update: upload fix

2026-03-10 14:25:21 +08:00
parent a3d928e697
commit fd2b574d5a
19 changed files with 575 additions and 156 deletions
--- a/.env.example
+++ b/.env.example
@@ -1,7 +1,14 @@
 DATABASE_URL=sqlite+aiosqlite:///./fund_tracer.db
-LLM_PROVIDER=openai

-# Optional: choose model names
+# --- OCR model (screenshot -> transactions) ---
+OCR_PROVIDER=openai
+OCR_MODEL=
+
+# --- Inference model (report generation, reasoning) ---
+INFERENCE_PROVIDER=openai
+INFERENCE_MODEL=
+
+# Provider default model names
 OPENAI_MODEL=gpt-4o
 ANTHROPIC_MODEL=claude-3-5-sonnet-20241022
 DEEPSEEK_MODEL=deepseek-chat
--- a/backend/.env.example
+++ b/backend/.env.example
@@ -0,0 +1,17 @@
+DATABASE_URL=sqlite+aiosqlite:///./fund_tracer.db
+LLM_PROVIDER=openai
+
+# Optional: choose model names
+OPENAI_MODEL=gpt-4o
+ANTHROPIC_MODEL=claude-3-5-sonnet-20241022
+DEEPSEEK_MODEL=deepseek-chat
+CUSTOM_OPENAI_MODEL=gpt-4o-mini
+
+# Custom OpenAI-compatible provider
+CUSTOM_OPENAI_BASE_URL=
+CUSTOM_OPENAI_API_KEY=
+
+# API keys
+OPENAI_API_KEY=
+ANTHROPIC_API_KEY=
+DEEPSEEK_API_KEY=
--- a/backend/app/api/screenshots.py
+++ b/backend/app/api/screenshots.py
@@ -1,6 +1,7 @@
 """Screenshot upload and extraction API."""

 import uuid
+from datetime import datetime
 from pathlib import Path

 from fastapi import APIRouter, Depends, HTTPException, UploadFile, File
@@ -84,18 +85,105 @@ async def extract_transactions(
    if not full_path.exists():
        raise HTTPException(status_code=404, detail="File not found on disk")
    image_bytes = full_path.read_bytes()
+    started_at = datetime.utcnow()
+    # 每次开始新一轮识别都重置计时，确保耗时是“本次分析”而不是历史累计
+    screenshot.started_at = started_at
+    screenshot.finished_at = None
+    screenshot.duration_ms = None
+    screenshot.error_message = None
+    screenshot.progress_step = "starting"
+    screenshot.progress_percent = 0
+    screenshot.progress_detail = "准备开始识别"
+    await db.commit()
+
+    async def update_progress(step: str, percent: int, detail: str):
+        screenshot.status = "processing"
+        screenshot.progress_step = step
+        screenshot.progress_percent = percent
+        screenshot.progress_detail = detail
+        await db.commit()
+
    try:
-        transactions = await extract_and_save(case_id, screenshot_id, image_bytes)
+        await update_progress("file_loaded", 10, "截图读取完成")
+        transactions = await extract_and_save(
+            case_id,
+            screenshot_id,
+            image_bytes,
+            progress_hook=update_progress,
+        )
    except Exception as e:
+        error_detail = _classify_error(e)
        r = await db.execute(select(Screenshot).where(Screenshot.id == screenshot_id))
        sc = r.scalar_one_or_none()
        if sc:
            sc.status = "failed"
+            sc.progress_step = "failed"
+            sc.progress_percent = 100
+            sc.progress_detail = "识别失败"
+            sc.finished_at = datetime.utcnow()
+            if sc.started_at:
+                sc.duration_ms = int((sc.finished_at - sc.started_at).total_seconds() * 1000)
+            sc.error_message = error_detail
            await db.commit()
-        raise HTTPException(status_code=502, detail=f"Extraction failed: {e!s}")
+        raise HTTPException(status_code=502, detail=error_detail)
    r = await db.execute(select(Screenshot).where(Screenshot.id == screenshot_id))
    sc = r.scalar_one_or_none()
    if sc:
        sc.status = "extracted"
+        sc.progress_step = "completed"
+        sc.progress_percent = 100
+        sc.progress_detail = "识别完成"
+        sc.finished_at = datetime.utcnow()
+        if sc.started_at:
+            sc.duration_ms = int((sc.finished_at - sc.started_at).total_seconds() * 1000)
+        sc.error_message = None
        await db.commit()
    return TransactionListResponse(items=transactions)
+
+
+def _classify_error(e: Exception) -> str:
+    """Produce a human-readable, categorized error message."""
+    name = type(e).__name__
+    msg = str(e)
+
+    if isinstance(e, ValueError):
+        return f"配置错误: {msg}"
+
+    # OpenAI SDK errors
+    try:
+        from openai import AuthenticationError, RateLimitError, APIConnectionError, BadRequestError, APIStatusError, APITimeoutError
+        if isinstance(e, AuthenticationError):
+            return f"API Key 无效或已过期 ({name}): {msg}"
+        if isinstance(e, RateLimitError):
+            return f"API 调用频率超限，请稍后重试 ({name}): {msg}"
+        if isinstance(e, APITimeoutError):
+            return f"模型服务响应超时，请检查 BaseURL/模型可用性或稍后重试 ({name}): {msg}"
+        if isinstance(e, APIConnectionError):
+            return f"无法连接到模型服务，请检查网络或 BaseURL ({name}): {msg}"
+        if isinstance(e, BadRequestError):
+            return f"请求被模型服务拒绝（可能模型名错误或不支持图片） ({name}): {msg}"
+        if isinstance(e, APIStatusError):
+            return f"模型服务返回错误 (HTTP {e.status_code}): {msg}"
+    except ImportError:
+        pass
+
+    # Anthropic SDK errors
+    try:
+        from anthropic import AuthenticationError as AnthAuthError, RateLimitError as AnthRateError
+        from anthropic import APIConnectionError as AnthConnError, BadRequestError as AnthBadReq
+        if isinstance(e, AnthAuthError):
+            return f"Anthropic API Key 无效或已过期: {msg}"
+        if isinstance(e, AnthRateError):
+            return f"Anthropic API 调用频率超限: {msg}"
+        if isinstance(e, AnthConnError):
+            return f"无法连接到 Anthropic 服务: {msg}"
+        if isinstance(e, AnthBadReq):
+            return f"Anthropic 请求被拒绝: {msg}"
+    except ImportError:
+        pass
+
+    # Connection / network
+    if "connect" in msg.lower() or "timeout" in msg.lower():
+        return f"网络连接失败或超时: {msg}"
+
+    return f"识别失败 ({name}): {msg}"
--- a/backend/app/api/settings.py
+++ b/backend/app/api/settings.py
@@ -9,7 +9,10 @@ router = APIRouter()


