[stat_visualization mask_user_info] タスクlist.csvのマスクに失敗する不具合の修正 (#809)

* version up * update * "number_of_rejections"を削除 * 習性 * update doc * update
kurusugawa-computer · Jun 30, 2022 · a35615c · a35615c
1 parent 974f3f6
commit a35615c
Show file tree

Hide file tree

Showing 8 changed files with 42 additions and 14 deletions.
diff --git a/annofabcli/__version__.py b/annofabcli/__version__.py
@@ -1 +1 @@
-__version__ = "1.66.1"
+__version__ = "1.66.2"
diff --git a/annofabcli/filesystem/mask_user_info.py b/annofabcli/filesystem/mask_user_info.py
@@ -136,7 +136,18 @@ def _get_tuple_column(df: pandas.DataFrame, column: str) -> Union[str, Tuple]:
         return column
 
 
-def replace_by_columns(df, replacement_dict: Dict[str, str], main_column: Any, sub_columns: Optional[List[Any]] = None):
+def replace_by_columns(
+    df: pandas.DataFrame, replacement_dict: Dict[str, str], main_column: Any, sub_columns: Optional[List[Any]] = None
+):
+    """引数dfの中のユーザ情報を、指定した列名を元に置換します。
+
+    Args:
+        df (pandas.DataFrame): _description_
+        replacement_dict (Dict[str, str]): 置換対象のuser_idと置換後のuser_id(username)。key: 置換対象のuser_id, value: 置換後のuser_id
+        main_column: 置換対象の列名(ex: user_id)
+        sub_column: main_columnと同じ値で置換する列(ex: username)
+    """
+
     def _get_username(row, main_column: Any, sub_column: Any) -> str:
         if row[main_column] in replacement_dict:
             return replacement_dict[row[main_column]]

diff --git a/annofabcli/stat_visualization/mask_visualization_dir.py b/annofabcli/stat_visualization/mask_visualization_dir.py
@@ -32,7 +32,6 @@
 
 def _replace_df_task(task: Task, replacement_dict_by_user_id: Dict[str, str]) -> Task:
     df_output = task.df.copy()
-    replace_by_columns(df_output, replacement_dict_by_user_id, main_column="user_id", sub_columns=["username"])
 
     replace_by_columns(
         df_output,

diff --git a/annofabcli/statistics/table.py b/annofabcli/statistics/table.py
@@ -144,11 +144,6 @@ def set_task_histories(cls, task: Task, task_histories: List[TaskHistory]):
             for e in task_histories
         )
 
-        # APIで取得した 'number_of_rejections' は非推奨で、number_of_rejections_by_inspection/acceptanceと矛盾する場合があるので、書き換える
-        task["number_of_rejections"] = (
-            task["number_of_rejections_by_inspection"] + task["number_of_rejections_by_acceptance"]
-        )
-
         return task
 
     @staticmethod

diff --git a/annofabcli/statistics/visualization/dataframe/task.py b/annofabcli/statistics/visualization/dataframe/task.py
@@ -242,10 +242,8 @@ def to_csv(self, output_file: Path) -> None:
             "phase",
             "phase_stage",
             "status",
-            "number_of_rejections",
             "number_of_rejections_by_inspection",
             "number_of_rejections_by_acceptance",
-            "first_acceptance_completed_datetime",
             # 1回目の教師付フェーズ
             "first_annotation_user_id",
             "first_annotation_username",
@@ -261,6 +259,8 @@ def to_csv(self, output_file: Path) -> None:
             "first_acceptance_username",
             "first_acceptance_worktime_hour",
             "first_acceptance_started_datetime",
+            # 最後の受入
+            "first_acceptance_completed_datetime",
             # 作業時間に関する内容
             "worktime_hour",
             "annotation_worktime_hour",

