annofabcli 1.101.0__tar.gz → 1.102.0__tar.gz

{annofabcli-1.101.0 → annofabcli-1.102.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: annofabcli
-Version: 1.101.0
+Version: 1.102.0
 Summary: Utility Command Line Interface for AnnoFab
 Author: Kurusugawa Computer Inc.
 License: MIT

annofabcli-1.102.0/annofabcli/statistics/list_annotation_area.py

@@ -0,0 +1,320 @@
+from __future__ import annotations
+
+import argparse
+import logging
+import sys
+import tempfile
+import zipfile
+from collections.abc import Collection, Iterator
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Any, Optional
+
+import numpy
+import pandas
+from annofabapi.models import InputDataType, ProjectMemberRole
+from annofabapi.parser import (
+    SimpleAnnotationParser,
+    lazy_parse_simple_annotation_dir,
+    lazy_parse_simple_annotation_zip,
+)
+from annofabapi.segmentation import read_binary_image
+from dataclasses_json import DataClassJsonMixin
+
+import annofabcli
+import annofabcli.common.cli
+from annofabcli.common.cli import (
+    COMMAND_LINE_ERROR_STATUS_CODE,
+    ArgumentParser,
+    CommandLine,
+    build_annofabapi_resource_and_login,
+)
+from annofabcli.common.download import DownloadingFile
+from annofabcli.common.enums import FormatArgument
+from annofabcli.common.facade import (
+    AnnofabApiFacade,
+    TaskQuery,
+    match_annotation_with_task_query,
+)
+from annofabcli.common.utils import print_csv, print_json
+
+logger = logging.getLogger(__name__)
+
+
+def lazy_parse_simple_annotation_by_input_data(annotation_path: Path) -> Iterator[SimpleAnnotationParser]:
+    if not annotation_path.exists():
+        raise RuntimeError(f"'{annotation_path}' は存在しません。")
+
+    if annotation_path.is_dir():
+        return lazy_parse_simple_annotation_dir(annotation_path)
+    elif zipfile.is_zipfile(str(annotation_path)):
+        return lazy_parse_simple_annotation_zip(annotation_path)
+    else:
+        raise RuntimeError(f"'{annotation_path}'は、zipファイルまたはディレクトリではありません。")
+
+
+@dataclass(frozen=True)
+class AnnotationAreaInfo(DataClassJsonMixin):
+    task_id: str
+    task_status: str
+    task_phase: str
+    task_phase_stage: int
+
+    input_data_id: str
+    input_data_name: str
+
+    label: str
+    annotation_id: str
+    annotation_area: int
+
+
+def calculate_bounding_box_area(data: dict[str, Any]) -> int:
+    width = abs(data["right_bottom"]["x"] - data["left_top"]["x"])
+    height = abs(data["right_bottom"]["y"] - data["left_top"]["y"])
+    return width * height
+
+
+def calculate_segmentation_area(outer_file: Path) -> int:
+    nd_array = read_binary_image(outer_file)
+    return numpy.count_nonzero(nd_array)
+
+
+def get_annotation_area_info_list(parser: SimpleAnnotationParser, simple_annotation: dict[str, Any]) -> list[AnnotationAreaInfo]:
+    result = []
+    for detail in simple_annotation["details"]:
+        if detail["data"]["_type"] in {"Segmentation", "SegmentationV2"}:
+            annotation_area = calculate_segmentation_area(parser.open_outer_file(detail["data"]["data_uri"]))
+        elif detail["data"]["_type"] == "BoundingBox":
+            annotation_area = calculate_bounding_box_area(detail["data"])
+
+        else:
+            continue
+
+        result.append(
+            AnnotationAreaInfo(
+                task_id=simple_annotation["task_id"],
+                task_phase=simple_annotation["task_phase"],
+                task_phase_stage=simple_annotation["task_phase_stage"],
+                task_status=simple_annotation["task_status"],
+                input_data_id=simple_annotation["input_data_id"],
+                input_data_name=simple_annotation["input_data_name"],
+                label=detail["label"],
+                annotation_id=detail["annotation_id"],
+                annotation_area=annotation_area,
+            )
+        )
+
+    return result
+
+
+def get_annotation_area_info_list_from_annotation_path(
+    annotation_path: Path,
+    *,