 class SettingsUpdate(BaseModel):
-    llm_provider: str | None = None
+    ocr_provider: str | None = None
+    ocr_model: str | None = None
+    inference_provider: str | None = None
+    inference_model: str | None = None
    openai_api_key: str | None = None
    anthropic_api_key: str | None = None
    deepseek_api_key: str | None = None
--- a/backend/app/config.py
+++ b/backend/app/config.py
@@ -20,18 +20,30 @@ class Settings(BaseSettings):
    max_upload_size_mb: int = 20
    allowed_extensions: set[str] = {"png", "jpg", "jpeg", "webp"}

-    # LLM
-    llm_provider: str = "openai"  # openai | anthropic | deepseek | custom_openai
+    # --- OCR (vision) model ---
+    ocr_provider: str = "openai"  # openai | anthropic | deepseek | custom_openai
+    ocr_model: str | None = None  # if None, falls back to provider default
+
+    # --- Inference (reasoning) model ---
+    inference_provider: str = "openai"
+    inference_model: str | None = None
+
+    # --- Provider credentials (shared between OCR and inference) ---
    openai_api_key: str | None = None
    anthropic_api_key: str | None = None
    deepseek_api_key: str | None = None
    custom_openai_api_key: str | None = None
    custom_openai_base_url: str | None = None
+
+    # Provider default model names (used when ocr_model / inference_model is None)
    openai_model: str = "gpt-4o"
    anthropic_model: str = "claude-3-5-sonnet-20241022"
    deepseek_model: str = "deepseek-chat"
    custom_openai_model: str = "gpt-4o-mini"

+    # Legacy compat: llm_provider maps to ocr_provider on load
+    llm_provider: str | None = None
+
    class Config:
        env_file = ".env"
        env_file_encoding = "utf-8"
@@ -40,8 +52,24 @@ class Settings(BaseSettings):

 _runtime_overrides: dict[str, str | None] = {}

+_ALLOWED_RUNTIME_KEYS = {
+    "ocr_provider",
+    "ocr_model",
+    "inference_provider",
+    "inference_model",
+    "openai_api_key",
+    "anthropic_api_key",
+    "deepseek_api_key",
+    "custom_openai_api_key",
+    "custom_openai_base_url",
+    "custom_openai_model",
+}
+

 def _apply_overrides(settings: Settings) -> Settings:
+    # Legacy: if llm_provider is set but ocr_provider is default, use it
+    if settings.llm_provider and settings.ocr_provider == "openai":
+        settings.ocr_provider = settings.llm_provider
    for key, value in _runtime_overrides.items():
        if hasattr(settings, key):
            setattr(settings, key, value)
@@ -53,19 +81,32 @@ def get_settings() -> Settings:
    return _apply_overrides(Settings())


