Update app.py

app.py

@@ -1,187 +1,358 @@
 import gradio as gr
+from openai import OpenAI
+from datasets import Dataset
+from datetime import datetime, timedelta
 import pandas as pd
-import time
-import random
-from datasets import load_dataset, Dataset
-
-# --- Load passages CSV ---
-try:
-    passages_df = pd.read_csv("passage.csv")
-except:
-    passages_df = pd.DataFrame(columns=["genre", "passage_id", "page", "text"])
-
-# --- Logging dataset ---
-log_repo = "Toya0421/reading_exercise_logging"
-
-
-def start_test(student_id, selected_genre):
-    if not student_id:
-        return gr.update(value="⚠️ 学籍番号を入力してください", visible=True), gr.update(visible=False)
-
-    # そのジャンルから教材リストを抽出
-    genre_passages = passages_df[passages_df["genre"] == selected_genre]
-
-    if genre_passages.empty:
-        return "⚠️ 選択ジャンルに教材がありません", gr.update(visible=False)
-
-    # 最初の教材を選ぶ
-    passage_id = random.choice(genre_passages["passage_id"].unique())
-    pages = genre_passages[genre_passages["passage_id"] == passage_id].sort_values("page")
-
-    # 初期状態
-    state = {
-        "student_id": student_id,
-        "genre": selected_genre,
-        "passage_id": passage_id,
-        "pages": pages,
-        "page_index": 0,
-        "logs": []
-    }
-
-    return "", state, pages.iloc[0]["text"], enable_buttons(state)
-
-
-def enable_buttons(state):
-    last_page = len(state["pages"]) - 1
-    pi = state["page_index"]
-
-    return {
-        "prev": gr.update(interactive=(pi > 0)),
-        "next": gr.update(interactive=(pi < last_page)),
-        "finish": gr.update(interactive=(pi == last_page)),  # ✅ 最終ページのみ押せる
-        "retire": gr.update(interactive=True)
-    }
-
-
-def next_page(state):
-    if state is None:
-        return None, "", enable_buttons(state)
-
-    if state["page_index"] < len(state["pages"]) - 1:
-        state["page_index"] += 1
-
-    text = state["pages"].iloc[state["page_index"]]["text"]
-    return state, text, enable_buttons(state)
-
-
-def prev_page(state):
-    if state is None:
-        return None, "", enable_buttons(state)
-
-    if state["page_index"] > 0:
-        state["page_index"] -= 1
-
-    text = state["pages"].iloc[state["page_index"]]["text"]
-    return state, text, enable_buttons(state)
-
-
-def choose_next_passage(state, finished):
-    """finished = True → 読み終えた、False → リタイア"""
-
-    # ログ追加
-    state["logs"].append({
-        "student_id": state["student_id"],
-        "genre": state["genre"],
-        "passage_id": state["passage_id"],
-        "completed": finished,
-        "timestamp": time.time()
-    })
-
-    # ログ保存
-    save_logs(state["logs"])
-
-    # 次の教材選択
-    genre_passages = passages_df[passages_df["genre"] == state["genre"]]
-    remaining = genre_passages[genre_passages["passage_id"] != state["passage_id"]]
-
-    if remaining.empty:
-        return state, "✅ 同ジャンルの教材は終了しました", disable_all()
-
-    # 次の教材
-    new_passage_id = random.choice(remaining["passage_id"].unique())
-    pages = genre_passages[genre_passages["passage_id"] == new_passage_id].sort_values("page")
-
-    state["passage_id"] = new_passage_id
-    state["pages"] = pages
-    state["page_index"] = 0
+import time, os, random, tempfile, json
+
+# --- API / HF 設定 ---
+API_KEY = os.getenv("API_KEY")
+BASE_URL = "https://openrouter.ai/api/v1"
+HF_TOKEN = os.getenv("HF_TOKEN")
+DATASET_REPO = "Toya0421/reading_exercise_logging"
+LOG_FILE = "reading_logs.csv"
+
+client = OpenAI(base_url=BASE_URL, api_key=API_KEY)
+
+# --- passage.csv 読み込み（columns: passage_id, genre, text, original_lexile_score）---
+passages_df = pd.read_csv("passage.csv")
+
+genres = [
+    "Literature","Science＆Technology","History","Social Science＆Society",
+    "Arts＆Culture","Religion＆Philosophy","Lifestyle＆Hobbies",
+    "Health＆Medicine","Education＆Reference"
+]
+
+# --- 状態変数 ---
+used_passages = set()
+current_user_id = None
+current_genre = None
+current_lexile = None
+action_log = []  # ページ操作ログ
+
+# --- ヘルパー関数 ---
+def rewrite_to_lexile(text, target_lexile):
+    prompt = f"""
+Rewrite the following passage so it fits about {target_lexile} Lexile.
