Update app.py

app.py

@@ -1,20 +1,25 @@
-# --- 変更1: Files の type を "filepath" にし、ハンドラをファイルパス対応へ ---
-with gr.Row():
-    in_files = gr.Files(
-        label="レジュメ類 (PDF/画像/Word/テキスト) 複数可",
-        file_count="multiple",
-        file_types=[".pdf", ".png", ".jpg", ".jpeg", ".tiff", ".bmp", ".docx", ".txt"],
-        type="filepath",  # ← "file" から修正
-    )
-    candidate_id = gr.Textbox(label="候補者ID（任意。未入力なら自動生成）")
-
-
-# --- 変更2: out_skills は schema 周りの不具合回避のため JSON → Code に ---
-with gr.Tab("抽出スキル"):
-    out_skills = gr.Code(label="スキル一覧(JSON)")  # ← gr.JSON から修正
+import os
+import io
+import json
+import hashlib
+import gradio as gr
+
+from pipelines.openai_ingest import (
+    extract_text_with_openai,
+    structure_with_openai,
+    summarize_with_openai,
+)
+from pipelines.parsing import normalize_resume
+from pipelines.merge import merge_normalized_records
+from pipelines.skills import extract_skills
+from pipelines.anonymize import anonymize_text, render_anonymized_pdf
+from pipelines.scoring import compute_quality_score
+from pipelines.storage import persist_to_hf
+from pipelines.utils import detect_filetype, load_doc_text
+
+APP_TITLE = "候補者インテーク & レジュメ標準化（OpenAI版）"
 
 
-# --- 変更3: ハンドラの files 取り扱いを「パス」前提に修正 ---
 def process_resumes(files, candidate_id: str, additional_notes: str = ""):
     if not files:
         raise gr.Error("少なくとも1ファイルをアップロードしてください。")
@@ -22,24 +27,24 @@ def process_resumes(files, candidate_id: str, additional_notes: str = ""):
     partial_records = []
     raw_texts = []
 
-    for path in files:  # ← UploadedFile ではなく filepath の配列
-        with open(path, "rb") as fh:
-            raw_bytes = fh.read()
-        fname = os.path.basename(path)
+    for f in files:
+        # gr.Files(type="binary") では .read() / .name が利用可能
+        raw_bytes = f.read()
+        filetype = detect_filetype(f.name, raw_bytes)
 
-        filetype = detect_filetype(fname, raw_bytes)
-
-        # 1) テキスト抽出
+        # 1) テキスト抽出：画像/PDFはOpenAI Vision OCR、docx/txtは生文面＋OpenAI整形
         if filetype in {"pdf", "image"}:
-            text = extract_text_with_openai(raw_bytes, filename=fname, filetype=filetype)
+            text = extract_text_with_openai(raw_bytes, filename=f.name, filetype=filetype)
         else:
             base_text = load_doc_text(filetype, raw_bytes)
-            text = extract_text_with_openai(base_text.encode("utf-8"), filename=fname, filetype="txt")
+            # 生テキストをそのままOpenAIへ渡し、軽く整形した全文を返す
+            text = extract_text_with_openai(base_text.encode("utf-8"), filename=f.name, filetype="txt")
 
-        raw_texts.append({"filename": fname, "text": text})
+        raw_texts.append({"filename": f.name, "text": text})
 
-        # 2) 構造化
+        # 2) OpenAIでセクション構造化
         structured = structure_with_openai(text)
+        # 念のためルールベース正規化も適用（期間抽出など補助）
         normalized = normalize_resume({
             "work_experience": structured.get("work_experience_raw", ""),
             "education": structured.get("education_raw", ""),
@@ -47,18 +52,38 @@ def process_resumes(files, candidate_id: str, additional_notes: str = ""):
             "skills": ", ".join(structured.get("skills_list", [])),
         })
         partial_records.append({
-            "source": fname,
+            "source": f.name,
             "text": text,
             "structured": structured,
             "normalized": normalized,
         })
 
-    # ...（中略：ロジックはそのまま）...
+    # 3) 統合（複数ファイル→1候補者）
+    merged = merge_normalized_records([r["normalized"] for r in partial_records])
+
+    # 4) スキル抽出（辞書/正規表現）
+    merged_text = "\n\n".join([r["text"] for r in partial_records])
+    skills = extract_skills(merged_text, {
+        "work_experience": merged.get("raw_sections", {}).get("work_experience", ""),
+        "education": merged.get("raw_sections", {}).get("education", ""),
+        "certifications": merged.get("raw_sections", {}).get("certifications", ""),
+        "skills": ", ".join(merged.get("skills", [])),
+    })
 