+    target_task_ids: Optional[Collection[str]] = None,
+    task_query: Optional[TaskQuery] = None,
+) -> list[AnnotationAreaInfo]:
+    annotation_area_list = []
+    target_task_ids = set(target_task_ids) if target_task_ids is not None else None
+    iter_parser = lazy_parse_simple_annotation_by_input_data(annotation_path)
+    logger.debug("アノテーションzip/ディレクトリを読み込み中")
+    for index, parser in enumerate(iter_parser):
+        if (index + 1) % 1000 == 0:
+            logger.debug(f"{index + 1}  件目のJSONを読み込み中")
+        if target_task_ids is not None and parser.task_id not in target_task_ids:
+            continue
+        simple_annotation_dict = parser.load_json()
+        if task_query is not None:  # noqa: SIM102
+            if not match_annotation_with_task_query(simple_annotation_dict, task_query):
+                continue
+        sub_annotation_area_list = get_annotation_area_info_list(parser, simple_annotation_dict)
+        annotation_area_list.extend(sub_annotation_area_list)
+    return annotation_area_list
+
+
+def create_df(
+    annotation_area_list: list[AnnotationAreaInfo],
+) -> pandas.DataFrame:
+    columns = [
+        "task_id",
+        "task_status",
+        "task_phase",
+        "task_phase_stage",
+        "input_data_id",
+        "input_data_name",
+        "label",
+        "annotation_id",
+        "annotation_area",
+    ]
+    df = pandas.DataFrame(e.to_dict() for e in annotation_area_list)
+    if len(df) == 0:
+        df = pandas.DataFrame(columns=columns)
+
+    df = df.fillna({"annotation_area": 0})
+    return df[columns]
+
+
+def print_annotation_area(
+    annotation_path: Path,
+    output_file: Path,
+    output_format: FormatArgument,
+    *,
+    target_task_ids: Optional[Collection[str]] = None,
+    task_query: Optional[TaskQuery] = None,
+) -> None:
+    annotation_area_list = get_annotation_area_info_list_from_annotation_path(
+        annotation_path,
+        target_task_ids=target_task_ids,
+        task_query=task_query,
+    )
+
+    logger.info(f"{len(annotation_area_list)} 件のタスクに含まれる塗りつぶしアノテーションのピクセル数情報を出力します。")
+
+    if output_format == FormatArgument.CSV:
+        df = create_df(annotation_area_list)
+        print_csv(df, output_file)
+
+    elif output_format in [FormatArgument.PRETTY_JSON, FormatArgument.JSON]:
+        json_is_pretty = output_format == FormatArgument.PRETTY_JSON
+        print_json(
+            [e.to_dict(encode_json=True) for e in annotation_area_list],
+            is_pretty=json_is_pretty,
+            output=output_file,
+        )
+
+    else:
+        raise RuntimeError(f"出力形式 '{output_format}' はサポートされていません。")
+
+
+class ListAnnotationArea(CommandLine):
+    COMMON_MESSAGE = "annofabcli statistics list_annotation_area: error:"
+
+    def validate(self, args: argparse.Namespace) -> bool:
+        if args.project_id is None and args.annotation is None:
+            print(  # noqa: T201
+                f"{self.COMMON_MESSAGE} argument --project_id: '--annotation'が未指定のときは、'--project_id' を指定してください。",
+                file=sys.stderr,
+            )
+            return False
+        return True
+
+    def main(self) -> None:
+        args = self.args
+
+        if not self.validate(args):
+            sys.exit(COMMAND_LINE_ERROR_STATUS_CODE)
+
+        project_id: Optional[str] = args.project_id
+        if project_id is not None:
+            super().validate_project(project_id, project_member_roles=[ProjectMemberRole.OWNER, ProjectMemberRole.TRAINING_DATA_USER])
+            project, _ = self.service.api.get_project(project_id)
+            if project["input_data_type"] != InputDataType.IMAGE.value:
+                logger.warning(f"project_id='{project_id}'であるプロジェクトは、画像プロジェクトでないので、出力されるデータは0件になります。")
+
+        annotation_path = Path(args.annotation) if args.annotation is not None else None
+
+        task_id_list = annofabcli.common.cli.get_list_from_args(args.task_id) if args.task_id is not None else None