diff --git a/docs/command_reference/statistics/visualize/out_dir/タスクlist.csv b/docs/command_reference/statistics/visualize/out_dir/タスクlist.csv
@@ -1,3 +1,3 @@
-project_id,task_id,phase,phase_stage,status,user_id,username,number_of_rejections,number_of_rejections_by_inspection,number_of_rejections_by_acceptance,started_datetime,updated_datetime,first_acceptance_completed_datetime,sampling,first_annotation_user_id,first_annotation_username,first_annotation_worktime_hour,first_annotation_started_datetime,first_inspection_user_id,first_inspection_username,first_inspection_worktime_hour,first_inspection_started_datetime,first_acceptance_user_id,first_acceptance_username,first_acceptance_worktime_hour,first_acceptance_started_datetime,worktime_hour,annotation_worktime_hour,inspection_worktime_hour,acceptance_worktime_hour,first_annotator_worktime_hour,first_inspector_worktime_hour,first_acceptor_worktime_hour,input_data_count,annotation_count,inspection_comment_count,input_data_count_of_inspection,annotator_is_changed,inspector_is_changed,acceptor_is_changed,inspection_is_skipped,acceptance_is_skipped,diff_days_to_first_acceptance_started,diff_days_to_first_inspection_started,diff_days_to_first_acceptance_completed,metadata
-prj1,task1,acceptance,1,complete,NZ,NZ,0,0,0,2020-09-01T11:08:47.598+09:00,2020-09-01T11:09:40.359+09:00,2020-09-01T11:09:40.351+09:00,,QU,QU,0.091450555555556,2020-09-01T09:44:35.431+09:00,,,0,,NZ,NZ,0.014653611111111,2020-09-01T11:08:47.598+09:00,0.106104166666667,0.091450555555556,0,0.014653611111111,0.091450555555556,0,0.014653611111111,20,194,0,0,False,False,False,False,False,0.058474155092593,,0.059084722222222,{}
-prj1,task2,acceptance,1,complete,NZ,NZ,0,0,0,2020-09-02T07:22:13.693+09:00,2020-09-02T07:23:38.295+09:00,2020-09-02T07:23:38.274+09:00,,KX,KX,0.533797777777778,2020-08-31T16:18:34.369+09:00,,,0,,NZ,NZ,0.023494722222222,2020-09-02T07:22:13.693+09:00,0.5572925,0.533797777777778,0,0.023494722222222,0.533797777777778,0,0.023494722222222,20,648,0,0,False,False,False,False,False,1.62753847222222,,1.62851741898148,{}
+project_id,task_id,phase,phase_stage,status,number_of_rejections_by_inspection,number_of_rejections_by_acceptance,first_annotation_user_id,first_annotation_username,first_annotation_worktime_hour,first_annotation_started_datetime,first_inspection_user_id,first_inspection_username,first_inspection_worktime_hour,first_inspection_started_datetime,first_acceptance_user_id,first_acceptance_username,first_acceptance_worktime_hour,first_acceptance_started_datetime,first_acceptance_completed_datetime,worktime_hour,annotation_worktime_hour,inspection_worktime_hour,acceptance_worktime_hour,input_data_count,annotation_count,inspection_comment_count,inspection_is_skipped,acceptance_is_skipped
+prj1,task1,acceptance,1,complete,0,0,TP,TP,0.091450555555556,2020-09-01T09:44:35.431+09:00,,,0,,QF,QF,0.014653611111111,2020-09-01T11:08:47.598+09:00,2020-09-01T11:09:40.351+09:00,0.106104166666667,0.091450555555556,0,0.014653611111111,20,194,0,False,False
+prj1,task2,acceptance,1,complete,0,0,MO,MO,0.533797777777778,2020-08-31T16:18:34.369+09:00,,,0,,QF,QF,0.023494722222222,2020-09-02T07:22:13.693+09:00,2020-09-02T07:23:38.274+09:00,0.5572925,0.533797777777778,0,0.023494722222222,20,648,0,False,False
diff --git a/docs/command_reference/statistics/visualize_output_rst/タスクlist_csv.rst b/docs/command_reference/statistics/visualize_output_rst/タスクlist_csv.rst
@@ -9,4 +9,27 @@
 * 教師付/検査/受入の作業時間
 * 最初に教師付/検査/受入したユーザ
 
+以下に、列名から内容が判断しづらい列の詳細を記載します。
+
+* number_of_rejections_by_inspection : 検査フェーズで差し戻された回数
+* number_of_rejections_by_acceptance : 受入フェーズで差し戻された回数
+
+* first_annotation_user_id : 最初に教師付フェーズを開始したユーザーのuser_id
+* first_annotation_username : 最初に教師付フェーズを開始したユーザーのusername
+* first_annotation_worktime_hour : 最初に教師付フェーズの作業時間[hour]
+* first_annotation_started_datetime : 最初に教師付フェーズを開始した日時
+
+* worktime_hour: 作業時間[hour]
+* annotation_worktime_hour: 教師付作業時間[hour]
+* inspection_worktime_hour: 検査作業時間[hour]
+* acceptance_worktime_hour: 受入作業時間[hour]
+
+* input_data_count : タスクに含まれる入力データ数
+* input_data_count : タスクに含まれるアノテーション数
+* inspection_comment_count : 指摘された検査コメント数（対応不要のコメントは除く）
+
+* inspection_is_skipped : 抜取検査により検査フェーズがスキップされたか
+* acceptance_is_skipped : 抜取受入により受入フェーズがスキップされたか
+
+
 `タスクlist.csv <https://github.com/kurusugawa-computer/annofab-cli/blob/main/docs/command_reference/statistics/visualize/out_dir/タスクlist.csv>`_
diff --git a/pyproject.toml b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "annofabcli"
-version = "1.66.1"
+version = "1.66.2"
 description = "Utility Command Line Interface for AnnoFab"
 authors = ["yuji38kwmt"]
 license = "MIT"