レビューを元に修正

.gitignore

@@ -163,7 +163,7 @@ cython_debug/
 .DS_Store
 
 # 申請書/報告書はリポジトリにはアップロードしない。
-data/inputsData.csv
+data/change_request_history.csv
 data/output.docx
 data/output_deviation.docx
 data/output_deviation_occurrence.docx

README.md

@@ -14,7 +14,7 @@ license: other
 製薬業務のドキュメント自動生成SaaSトライアル版
 
 # 使い方
-## 1. .env.exampleを参考に.envを作成し、OpenAIとPINECONEのAPIキーを設定
+## 1. .env.exampleを参考に.envを作成し、OpenAIとPINECONE、AWS_ACCESS_KEY_ID、AWS_SECRET_ACCESS_KEYのAPIキーを設定
 ```
 OPENAI_API_KEY =
 PINECONE_API_KEY =
@@ -24,6 +24,7 @@ AWS_SECRET_ACCESS_KEY =
 値はnotionで管理されているので、参照する。
 [OpenAI API](https://www.notion.so/equesai/OpenAI-API-b2029c9b5d774eaebd7318a702ec31f1?pvs=4)
 [PINECONE](https://www.notion.so/equesai/API-61268f5a9218495ca96ee0bb176fa13a?pvs=4)
+[AWS_ACCESS_KEY_ID/AWS_SECRET_ACCESS_KEY](https://www.notion.so/equesai/API-61268f5a9218495ca96ee0bb176fa13a?pvs=4)
 
 ## 2. 仮想環境を作成・プレコミットの導入
 `pipenv install -r ./requirements.txt`

main.py

@@ -9,7 +9,7 @@ pages = {
     "変更申請書": change_request_form,
     "逸脱発生報告書": deviation_occurrence_report_form,
     "逸脱報告書": deviation_report_form,
-    "データベース確認": database_check_form,
+    "申請履歴": database_check_form,
 }
 
 selected_page = st.sidebar.selectbox(
@@ -22,5 +22,5 @@ match selected_page:
         pages["逸脱発生報告書"]()
     case "逸脱報告書":
         pages["逸脱報告書"]()
-    case "データベース確認":
-        pages["データベース確認"]()
+    case "申請履歴":
+        pages["申請履歴"]()

src/aws/s3.py

@@ -1,37 +1,23 @@
 from io import StringIO
-from typing import Any, List
 
 import boto3
 import pandas as pd
 
 s3 = boto3.client("s3")
-bucket_name = "bankyo-pharmaceutical"
 
 
 # Download the CSV file from S3
-def download_csv_from_s3(file_name: str) -> pd.DataFrame:
+def download_csv_panda_dataframe_from_s3(
+    bucket_name: str, file_name: str
+) -> pd.DataFrame:
     obj = s3.get_object(Bucket=bucket_name, Key=file_name)
     data = pd.read_csv(obj["Body"])
     return data
 
 
-def save_to_s3(data: pd.DataFrame, file_name: str) -> None:
+def save_csv_panda_dataframe_to_s3(
+    data: pd.DataFrame, bucket_name: str, file_name: str
+) -> None:
     csv_buffer = StringIO()
     data.to_csv(csv_buffer, index=False)
     s3.put_object(Bucket=bucket_name, Key=file_name, Body=csv_buffer.getvalue())
-
-
-def get_all_objects() -> List[str]:
-    response = s3.list_objects_v2(Bucket=bucket_name)
-    csv_files = [
-        obj["Key"]
-        for obj in response.get("Contents", [])
-        if obj["Key"].endswith(".csv")
-    ]
-    return csv_files
-
-
-def get_csv_content(file_name: str) -> Any:
-    obj = s3.get_object(Bucket=bucket_name, Key=file_name)
-    df = pd.read_csv(obj["Body"])
-    return df.iloc[0]

src/database.py

@@ -1,12 +1,16 @@
-import csv
 from datetime import datetime
 from typing import List
 
-# import boto3
 import pandas as pd
-import streamlit as st
+import streamlit
 
-from src.aws.s3 import get_all_objects, get_csv_content, save_to_s3
+from src.aws.s3 import (
+    download_csv_panda_dataframe_from_s3,
+    save_csv_panda_dataframe_to_s3,
+)
+
+bucket_name = "pharma-doc-trials"
+path_name = "bankyo/change_request_history.csv"
 
 
 def write_csv(
@@ -14,88 +18,47 @@ def write_csv(
     change_type: List[str],
     category: List[str],
     subcategory: List[str],
-    input: str,
-    output_factory: str,
-    output_username: str,
-    output_date: str,
-    output_change_target: str,
-    output_change_period: str,
+    questions: List[str],
+    input: List[str],
     output_change_content: str,
     output_change_reason: str,
     output_effect: str,
-    output_resource: str,
-    output_other: str,
 ) -> None:
     all_types = []
     for i in range(len(change_type)):
         all_type = [i + 1, change_type[i], category[i], subcategory[i]]
         all_types.append(all_type)
-
-    column_names = [
-        "申請者",
-        "全ての変更タイプ",
-        "入力",
-        "出力:工場",
-        "出力:申請者名",
-        "出力:申請日",
-        "出力:変更対象",
-        "出力:実施予定時期",
-        "出力:変更内容",
-        "出力:変更理由",
-        "出力:品質への影響",
-        "出力:参考資料",
-        "出力:備考",
-    ]
-    new_data = pd.DataFrame(
-        [
-            [
-                applicant_name,
-                all_types,
-                input,
-                output_factory,
-                output_username,
-                output_date,
-                output_change_target,
-                output_change_period,
-                output_change_content,
-                output_change_reason,
-                output_effect,
-                output_resource,
-                output_other,
-            ]
-        ],
-        columns=column_names,
-    )
     timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
-    file_name = f"{timestamp}.csv"
-    save_to_s3(new_data, file_name)
+    data = download_csv_panda_dataframe_from_s3(bucket_name, path_name)
+    new_row = [
+        timestamp,
