Update app.py

app.py

@@ -4,6 +4,7 @@ import glob
 import csv
 import threading
 from datetime import datetime, timedelta
+from typing import Optional
 
 import gradio as gr
 import textstat
@@ -16,12 +17,18 @@ API_KEY = os.getenv("API_KEY")
 BASE_URL = os.getenv("BASE_URL", "https://openrouter.ai/api/v1")
 MODEL = os.getenv("MODEL", "google/gemini-2.5-flash")
 
-# Hugging Face Spaces 永続ストレージ（推奨）
-# Spaces の Persistent Storage を有効化している前提（/data が使える）
+# Hugging Face Spaces 永続ストレージ（Persistent Storage 有効なら /data が永続）
 OUT_DIR = os.getenv("OUT_DIR", "/data")
 os.makedirs(OUT_DIR, exist_ok=True)
+
+# スコアCSV（任意だが便利）
 CSV_PATH = os.path.join(OUT_DIR, "rewrite_scores.csv")
 
+# ★追加：書き換え本文をtxtで蓄積するフォルダ（要件：rewrite_passages）
+# 既定は /data/rewrite_passages（永続に残る）
+REWRITE_DIR = os.getenv("REWRITE_DIR", os.path.join(OUT_DIR, "rewrite_passages"))
+os.makedirs(REWRITE_DIR, exist_ok=True)
+
 PASSAGES_DIR = os.getenv("PASSAGES_DIR", "passages")
 
 if not API_KEY:
@@ -36,7 +43,6 @@ _rewrite_sem = threading.Semaphore(REWRITE_CONCURRENCY)
 _stop_flag_lock = threading.Lock()
 _stop_flag = False
 
-
 # =========================
 # passages の列挙
 # =========================
@@ -44,7 +50,7 @@ def list_passage_files_sorted(passages_dir: str) -> list[tuple[int, str]]:
     pattern = os.path.join(passages_dir, "pg*.txt")
     files = glob.glob(pattern)
 
-    items = []
+    items: list[tuple[int, str]] = []
     for fp in files:
         name = os.path.basename(fp)
         m = re.match(r"pg(\d+)\.txt$", name)
@@ -53,12 +59,10 @@ def list_passage_files_sorted(passages_dir: str) -> list[tuple[int, str]]:
     items.sort(key=lambda x: x[0])
     return items
 
-
 def load_text(path: str) -> str:
     with open(path, "r", encoding="utf-8") as f:
         return f.read()
 
-
 # =========================
 # 書き換え（プロンプト同一）
 # =========================
@@ -86,7 +90,6 @@ excluding the title, author name, source information, chapter number, annotation
         )
     return resp.choices[0].message.content.strip()
 
-
 # =========================
 # 指標（FRE + 単語数）
 # =========================
@@ -100,16 +103,13 @@ def compute_metrics(text: str) -> tuple[float, int]:
     wc = count_words_english(text)
     return fre, wc
 
-
 # =========================
 # CSV追記（軽量・永続）
 # =========================
 _csv_lock = threading.Lock()
 
 def append_csv_row(row: dict):
-    """
-    /data/rewrite_scores.csv に1行追記（ヘッダ無ければ作成）
-    """
+    """rewrite_scores.csv に1行追記（ヘッダ無ければ作成）"""
     fieldnames = ["timestamp_jst", "Text#", "target_level", "flesch_reading_ease", "word_count", "rewritten_text"]
     with _csv_lock:
         exists = os.path.exists(CSV_PATH)
@@ -119,6 +119,41 @@ def append_csv_row(row: dict):
                 w.writeheader()
             w.writerow({k: row.get(k, "") for k in fieldnames})
 