+- Keep original meaning and length
+- Avoid figurative language
+- Use simple syntax
+- Output only the rewritten passage
+
+{text}
+"""
+    resp = client.chat.completions.create(
+        model="google/gemma-3-27b-it:free",
+        messages=[{"role": "user", "content": prompt}],
+        temperature=0.4,
+        max_tokens=1000
+    )
+    return resp.choices[0].message.content.strip()
+
+def split_pages(text, words=120):
+    w = text.split()
+    pages = [" ".join(w[i:i+words]) for i in range(0, len(w), words)]
+    return pages if pages else [text]
+
+def get_new_passage(same_genre=True):
+    """同ジャンル or 別ジャンルでランダムに passage を返す (passage_id, text, original_lexile)"""
+    global used_passages, current_genre
+    if same_genre:
+        df = passages_df[passages_df["genre"] == current_genre]
+    else:
+        df = passages_df[passages_df["genre"] != current_genre]
+
+    if df.empty:
+        return None, None, None
+
+    available = [pid for pid in df["passage_id"].unique() if pid not in used_passages]
+    if not available:
+        # 使い切ったら既読クリアして再利用
+        used_passages.clear()
+        available = list(df["passage_id"].unique())
+
+    pid = random.choice(available)
+    row = df[df["passage_id"] == pid].iloc[0]
+    used_passages.add(pid)
+    return pid, row["text"], row.get("original_lexile_score", None)
+
+def save_log(entry):
+    df = pd.DataFrame([entry])
+    if os.path.exists(LOG_FILE):
+        df.to_csv(LOG_FILE, mode="a", index=False, header=False)
+    else:
+        df.to_csv(LOG_FILE, index=False)
+    # push to HF
+    all_logs = pd.read_csv(LOG_FILE)
+    tmp_dir = tempfile.mkdtemp()
+    tmp_path = os.path.join(tmp_dir, "data.parquet")
+    all_logs.to_parquet(tmp_path)
+    dataset = Dataset.from_parquet(tmp_path)
+    dataset.push_to_hub(DATASET_REPO, token=HF_TOKEN)
+
+# ============================
+# Gradio コールバック関数
+# ============================
+
+def start_test(student_id, genre, lexile_input):
+    """開始：学生番号・ジャンル・Lexile入力は必須（仕様どおり）"""
+    global current_user_id, current_genre, current_lexile, used_passages, action_log
+    used_passages = set()
+    action_log = []
+
+    if not student_id or str(student_id).strip() == "":
+        # テキストとページ表示だけ空で返し、ボタンはすべて disabled
+        return (
+            "",  # text_display
+            "",  # page_display
+            json.dumps([]),  # hidden_pages
+            0,  # hidden_page_index
+            "",  # start_time
+            0,  # total_pages
+            "", None, None,  # pid, orig_lex, assigned_lex
+            gr.update(interactive=False),  # prev_btn update
+            gr.update(interactive=False),  # next_btn update
+            gr.update(interactive=False)   # finish_btn update
+        )
+
+    current_user_id = str(student_id).strip()
+    current_genre = genre
+    current_lexile = int(lexile_input)
+
+    pid, text, orig_lex = get_new_passage(same_genre=True)
+    if text is None:
+        return (
+            "教材が見つかりません", "", json.dumps([]), 0, "", 0, "", None, None,
+            gr.update(interactive=False), gr.update(interactive=False), gr.update(interactive=False)
+        )
+
+    rewritten = rewrite_to_lexile(text, current_lexile)
+    pages = split_pages(rewritten)
+    start_time = (datetime.utcnow() + timedelta(hours=9)).isoformat()
+    total = len(pages)
+
+    # ボタン状態（最初は prev 無効、 next 有効 if pages>1、 finish 有効 only if single page）
+    prev_upd = gr.update(interactive=False)
+    next_upd = gr.update(interactive=(total > 1))
+    finish_upd = gr.update(interactive=(0 == total - 1))  # True if only 1 page
+
+    return (
+        pages[0],  # text_display
+        f"1 / {total}",  # page_display
+        json.dumps(pages, ensure_ascii=False),  # hidden_pages
+        0,  # hidden_page_index
+        start_time,  # hidden_start_time
+        total,  # hidden_total_pages
+        pid,  # hidden_passage_id
+        orig_lex,  # hidden_orig_lex
+        current_lexile,  # hidden_assigned_lex
+        prev_upd,
+        next_upd,
+        finish_upd
+    )
 