+        applicant_name,
+        all_types,
+        output_change_content,
+        output_change_reason,
+        output_effect,
+    ]
+    for change_serial_number in range(len(questions)):
+        new_row.append(questions[change_serial_number])
+        new_row.append(input[change_serial_number])
 
+    max_length = max(len(data.columns), len(new_row))
+    number_of_qeustions = len(questions)
+    column_existed_number_of_question = (len(data.columns) - 6) // 2
+    if len(data.columns) < max_length:
+        for i in range(column_existed_number_of_question, number_of_qeustions):
+            data[f"質問{i+1}"] = None
+            data[f"回答{i+1}"] = None
+    new_columns = data.columns.tolist()
+    while len(new_row) < len(new_columns):
+        new_row.append("")
+    new_data = pd.DataFrame([new_row], columns=new_columns)
+    data = pd.concat([data, new_data], ignore_index=True)
+    save_csv_panda_dataframe_to_s3(data, bucket_name, path_name)
 
-def display_csv() -> None:
-    datas = get_all_objects()
-    header = [
-        "申請者",
-        "全ての変更タイプ",
-        "入力",
-        "出力:工場",
-        "出力:申請者名",
-        "出力:申請日",
-        "出力:変更対象",
-        "出力:実施予定時期",
-        "出力:変更内容",
-        "出力:変更理由",
-        "出力:品質への影響",
-        "出力:参考資料",
-        "出力:備考",
-    ]
-    path = "data/inputsData.csv"
-    with open(path, mode="w", newline="") as file:
-        writer = csv.writer(file)
-        writer.writerow(header)
-        if datas:
-            for data in datas:
-                data = get_csv_content(data)
-                writer.writerow(data)
+
+def display_csv(file_path: str, st: streamlit) -> None:
+    data = download_csv_panda_dataframe_from_s3(bucket_name, path_name)
+    data.to_csv(file_path, index=False)
     st.write("**データベース**")
-    df = pd.read_csv(path)
-    st.dataframe(df)
+    st.dataframe(data)

src/forms/change_request.py

@@ -358,6 +358,9 @@ def change_request_form() -> None:
                     + systemprompt_template_2
                 )
                 user_prompt = ""
+                # db用
+                user_input = []
+
                 for change_serial_number in range(
                     start_from_one, change_number + upper_bound_correction
                 ):
@@ -380,6 +383,7 @@ def change_request_form() -> None:
                         # question は「Q1-12...」などの形式
                         answer_id = question.split(" : ")[0].replace("Q", "A")
                         user_prompt += f"{answer_id} :  {answer}\n"
+                        user_input.append(f"{answer_id} :  {answer}\n")
 
                 system_prompt += systemprompt_template_3
 
@@ -446,17 +450,11 @@ def change_request_form() -> None:
                     type,
                     category,
                     subcategory,
-                    user_prompt,
-                    st.session_state["contents"]["工場"],
-                    st.session_state["contents"]["申請者"],
-                    st.session_state["contents"]["申請日"],
-                    st.session_state["contents"]["変更対象"],
-                    st.session_state["contents"]["実施予定時期"],
-                    st.session_state["contents"]["変更内容"],
-                    st.session_state["contents"]["変更理由"],
-                    st.session_state["contents"]["品質への影響"],
-                    st.session_state["contents"]["添付資料"],
-                    st.session_state["contents"]["備考"],
+                    question_about_this_change,
+                    user_input,
+                    str(response_data["変更内容"]),
+                    str(response_data["変更理由"]),
+                    str(response_data["品質への影響"]),
                 )
                 st.session_state["document_generated"] = True
 
@@ -486,7 +484,7 @@ def change_request_form() -> None:
         st.markdown(f'**備考** : {st.session_state["contents"]["備考"]}')
 
         review_button = st.button("レビューする")
-        display_csv()
+        display_csv("data/change_request_history.csv", st)
         if review_button:
             query_content = st.session_state["contents"]["変更内容"]
             query_reason = st.session_state["contents"]["変更理由"]

src/forms/database_check.py

@@ -6,14 +6,14 @@ from src.database import display_csv
 
 
 def database_check_form() -> None:
-    display_csv()
-    file_path = "data/inputInfo.csv"
+    file_path = "data/change_request_history.csv"
+    display_csv(file_path, st)
     file_exists = os.path.isfile(file_path)
     if file_exists:
-        with open("data/inputInfo.csv", "rb") as file:
+        with open(file_path, "rb") as file:
             st.download_button(
                 label="ダウンロード",
                 data=file,
-                file_name="inputInfo.csv",
+                file_name="change_request_history.csv",
                 mime="application/vnd.openxmlformats-officedocument.wordprocessingml.document",
             )