+# =========================
+# ★追加：rewrite_passages に txt 追記
+# =========================
+_txt_lock = threading.Lock()
+
+def append_rewrite_txt(
+    text_id: int,
+    target_level: int,
+    fre: float,
+    word_count: int,
+    rewritten_text: str,
+):
+    """
+    rewrite_passages/Text_{id}.txt に結果を追記（同じText#の再実行も追記で残す）
+    """
+    ts = (datetime.utcnow() + timedelta(hours=9)).strftime("%Y-%m-%d %H:%M:%S")
+    path = os.path.join(REWRITE_DIR, f"Text_{text_id}.txt")
+
+    block = (
+        f"# Text {text_id}\n"
+        f"Target Level: {target_level}\n"
+        f"Flesch Reading Ease: {fre:.2f}\n"
+        f"Word Count: {word_count}\n"
+        f"Timestamp (JST): {ts}\n"
+        f"Model: {MODEL}\n"
+        f"\n"
+        f"---- Rewritten Text ----\n"
+        f"{rewritten_text}\n"
+        f"\n"
+        f"{'=' * 80}\n"
+    )
+
+    with _txt_lock:
+        with open(path, "a", encoding="utf-8") as f:
+            f.write(block)
 
 # =========================
 # 停止フラグ
@@ -132,16 +167,12 @@ def get_stop() -> bool:
     with _stop_flag_lock:
         return _stop_flag
 
-
 # =========================
 # UIロジック
 # =========================
 def init_state():
     files = list_passage_files_sorted(PASSAGES_DIR)
-    return {
-        "files": files,     # [(text_id, path), ...]
-        "idx": 0,           # 次に処理する位置
-    }
+    return {"files": files, "idx": 0}
 
 def start(level: int):
     set_stop(False)
@@ -172,6 +203,7 @@ def run_one(level: int, state: dict):
     # JSTタイムスタンプ（+9）
     ts = (datetime.utcnow() + timedelta(hours=9)).strftime("%Y-%m-%d %H:%M:%S")
 
+    # CSV追記
     append_csv_row({
         "timestamp_jst": ts,
         "Text#": text_id,
@@ -181,16 +213,30 @@ def run_one(level: int, state: dict):
         "rewritten_text": rewritten
     })
 
+    # ★txt追記
+    append_rewrite_txt(
+        text_id=text_id,
+        target_level=level,
+        fre=fre,
+        word_count=wc,
+        rewritten_text=rewritten,
+    )
+
     state["idx"] = idx + 1
 
-    header = f"#Text {text_id}\nTarget Level: {level}\nFlesch Reading Ease: {fre:.2f}\nWord Count: {wc}\nSaved: {CSV_PATH}"
+    header = (
+        f"#Text {text_id}\n"
+        f"Target Level: {level}\n"
+        f"Flesch Reading Ease: {fre:.2f}\n"
+        f"Word Count: {wc}\n"
+        f"Saved CSV: {CSV_PATH}\n"
+        f"Saved TXT: {os.path.join(REWRITE_DIR, f'Text_{text_id}.txt')}"
+    )
     progress = f"{state['idx']} / {total}"
     return state, "1件処理しました。", header, progress, rewritten, gr.update(visible=True)
 
 def run_all(level: int, state: dict):
-    """
-    全件（または残り）を順次処理。途中で「停止」ボタンで止められる。
-    """
+    """全件（または残り）を順次処理。途中で「停止」ボタンで止められる。"""
     set_stop(False)
 
     files = state.get("files", [])
@@ -225,7 +271,23 @@ def run_all(level: int, state: dict):
             "rewritten_text": rewritten
         })
 
