |
|
132
|
|
|
) |
|
133
|
|
|
|
|
134
|
|
|
help_texts = { |
|
135
|
|
|
'uploaded_file': 'Need to be in UTF-8 or ASCII format', |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
|
|
139
|
|
|
# I use forms.Form since I need to pass primary key as a field, |
|
140
|
|
|
# and I can't use it with a modelform |
|
141
|
|
|
class ReloadForm(SubmissionFormMixin, RequestFormMixin, forms.ModelForm): |
|
142
|
|
|
# custom attributes |
|
143
|
|
|
agree_reload = forms.BooleanField( |
|
144
|
|
|
label="That's fine. Replace my submission data with this file", |
|
145
|
|
|
help_text="You have to check this box to reload your data") |
|
146
|
|
|
|
|
147
|
|
|
class Meta: |
|
148
|
|
|
model = Submission |
|
149
|
|
|
fields = ( |
|
150
|
|
|
'datasource_type', |
|
151
|
|
|
'datasource_version', |
|
152
|
|
|
'uploaded_file', |
|
153
|
|
|
) |
|
154
|
|
|
|
|
155
|
|
|
help_texts = { |
|
156
|
|
|
'uploaded_file': 'Need to be in UTF-8 or ASCII format', |
|
157
|
|
|
} |
|
158
|
|
|
|
cnr-ibba /
IMAGE-InjectTool
