annif.rest.list_vocabs() - Code Metrics - Inspection of "Add vocabulary information to REST API" - NatLibFi/Annif - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — main (#839)

by Osma

created 2025-03-21 14:08 UTC

annif.rest.list_vocabs() A

↳ Parent: annif.rest

Complexity

Conditions

Size

Total Lines	10
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
eloc	6
nop	0
dl	0
loc	10
rs	10
c	0
b	0
f	0

"""Definitions for REST API operations. These are wired via Connexion to
methods defined in the OpenAPI specification."""

from __future__ import annotations

import importlib
from typing import TYPE_CHECKING, Any

import connexion

import annif.registry
import annif.simplemma_util
from annif.corpus import Document, DocumentList, SubjectSet
from annif.exception import AnnifException
from annif.project import Access

if TYPE_CHECKING:
    from connexion.lifecycle import ConnexionResponse

    from annif.corpus.subject import SubjectIndex
    from annif.suggestion import SubjectSuggestion, SuggestionResults


def project_not_found_error(project_id: str) -> ConnexionResponse:
    """return a Connexion error object when a project is not found"""

    return connexion.problem(
        status=404,
        title="Project not found",
        detail="Project '{}' not found".format(project_id),
    )


def server_error(
    err: AnnifException,
) -> ConnexionResponse:
    """return a Connexion error object when there is a server error (project
    or backend problem)"""

    return connexion.problem(
        status=503, title="Service unavailable", detail=err.format_message()
    )


def show_info() -> tuple:
    """return version of annif and a title for the api according to OpenAPI spec"""

    result = {"title": "Annif REST API", "version": importlib.metadata.version("annif")}
    return result, 200, {"Content-Type": "application/json"}


def language_not_supported_error(lang: str) -> ConnexionResponse:
    """return a Connexion error object when attempting to use unsupported language"""

    return connexion.problem(
        status=400,
        title="Bad Request",
        detail=f'language "{lang}" not supported by vocabulary',
    )


def list_vocabs() -> tuple:
    """return a dict with vocabularies formatted according to OpenAPI spec"""

    result = {
        "vocabs": [
            vocab.dump()
            for vocab in annif.registry.get_vocabs(min_access=Access.public).values()
        ]
    }
    return result, 200, {"Content-Type": "application/json"}


def list_projects() -> tuple:
    """return a dict with projects formatted according to OpenAPI spec"""

    result = {
        "projects": [
            proj.dump()
            for proj in annif.registry.get_projects(min_access=Access.public).values()
        ]
    }
    return result, 200, {"Content-Type": "application/json"}


def show_project(
    project_id: str,
) -> dict | ConnexionResponse:
    """return a single project formatted according to OpenAPI spec"""

    try:
        project = annif.registry.get_project(project_id, min_access=Access.hidden)
    except ValueError:
        return project_not_found_error(project_id)
    return project.dump(), 200, {"Content-Type": "application/json"}


def detect_language(body: dict[str, Any]):
    """return scores for detected languages formatted according to Swagger spec"""

    text = body.get("text")
    languages = body.get("languages")

    try:
        proportions = annif.simplemma_util.detect_language(text, tuple(languages))
    except ValueError:
        return connexion.problem(
            status=400,
            title="Bad Request",
            detail="unsupported candidate languages",
        )

    result = {
        "results": [
            {"language": lang if lang != "unk" else None, "score": score}
            for lang, score in proportions.items()
        ]
    }
    return result, 200, {"Content-Type": "application/json"}


def _suggestion_to_dict(
    suggestion: SubjectSuggestion, subject_index: SubjectIndex, language: str
) -> dict[str, str | float | None]:
    subject = subject_index[suggestion.subject_id]
    return {
        "uri": subject.uri,
        "label": subject.labels[language],
        "notation": subject.notation,
        "score": suggestion.score,
    }


def _hit_sets_to_list(
    hit_sets: SuggestionResults, subjects: SubjectIndex, lang: str
) -> list[dict[str, list]]:
    return [
        {"results": [_suggestion_to_dict(hit, subjects, lang) for hit in hits]}
        for hits in hit_sets
    ]


def _is_error(result: list[dict[str, list]] | ConnexionResponse) -> bool:
    return (
        isinstance(result, connexion.lifecycle.ConnexionResponse)
        and result.status_code >= 400
    )


def suggest(
    project_id: str, body: dict[str, Any]
) -> dict[str, list] | ConnexionResponse:
    """suggest subjects for the given text and return a dict with results
    formatted according to OpenAPI spec"""

    parameters = dict(
        (key, body[key]) for key in ["language", "limit", "threshold"] if key in body
    )
    documents = [{"text": body["text"]}]
    result = _suggest(project_id, documents, parameters)

    if _is_error(result):
        return result
    return result[0], 200, {"Content-Type": "application/json"}


def suggest_batch(
    project_id: str,
    body: dict[str, list],
    **query_parameters,
) -> list[dict[str, Any]] | ConnexionResponse:
    """suggest subjects for the given documents and return a list of dicts with results
    formatted according to OpenAPI spec"""

