|
1
|
|
|
"""HTTP/REST client backend that makes calls to a web service |
|
2
|
|
|
and returns the results""" |
|
3
|
|
|
|
|
4
|
|
|
from __future__ import annotations |
|
5
|
|
|
|
|
6
|
|
|
import importlib |
|
7
|
|
|
from typing import TYPE_CHECKING, Any |
|
8
|
|
|
|
|
9
|
|
|
import dateutil.parser |
|
10
|
|
|
import requests |
|
11
|
|
|
import requests.exceptions |
|
12
|
|
|
|
|
13
|
|
|
from annif.exception import OperationFailedException |
|
14
|
|
|
from annif.suggestion import SubjectSuggestion |
|
15
|
|
|
|
|
16
|
|
|
from . import backend |
|
17
|
|
|
|
|
18
|
|
|
if TYPE_CHECKING: |
|
19
|
|
|
from datetime import datetime |
|
20
|
|
|
|
|
21
|
|
|
from annif.corpus import Document |
|
22
|
|
|
|
|
23
|
|
|
|
|
24
|
|
|
class HTTPBackend(backend.AnnifBackend): |
|
25
|
|
|
name = "http" |
|
26
|
|
|
_headers = None |
|
27
|
|
|
|
|
28
|
|
|
@property |
|
29
|
|
|
def headers(self) -> dict[str, str]: |
|
30
|
|
|
if self._headers is None: |
|
31
|
|
|
version = importlib.metadata.version("annif") |
|
32
|
|
|
self._headers = { |
|
33
|
|
|
"User-Agent": f"Annif/{version}", |
|
34
|
|
|
} |
|
35
|
|
|
return self._headers |
|
36
|
|
|
|
|
37
|
|
|
@property |
|
38
|
|
|
def is_trained(self) -> bool | None: |
|
39
|
|
|
return self._get_project_info("is_trained") |
|
40
|
|
|
|
|
41
|
|
|
@property |
|
42
|
|
|
def modification_time(self) -> datetime | None: |
|
43
|
|
|
mtime = self._get_project_info("modification_time") |
|
44
|
|
|
if mtime is None: |
|
45
|
|
|
return None |
|
46
|
|
|
return dateutil.parser.parse(mtime) |
|
47
|
|
|
|
|
48
|
|
|
def _get_project_info(self, key: str) -> bool | str | None: |
|
49
|
|
|
params = self._get_backend_params(None) |
|
50
|
|
|
try: |
|
51
|
|
|
req = requests.get( |
|
52
|
|
|
params["endpoint"].replace("/suggest", ""), headers=self.headers |
|
53
|
|
|
) |
|
54
|
|
|
req.raise_for_status() |
|
55
|
|
|
except requests.exceptions.RequestException as err: |
|
56
|
|
|
msg = f"HTTP request failed: {err}" |
|
57
|
|
|
raise OperationFailedException(msg) from err |
|
58
|
|
|
try: |
|
59
|
|
|
response = req.json() |
|
60
|
|
|
except ValueError as err: |
|
61
|
|
|
msg = f"JSON decode failed: {err}" |
|
62
|
|
|
raise OperationFailedException(msg) from err |
|
63
|
|
|
|
|
64
|
|
|
if key in response: |
|
65
|
|
|
return response[key] |
|
66
|
|
|
else: |
|
67
|
|
|
return None |
|
68
|
|
|
|
|
69
|
|
|
def _suggest( |
|
70
|
|
|
self, doc: Document, params: dict[str, Any] |
|
71
|
|
|
) -> list[SubjectSuggestion]: |
|
72
|
|
|
data = {"text": doc.text} |
|
73
|
|
|
for key, value in doc.metadata.items(): |
|
74
|
|
|
data[f"metadata_{key}"] = value |
|
75
|
|
|
if "project" in params: |
|
76
|
|
|
data["project"] = params["project"] |
|
77
|
|
|
if "limit" in params: |
|
78
|
|
|
data["limit"] = params["limit"] |
|
79
|
|
|
|
|
80
|
|
|
try: |
|
81
|
|
|
req = requests.post(params["endpoint"], data=data, headers=self.headers) |
|
82
|
|
|
req.raise_for_status() |
|
83
|
|
|
except requests.exceptions.RequestException as err: |
|
84
|
|
|
self.warning("HTTP request failed: {}".format(err)) |
|
85
|
|
|
return [] |
|
86
|
|
|
|
|
87
|
|
|
try: |
|
88
|
|
|
response = req.json() |
|
89
|
|
|
except ValueError as err: |
|
90
|
|
|
self.warning("JSON decode failed: {}".format(err)) |
|
91
|
|
|
return [] |
|
92
|
|
|
|
|
93
|
|
|
if "results" in response: |
|
94
|
|
|
results = response["results"] |
|
95
|
|
|
else: |
|
96
|
|
|
results = response |
|
97
|
|
|
|
|
98
|
|
|
try: |
|
99
|
|
|
subject_suggestions = [ |
|
100
|
|
|
SubjectSuggestion( |
|
101
|
|
|
subject_id=self.project.subjects.by_uri(hit["uri"]), |
|
102
|
|
|
score=hit["score"], |
|
103
|
|
|
) |
|
104
|
|
|
for hit in results |
|
105
|
|
|
if hit["score"] > 0.0 |
|
106
|
|
|
] |
|
107
|
|
|
except (TypeError, ValueError) as err: |
|
108
|
|
|
self.warning("Problem interpreting JSON data: {}".format(err)) |
|
109
|
|
|
return [] |
|
110
|
|
|
|
|
111
|
|
|
return subject_suggestions |
|
112
|
|
|
|