-def update_runtime_settings(payload: dict[str, str | None]) -> Settings:
-    """Update runtime settings and refresh cached Settings object."""
-    allowed = {
-        "llm_provider",
-        "openai_api_key",
-        "anthropic_api_key",
-        "deepseek_api_key",
-        "custom_openai_api_key",
-        "custom_openai_base_url",
-        "custom_openai_model",
+def _resolve_model(provider: str, explicit_model: str | None, settings: Settings) -> str:
+    """Return the model name to use for a given provider."""
+    if explicit_model:
+        return explicit_model
+    defaults = {
+        "openai": settings.openai_model,
+        "anthropic": settings.anthropic_model,
+        "deepseek": settings.deepseek_model,
+        "custom_openai": settings.custom_openai_model,
    }
+    return defaults.get(provider, settings.openai_model)
+
+
+def get_ocr_model() -> str:
+    s = get_settings()
+    return _resolve_model(s.ocr_provider, s.ocr_model, s)
+
+
+def get_inference_model() -> str:
+    s = get_settings()
+    return _resolve_model(s.inference_provider, s.inference_model, s)
+
+
+def update_runtime_settings(payload: dict[str, str | None]) -> Settings:
    for key, value in payload.items():
-        if key in allowed:
+        if key in _ALLOWED_RUNTIME_KEYS:
            _runtime_overrides[key] = value
    get_settings.cache_clear()
    return get_settings()
@@ -74,9 +115,12 @@ def update_runtime_settings(payload: dict[str, str | None]) -> Settings:
 def public_settings() -> dict:
    s = get_settings()
    return {
-        "llm_provider": s.llm_provider,
+        "ocr_provider": s.ocr_provider,
+        "ocr_model": get_ocr_model(),
+        "inference_provider": s.inference_provider,
+        "inference_model": get_inference_model(),
        "providers": ["openai", "anthropic", "deepseek", "custom_openai"],
-        "models": {
+        "provider_defaults": {
            "openai": s.openai_model,
            "anthropic": s.anthropic_model,
            "deepseek": s.deepseek_model,
--- a/backend/app/models/screenshot.py
+++ b/backend/app/models/screenshot.py
@@ -2,7 +2,7 @@

 from __future__ import annotations
 from datetime import datetime
-from sqlalchemy import String, DateTime, ForeignKey
+from sqlalchemy import String, Text, DateTime, ForeignKey
 from sqlalchemy.orm import Mapped, mapped_column, relationship

 from app.models.database import Base
@@ -15,7 +15,14 @@ class Screenshot(Base):
    case_id: Mapped[int] = mapped_column(ForeignKey("cases.id", ondelete="CASCADE"), index=True)
    filename: Mapped[str] = mapped_column(String(255))
    file_path: Mapped[str] = mapped_column(String(512))
-    status: Mapped[str] = mapped_column(String(32), default="pending")  # pending | extracted | failed
+    status: Mapped[str] = mapped_column(String(32), default="pending")  # pending | processing | extracted | failed
+    progress_step: Mapped[str | None] = mapped_column(String(64), nullable=True, default=None)
+    progress_percent: Mapped[int] = mapped_column(default=0)
+    progress_detail: Mapped[str | None] = mapped_column(String(255), nullable=True, default=None)
+    started_at: Mapped[datetime | None] = mapped_column(DateTime, nullable=True, default=None)
+    finished_at: Mapped[datetime | None] = mapped_column(DateTime, nullable=True, default=None)
+    duration_ms: Mapped[int | None] = mapped_column(nullable=True, default=None)
+    error_message: Mapped[str | None] = mapped_column(Text, nullable=True, default=None)
    created_at: Mapped[datetime] = mapped_column(DateTime, default=datetime.utcnow)

    case: Mapped["Case"] = relationship("Case", back_populates="screenshots")
--- a/backend/app/schemas/screenshot.py
+++ b/backend/app/schemas/screenshot.py
@@ -11,6 +11,13 @@ class ScreenshotResponse(BaseModel):
    filename: str
    file_path: str
    status: str
+    progress_step: str | None = None
+    progress_percent: int = 0
+    progress_detail: str | None = None
+    started_at: datetime | None = None
+    finished_at: datetime | None = None
+    duration_ms: int | None = None
+    error_message: str | None = None
    created_at: datetime


--- a/backend/app/services/extractor.py
+++ b/backend/app/services/extractor.py
@@ -1,7 +1,7 @@
 """Transaction data extraction: LLM Vision + persistence."""

 from app.models import Transaction
-from app.models.database import async_session_maker
+import app.models.database as db_module
 from app.schemas.transaction import TransactionExtractItem, TransactionResponse
 from app.services.llm import get_llm_provider

@@ -10,15 +10,26 @@ async def extract_and_save(
    case_id: int,
    screenshot_id: int,
    image_bytes: bytes,
+    progress_hook=None,
 ) -> list[TransactionResponse]:
    """
    Run vision extraction on image and persist transactions to DB.
    Returns list of created transactions; low-confidence items are still saved but flagged.
    """
+    if progress_hook:
+        await progress_hook("init", 5, "初始化识别上下文")
    provider = get_llm_provider()
+    if progress_hook:
+        await progress_hook("provider_ready", 15, f"已加载模型提供商: {type(provider).__name__}")
+    if progress_hook:
+        await progress_hook("calling_model", 35, "调用视觉模型识别截图中交易")
    items: list[TransactionExtractItem] = await provider.extract_from_image(image_bytes)
+    if progress_hook:
+        await progress_hook("model_returned", 70, f"模型返回 {len(items)} 条交易")
    results: list[TransactionResponse] = []
-    async with async_session_maker() as session:
+    async with db_module.async_session_maker() as session:
+        if progress_hook:
+            await progress_hook("db_writing", 85, "写入交易记录到数据库")
        for it in items:
            t = Transaction(
                case_id=case_id,
@@ -39,4 +50,6 @@ async def extract_and_save(
            await session.flush()
            results.append(TransactionResponse.model_validate(t))
        await session.commit()
+    if progress_hook:
+        await progress_hook("completed", 100, "识别完成")
    return results
--- a/backend/app/services/llm/base.py
+++ b/backend/app/services/llm/base.py
@@ -6,13 +6,16 @@ from app.schemas.transaction import TransactionExtractItem