    documents = body["documents"]
    result = _suggest(project_id, documents, query_parameters)

    if _is_error(result):
        return result
    for document_results, document in zip(result, documents):
        document_results["document_id"] = document.get("document_id")
    return result, 200, {"Content-Type": "application/json"}


def _suggest(
    project_id: str,
    documents: list[dict[str, str]],
    parameters: dict[str, Any],
) -> list[dict[str, list]] | ConnexionResponse:
    corpus = _documents_to_corpus(documents, subject_index=None)
    try:
        project = annif.registry.get_project(project_id, min_access=Access.hidden)
    except ValueError:
        return project_not_found_error(project_id)

    try:
        lang = parameters.get("language") or project.vocab_lang
    except AnnifException as err:
        return server_error(err)

    if lang not in project.vocab.languages:
        return language_not_supported_error(lang)

    limit = parameters.get("limit", 10)
    threshold = parameters.get("threshold", 0.0)

    try:
        hit_sets = project.suggest_corpus(corpus).filter(limit, threshold)
    except AnnifException as err:
        return server_error(err)

    return _hit_sets_to_list(hit_sets, project.subjects, lang)


def _documents_to_corpus(
    documents: list[dict[str, Any]],
    subject_index: SubjectIndex | None,
) -> annif.corpus.document.DocumentList:
    if subject_index is not None:
        corpus = [
            Document(
                text=d["text"],
                subject_set=SubjectSet(
                    [subject_index.by_uri(subj["uri"]) for subj in d["subjects"]]
                ),
            )
            for d in documents
            if "text" in d and "subjects" in d
        ]
    else:
        corpus = [
            Document(text=d["text"], subject_set=None) for d in documents if "text" in d
        ]
    return DocumentList(corpus)


def learn(
    project_id: str,
    body: list[dict[str, Any]],
) -> ConnexionResponse | tuple[None, int]:
    """learn from documents and return an empty 204 response if succesful"""

    try:
        project = annif.registry.get_project(project_id, min_access=Access.hidden)
    except ValueError:
        return project_not_found_error(project_id)

    try:
        corpus = _documents_to_corpus(body, project.subjects)
        project.learn(corpus)
    except AnnifException as err:
        return server_error(err)