+        task_query = TaskQuery.from_dict(annofabcli.common.cli.get_json_from_args(args.task_query)) if args.task_query is not None else None
+
+        output_file: Path = args.output
+        output_format = FormatArgument(args.format)
+
+        downloading_obj = DownloadingFile(self.service)
+
+        def download_and_print_annotation_area(project_id: str, temp_dir: Path, *, is_latest: bool, annotation_path: Optional[Path]) -> None:
+            if annotation_path is None:
+                annotation_path = temp_dir / f"{project_id}__annotation.zip"
+                downloading_obj.download_annotation_zip(
+                    project_id,
+                    dest_path=annotation_path,
+                    is_latest=is_latest,
+                )
+            print_annotation_area(
+                output_format=output_format,
+                output_file=output_file,
+                target_task_ids=task_id_list,
+                task_query=task_query,
+                annotation_path=annotation_path,
+            )
+
+        if project_id is not None:
+            if args.temp_dir is not None:
+                download_and_print_annotation_area(
+                    project_id=project_id, temp_dir=args.temp_dir, is_latest=args.latest, annotation_path=annotation_path
+                )
+            else:
+                with tempfile.TemporaryDirectory() as str_temp_dir:
+                    download_and_print_annotation_area(
+                        project_id=project_id, temp_dir=Path(str_temp_dir), is_latest=args.latest, annotation_path=annotation_path
+                    )
+        else:
+            assert annotation_path is not None
+            print_annotation_area(
+                output_format=output_format,
+                output_file=output_file,
+                target_task_ids=task_id_list,
+                task_query=task_query,
+                annotation_path=annotation_path,
+            )
+
+
+def parse_args(parser: argparse.ArgumentParser) -> None:
+    argument_parser = ArgumentParser(parser)
+
+    parser.add_argument(
+        "--annotation",
+        type=str,
+        help="アノテーションzip、またはzipを展開したディレクトリを指定します。指定しない場合はAnnofabからダウンロードします。",
+    )
+
+    parser.add_argument(
+        "-p",
+        "--project_id",
+        type=str,
+        help="project_id。``--annotation`` が未指定のときは必須です。\n"
+        "``--annotation`` が指定されているときに ``--project_id`` を指定すると、アノテーション仕様を参照して、集計対象の属性やCSV列順が決まります。",
+    )
+
+    argument_parser.add_format(
+        choices=[FormatArgument.CSV, FormatArgument.JSON, FormatArgument.PRETTY_JSON],
+        default=FormatArgument.CSV,
+    )
+
+    argument_parser.add_output()
+
+    parser.add_argument(
+        "-tq",
+        "--task_query",
+        type=str,
+        help="集計対象タスクを絞り込むためのクエリ条件をJSON形式で指定します。使用できるキーは task_id, status, phase, phase_stage です。"
+        " ``file://`` を先頭に付けると、JSON形式のファイルを指定できます。",
+    )
+    argument_parser.add_task_id(required=False)
+
+    parser.add_argument(
+        "--latest",
+        action="store_true",
+        help="``--annotation`` を指定しないとき、最新のアノテーションzipを参照します。このオプションを指定すると、アノテーションzipを更新するのに数分待ちます。",  # noqa: E501
+    )
+
+    parser.add_argument(
+        "--temp_dir",
+        type=Path,
+        help="指定したディレクトリに、アノテーションZIPなどの一時ファイルをダウンロードします。",
+    )
+
+    parser.set_defaults(subcommand_func=main)
+
+
+def main(args: argparse.Namespace) -> None:
+    service = build_annofabapi_resource_and_login(args)
+    facade = AnnofabApiFacade(service)
+    ListAnnotationArea(service, facade, args).main()
+
+
+def add_parser(subparsers: Optional[argparse._SubParsersAction] = None) -> argparse.ArgumentParser:
+    subcommand_name = "list_annotation_area"
+    subcommand_help = "塗りつぶしアノテーションまたは矩形アノテーションの面積を出力します。"
+    epilog = "オーナロールまたはアノテーションユーザロールを持つユーザで実行してください。"
+    parser = annofabcli.common.cli.add_parser(subparsers, subcommand_name, subcommand_help, description=subcommand_help, epilog=epilog)
+    parse_args(parser)
+    return parser

