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
|
|
|
from django.conf import settings |
14
|
|
|
|
15
|
|
|
from common.constants import CRB_ANIM_TYPE, TEMPLATE_TYPE |
16
|
|
|
from common.forms import RequestFormMixin |
17
|
|
|
from common.helpers import get_admin_emails |
18
|
|
|
from uid.models import Submission |
19
|
|
|
from crbanim.helpers import CRBAnimReader |
20
|
|
|
from excel.helpers import ExcelTemplateReader |
21
|
|
|
|
22
|
|
|
|
23
|
|
|
class UniqueSubmissionMixin(): |
24
|
|
|
def check_submission_exists(self): |
25
|
|
|
"""Test if I already have a submission with the same data""" |
26
|
|
|
|
27
|
|
|
# get unique attributes |
28
|
|
|
unique_together = Submission._meta.unique_together[0] |
29
|
|
|
|
30
|
|
|
# get submitted attributes |
31
|
|
|
data = {key: self.cleaned_data.get(key) for key in unique_together |
32
|
|
|
if self.cleaned_data.get(key) is not None} |
33
|
|
|
|
34
|
|
|
# ovverride owner attribute |
35
|
|
|
data['owner'] = self.request.user |
36
|
|
|
|
37
|
|
|
# test for a submission object with the same attributes |
38
|
|
|
if Submission.objects.filter(**data).exists(): |
39
|
|
|
msg = ( |
40
|
|
|
"Error: There is already a submission with the same " |
41
|
|
|
"attributes. Please change one of the following: " |
42
|
|
|
"Gene bank name, Gene bank country, Data source type and " |
43
|
|
|
"Data source version") |
44
|
|
|
|
45
|
|
|
# raising an exception: |
46
|
|
|
raise forms.ValidationError(msg, code='invalid') |
47
|
|
|
|
48
|
|
|
|
49
|
|
|
class SubmissionFormMixin(UniqueSubmissionMixin): |
50
|
|
|
def clean(self): |
51
|
|
|
# test if I have a submission with the provided data |
52
|
|
|
self.check_submission_exists() |
53
|
|
|
|
54
|
|
|
# I can call this method without providing a 'uploaded file' |
55
|
|
|
# (for instance, when omitting uploaded file) |
56
|
|
|
if "uploaded_file" in self.cleaned_data: |
57
|
|
|
# avoid file type for excel types (is not an text file) |
58
|
|
|
if ("datasource_type" in self.cleaned_data and |
59
|
|
|
self.cleaned_data["datasource_type"] != TEMPLATE_TYPE): |
60
|
|
|
self.check_file_encoding() |
61
|
|
|
|
62
|
|
|
# check crbanim files only if provided |
63
|
|
|
if ("datasource_type" in self.cleaned_data and |
64
|
|
|
self.cleaned_data["datasource_type"] == CRB_ANIM_TYPE): |
65
|
|
|
self.check_crbanim_columns() |
66
|
|
|
|
67
|
|
|
# check template files only if provided |
68
|
|
|
if ("datasource_type" in self.cleaned_data and |
69
|
|
|
self.cleaned_data["datasource_type"] == TEMPLATE_TYPE): |
70
|
|
|
self.check_template_file() |
71
|
|
|
|
72
|
|
|
def check_file_encoding(self): |
73
|
|
|
uploaded_file = self.cleaned_data['uploaded_file'] |
74
|
|
|
|
75
|
|
|
# read one chunk of such file |
76
|
|
|
chunk = next(uploaded_file.chunks()) |
77
|
|
|
magic_line = magic.from_buffer(chunk) |
78
|
|
|
file_type = magic_line.split(",")[0] |
79
|
|
|
|
80
|
|
|
if "UTF-8" not in file_type and "ASCII" not in file_type: |
81
|
|
|
# create message and add error |
82
|
|
|
msg = ( |
83
|
|
|
"Error: file not in UTF-8 nor ASCII format: " |
84
|
|
|
"format was %s" % file_type) |
85
|
|
|
|
86
|
|
|
# raising an exception: |
87
|
|
|
raise forms.ValidationError(msg, code='invalid') |
88
|
|
|
|
89
|
|
|
def check_crbanim_columns(self): |
90
|
|
|
"""Check if a CRBanim file has mandatory columns""" |
91
|
|
|
|
92
|
|
|
uploaded_file = self.cleaned_data['uploaded_file'] |
93
|
|
|
|
94
|
|
|
# read one chunk of such file |
95
|
|
|
chunk = next(uploaded_file.chunks()) |
96
|
|
|
|
97
|
|
|
# now determine if CRBanim file is valid. chunk is in binary format |
98
|
|
|
# neet to convert to a string, fortunately I've already check that |
99
|
|
|
# file is in UTF-8 |
100
|
|
|
check, not_found = CRBAnimReader.is_valid(chunk.decode("utf-8")) |
101
|
|
|
|
102
|
|
|
if check is False: |
103
|
|
|
msg = "Error: file lacks of CRBanim mandatory columns: %s" % ( |
104
|
|
|
not_found) |
105
|
|
|
|
106
|
|
|
# raising an exception: |
107
|
|
|
raise forms.ValidationError(msg, code='invalid') |
108
|
|
|
|
109
|
|
|
def check_template_file(self): |
110
|
|
|
"""Check if template file has columns and sheets""" |
111
|
|
|
|
112
|
|
|
uploaded_file = self.cleaned_data['uploaded_file'] |
113
|
|
|
|
114
|
|
|
chunk = next(uploaded_file.chunks()) |
115
|
|
|
magic_line = magic.from_buffer(chunk) |