 class BaseLLMProvider(ABC):
-    """Abstract base for LLM vision providers. Each provider implements extract_from_image."""
+    """Abstract base for LLM providers. Supports optional model override."""
+
+    def __init__(self, model_override: str | None = None):
+        self._model_override = model_override

    @abstractmethod
    async def extract_from_image(self, image_bytes: bytes) -> list[TransactionExtractItem]:
-        """
-        Analyze a billing screenshot and return structured transaction list.
-        :param image_bytes: Raw image file content (PNG/JPEG)
-        :return: List of extracted transactions (may be empty or partial on failure)
-        """
+        pass
+
+    @abstractmethod
+    async def chat(self, system: str, user: str) -> str:
+        """Plain text chat (for inference/reasoning tasks like report generation)."""
        pass
--- a/backend/app/services/llm/claude_vision.py
+++ b/backend/app/services/llm/claude_vision.py
@@ -1,8 +1,6 @@
 """Anthropic Claude Vision provider."""

 import base64
-import json
-import re
 from anthropic import AsyncAnthropic

 from app.config import get_settings
@@ -13,6 +11,9 @@ from app.services.llm.openai_vision import _parse_json_array


 class ClaudeVisionProvider(BaseLLMProvider):
+    def _get_model(self) -> str:
+        return self._model_override or get_settings().anthropic_model
+
    async def extract_from_image(self, image_bytes: bytes) -> list[TransactionExtractItem]:
        settings = get_settings()
        if not settings.anthropic_api_key:
@@ -20,14 +21,12 @@ class ClaudeVisionProvider(BaseLLMProvider):
        client = AsyncAnthropic(api_key=settings.anthropic_api_key)
        b64 = base64.standard_b64encode(image_bytes).decode("ascii")
        messages = get_extract_messages(b64)
-        # Claude API: user message with content block list
        user_content = messages[1]["content"]
        content_blocks = []
        for block in user_content:
            if block["type"] == "text":
                content_blocks.append({"type": "text", "text": block["text"]})
            elif block["type"] == "image_url":
-                # Claude expects base64 without data URL prefix
                content_blocks.append({
                    "type": "image",
                    "source": {
@@ -37,7 +36,7 @@ class ClaudeVisionProvider(BaseLLMProvider):
                    },
                })
        response = await client.messages.create(
-            model=settings.anthropic_model,
+            model=self._get_model(),
            max_tokens=4096,
            system=messages[0]["content"],
            messages=[{"role": "user", "content": content_blocks}],
@@ -47,3 +46,20 @@ class ClaudeVisionProvider(BaseLLMProvider):
            if hasattr(block, "text"):
                text += block.text
        return _parse_json_array(text or "[]")
+
+    async def chat(self, system: str, user: str) -> str:
+        settings = get_settings()
+        if not settings.anthropic_api_key:
+            raise ValueError("ANTHROPIC_API_KEY is not set")
+        client = AsyncAnthropic(api_key=settings.anthropic_api_key)
+        response = await client.messages.create(
+            model=self._get_model(),
+            max_tokens=4096,
+            system=system,
+            messages=[{"role": "user", "content": user}],
+        )
+        text = ""
+        for block in response.content:
+            if hasattr(block, "text"):
+                text += block.text
+        return text or ""
--- a/backend/app/services/llm/custom_openai_vision.py
+++ b/backend/app/services/llm/custom_openai_vision.py
@@ -11,22 +11,44 @@ from app.services.llm.openai_vision import _parse_json_array


 class CustomOpenAICompatibleProvider(BaseLLMProvider):
-    async def extract_from_image(self, image_bytes: bytes) -> list[TransactionExtractItem]:
+    def _get_client(self) -> AsyncOpenAI:
        settings = get_settings()
        if not settings.custom_openai_api_key:
            raise ValueError("CUSTOM_OPENAI_API_KEY is not set")
        if not settings.custom_openai_base_url:
            raise ValueError("CUSTOM_OPENAI_BASE_URL is not set")
-        client = AsyncOpenAI(
+        return AsyncOpenAI(
            api_key=settings.custom_openai_api_key,
            base_url=settings.custom_openai_base_url,
+            timeout=45.0,
+            max_retries=1,
        )
+
+    def _get_model(self) -> str:
+        return self._model_override or get_settings().custom_openai_model
+
+    async def extract_from_image(self, image_bytes: bytes) -> list[TransactionExtractItem]:
+        client = self._get_client()
        b64 = base64.standard_b64encode(image_bytes).decode("ascii")
        messages = get_extract_messages(b64)
        response = await client.chat.completions.create(
-            model=settings.custom_openai_model,
+            model=self._get_model(),
            messages=messages,
            max_tokens=4096,
+            timeout=45.0,
        )
        text = response.choices[0].message.content or "[]"
        return _parse_json_array(text)
+
+    async def chat(self, system: str, user: str) -> str:
+        client = self._get_client()
+        response = await client.chat.completions.create(
+            model=self._get_model(),
+            messages=[
+                {"role": "system", "content": system},
+                {"role": "user", "content": user},
+            ],
+            max_tokens=4096,
+            timeout=45.0,
+        )
+        return response.choices[0].message.content or ""
--- a/backend/app/services/llm/deepseek_vision.py
+++ b/backend/app/services/llm/deepseek_vision.py
@@ -9,26 +9,39 @@ from app.services.llm.base import BaseLLMProvider
 from app.prompts.extract_transaction import get_extract_messages
 from app.services.llm.openai_vision import _parse_json_array