-    # 8) まとめ
+    # 5) 匿名化
+    anonymized_text, anon_map = anonymize_text(merged_text)
+    anon_pdf_bytes = render_anonymized_pdf(anonymized_text)
+
+    # 6) 品質スコア
+    score = compute_quality_score(merged_text, merged)
+
+    # 7) 要約（300/100/1文）
+    summaries = summarize_with_openai(merged_text)
+
+    # 8) 構造化出力
     result_json = {
         "candidate_id": candidate_id or hashlib.sha256(merged_text.encode("utf-8")).hexdigest()[:16],
-        "files": [os.path.basename(p) for p in files],  # ← fname ���列へ
+        "files": [f.name for f in files],
         "merged": merged,
         "skills": skills,
         "quality_score": score,
@@ -67,12 +92,25 @@ def process_resumes(files, candidate_id: str, additional_notes: str = ""):
         "notes": additional_notes,
     }
 
-    # ...（Datasets 保存処理はそのまま）...
+    # 9) HF Datasets 保存
+    dataset_repo = os.environ.get("DATASET_REPO")
+    commit_info = None
+    if dataset_repo:
+        file_hash = result_json["candidate_id"]
+        commit_info = persist_to_hf(
+            dataset_repo=dataset_repo,
+            record=result_json,
+            anon_pdf_bytes=anon_pdf_bytes,
+            parquet_path=f"candidates/{file_hash}.parquet",
+            json_path=f"candidates/{file_hash}.json",
+            pdf_path=f"candidates/{file_hash}.anon.pdf",
+        )
+
+    anon_pdf = (result_json["candidate_id"] + ".anon.pdf", anon_pdf_bytes)
 
-    # 返り値：out_skills を Code にしたので JSON 文字列で返す
     return (
         json.dumps(result_json, ensure_ascii=False, indent=2),
-        json.dumps(skills, ensure_ascii=False, indent=2),  # ← 文字列化
+        skills,
         json.dumps(score, ensure_ascii=False, indent=2),
         summaries["300chars"],
         summaries["100chars"],
@@ -82,6 +120,48 @@ def process_resumes(files, candidate_id: str, additional_notes: str = ""):
     )
 
 
-# --- 変更4: HF Spaces での起動を安定させるため launch の引数を明示 ---
+with gr.Blocks(title=APP_TITLE) as demo:
+    gr.Markdown(f"# {APP_TITLE}\n複数ファイルを統合→OpenAIで読み込み/構造化/要約→匿名化→Datasets保存")
+
+    with gr.Row():
+        in_files = gr.Files(
+            label="レジュメ類 (PDF/画像/Word/テキスト) 複数可",
+            file_count="multiple",
+            file_types=[".pdf", ".png", ".jpg", ".jpeg", ".tiff", ".bmp", ".docx", ".txt"],
+            type="binary",  # Gradio 4.44系：'binary' or 'filepath'
+        )
+        candidate_id = gr.Textbox(label="候補者ID（任意。未入力なら自動生成）")
+    notes = gr.Textbox(label="補足メモ（任意）", lines=3)
+
+    run_btn = gr.Button("実行")
+
+    with gr.Tab("構造化JSON"):
+        out_json = gr.Code(label="統合出力 (JSON)")
+
+    with gr.Tab("抽出スキル"):
+        out_skills = gr.JSON(label="スキル一覧")
+
+    with gr.Tab("品質スコア"):
+        out_score = gr.Code(label="品質評価")
+
+    with gr.Tab("要約 (300/100/1文)"):
+        out_sum_300 = gr.Textbox(label="300字要約")
+        out_sum_100 = gr.Textbox(label="100字要約")
+        out_sum_1 = gr.Textbox(label="1文要約")
+
+    with gr.Tab("匿名PDF"):
+        out_pdf = gr.File(label="匿名PDFダウンロード")
+
+    with gr.Tab("Datasets 保存ログ"):
+        out_commit = gr.Code(label="コミット情報")
+
+    run_btn.click(
+        process_resumes,
+        inputs=[in_files, candidate_id, notes],
+        outputs=[out_json, out_skills, out_score, out_sum_300, out_sum_100, out_sum_1, out_pdf, out_commit],
+    )
+
+
 if __name__ == "__main__":
-    demo.launch(server_name="0.0.0.0", server_port=7860)
+    # Hugging Face Spaces では share=True は不要／非推奨
+    demo.launch()