{annofabcli-1.101.0 → annofabcli-1.102.0}/annofabcli/statistics/subcommand_statistics.py

@@ -4,6 +4,7 @@ from typing import Optional
 import annofabcli
 import annofabcli.common.cli
 import annofabcli.stat_visualization.merge_visualization_dir
+import annofabcli.statistics.list_annotation_area
 import annofabcli.statistics.list_annotation_attribute
 import annofabcli.statistics.list_annotation_attribute_filled_count
 import annofabcli.statistics.list_annotation_count
@@ -27,6 +28,7 @@ def parse_args(parser: argparse.ArgumentParser) -> None:
     annofabcli.statistics.list_annotation_attribute_filled_count.add_parser(subparsers)
     annofabcli.statistics.list_annotation_count.add_parser(subparsers)
     annofabcli.statistics.list_annotation_duration.add_parser(subparsers)
+    annofabcli.statistics.list_annotation_area.add_parser(subparsers)
 
     annofabcli.statistics.list_video_duration.add_parser(subparsers)
     annofabcli.statistics.list_worktime.add_parser(subparsers)

{annofabcli-1.101.0 → annofabcli-1.102.0}/docs/command_reference/statistics/index.rst

@@ -20,6 +20,7 @@ Available Commands
    list_annotation_attribute_filled_count
    list_annotation_count
    list_annotation_duration
+   list_annotation_area
    list_video_duration
    list_worktime
    summarize_task_count

annofabcli-1.102.0/docs/command_reference/statistics/list_annotation_area.rst

@@ -0,0 +1,46 @@
+==========================================
+statistics list_annotation_area
+==========================================
+
+Description
+=================================
+
+アノテーションzipまたはディレクトリに含まれる、塗りつぶしアノテーションまたは矩形アノテーションの面積を、アノテーションごとに出力します。
+
+Examples
+=================================
+
+
+.. code-block::
+
+    $ annofabcli statistics list_annotation_area --project_id prj1 \
+     --output out.json --format pretty_json
+
+.. code-block:: json
+    :caption: out.json
+
+    [
+      {
+        "task_id": "task_00",
+        "task_status": "complete",
+        "task_phase": "annotation",
+        "task_phase_stage": 1,
+        "input_data_id": "i1",
+        "input_data_name": "i1.jpg",
+        "label": "cat",
+        "annotation_id": "ann1",
+        "annotation_area": 1234
+      },
+    ]
+
+
+
+Usage Details
+=================================
+
+.. argparse::
+   :ref: annofabcli.statistics.list_annotation_area.add_parser
+   :prog: annofabcli statistics list_annotation_area
+   :nosubcommands:
+   :nodefaultconst:
+

annofabcli-1.102.0/tests/data/statistics/list_annotation_area/annotation/test_17/6026efe3-7eea-47c2-9dc6-83f452b22900.json

