Refactor database connection to use Supabase DB and Enhance project structure

.gitignore

@@ -19,6 +19,10 @@ wheels/
 *.egg-info/
 .installed.cfg
 *.egg
+MANIFEST
+
+# Package build
+src/*.egg-info/
 
 # Virtual environments
 venv/
@@ -30,6 +34,9 @@ env/
 .env
 .env.local
 
+# Cache directories
+.cache/
+
 # UV
 .uv/
 uv.lock
@@ -41,8 +48,9 @@ uv.lock
 *.swo
 *~
 
-# Streamlit
-.streamlit/
+# Streamlit (but keep config.toml)
+.streamlit/*
+!.streamlit/config.toml
 
 # JSON data output
 json_data/

.streamlit/config.toml

@@ -0,0 +1,13 @@
+[server]
+headless = true
+port = 7860
+address = "0.0.0.0"
+fileWatcherType = "none"
+
+[browser]
+gatherUsageStats = false
+serverAddress = "0.0.0.0"
+serverPort = 7860
+
+[theme]
+base = "light"

Dockerfile

@@ -9,20 +9,38 @@ RUN apt-get update && apt-get install -y \
     && rm -rf /var/lib/apt/lists/* \
     && curl -LsSf https://astral.sh/uv/install.sh | sh
 
-# Add uv to PATH
-ENV PATH="/root/.cargo/bin:$PATH"
+# Add uv to PATH (it installs to /root/.local/bin)
+ENV PATH="/root/.local/bin:$PATH"
 
-# Copy dependency files
+# Set Streamlit config to avoid permission issues
+ENV STREAMLIT_SERVER_HEADLESS=true
+ENV STREAMLIT_SERVER_FILE_WATCHER_TYPE=none
+ENV STREAMLIT_BROWSER_GATHER_USAGE_STATS=false
+
+# Set HuggingFace cache to writable directory
+ENV HF_HOME=/app/.cache
+ENV SENTENCE_TRANSFORMERS_HOME=/app/.cache/sentence-transformers
+
+# Copy project files for installation
 COPY pyproject.toml .
+COPY src/ ./src/
 
-# Install dependencies using modern uv (directly from pyproject.toml)
+# Install the package with dependencies using modern uv
 RUN uv pip install --system --no-cache .
 
-# Copy application code
+# Create cache directory and pre-download ML models
+RUN mkdir -p /app/.cache /app/json_data && \
+    python3 -c "from sentence_transformers import SentenceTransformer; SentenceTransformer('all-MiniLM-L12-v2')" && \
+    python3 -c "from sudachipy import dictionary; dictionary.Dictionary().create()" && \
+    chmod -R 777 /app/.cache /app/json_data
+
+# Copy remaining application files (env.example, schema.sql, etc.)
 COPY . .
 
 EXPOSE 7860
 
-HEALTHCHECK CMD curl --fail http://localhost:7860/_stcore/health || exit 1
+# Healthcheck (optional - comment out if causing issues)
+HEALTHCHECK --interval=30s --timeout=10s --start-period=60s --retries=3 \
+    CMD curl --fail http://localhost:7860/_stcore/health || exit 1
 
-ENTRYPOINT ["streamlit", "run", "日次RA.py", "--server.port=7860", "--server.address=0.0.0.0"]
+CMD ["streamlit", "run", "src/daily_ra/app.py", "--server.port=7860", "--server.address=0.0.0.0"]

README.md

@@ -36,7 +36,7 @@ See [DEPLOYMENT.md](./DEPLOYMENT.md) for detailed setup instructions.
 # Install uv
 curl -LsSf https://astral.sh/uv/install.sh | sh
 
-# Install dependencies (modern way - directly from pyproject.toml)
+# Install package in editable mode (installs dependencies + package)
 uv pip install -e .
 