-    return state, pages.iloc[0]["text"], enable_buttons(state)
+def next_page(pages_json, current_page, total_pages):
+    pages = json.loads(pages_json)
+    if not pages:
+        return "", "", json.dumps([]), 0, gr.update(interactive=False), gr.update(interactive=False), gr.update(interactive=False)
+    # advance safely
+    new_page = min(current_page + 1, total_pages - 1)
+    if new_page != current_page:
+        action_log.append({"action": "next", "time": (datetime.utcnow() + timedelta(hours=9)).isoformat()})
+    finish_enabled = (new_page == total_pages - 1)
+    prev_enabled = (new_page > 0)
+    next_enabled = (new_page < total_pages - 1)
+    return (
+        pages[new_page],
+        f"{new_page+1} / {total_pages}",
+        json.dumps(pages, ensure_ascii=False),
+        new_page,
+        gr.update(interactive=prev_enabled),
+        gr.update(interactive=next_enabled),
+        gr.update(interactive=finish_enabled)
+    )
 
+def prev_page(pages_json, current_page, total_pages):
+    pages = json.loads(pages_json)
+    if not pages:
+        return "", "", json.dumps([]), 0, gr.update(interactive=False), gr.update(interactive=False), gr.update(interactive=False)
+    new_page = max(current_page - 1, 0)
+    if new_page != current_page:
+        action_log.append({"action": "prev", "time": (datetime.utcnow() + timedelta(hours=9)).isoformat()})
+    finish_enabled = (new_page == total_pages - 1)
+    prev_enabled = (new_page > 0)
+    next_enabled = (new_page < total_pages - 1)
+    return (
+        pages[new_page],
+        f"{new_page+1} / {total_pages}",
+        json.dumps(pages, ensure_ascii=False),
+        new_page,
+        gr.update(interactive=prev_enabled),
+        gr.update(interactive=next_enabled),
+        gr.update(interactive=finish_enabled)
+    )
 