@@ -0,0 +1,99 @@
+{
+    "project_id": "87dde4e1-f434-444c-8cd9-c262bf02999b",
+    "annotation_format_version": "1.2.0",
+    "task_id": "test_17",
+    "task_phase": "annotation",
+    "task_phase_stage": 1,
+    "task_status": "break",
+    "input_data_id": "6026efe3-7eea-47c2-9dc6-83f452b22900",
+    "input_data_name": "val.zip/val/ab1d5abf-00000000.jpg",
+    "details": [
+        {
+            "label": "car",
+            "annotation_id": "7168a1be-684a-48ca-a0ba-167fd49b76ce",
+            "data": {
+                "left_top": {
+                    "x": 667,
+                    "y": 143
+                },
+                "right_bottom": {
+                    "x": 709,
+                    "y": 183
+                },
+                "_type": "BoundingBox"
+            },
+            "attributes": {
+                "number": 0,
+                "20240906-text": "",
+                "direction": "aslant",
+                "comment": "",
+                "link": "",
+                "truncation": false,
+                "tracking_id": "7168a1be-684a-48ca-a0ba-167fd49b76ce",
+                "20240906-check": false
+            }
+        },
+        {
+            "label": "road_segment",
+            "annotation_id": "7604d4ea-4288-4f31-bed9-d5624571e52f",
+            "data": {
+                "data_uri": "7604d4ea-4288-4f31-bed9-d5624571e52f",
+                "_type": "SegmentationV2"
+            },
+            "attributes": {}
+        },
+        {
+            "label": "road_instance",
+            "annotation_id": "2d9884e3-6926-47dc-97d4-d9f55d113625",
+            "data": {
+                "data_uri": "2d9884e3-6926-47dc-97d4-d9f55d113625",
+                "_type": "Segmentation"
+            },
+            "attributes": {
+                "type": "a"
+            }
+        },
+        {
+            "label": "road_instance",
+            "annotation_id": "1b3f0069-9d88-4781-8406-023180822407",
+            "data": {
+                "data_uri": "1b3f0069-9d88-4781-8406-023180822407",
+                "_type": "Segmentation"
+            },
+            "attributes": {
+                "type": "a"
+            }
+        },
+        {
+            "label": "天気",
+            "annotation_id": "46069020-12ab-4698-92d0-818d8cf0cf3d",
+            "data": {
+                "_type": "Classification"
+            },
+            "attributes": {
+                "truncation": false
+            }
+        },
+        {
+            "label": "現場",
+            "annotation_id": "dbcff52c-fb2d-4a76-b5fc-2af8e9e56cc6",
+            "data": {
+                "_type": "Classification"
+            },
+            "attributes": {
+                "type": "a"
+            }
+        },
+        {
+            "label": "foo",
+            "annotation_id": "53530a8f-e48f-46e9-833f-8e48becb25af",
+            "data": {
+                "_type": "Classification"
+            },
+            "attributes": {
+                "truncation": false
+            }
+        }
+    ],
+    "updated_datetime": "2025-05-21T12:37:42.187+09:00"
+}

annofabcli-1.102.0/tests/statistics/test_list_annotation_area.py

@@ -0,0 +1,17 @@
+from pathlib import Path
+
+# ここから list_annotation_area.py のテスト
+from annofabcli.statistics.list_annotation_area import get_annotation_area_info_list_from_annotation_path
+
+output_dir = Path("./tests/out/statistics/list_annotation_count")
+data_dir = Path("./tests/data/statistics/list_annotation_area")
+output_dir.mkdir(exist_ok=True, parents=True)
+
+
+def test__get_annotation_area_info_list_from_annotation_path():
+    # テスト用のアノテーションファイルを作成
+    annotation_area_info_list = get_annotation_area_info_list_from_annotation_path(data_dir / "annotation")
+    assert len(annotation_area_info_list) == 4
+    assert annotation_area_info_list[0].annotation_area == 1680
+    # 1pxも塗られていないアノテーション
+    assert annotation_area_info_list[2].annotation_area == 0