annif.rest.project_not_found_error() - Code Metrics - Inspection of "Add CI/CD jobs for testing on Windows and MacOS" - NatLibFi/Annif - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — testing-on-windows-and-macos ( 782857...ea99ad )

by Juho

created 2025-02-11 13:59 UTC

annif.rest.project_not_found_error() A

↳ Parent: annif.rest

Complexity

Conditions

Size

Total Lines	7
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	5
dl	0
loc	7
rs	10
c	0
b	0
f	0
cc	1
nop	1

"""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_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_projects() -> tuple:
63			"""return a dict with projects formatted according to OpenAPI spec"""
64
65			result = {
66			"projects": [
67			proj.dump()
68			for proj in annif.registry.get_projects(min_access=Access.public).values()
69			]
70			}
71			return result, 200, {"Content-Type": "application/json"}
72
73
74			def show_project(
75			project_id: str,
76			) -> dict \| ConnexionResponse:
77			"""return a single project formatted according to OpenAPI spec"""
78
79			try:
80			project = annif.registry.get_project(project_id, min_access=Access.hidden)
81			except ValueError:
82			return project_not_found_error(project_id)
83			return project.dump(), 200, {"Content-Type": "application/json"}
84
85
86			def detect_language(body: dict[str, Any]):
87			"""return scores for detected languages formatted according to Swagger spec"""
88
89			text = body.get("text")
90			languages = body.get("languages")
91
92			try:
93			proportions = annif.simplemma_util.detect_language(text, tuple(languages))
94			except ValueError:
95			return connexion.problem(
96			status=400,
97			title="Bad Request",
98			detail="unsupported candidate languages",
99			)
100
101			result = {
102			"results": [
103			{"language": lang if lang != "unk" else None, "score": score}
104			for lang, score in proportions.items()
105			]
106			}
107			return result, 200, {"Content-Type": "application/json"}
108
109
110			def _suggestion_to_dict(
111			suggestion: SubjectSuggestion, subject_index: SubjectIndex, language: str
112			) -> dict[str, str \| float \| None]:
113			subject = subject_index[suggestion.subject_id]
114			return {
115			"uri": subject.uri,
116			"label": subject.labels[language],
117			"notation": subject.notation,
118			"score": suggestion.score,
119			}
120
121
122			def _hit_sets_to_list(
123			hit_sets: SuggestionResults, subjects: SubjectIndex, lang: str
124			) -> list[dict[str, list]]:
125			return [
126			{"results": [_suggestion_to_dict(hit, subjects, lang) for hit in hits]}
127			for hits in hit_sets
128			]
129
130
131			def _is_error(result: list[dict[str, list]] \| ConnexionResponse) -> bool:
132			return (
133			isinstance(result, connexion.lifecycle.ConnexionResponse)
134			and result.status_code >= 400
135			)
136
137
138			def suggest(
139			project_id: str, body: dict[str, Any]
140			) -> dict[str, list] \| ConnexionResponse:
141			"""suggest subjects for the given text and return a dict with results
142			formatted according to OpenAPI spec"""
143
144			parameters = dict(
145			(key, body[key]) for key in ["language", "limit", "threshold"] if key in body
146			)
147			documents = [{"text": body["text"]}]
148			result = _suggest(project_id, documents, parameters)
149
150			if _is_error(result):
151			return result
152			return result[0], 200, {"Content-Type": "application/json"}
153
154
155			def suggest_batch(
156			project_id: str,
157			body: dict[str, list],
158			**query_parameters,
159			) -> list[dict[str, Any]] \| ConnexionResponse:
160			"""suggest subjects for the given documents and return a list of dicts with results
161			formatted according to OpenAPI spec"""
162
163			documents = body["documents"]
164			result = _suggest(project_id, documents, query_parameters)
165
166			if _is_error(result):
167			return result
168			for document_results, document in zip(result, documents):
169			document_results["document_id"] = document.get("document_id")
170			return result, 200, {"Content-Type": "application/json"}
171
172
173			def _suggest(
174			project_id: str,
175			documents: list[dict[str, str]],
176			parameters: dict[str, Any],
177			) -> list[dict[str, list]] \| ConnexionResponse:
178			corpus = _documents_to_corpus(documents, subject_index=None)
179			try:
180			project = annif.registry.get_project(project_id, min_access=Access.hidden)
181			except ValueError:
182			return project_not_found_error(project_id)
183
184			try:
185			lang = parameters.get("language") or project.vocab_lang
186			except AnnifException as err:
187			return server_error(err)
188
189			if lang not in project.vocab.languages:
190			return language_not_supported_error(lang)
191
192			limit = parameters.get("limit", 10)
193			threshold = parameters.get("threshold", 0.0)
194
195			try:
196			hit_sets = project.suggest_corpus(corpus).filter(limit, threshold)
197			except AnnifException as err:
198			return server_error(err)
199
200			return _hit_sets_to_list(hit_sets, project.subjects, lang)
201
202
203			def _documents_to_corpus(
204			documents: list[dict[str, Any]],
205			subject_index: SubjectIndex \| None,
206			) -> annif.corpus.document.DocumentList:
207			if subject_index is not None:
208			corpus = [
209			Document(
210			text=d["text"],
211			subject_set=SubjectSet(
212			[subject_index.by_uri(subj["uri"]) for subj in d["subjects"]]
213			),
214			)
215			for d in documents
216			if "text" in d and "subjects" in d
217			]
218			else:
219			corpus = [
220			Document(text=d["text"], subject_set=None) for d in documents if "text" in d
221			]
222			return DocumentList(corpus)
223
224
225			def learn(
226			project_id: str,
227			body: list[dict[str, Any]],
228			) -> ConnexionResponse \| tuple[None, int]:
229			"""learn from documents and return an empty 204 response if succesful"""
230
231			try:
232			project = annif.registry.get_project(project_id, min_access=Access.hidden)
233			except ValueError:
234			return project_not_found_error(project_id)
235
236			try:
237			corpus = _documents_to_corpus(body, project.subjects)
238			project.learn(corpus)
239			except AnnifException as err:
240			return server_error(err)
241
242			return None, 204, {"Content-Type": "application/json"}
243

NatLibFi / Annif

Push — testing-on-windows-and-macos ( 782857...ea99ad )

annif.rest.project_not_found_error() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like