-def disable_all():
-    return {
-        "prev": gr.update(interactive=False),
-        "next": gr.update(interactive=False),
-        "finish": gr.update(interactive=False),
-        "retire": gr.update(interactive=False)
+def finish_or_retire(pages_json, current_page, pid, orig_lex, start_time, action):
+    """action == 'finished' or 'retire'"""
+    pages = json.loads(pages_json)
+    now = (datetime.utcnow() + timedelta(hours=9)).isoformat()
+
+    # ログ
+    entry = {
+        "user_id": current_user_id,
+        "genre": current_genre,
+        "assigned_lexile": current_lexile,
+        "passage_id": pid,
+        "original_lexile": orig_lex,
+        "first_page_displayed": start_time,
+        "finished_time": now,
+        "actions": json.dumps(action_log, ensure_ascii=False),
+        "result": action
     }
-
-
-def save_logs(logs):
-    if not logs:
-        return
-    df = pd.DataFrame(logs)
-    dataset = Dataset.from_pandas(df)
-    dataset.push_to_hub(log_repo, private=False)
-
-
-# ==== UI ====
-
-with gr.Blocks() as demo:
-    gr.Markdown("### 📘 リーディングアプリ")
-
-    student_id = gr.Textbox(label="学籍番号を入力", placeholder="例: A12345")
-
-    genre_select = gr.Dropdown(
-        choices=[
-            "Literature", "Science＆Technology", "History",
-            "Social Science＆Society", "Arts＆Culture",
-            "Religion＆Philosophy", "Lifestyle＆Hobbies",
-            "Health＆Medicine", "Education＆Reference"
-        ],
-        label="ジャンルを選択"
+    save_log(entry=entry) if 'save_log' in globals() else save_log_local(entry)
+
+    # 次教材を取得
+    if action == "finished":
+        new_pid, new_text, new_orig_lex = get_new_passage(same_genre=True)
+    else:
+        new_pid, new_text, new_orig_lex = get_new_passage(same_genre=False)
+
+    if new_text is None:
+        # 全部終わった or 該当なし
+        return (
+            "教材がありません",  # text_display
+            "",  # page_display
+            json.dumps([]),  # hidden_pages
+            0,  # page_index
+            "",  # start_time
+            0,  # total_pages
+            "", None, None,  # pid, orig_lex, assigned_lex
+            gr.update(interactive=False),
+            gr.update(interactive=False),
+            gr.update(interactive=False)
+        )
+
+    rewritten = rewrite_to_lexile(new_text, current_lexile)
+    new_pages = split_pages(rewritten)
+    new_start = (datetime.utcnow() + timedelta(hours=9)).isoformat()
+    total = len(new_pages)
+
+    # reset action log for new passage
+    action_log.clear()
+
+    prev_upd = gr.update(interactive=False)
+    next_upd = gr.update(interactive=(total > 1))
+    finish_upd = gr.update(interactive=(0 == total - 1))
+
+    return (
+        new_pages[0],
+        f"1 / {total}",
+        json.dumps(new_pages, ensure_ascii=False),
+        0,
+        new_start,
+        total,
+        new_pid,
+        new_orig_lex,
+        current_lexile,
+        prev_upd, next_upd, finish_upd
     )
 
-    start_btn = gr.Button("スタート")
-    warn = gr.Textbox(label="警告", visible=False)
+# fallback local saver if earlier definitions differ
+def save_log_local(entry):
+    df = pd.DataFrame([entry])
+    if os.path.exists(LOG_FILE):
+        df.to_csv(LOG_FILE, mode="a", index=False, header=False)
+    else:
+        df.to_csv(LOG_FILE, index=False)
+
+# ============================
+# UI 部分（Lexile 入力は復活）
+# ============================
+with gr.Blocks() as demo:
+    gr.Markdown("# 📚 Reading Exercise")
 
-    passage_state = gr.State()
-    passage_text = gr.Textbox(label="教材", lines=15, interactive=False)
+    student_id_input = gr.Textbox(label="学生番号（必須）")
+    lexile_input = gr.Number(label="あなたの Lexile スコア（例: 900）", precision=0)
+    genre_input = gr.Dropdown(choices=genres, label="ジャンル（1つ選択）")
 
-    prev_btn = gr.Button("◀ 前へ")
-    next_btn = gr.Button("次へ ▶")
-    finish_btn = gr.Button("読み終えた")
-    retire_btn = gr.Button("リタイア")
+    start_btn = gr.Button("スタート")
 