 # Set up environment variables
@@ -44,7 +44,7 @@ cp env.example .env
 # Edit .env with your Supabase credentials
 
 # Run the app
-streamlit run 日次RA.py
+streamlit run src/daily_ra/app.py
 ```
 
 ### Docker
@@ -108,18 +108,14 @@ uv pip install --upgrade -e .
    POOL_NAME=vvkubjwatpdqnkmgmntf
    ```
 
-4. Install dependencies using uv (modern way):
+4. Install package in editable mode:
    ```bash
-   # Option 1: Install in editable mode (recommended for development)
    uv pip install -e .
-   
-   # Option 2: Sync dependencies (if using uv.lock)
-   uv sync
    ```
 
 5. Run the app locally:
    ```bash
-   streamlit run 日次RA.py
+   streamlit run src/daily_ra/app.py
    ```
 
 ### 3️⃣ **Deploy to Hugging Face Spaces**
@@ -128,13 +124,12 @@ uv pip install --upgrade -e .
 
 2. Go to your Space Settings → **Repository secrets**
 
-3. Add the following secrets:
-   - `HOST` = `aws-1-ap-northeast-1.pooler.supabase.com`
+3. Add the following secrets (get these from your Supabase project settings):
+   - `HOST` = `aws-1-ap-northeast-1.pooler.supabase.com` (your pooler host)
    - `PORT` = `5432`
    - `DATABASE` = `postgres`
-   - `USERNAME` = `postgres.vvkubjwatpdqnkmgmntf`
+   - `USERNAME` = `postgres.vvkubjwatpdqnkmgmntf` (format: postgres.YOUR_PROJECT_REF)
    - `PASSWORD` = `your-actual-password`
-   - `POOL_NAME` = `vvkubjwatpdqnkmgmntf`
 
 4. The app will automatically rebuild and deploy
 

env.example

@@ -1,8 +1,22 @@
 # Supabase PostgreSQL Connection Details
+# Copy this file to .env and fill in your actual values
+
+# Example connection string from Supabase:
+# postgresql://postgres.PROJECT_REF:[PASSWORD]@aws-0-region.pooler.supabase.com:5432/postgres
+
+# Database host (get from Supabase project settings)
 HOST=aws-1-ap-northeast-1.pooler.supabase.com
+
+# Database port (default is 5432 for PostgreSQL)
 PORT=5432
+
+# Database name (usually 'postgres' for Supabase)
 DATABASE=postgres
+
+# Username - IMPORTANT: Use the FULL format from Supabase
+# Format: postgres.YOUR_PROJECT_REF (e.g., postgres.vvkubjwatpdqnkmgmntf)
+# This is a SINGLE username, not two separate parts!
 USERNAME=postgres.vvkubjwatpdqnkmgmntf
-PASSWORD=your-password-here
-POOL_NAME=vvkubjwatpdqnkmgmntf
 
+# Your database password (from Supabase project settings)
+PASSWORD=your-password-here

pyproject.toml

@@ -9,6 +9,7 @@ dependencies = [
     "pandas>=2.0.0",
     "sentence-transformers>=2.2.0",
     "sudachipy>=0.6.0",
+    "sudachidict-core>=20240716",
     "python-dotenv>=1.0.0",
 ]
 
@@ -16,5 +17,5 @@ dependencies = [
 requires = ["hatchling"]
 build-backend = "hatchling.build"
 
-[tool.uv]
-dev-dependencies = []
+[tool.hatch.build.targets.wheel]
+packages = ["src/daily_ra"]

src/daily_ra/__init__.py

@@ -0,0 +1,9 @@
+"""
+Daily Risk Assessment (日次RA) Application
+
+A Streamlit application for daily risk assessment in construction and industrial settings,
+with automatic safety rule generation.
+"""
+
+__version__ = "0.1.0"
+

src/daily_ra/app.py

@@ -0,0 +1,212 @@
+import streamlit as st
+import psycopg2
+import json
+from datetime import datetime
+from sudachipy import dictionary, tokenizer
+from sentence_transformers import SentenceTransformer, util