-
-# DeepSeek vision endpoint (OpenAI-compatible)
 DEEPSEEK_BASE = "https://api.deepseek.com"


 class DeepSeekVisionProvider(BaseLLMProvider):
-    async def extract_from_image(self, image_bytes: bytes) -> list[TransactionExtractItem]:
+    def _get_client(self) -> AsyncOpenAI:
        settings = get_settings()
        if not settings.deepseek_api_key:
            raise ValueError("DEEPSEEK_API_KEY is not set")
-        client = AsyncOpenAI(
-            api_key=settings.deepseek_api_key,
-            base_url=DEEPSEEK_BASE,
-        )
+        return AsyncOpenAI(api_key=settings.deepseek_api_key, base_url=DEEPSEEK_BASE)
+
+    def _get_model(self) -> str:
+        return self._model_override or get_settings().deepseek_model
+
+    async def extract_from_image(self, image_bytes: bytes) -> list[TransactionExtractItem]:
+        client = self._get_client()
        b64 = base64.standard_b64encode(image_bytes).decode("ascii")
        messages = get_extract_messages(b64)
        response = await client.chat.completions.create(
-            model=settings.deepseek_model,
+            model=self._get_model(),
            messages=messages,
            max_tokens=4096,
        )
        text = response.choices[0].message.content or "[]"
        return _parse_json_array(text)
+
+    async def chat(self, system: str, user: str) -> str:
+        client = self._get_client()
+        response = await client.chat.completions.create(
+            model=self._get_model(),
+            messages=[
+                {"role": "system", "content": system},
+                {"role": "user", "content": user},
+            ],
+            max_tokens=4096,
+        )
+        return response.choices[0].message.content or ""
--- a/backend/app/services/llm/openai_vision.py
+++ b/backend/app/services/llm/openai_vision.py
@@ -12,6 +12,9 @@ from app.prompts.extract_transaction import get_extract_messages


 class OpenAIVisionProvider(BaseLLMProvider):
+    def _get_model(self) -> str:
+        return self._model_override or get_settings().openai_model
+
    async def extract_from_image(self, image_bytes: bytes) -> list[TransactionExtractItem]:
        settings = get_settings()
        if not settings.openai_api_key:
@@ -20,18 +23,32 @@ class OpenAIVisionProvider(BaseLLMProvider):
        b64 = base64.standard_b64encode(image_bytes).decode("ascii")
        messages = get_extract_messages(b64)
        response = await client.chat.completions.create(
-            model=settings.openai_model,
+            model=self._get_model(),
            messages=messages,
            max_tokens=4096,
        )
        text = response.choices[0].message.content or "[]"
        return _parse_json_array(text)

+    async def chat(self, system: str, user: str) -> str:
+        settings = get_settings()
+        if not settings.openai_api_key:
+            raise ValueError("OPENAI_API_KEY is not set")
+        client = AsyncOpenAI(api_key=settings.openai_api_key)
+        response = await client.chat.completions.create(
+            model=self._get_model(),
+            messages=[
+                {"role": "system", "content": system},
+                {"role": "user", "content": user},
+            ],
+            max_tokens=4096,
+        )
+        return response.choices[0].message.content or ""
+

 def _parse_json_array(text: str) -> list[TransactionExtractItem]:
    """Parse LLM response into list of TransactionExtractItem. Tolerates markdown and extra text."""
    text = text.strip()
-    # Remove optional markdown code block
    if text.startswith("```"):
        text = re.sub(r"^```(?:json)?\s*", "", text)
        text = re.sub(r"\s*```\s*$", "", text)
@@ -46,7 +63,6 @@ def _parse_json_array(text: str) -> list[TransactionExtractItem]:
        if not isinstance(item, dict):
            continue
        try:
-            # Normalize transaction_time: allow string -> datetime
            if isinstance(item.get("transaction_time"), str) and item["transaction_time"]:
                from dateutil import parser as date_parser
                item["transaction_time"] = date_parser.isoparse(item["transaction_time"])
--- a/backend/app/services/llm/router.py
+++ b/backend/app/services/llm/router.py
@@ -1,6 +1,6 @@
-"""LLM provider factory - returns provider by config."""
+"""LLM provider factory - returns provider by config, split by role (ocr / inference)."""