-    # --- Event binding ---
+    text_display = gr.Textbox(label="教材", lines=18, interactive=False)
+    page_display = gr.Textbox(label="進行状況", lines=1, interactive=False)
+
+    # hidden state
+    hidden_pages = gr.Textbox(visible=False)
+    hidden_page_index = gr.Number(visible=False)
+    hidden_start_time = gr.Textbox(visible=False)
+    hidden_total_pages = gr.Number(visible=False)
+    hidden_passage_id = gr.Textbox(visible=False)
+    hidden_orig_lex = gr.Textbox(visible=False)
+    hidden_assigned_lex = gr.Textbox(visible=False)
+
+    # buttons row (always visible)
+    with gr.Row():
+        prev_btn = gr.Button("◀ 前へ", interactive=False)
+        next_btn = gr.Button("次へ ▶", interactive=False)
+        finish_btn = gr.Button("読み終えた", interactive=False)
+        retire_btn = gr.Button("リタイア")
+
+    # Start click
     start_btn.click(
-        start_test,
-        [student_id, genre_select],
-        [warn, passage_state, passage_text, prev_btn, next_btn, finish_btn, retire_btn]
+        fn=start_test,
+        inputs=[student_id_input, genre_input, lexile_input],
+        outputs=[
+            text_display, page_display,
+            hidden_pages, hidden_page_index,
+            hidden_start_time, hidden_total_pages, hidden_passage_id,
+            hidden_orig_lex, hidden_assigned_lex,
+            prev_btn, next_btn, finish_btn
+        ]
     )
 
+    # Next / Prev click
     next_btn.click(
-        next_page,
-        passage_state,
-        [passage_state, passage_text, prev_btn, next_btn, finish_btn, retire_btn]
+        fn=next_page,
+        inputs=[hidden_pages, hidden_page_index, hidden_total_pages],
+        outputs=[
+            text_display, page_display,
+            hidden_pages, hidden_page_index,
+            prev_btn, next_btn, finish_btn
+        ]
     )
 
     prev_btn.click(
-        prev_page,
-        passage_state,
-        [passage_state, passage_text, prev_btn, next_btn, finish_btn, retire_btn]
+        fn=prev_page,
+        inputs=[hidden_pages, hidden_page_index, hidden_total_pages],
+        outputs=[
+            text_display, page_display,
+            hidden_pages, hidden_page_index,
+            prev_btn, next_btn, finish_btn
+        ]
     )
 
+    # Finish / Retire click
     finish_btn.click(
-        lambda s: choose_next_passage(s, True),
-        passage_state,
-        [passage_state, passage_text, prev_btn, next_btn, finish_btn, retire_btn]
+        fn=lambda p, i, pid, o, st: finish_or_retire(p, i, pid, o, st, "finished"),
+        inputs=[hidden_pages, hidden_page_index, hidden_passage_id, hidden_orig_lex, hidden_start_time],
+        outputs=[
+            text_display, page_display,
+            hidden_pages, hidden_page_index,
+            hidden_start_time, hidden_total_pages, hidden_passage_id,
+            hidden_orig_lex, hidden_assigned_lex,
+            prev_btn, next_btn, finish_btn
+        ]
     )
 
     retire_btn.click(
-        lambda s: choose_next_passage(s, False),
-        passage_state,
-        [passage_state, passage_text, prev_btn, next_btn, finish_btn, retire_btn]
+        fn=lambda p, i, pid, o, st: finish_or_retire(p, i, pid, o, st, "retire"),
+        inputs=[hidden_pages, hidden_page_index, hidden_passage_id, hidden_orig_lex, hidden_start_time],
+        outputs=[
+            text_display, page_display,
+            hidden_pages, hidden_page_index,
+            hidden_start_time, hidden_total_pages, hidden_passage_id,
+            hidden_orig_lex, hidden_assigned_lex,
+            prev_btn, next_btn, finish_btn
+        ]
     )
 
 demo.launch()