+import pandas as pd  
+import os
+from dotenv import load_dotenv
+
+# ============================
+# 🔧 1. 設定
+# ============================
+
+# Load environment variables from .env file (for local development)
+load_dotenv()
+
+# DB接続設定
+DB_HOST = os.environ.get("DB_HOST")
+DB_PORT = os.environ.get("DB_PORT", "5432")
+DB_USER = os.environ.get("DB_USERNAME")
+DB_PASSWORD = os.environ.get("DB_PASSWORD")
+DB_NAME = os.environ.get("DB_NAME", "postgres")
+
+print(DB_HOST, DB_PORT, DB_USER, DB_NAME)
+
+# Construct PostgreSQL connection URL for Supabase
+if all([DB_HOST, DB_USER, DB_PASSWORD, DB_NAME]):
+    DB_URL = f"postgresql://{DB_USER}:{DB_PASSWORD}@{DB_HOST}:{DB_PORT}/{DB_NAME}"
+    print(f"🔗 Connecting to Supabase database as user '{DB_USER}'...")
+else:
+    st.error("❌ Database connection details not found in environment variables!")
+    st.stop()
+
+# ============================
+# DB 接続
+# ============================
+try:
+    conn = psycopg2.connect(DB_URL)
+    conn.autocommit = True
+    cur = conn.cursor()
+    print(f"✅ Connected to Supabase database '{DB_NAME}'")
+except Exception as e:
+    st.error(f"❌ Database connection failed: {str(e)}")
+    st.stop()
+
+# SudachiPy セットアップ
+sudachi_tokenizer = dictionary.Dictionary().create()
+def sudachi_tokenizer_func(text):
+    tokens = sudachi_tokenizer.tokenize(text, tokenizer.Tokenizer.SplitMode.C)
+    return [t.surface() for t in tokens]
+
+# SentenceTransformerモデル
+model = SentenceTransformer("all-MiniLM-L12-v2")
+
+# 正規化辞書
+NORMALIZE = {
+    "重機": ["ショベルカー", "ユンボ", "バックホウ", "グレーダー"],
+    "作業員": ["作業者", "職人", "人"],
+    "クレーン": ["クレーン車", "吊り上げ機"],
+    "足場": ["仮設足場", "高所足場"],
+    "吊荷": ["荷", "吊り荷", "吊下げ物"]
+}
+
+# 分類キーワード
+OBJECTS = ["作業員", "重機", "クレーン", "吊荷", "足場", "ダンプ"]
+RISKS = ["挟まれ", "接触", "墜落", "転倒", "感電", "落下", "衝突"]
+
+POTENTIAL_RISKS = {
+    ("作業員", "重機"): "作業員と重機が近接している状態",
+    ("作業員", "足場"): "作業員が高所作業中の可能性",
+    ("クレーン", "吊荷"): "吊荷の下に人がいる可能性",
+    ("作業員", "吊荷"): "作業員が吊荷の下にいる可能性",
+}
+
+# ============================
+# 🧩 2. 関数群
+# ============================
+
+def normalize_text(text):
+    """表記ゆれ統一"""
+    for base, words in NORMALIZE.items():
+        for w in words:
+            text = text.replace(w, base)
+    return text
+
+def extract_relations(text):
+    """
+    文中の対象物とリスクを組み合わせて簡易ペア抽出
+    """
+    pairs = []
+    text_norm = normalize_text(text)
+
+    # 文中の対象物を検出
+    found_objects = [obj for obj in OBJECTS if obj in text_norm]
+
+    # 文中のリスクワードを検出
+    found_risks = [risk for risk in RISKS if risk in text_norm]
+
+    # 複数対象物とリスクがある場合にペア化
+    if len(found_objects) >= 2 and found_risks:
+        for i in range(len(found_objects)):
+            for j in range(i+1, len(found_objects)):
+                pairs.append((found_objects[i], found_objects[j], found_risks))
+    return pairs
+
+def generate_rules(data):
+    """ルールベース生成"""
+    text = normalize_text(" ".join([
+        data["work_content"],
+        data["hazard_points"],
+        data["risk_identification"],
+        data["mitigation_measures"]
+    ]))
+
+    # 構文関係抽出
+    relations = extract_relations(text)
+