-from app.config import get_settings
+from app.config import get_settings, get_ocr_model, get_inference_model
 from app.services.llm.base import BaseLLMProvider
 from app.services.llm.openai_vision import OpenAIVisionProvider
 from app.services.llm.claude_vision import ClaudeVisionProvider
@@ -8,15 +8,25 @@ from app.services.llm.deepseek_vision import DeepSeekVisionProvider
 from app.services.llm.custom_openai_vision import CustomOpenAICompatibleProvider


-def get_llm_provider() -> BaseLLMProvider:
+def _make_provider(provider_name: str, model_override: str | None = None) -> BaseLLMProvider:
+    name = (provider_name or "openai").lower()
+    if name == "openai":
+        return OpenAIVisionProvider(model_override=model_override)
+    if name == "anthropic":
+        return ClaudeVisionProvider(model_override=model_override)
+    if name == "deepseek":
+        return DeepSeekVisionProvider(model_override=model_override)
+    if name == "custom_openai":
+        return CustomOpenAICompatibleProvider(model_override=model_override)
+    return OpenAIVisionProvider(model_override=model_override)
+
+
+def get_llm_provider(role: str = "ocr") -> BaseLLMProvider:
+    """
+    role="ocr"       -> uses ocr_provider + ocr_model
+    role="inference"  -> uses inference_provider + inference_model
+    """
    settings = get_settings()
-    provider = (settings.llm_provider or "openai").lower()
-    if provider == "openai":
-        return OpenAIVisionProvider()
-    if provider == "anthropic":
-        return ClaudeVisionProvider()
-    if provider == "deepseek":
-        return DeepSeekVisionProvider()
-    if provider == "custom_openai":
-        return CustomOpenAICompatibleProvider()
-    return OpenAIVisionProvider()
+    if role == "inference":
+        return _make_provider(settings.inference_provider, get_inference_model())
+    return _make_provider(settings.ocr_provider, get_ocr_model())
--- a/frontend/src/App.tsx
+++ b/frontend/src/App.tsx
@@ -8,7 +8,17 @@ import Settings from "./pages/Settings";

 function App() {
  return (
-    <ConfigProvider locale={zhCN}>
+    <ConfigProvider
+      locale={zhCN}
+      theme={{
+        token: {
+          fontSize: 16,
+          fontSizeSM: 14,
+          fontSizeLG: 18,
+          lineHeight: 1.6,
+        },
+      }}
+    >
      <Routes>
        <Route element={<AppLayout />}>
          <Route path="/" element={<CaseList />} />
--- a/frontend/src/components/ScreenshotUploader.tsx
+++ b/frontend/src/components/ScreenshotUploader.tsx
@@ -1,6 +1,8 @@
-import { useState, useEffect } from "react";
-import { Upload, List, Button, Card, Tag, message } from "antd";
-import { InboxOutlined, ThunderboltOutlined } from "@ant-design/icons";
+import { useEffect, useMemo, useState } from "react";
+import type { Key } from "react";
+import { Upload, Table, Button, Tag, Alert, message, Space, Progress } from "antd";
+import type { ColumnsType } from "antd/es/table";
+import { InboxOutlined, ThunderboltOutlined, ReloadOutlined } from "@ant-design/icons";
 import { api, type ScreenshotItem } from "../services/api";

 const { Dragger } = Upload;
@@ -10,10 +12,24 @@ interface Props {
  onExtracted?: () => void;
 }

+function statusTag(item: ScreenshotItem) {
+  if (item.status === "extracted") return <Tag color="green">已识别</Tag>;
+  if (item.status === "processing") return <Tag color="blue">识别中</Tag>;
+  if (item.status === "failed") return <Tag color="red">失败</Tag>;
+  return <Tag>待识别</Tag>;
+}
+
+function formatDuration(ms: number | null) {
+  if (ms == null) return "-";
+  if (ms < 1000) return `${ms} ms`;
+  return `${(ms / 1000).toFixed(2)} s`;
+}
+
 export default function ScreenshotUploader({ caseId, onExtracted }: Props) {
  const [screenshots, setScreenshots] = useState<ScreenshotItem[]>([]);
  const [loading, setLoading] = useState(false);
-  const [extractingId, setExtractingId] = useState<number | null>(null);
+  const [extractingIds, setExtractingIds] = useState<number[]>([]);
+  const [selectedRowKeys, setSelectedRowKeys] = useState<Key[]>([]);

  const loadScreenshots = async () => {
    try {
@@ -35,73 +51,168 @@ export default function ScreenshotUploader({ caseId, onExtracted }: Props) {
    } finally {
      setLoading(false);
    }
-    return false; // prevent default upload
+    return false;
  };