-        last_header = f"#Text {text_id}\nTarget Level: {level}\nFlesch Reading Ease: {fre:.2f}\nWord Count: {wc}\nSaved: {CSV_PATH}"
+        # ★txt追記
+        append_rewrite_txt(
+            text_id=text_id,
+            target_level=level,
+            fre=fre,
+            word_count=wc,
+            rewritten_text=rewritten,
+        )
+
+        last_header = (
+            f"#Text {text_id}\n"
+            f"Target Level: {level}\n"
+            f"Flesch Reading Ease: {fre:.2f}\n"
+            f"Word Count: {wc}\n"
+            f"Saved CSV: {CSV_PATH}\n"
+            f"Saved TXT: {os.path.join(REWRITE_DIR, f'Text_{text_id}.txt')}"
+        )
         last_text = rewritten
 
         idx += 1
@@ -243,18 +305,31 @@ def reset_csv():
             os.remove(CSV_PATH)
     return f"CSVを削除しました: {CSV_PATH}"
 
+def reset_rewrite_folder():
+    """
+    rewrite_passages を全消しは危険なので、ここでは「中の txt を削除」する実装。
+    """
+    removed = 0
+    with _txt_lock:
+        for fp in glob.glob(os.path.join(REWRITE_DIR, "Text_*.txt")):
+            try:
+                os.remove(fp)
+                removed += 1
+            except Exception:
+                pass
+    return f"rewrite_passages の Text_*.txt を削除しました（{removed}件）: {REWRITE_DIR}"
 
 # =========================
 # Gradio UI（Spaces向け）
 # =========================
 with gr.Blocks() as demo:
-    gr.Markdown("# 🔁 Passage Rewrite + FRE Scoring (HF Spaces)")
+    gr.Markdown("# 🔁 Passage Rewrite + FRE/Word Scoring (HF Spaces)")
 
     state = gr.State(init_state())
 
     level = gr.Dropdown(choices=[1, 2, 3, 4, 5], value=1, label="Target Level (1..5)")
     status = gr.Textbox(label="Status", interactive=False)
-    header = gr.Textbox(label="Result Header (#Text / FRE / Words)", lines=5, interactive=False)
+    header = gr.Textbox(label="Result Header (#Text / FRE / Words)", lines=6, interactive=False)
     progress = gr.Textbox(label="Progress", interactive=False)
     output_text = gr.Textbox(label="Rewritten Text", lines=18, interactive=False)
 
@@ -263,18 +338,20 @@ with gr.Blocks() as demo:
         one_btn = gr.Button("次へ（1件処理）")
         all_btn = gr.Button("全件実行（残りを処理）")
         stop_btn = gr.Button("停止")
+
     with gr.Row():
         reset_btn = gr.Button("CSVリセット（削除）")
-        csv_hint = gr.Markdown(f"📄 CSV保存先: `{CSV_PATH}`（SpacesのFilesに出ます）")
+        reset_txt_btn = gr.Button("rewrite_passagesリセット（Text_*.txt削除）")
 
-    # stop通知を見せる用
-    stop_note = gr.Markdown(visible=False)
+    gr.Markdown(f"📄 CSV保存先: `{CSV_PATH}`（SpacesのFilesに出ます）")
+    gr.Markdown(f"📝 TXT保存先: `{REWRITE_DIR}`（SpacesのFilesに出ます）")
 
-    start_btn.click(fn=start, inputs=[level], outputs=[state, status, header, progress, output_text, stop_note])
-    one_btn.click(fn=run_one, inputs=[level, state], outputs=[state, status, header, progress, output_text, stop_note])
-    all_btn.click(fn=run_all, inputs=[level, state], outputs=[state, status, header, progress, output_text, stop_note])
+    start_btn.click(fn=start, inputs=[level], outputs=[state, status, header, progress, output_text])
+    one_btn.click(fn=run_one, inputs=[level, state], outputs=[state, status, header, progress, output_text])
+    all_btn.click(fn=run_all, inputs=[level, state], outputs=[state, status, header, progress, output_text])
     stop_btn.click(fn=stop, inputs=[], outputs=[status])
     reset_btn.click(fn=reset_csv, inputs=[], outputs=[status])
+    reset_txt_btn.click(fn=reset_rewrite_folder, inputs=[], outputs=[status])
 
 demo.queue(max_size=64)
 demo.launch()