+    rules = []
+    for subj, obj, _ in relations:
+        # 潜在リスクを確認
+        risk_desc = POTENTIAL_RISKS.get((subj, obj)) or POTENTIAL_RISKS.get((obj, subj)) or []
+        rules.append({
+            "object1": subj,
+            "object2": obj,
+            "risk": risk_desc  
+        })
+    return rules
+
+# ============================
+# 🖥 3. Streamlit UI
+# ============================
+
+st.title("日次RA入力")
+
+with st.form("ra_form"):
+    work_date = st.date_input("作業日")
+    work_content = st.text_area("作業内容")
+    hazard_points = st.text_area("作業危険ポイント")
+    general_comments = st.text_area("元請コメント")
+    risk_identification = st.text_area("危険性・有害性の特定")
+    mitigation_measures = st.text_area("危険性・有害性の低減策")
+    inspection_items = st.text_area("点検事項")
+
+    submitted = st.form_submit_button("保存")
+
+# ============================
+# フォーム送信後の処理
+# ============================
+if submitted:
+    # --- 入力データを辞書にまとめる ---
+    form_data = {
+        "work_date": str(work_date),
+        "work_content": work_content,
+        "hazard_points": hazard_points,
+        "general_comments": general_comments,
+        "risk_identification": risk_identification,
+        "mitigation_measures": mitigation_measures,
+        "inspection_items": inspection_items
+    }
+
+    # --- ルール生成 ---
+    rules = generate_rules(form_data)
+
+    # --- PostgreSQL 保存 ---
+    sql = """INSERT INTO daily_ra 
+        (work_date, work_content, hazard_points, general_comments, risk_identification, mitigation_measures, inspection_items, created_at)
+        VALUES (%s,%s,%s,%s,%s,%s,%s,NOW()) RETURNING id"""
+    cur.execute(sql, tuple(form_data.values()))
+    daily_id = cur.fetchone()[0]  # PostgreSQL uses RETURNING to get the inserted ID
+
+    for r in rules:
+        sql_rule = """INSERT INTO rule_base (daily_ra_id, object1, object2, risk, created_at)
+                      VALUES (%s,%s,%s,%s,NOW())"""
+        cur.execute(sql_rule, (daily_id, r["object1"], r["object2"], json.dumps(r["risk"], ensure_ascii=False)))
+
+    # No need to commit with autocommit=True, but keeping for clarity
+    conn.commit()
+    st.success("✅ 入力内容とルールベースの生成・保存が完了しました！")
+
+    # --- 表形式でルール表示 ---
+    if rules:
+        df = pd.DataFrame(rules)
+        st.subheader("🔍 生成されたルール（テーブル形式）")
+        st.dataframe(df)
+
+    # --- JSON作成（LLM連携用）＆保存 ---
+    json_data = {
+        "daily_id": daily_id,
+        "rules": rules
+    }
+
+    # JSON保存用ディレクトリ作成
+    # Use absolute path in Docker, relative path locally
+    if os.path.exists("/app"):
+        json_dir = "/app/json_data"
+    else:
+        json_dir = "json_data"
+    
+    try:
+        os.makedirs(json_dir, exist_ok=True)
+        
+        # ファイル名に daily_id とタイムスタンプを付与
+        json_path = os.path.join(json_dir, f"daily_ra_{daily_id}_{datetime.now().strftime('%Y%m%d_%H%M%S')}.json")
+
+        # JSONファイルとして保存
+        with open(json_path, "w", encoding="utf-8") as f:
+            json.dump(json_data, f, ensure_ascii=False, indent=2)
+
+        st.success(f"✅ JSONファイルを保存しました: {json_path}")
+    except PermissionError:
+        st.warning("⚠️ JSONファイルの保存はスキップされました（データベースには正常に保存されています）")