-  const handleExtract = async (screenshotId: number) => {
-    setExtractingId(screenshotId);
+  const handleExtractSingle = async (screenshotId: number) => {
+    setExtractingIds((prev) => Array.from(new Set([...prev, screenshotId])));
    try {
      await api.screenshots.extract(caseId, screenshotId);
-      message.success("识别完成");
      await loadScreenshots();
      onExtracted?.();
+      message.success(`截图 ${screenshotId} 识别完成`);
    } catch (e: unknown) {
-      const msg = e && typeof e === "object" && "response" in e
+      const detail =
+        e && typeof e === "object" && "response" in e
          ? (e as { response?: { data?: { detail?: string } } }).response?.data?.detail
-        : "识别失败";
-      message.error(msg || "识别失败");
+          : undefined;
+      message.error(detail || `截图 ${screenshotId} 识别失败`);
+      await loadScreenshots();
    } finally {
-      setExtractingId(null);
+      setExtractingIds((prev) => prev.filter((id) => id !== screenshotId));
    }
  };

+  const handleBatchExtract = async () => {
+    const ids = selectedRowKeys.map((k) => Number(k)).filter((id) => Number.isFinite(id));
+    if (!ids.length) {
+      message.warning("请先勾选要识别的截图");
+      return;
+    }
+    setExtractingIds((prev) => Array.from(new Set([...prev, ...ids])));
+    const started = Date.now();
+    let ok = 0;
+    let fail = 0;
+    for (const id of ids) {
+      try {
+        await api.screenshots.extract(caseId, id);
+        ok += 1;
+      } catch {
+        fail += 1;
+      }
+      await loadScreenshots();
+    }
+    setExtractingIds((prev) => prev.filter((id) => !ids.includes(id)));
+    onExtracted?.();
+    const elapsed = Date.now() - started;
+    message.info(`批量识别结束：成功 ${ok}，失败 ${fail}，总耗时 ${(elapsed / 1000).toFixed(2)} s`);
+  };
+
  useEffect(() => {
-    if (caseId) loadScreenshots();
+    if (!caseId) return;
+    loadScreenshots();
+    const timer = window.setInterval(loadScreenshots, 1500);
+    return () => window.clearInterval(timer);
  }, [caseId]);

+  const rowSelection = {
+    selectedRowKeys,
+    onChange: (keys: Key[]) => setSelectedRowKeys(keys),
+    getCheckboxProps: (record: ScreenshotItem) => ({
+      disabled: record.status === "processing",
+    }),
+  };
+
+  const columns: ColumnsType<ScreenshotItem> = useMemo(() => [
+    {
+      title: "截图",
+      dataIndex: "filename",
+      key: "filename",
+      render: (v: string) => (
+        <span style={{ display: "block", maxWidth: 260, overflow: "hidden", textOverflow: "ellipsis", whiteSpace: "nowrap" }} title={v}>
+          {v}
+        </span>
+      ),
+    },
+    {
+      title: "状态",
+      key: "status",
+      width: 110,
+      render: (_, r) => statusTag(r),
+    },
+    {
+      title: "识别进度",
+      key: "progress",
+      width: 280,
+      render: (_, r) => (
+        <div>
+          <Progress percent={r.progress_percent || 0} size="small" status={r.status === "failed" ? "exception" : r.status === "extracted" ? "success" : "active"} />
+          <div style={{ fontSize: 12, color: "#666" }}>{r.progress_detail || "-"}</div>
+          {r.progress_step && <div style={{ fontSize: 12, color: "#999" }}>步骤: {r.progress_step}</div>}
+        </div>
+      ),
+    },
+    {
+      title: "耗时",
+      key: "duration",
+      width: 120,
+      render: (_, r) => formatDuration(r.duration_ms),
+    },
+    {
+      title: "错误信息",
+      key: "error_message",
+      render: (_, r) =>
+        r.status === "failed" && r.error_message ? (
+          <Alert type="error" showIcon message={r.error_message} />
+        ) : (
+          "-"
+        ),
+    },
+    {
+      title: "操作",
+      key: "action",
+      width: 120,
+      render: (_, r) => (
+        (r.status === "pending" || r.status === "failed") && (
+          <Button
+            type="primary"
+            size="small"
+            icon={r.status === "failed" ? <ReloadOutlined /> : <ThunderboltOutlined />}
+            loading={extractingIds.includes(r.id)}
+            onClick={() => handleExtractSingle(r.id)}
+          >
+            {r.status === "failed" ? "重试" : "识别"}
+          </Button>
+        )
+      ),
+    },
+  ], [extractingIds]);
+
  return (
    <div>
      <Dragger
        multiple
        accept=".png,.jpg,.jpeg,.webp"
        showUploadList={false}
-        beforeUpload={(file) => { handleUpload(file as File); return false; }}
+        beforeUpload={(file) => {
+          handleUpload(file as File);
+          return false;
+        }}
        disabled={loading}
      >
        <p className="ant-upload-drag-icon"><InboxOutlined /></p>
        <p className="ant-upload-text">点击或拖拽账单截图到此处上传</p>
        <p className="ant-upload-hint">支持 png / jpg / webp，单次可多选</p>
      </Dragger>
-      <div style={{ marginTop: 16 }}>
+
+      <div style={{ marginTop: 12 }}>
+        <Space>
          <Button type="link" onClick={loadScreenshots} style={{ padding: 0 }}>刷新截图列表</Button>
-        <List
-          style={{ marginTop: 8 }}
-          grid={{ gutter: 16, column: 4 }}
-          dataSource={screenshots}
-          renderItem={(item) => (
-            <List.Item>
-              <Card size="small" title={item.filename}>
-                <div style={{ marginBottom: 8 }}>
-                  <Tag color={item.status === "extracted" ? "green" : item.status === "failed" ? "red" : "default"}>
-                    {item.status === "extracted" ? "已识别" : item.status === "failed" ? "失败" : "待识别"}
-                  </Tag>
-                </div>
-                {item.status === "pending" && (
-                  <Button
-                    type="primary"
-                    size="small"
-                    icon={<ThunderboltOutlined />}
-                    loading={extractingId === item.id}
-                    onClick={() => handleExtract(item.id)}
-                  >
-                    识别交易
+          <Button type="primary" onClick={handleBatchExtract} disabled={!selectedRowKeys.length} loading={extractingIds.length > 0}>
+            一键识别（已勾选）
          </Button>
-                )}
-              </Card>
-            </List.Item>
-          )}
-        />
+        </Space>
      </div>
+
+      <Table
+        style={{ marginTop: 8 }}
+        rowKey="id"
+        dataSource={screenshots}
+        columns={columns}
+        rowSelection={rowSelection}
+        pagination={false}
+        size="small"
+      />
    </div>
  );
 }
--- a/frontend/src/index.css
+++ b/frontend/src/index.css
@@ -3,5 +3,7 @@
 }
 body {
  margin: 0;
+  font-size: 16px;
+  line-height: 1.6;
  font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Helvetica Neue", Arial, sans-serif;
 }
--- a/frontend/src/pages/Settings.tsx
+++ b/frontend/src/pages/Settings.tsx
@@ -1,12 +1,20 @@
 import { useEffect, useState } from "react";
-import { Card, Form, Input, Select, Button, Alert, Space, message } from "antd";
+import { Card, Form, Input, Select, Button, Alert, Space, Divider, Descriptions, message } from "antd";
 import {
  api,
  type RuntimeSettings,
+  type ProviderKey,
  getApiBaseUrl,
  setApiBaseUrl,
 } from "../services/api";

