Docker SDK → Gradio SDK 전환

.dockerignore

@@ -1,17 +0,0 @@
-.git
-.github
-.venv
-.venv-wsl
-__pycache__
-*.pyc
-.pytest_cache
-.ruff_cache
-experiments
-notebooks
-tests
-docs
-blog
-landing
-data
-.claude
-*.egg-info

Dockerfile

@@ -1,24 +0,0 @@
-# HuggingFace Spaces — DartLab 데모
-# CPU free tier (16GB RAM), port 7860
-
-FROM python:3.12-slim
-
-WORKDIR /app
-
-# 시스템 의존성
-RUN apt-get update && apt-get install -y --no-install-recommends \
-    build-essential \
-    && rm -rf /var/lib/apt/lists/*
-
-# dartlab 설치
-RUN pip install --no-cache-dir dartlab
-
-# 샘플 데이터 프리로드 스크립트
-COPY scripts/hfPreload.py scripts/hfPreload.py
-RUN python scripts/hfPreload.py
-
-# HF Spaces 기본 포트
-EXPOSE 7860
-
-# 서버 실행
-CMD ["python", "-m", "dartlab.server"]

README.md

@@ -1,11 +1,12 @@
 ---
 title: DartLab
-emoji: 🗺️
+emoji: 📊
 colorFrom: red
 colorTo: yellow
-sdk: docker
-app_port: 7860
+sdk: gradio
+sdk_version: "5.0"
+app_file: app.py
 pinned: true
 license: mit
-short_description: "DART + EDGAR disclosure analysis — try without install"
+short_description: DART + EDGAR disclosure analysis — try without install
 ---