    return None, 204, {"Content-Type": "application/json"}


1			"""Definitions for REST API operations. These are wired via Connexion to
2			methods defined in the OpenAPI specification."""
3
4			from __future__ import annotations
5
6			import importlib
7			from typing import TYPE_CHECKING, Any
8
9			import connexion
10
11			import annif.registry
12			import annif.simplemma_util
13			from annif.corpus import Document, DocumentList, SubjectSet
14			from annif.exception import AnnifException
15			from annif.project import Access
16
17			if TYPE_CHECKING:
18			from connexion.lifecycle import ConnexionResponse
19
20			from annif.corpus.subject import SubjectIndex
21			from annif.suggestion import SubjectSuggestion, SuggestionResults
22
23
24			def project_not_found_error(project_id: str) -> ConnexionResponse:
25			"""return a Connexion error object when a project is not found"""
26
27			return connexion.problem(
28			status=404,
29			title="Project not found",
30			detail="Project '{}' not found".format(project_id),
31			)
32
33
34			def server_error(
35			err: AnnifException,
36			) -> ConnexionResponse:
37			"""return a Connexion error object when there is a server error (project
38			or backend problem)"""
39
40			return connexion.problem(
41			status=503, title="Service unavailable", detail=err.format_message()
42			)
43
44
45			def show_info() -> tuple:
46			"""return version of annif and a title for the api according to OpenAPI spec"""
47
48			result = {"title": "Annif REST API", "version": importlib.metadata.version("annif")}
49			return result, 200, {"Content-Type": "application/json"}
50
51
52			def language_not_supported_error(lang: str) -> ConnexionResponse:
53			"""return a Connexion error object when attempting to use unsupported language"""
54
55			return connexion.problem(
56			status=400,
57			title="Bad Request",
58			detail=f'language "{lang}" not supported by vocabulary',
59			)
60
61
62			def list_vocabs() -> tuple:
63			"""return a dict with vocabularies formatted according to OpenAPI spec"""
64
65			result = {
66			"vocabs": [
67			vocab.dump()
68			for vocab in annif.registry.get_vocabs(min_access=Access.public).values()
69			]
70			}
71			return result, 200, {"Content-Type": "application/json"}
72
73
74			def list_projects() -> tuple:
75			"""return a dict with projects formatted according to OpenAPI spec"""
76
77			result = {
78			"projects": [
79			proj.dump()
80			for proj in annif.registry.get_projects(min_access=Access.public).values()
81			]
82			}
83			return result, 200, {"Content-Type": "application/json"}
84
85
86			def show_project(
87			project_id: str,
88			) -> dict \| ConnexionResponse:
89			"""return a single project formatted according to OpenAPI spec"""
90
91			try:
92			project = annif.registry.get_project(project_id, min_access=Access.hidden)
93			except ValueError:
94			return project_not_found_error(project_id)
95			return project.dump(), 200, {"Content-Type": "application/json"}
96
97
98			def detect_language(body: dict[str, Any]):
99			"""return scores for detected languages formatted according to Swagger spec"""
100
101			text = body.get("text")
102			languages = body.get("languages")
103
104			try:
105			proportions = annif.simplemma_util.detect_language(text, tuple(languages))
106			except ValueError:
107			return connexion.problem(
108			status=400,
109			title="Bad Request",
110			detail="unsupported candidate languages",
111			)
112
113			result = {
114			"results": [
115			{"language": lang if lang != "unk" else None, "score": score}
116			for lang, score in proportions.items()
117			]
118			}
119			return result, 200, {"Content-Type": "application/json"}
120
121
122			def _suggestion_to_dict(
123			suggestion: SubjectSuggestion, subject_index: SubjectIndex, language: str
124			) -> dict[str, str \| float \| None]:
125			subject = subject_index[suggestion.subject_id]
126			return {
127			"uri": subject.uri,
128			"label": subject.labels[language],
129			"notation": subject.notation,
130			"score": suggestion.score,
131			}
132
133
134			def _hit_sets_to_list(
135			hit_sets: SuggestionResults, subjects: SubjectIndex, lang: str
136			) -> list[dict[str, list]]:
137			return [
138			{"results": [_suggestion_to_dict(hit, subjects, lang) for hit in hits]}
139			for hits in hit_sets
140			]
141
142
143			def _is_error(result: list[dict[str, list]] \| ConnexionResponse) -> bool:
144			return (
145			isinstance(result, connexion.lifecycle.ConnexionResponse)
146			and result.status_code >= 400
147			)
148
149
150			def suggest(
151			project_id: str, body: dict[str, Any]
152			) -> dict[str, list] \| ConnexionResponse:
153			"""suggest subjects for the given text and return a dict with results
154			formatted according to OpenAPI spec"""
155
156			parameters = dict(
157			(key, body[key]) for key in ["language", "limit", "threshold"] if key in body
158			)
159			documents = [{"text": body["text"]}]
160			result = _suggest(project_id, documents, parameters)
161
162			if _is_error(result):
163			return result
164			return result[0], 200, {"Content-Type": "application/json"}
165
166
167			def suggest_batch(
168			project_id: str,
169			body: dict[str, list],
170			**query_parameters,
171			) -> list[dict[str, Any]] \| ConnexionResponse:
172			"""suggest subjects for the given documents and return a list of dicts with results
173			formatted according to OpenAPI spec"""
174
175			documents = body["documents"]
176			result = _suggest(project_id, documents, query_parameters)
177
178			if _is_error(result):
179			return result
180			for document_results, document in zip(result, documents):
181			document_results["document_id"] = document.get("document_id")
182			return result, 200, {"Content-Type": "application/json"}
183
184
185			def _suggest(
186			project_id: str,
187			documents: list[dict[str, str]],
188			parameters: dict[str, Any],
189			) -> list[dict[str, list]] \| ConnexionResponse:
190			corpus = _documents_to_corpus(documents, subject_index=None)
191			try:
192			project = annif.registry.get_project(project_id, min_access=Access.hidden)
193			except ValueError:
194			return project_not_found_error(project_id)
195
196			try:
197			lang = parameters.get("language") or project.vocab_lang
198			except AnnifException as err:
199			return server_error(err)
200
201			if lang not in project.vocab.languages:
202			return language_not_supported_error(lang)
203
204			limit = parameters.get("limit", 10)
205			threshold = parameters.get("threshold", 0.0)
206
207			try:
208			hit_sets = project.suggest_corpus(corpus).filter(limit, threshold)
209			except AnnifException as err:
210			return server_error(err)
211
212			return _hit_sets_to_list(hit_sets, project.subjects, lang)
213
214
215			def _documents_to_corpus(
216			documents: list[dict[str, Any]],
217			subject_index: SubjectIndex \| None,
218			) -> annif.corpus.document.DocumentList:
219			if subject_index is not None:
220			corpus = [
221			Document(
222			text=d["text"],
223			subject_set=SubjectSet(
224			[subject_index.by_uri(subj["uri"]) for subj in d["subjects"]]
225			),
226			)
227			for d in documents
228			if "text" in d and "subjects" in d
229			]
230			else:
231			corpus = [
232			Document(text=d["text"], subject_set=None) for d in documents if "text" in d
233			]
234			return DocumentList(corpus)
235
236
237			def learn(
238			project_id: str,
239			body: list[dict[str, Any]],
240			) -> ConnexionResponse \| tuple[None, int]:
241			"""learn from documents and return an empty 204 response if succesful"""
242
243			try:
244			project = annif.registry.get_project(project_id, min_access=Access.hidden)
245			except ValueError:
246			return project_not_found_error(project_id)
247
248			try:
249			corpus = _documents_to_corpus(body, project.subjects)
250			project.learn(corpus)
251			except AnnifException as err:
252			return server_error(err)
253
254			return None, 204, {"Content-Type": "application/json"}
255

NatLibFi / Annif

Pull Request — main (#839)

annif.rest.list_vocabs() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like