1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
|
# pylint: disable=[R0902, R0903]
"""This contains all the necessary functions that access the phenotypes from
the db"""
from dataclasses import dataclass, asdict, astuple
from typing import Any, Dict, Optional
from MySQLdb import escape_string
from typing_extensions import Protocol
class Dataclass(Protocol):
"""Type Definition for a Dataclass"""
__dataclass_fields__: Dict
@dataclass(frozen=True)
class Phenotype:
"""Data Type that represents a Phenotype"""
id_: Optional[int] = None
pre_pub_description: Optional[str] = None
post_pub_description: Optional[str] = None
original_description: Optional[str] = None
units: Optional[str] = None
pre_pub_abbrevition: Optional[str] = None
post_pub_abbreviation: Optional[str] = None
lab_code: Optional[str] = None
submitter: Optional[str] = None
owner: Optional[str] = None
authorized_users: Optional[str] = None
# Mapping from the Phenotype dataclass to the actual column names in the
# database
phenotype_column_mapping = {
"id_": "id",
"pre_pub_description": "Pre_publication_description",
"post_pub_description": "Post_publication_description",
"original_description": "Original_description",
"units": "Units",
"pre_pub_abbrevition": "Pre_publication_abbreviation",
"post_pub_abbreviation": "Post_publication_abbreviation",
"lab_code": "Lab_code",
"submitter": "Submitter",
"owner": "Owner",
"authorized_users": "Authorized_Users",
}
@dataclass(frozen=True)
class PublishXRef:
"""Data Type that represents the table PublishXRef"""
id_: Optional[int] = None
inbred_set_id: Optional[str] = None
phenotype_id: Optional[int] = None
publication_id: Optional[str] = None
data_id: Optional[int] = None
mean: Optional[float] = None
locus: Optional[str] = None
lrs: Optional[float] = None
additive: Optional[float] = None
sequence: Optional[int] = None
comments: Optional[str] = None
# Mapping from the PublishXRef dataclass to the actual column names in the
# database
publish_x_ref_mapping = {
"id_": "Id",
"inbred_set_id": "InbredSetId",
"phenotype_id": "PhenotypeId",
"publication_id": "PublicationId",
"data_id": "DataId",
"mean": "mean",
"locus": "locus",
"lrs": "lrs",
"additive": "additive",
"sequence": "sequence",
"comments": "comments",
}
TABLEMAP = {
"Phenotype": phenotype_column_mapping,
"PublishXRef": publish_x_ref_mapping,
}
def update(conn: Any,
table: str,
data: Dataclass,
where: Dataclass) -> Optional[int]:
"""Run an UPDATE on a table"""
if not any(astuple(data) + astuple(where)):
return None
sql = f"UPDATE {table} SET "
sql += ", ".join(f"{TABLEMAP[table].get(k)} "
f"= '{escape_string(str(v)).decode('utf-8')}'" for
k, v in asdict(data).items()
if v is not None and k in TABLEMAP[table])
sql += " WHERE "
sql += "AND ".join(f"{TABLEMAP[table].get(k)} = "
f"'{escape_string(str(v)).decode('utf-8')}'" for
k, v in asdict(where).items()
if v is not None and k in TABLEMAP[table])
with conn.cursor() as cursor:
cursor.execute(sql)
return cursor.rowcount
|