app.py

@@ -0,0 +1,305 @@
+"""DartLab Gradio Demo — DART/EDGAR 공시 분석."""
+
+from __future__ import annotations
+
+import gc
+import os
+import threading
+from collections import OrderedDict
+
+import gradio as gr
+
+import dartlab
+
+# ── 설정 ──────────────────────────────────────────────
+
+_MAX_CACHE = 2
+_companyCache: OrderedDict = OrderedDict()
+_HAS_AI = bool(os.environ.get("OPENAI_API_KEY"))
+
+if _HAS_AI:
+    dartlab.llm.configure(provider="openai", api_key=os.environ["OPENAI_API_KEY"])
+
+
+# ── 유틸 ──────────────────────────────────────────────
+
+
+def _toPandas(df):
+    """Polars/pandas DataFrame → pandas 변환."""
+    if df is None:
+        return None
+    if hasattr(df, "to_pandas"):
+        return df.to_pandas()
+    return df
+
+
+def _getCompany(code: str):
+    """캐시된 Company 반환, 최대 2개 유지."""
+    code = code.strip()
+    if code in _companyCache:
+        _companyCache.move_to_end(code)
+        return _companyCache[code]
+    while len(_companyCache) >= _MAX_CACHE:
+        _companyCache.popitem(last=False)
+        gc.collect()
+    c = dartlab.Company(code)
+    _companyCache[code] = c
+    return c
+
+
+# ── 프리로드 ──────────────────────────────────────────
+
+
+def _warmup():
+    """백그라운드 listing 캐시 워밍업."""
+    try:
+        dartlab.search("삼성전자")
+    except Exception:
+        pass
+
+
+threading.Thread(target=_warmup, daemon=True).start()
+
+
+# ── 핸들러: 검색 ─────────────────────────────────────
+
+
+def handleSearch(keyword: str):
+    """종목 검색."""
+    if not keyword or not keyword.strip():
+        return None
+    df = dartlab.search(keyword.strip())
+    return _toPandas(df)
+
+
+# ── 핸들러: 기업 개요 + 재무 ─────────────────────────
+
+
+def handleLoadCompany(code: str):
+    """기업 개요 로드 → (기본정보 markdown, topics 리스트, 재무 DataFrame)."""
+    if not code or not code.strip():
+        return "종목코드를 입력하세요.", gr.update(choices=[]), None
+    try:
+        c = _getCompany(code)
+    except Exception as e:
+        return f"로드 실패: {e}", gr.update(choices=[]), None
+
+    # 기본정보
+    info = f"## {c.corpName} ({c.stockCode})\n"
+    info += f"- 시장: {c.market}\n"
+    if hasattr(c, "currency") and c.currency:
+        info += f"- 통화: {c.currency}\n"
+
+    # topics
+    topics = []
+    try:
+        topics = list(c.topics) if c.topics else []
+    except Exception:
+        pass
+
+    # IS 기본 로드
+    finance = None
+    try:
+        finance = _toPandas(c.IS)
+    except Exception:
+        pass
+
+    return info, gr.update(choices=topics, value=topics[0] if topics else None), finance
+
+
+def handleFinance(code: str, sheet: str):
+    """재무제표 시트 전환."""
+    if not code or not code.strip():
+        return None
+    try:
+        c = _getCompany(code)
+    except Exception:
+        return None
+
+    lookup = {"BS": "BS", "IS": "IS", "CF": "CF", "ratios": "ratios"}
+    attr = lookup.get(sheet, "IS")
+    try:
+        result = getattr(c, attr, None)
+        return _toPandas(result)
+    except Exception:
+        return None
+
+
+# ── 핸들러: Sections ──────────────────────────────────
+
+
+def handleShow(code: str, topic: str):
+    """sections show → DataFrame 또는 Markdown."""
+    if not code or not topic:
+        return None, ""
+    try:
+        c = _getCompany(code)
+        result = c.show(topic)
+    except Exception as e:
+        return None, f"조회 실패: {e}"
+
+    if result is None:
+        return None, "데이터 없음"
+    if hasattr(result, "to_pandas"):
+        return _toPandas(result), ""
+    return None, str(result)
+
+
+# ── 핸들러: AI Chat ───────────────────────────────────
+
+
+def handleChat(message: str, history: list, code: str):
+    """AI 분석 대화."""
+    if not _HAS_AI:
+        history = history + [
+            {"role": "user", "content": message},
+            {"role": "assistant", "content": "OPENAI_API_KEY가 설정되지 않았습니다.\n\nHuggingFace Spaces Settings → Variables and secrets에서 설정하세요."},
+        ]
+        return history, ""
+
+    stockCode = code.strip() if code else None
+    if not stockCode:
+        history = history + [
+            {"role": "user", "content": message},
+            {"role": "assistant", "content": "먼저 Company 탭에서 종목을 로드하세요."},
+        ]
+        return history, ""
+
+    # ask() 호출 — raw=True로 Generator 받아서 스트리밍
+    try:
+        if stockCode:
+            query = f"{stockCode} {message}"
+        else:
+            query = message
+        answer = dartlab.ask(query, stream=False, raw=False)
+    except Exception as e:
+        answer = f"분석 실패: {e}"
+
+    history = history + [
+        {"role": "user", "content": message},
+        {"role": "assistant", "content": answer if answer else "응답 없음"},
+    ]
+    return history, ""
+
+
+# ── UI ────────────────────────────────────────────────
+
+_CSS = """