116
|
|
|
|
117
|
|
|
if 'Microsoft' not in magic_line: |
118
|
|
|
msg = "The file you provided is not a Template file" |
119
|
|
|
raise forms.ValidationError(msg, code='invalid') |
120
|
|
|
|
121
|
|
|
# xlrd can manage only files. Write a temporary file |
122
|
|
|
with tempfile.NamedTemporaryFile(delete=True) as tmpfile: |
123
|
|
|
for chunk in uploaded_file.chunks(): |
124
|
|
|
tmpfile.write(chunk) |
125
|
|
|
|
126
|
|
|
# open the file with proper model |
127
|
|
|
reader = ExcelTemplateReader() |
128
|
|
|
reader.read_file(tmpfile.name) |
129
|
|
|
|
130
|
|
|
# check that template has at least breed, animal, sample sheets |
131
|
|
|
check, not_found = reader.check_sheets() |
132
|
|
|
|
133
|
|
|
if check is False: |
134
|
|
|
msg = "Error: file lacks of Template mandatory sheets: %s" % ( |
135
|
|
|
not_found) |
136
|
|
|
|
137
|
|
|
# raising an exception: |
138
|
|
|
raise forms.ValidationError(msg, code='invalid') |
139
|
|
|
|
140
|
|
|
# check that template has at least breed, animal, sample sheets |
141
|
|
|
check, not_found = reader.check_columns() |
142
|
|
|
|
143
|
|
|
if check is False: |
144
|
|
|
msg = "Error: file lacks of Template mandatory columns: %s" % ( |
145
|
|
|
not_found) |
146
|
|
|
|
147
|
|
|
# raising an exception: |
148
|
|
|
raise forms.ValidationError(msg, code='invalid') |
149
|
|
|
|
150
|
|
|
|
151
|
|
|
class SubmissionForm(SubmissionFormMixin, RequestFormMixin, forms.ModelForm): |
152
|
|
|
class Meta: |
153
|
|
|
model = Submission |
154
|
|
|
fields = ( |
155
|
|
|
'title', |
156
|
|
|
'description', |
157
|
|
|
'gene_bank_name', |
158
|
|
|
'gene_bank_country', |
159
|
|
|
'organization', |
160
|
|
|
'datasource_type', |
161
|
|
|
'datasource_version', |
162
|
|
|
'uploaded_file' |
163
|
|
|
) |
164
|
|
|
|
165
|
|
|
help_texts = { |
166
|
|
|
'uploaded_file': 'Need to be in UTF-8 or ASCII format', |
167
|
|
|
'organization': ( |
168
|
|
|
"""Who owns the data. Not listed? please """ |
169
|
|
|
"""<a href="mailto:{0}?subject=please add my organization">""" |
170
|
|
|
"""contact us</a>""".format(get_admin_emails()[0]) |
171
|
|
|
), |
172
|
|
|
'datasource_type': ( |
173
|
|
|
"""example: CryoWeb. Need an empty template file? """ |
174
|
|
|
"""download it from <a href="%s%s">here</a>""" % ( |
175
|
|
|
settings.MEDIA_URL, |
176
|
|
|
"Image_sample_empty_template_20191002.xlsx") |
177
|
|
|
) |
178
|
|
|
} |
179
|
|
|
|
180
|
|
|
|
181
|
|
|
# I use forms.Form since I need to pass primary key as a field, |
182
|
|
|
# and I can't use it with a modelform |
183
|
|
|
class ReloadForm(SubmissionFormMixin, RequestFormMixin, forms.ModelForm): |
184
|
|
|
# custom attributes |
185
|
|
|
agree_reload = forms.BooleanField( |
186
|
|
|
label="That's fine. Replace my submission data with this file", |
187
|
|
|
help_text="You have to check this box to reload your data") |
188
|
|
|
|
189
|
|
|
class Meta: |
190
|
|
|
model = Submission |
191
|
|
|
fields = ( |
192
|
|
|
'datasource_type', |
193
|
|
|
'datasource_version', |
194
|
|
|
'uploaded_file', |
195
|
|
|
) |
196
|
|
|
|
197
|
|
|
help_texts = { |
198
|
|
|
'uploaded_file': 'Need to be in UTF-8 or ASCII format', |
199
|
|
|
'datasource_type': ( |
200
|
|
|
"""example: CryoWeb. Need an empty template file? """ |
201
|
|
|
"""download it from <a href="%s%s">here</a>""" % ( |
202
|
|
|
settings.MEDIA_URL, |
203
|
|
|
"Image_sample_empty_template_20191002.xlsx") |
204
|
|
|
) |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
|
208
|
|
|
class UpdateSubmissionForm( |
209
|
|
|
UniqueSubmissionMixin, RequestFormMixin, forms.ModelForm): |
210
|
|
|
class Meta: |
211
|
|
|
model = Submission |
212
|
|
|
fields = ( |
213
|
|
|
'title', |
214
|
|
|
'description', |
215
|
|
|
'gene_bank_name', |
216
|
|
|
'gene_bank_country', |
217
|
|
|
'organization', |
218
|
|
|
"datasource_type", |
219
|
|
|
"datasource_version", |
220
|
|
|
) |
221
|
|
|
|
222
|
|
|
help_texts = { |
223
|
|
|
'organization': ( |
224
|
|
|
"""Who owns the data. Not listed? please """ |
225
|
|
|
"""<a href="mailto:{0}?subject=please add my organization">""" |
226
|
|
|
"""contact us</a>""".format(get_admin_emails()[0]) |
227
|
|
|
) |
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
def clean(self): |
231
|
|
|
# test if I have a submission with the provided data |
232
|
|
|
self.check_submission_exists() |
233
|
|
|
|