+.main-title { text-align: center; margin-bottom: 0.5em; }
+.subtitle { text-align: center; color: #666; margin-top: 0; }
+"""
+
+with gr.Blocks(
+    title="DartLab — DART/EDGAR 공시 분석",
+    theme=gr.themes.Soft(),
+    css=_CSS,
+) as demo:
+    gr.Markdown("# DartLab", elem_classes="main-title")
+    gr.Markdown(
+        "DART 전자공시 + EDGAR — 종목코드 하나로 기업 분석",
+        elem_classes="subtitle",
+    )
+
+    # 공유 state
+    codeState = gr.State("")
+
+    with gr.Tabs():
+        # ── Tab 1: 검색 ──
+        with gr.Tab("Search"):
+            with gr.Row():
+                searchInput = gr.Textbox(
+                    label="종목 검색",
+                    placeholder="삼성전자, AAPL, 005930 ...",
+                    scale=4,
+                )
+                searchBtn = gr.Button("검색", scale=1, variant="primary")
+            searchResult = gr.DataFrame(label="검색 결과", interactive=False)
+
+            with gr.Row():
+                selectedCode = gr.Textbox(
+                    label="종목코드 입력 후 분석",
+                    placeholder="005930",
+                    scale=4,
+                )
+                loadBtn = gr.Button("분석하기", scale=1, variant="primary")
+
+            searchBtn.click(handleSearch, inputs=searchInput, outputs=searchResult)
+            searchInput.submit(handleSearch, inputs=searchInput, outputs=searchResult)
+
+        # ── Tab 2: 기업 + 재무 ──
+        with gr.Tab("Company"):
+            companyInfo = gr.Markdown("종목코드를 입력하고 '분석하기'를 클릭하세요.")
+
+            with gr.Row():
+                sheetSelect = gr.Dropdown(
+                    choices=["IS", "BS", "CF", "ratios"],
+                    value="IS",
+                    label="재무제표",
+                    scale=1,
+                )
+            financeTable = gr.DataFrame(label="재무제표", interactive=False)
+
+            sheetSelect.change(
+                handleFinance,
+                inputs=[codeState, sheetSelect],
+                outputs=financeTable,
+            )
+
+        # ── Tab 3: Sections ──
+        with gr.Tab("Sections"):
+            gr.Markdown("topic × period 수평화 — 기업의 전체 지도")
+            topicSelect = gr.Dropdown(
+                choices=[],
+                label="Topic",
+                interactive=True,
+            )
+            sectionTable = gr.DataFrame(label="Section 데이터", interactive=False)
+            sectionText = gr.Markdown("")
+
+            topicSelect.change(
+                handleShow,
+                inputs=[codeState, topicSelect],
+                outputs=[sectionTable, sectionText],
+            )
+
+        # ── Tab 4: AI Chat ──
+        with gr.Tab("AI Chat"):
+            if not _HAS_AI:
+                gr.Markdown(
+                    "**AI 분석을 사용하려면** HuggingFace Spaces Settings → "
+                    "Variables and secrets에서 `OPENAI_API_KEY`를 설정하세요."
+                )
+            chatbot = gr.Chatbot(label="AI 분석", type="messages", height=500)
+            with gr.Row():
+                chatInput = gr.Textbox(
+                    label="질문",
+                    placeholder="재무건전성 분석해줘, 배당 이상 징후 찾아줘 ...",
+                    scale=5,
+                )
+                chatBtn = gr.Button("전송", scale=1, variant="primary")
+
+            chatBtn.click(
+                handleChat,
+                inputs=[chatInput, chatbot, codeState],
+                outputs=[chatbot, chatInput],
+            )
+            chatInput.submit(
+                handleChat,
+                inputs=[chatInput, chatbot, codeState],
+                outputs=[chatbot, chatInput],
+            )
+
+    # ── 분석하기 버튼 → Company 탭 로드 ──
+    loadBtn.click(
+        lambda code: code.strip(),
+        inputs=selectedCode,
+        outputs=codeState,
+    ).then(
+        handleLoadCompany,
+        inputs=selectedCode,
+        outputs=[companyInfo, topicSelect, financeTable],
+    )
+
+
+if __name__ == "__main__":
+    demo.launch(server_name="0.0.0.0", server_port=7860)

requirements.txt

@@ -0,0 +1,2 @@
+dartlab>=0.7.8
+gradio>=5.0,<6

scripts/hfPreload.py

@@ -1,31 +0,0 @@
-"""HuggingFace Spaces Docker 빌드 시 샘플 데이터 프리로드.
-
-삼성전자(005930)와 AAPL 데이터를 미리 다운로드하여
-cold start 시간을 줄인다.
-"""
-
-from __future__ import annotations
-
-import os
-import sys
-
-# Docker 빌드 환경에서는 dartlab이 이미 pip install된 상태
-sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", "src"))
-
-SAMPLES = ["005930", "AAPL"]
-
-
-def main() -> None:
-    from dartlab.core.dataLoader import download
-
-    for code in SAMPLES:
-        print(f"  preloading {code}...")
-        try:
-            download(code)
-            print(f"  {code} done")
-        except Exception as e:  # noqa: BLE001
-            print(f"  {code} skipped: {e}")
-
-
-if __name__ == "__main__":
-    main()