+const PROVIDER_OPTIONS = [
+  { label: "OpenAI", value: "openai" as ProviderKey },
+  { label: "Anthropic", value: "anthropic" as ProviderKey },
+  { label: "DeepSeek", value: "deepseek" as ProviderKey },
+  { label: "自定义(OpenAI兼容)", value: "custom_openai" as ProviderKey },
+];
+
 export default function Settings() {
  const [form] = Form.useForm();
  const [loading, setLoading] = useState(false);
@@ -20,9 +28,11 @@ export default function Settings() {
      setRuntime(data);
      form.setFieldsValue({
        system_api_base_url: getApiBaseUrl(),
-        llm_provider: data.llm_provider,
+        ocr_provider: data.ocr_provider,
+        ocr_model: data.ocr_model,
+        inference_provider: data.inference_provider,
+        inference_model: data.inference_model,
        custom_openai_base_url: data.base_urls?.custom_openai || "",
-        custom_openai_model: data.models?.custom_openai || "gpt-4o-mini",
      });
    } catch {
      message.error("加载设置失败");
@@ -35,21 +45,15 @@ export default function Settings() {
    loadSettings();
  }, []);

-  const onFinish = async (values: {
-    system_api_base_url?: string;
-    llm_provider: "openai" | "anthropic" | "deepseek" | "custom_openai";
-    openai_api_key?: string;
-    anthropic_api_key?: string;
-    deepseek_api_key?: string;
-    custom_openai_api_key?: string;
-    custom_openai_base_url?: string;
-    custom_openai_model?: string;
-  }) => {
+  const onFinish = async (values: Record<string, string | undefined>) => {
    setSaving(true);
    try {
      setApiBaseUrl(values.system_api_base_url || "");
-      const payload = {
-        llm_provider: values.llm_provider,
+      const payload: Record<string, string | undefined> = {
+        ocr_provider: values.ocr_provider,
+        ocr_model: values.ocr_model?.trim() || undefined,
+        inference_provider: values.inference_provider,
+        inference_model: values.inference_model?.trim() || undefined,
        openai_api_key: values.openai_api_key?.trim() || undefined,
        anthropic_api_key: values.anthropic_api_key?.trim() || undefined,
        deepseek_api_key: values.deepseek_api_key?.trim() || undefined,
@@ -59,7 +63,7 @@ export default function Settings() {
      };
      const data = await api.settings.update(payload);
      setRuntime(data);
-      message.success("设置已保存并生效（含系统 API BaseURL）");
+      message.success("设置已保存");
    } catch {
      message.error("保存失败");
    } finally {
@@ -68,35 +72,42 @@ export default function Settings() {
  };

  return (
-    <Card title="LLM 设置" loading={loading}>
+    <Card title="模型与接口设置" loading={loading}>
      <Alert
        type="info"
        showIcon
        style={{ marginBottom: 16 }}
-        message="LLM API Key 仅在当前服务进程运行期内生效，不会自动写入磁盘。"
+        message="API Key 仅在当前服务进程运行期内生效，不写入磁盘。OCR 模型用于从截图中提取交易，推理模型用于生成报告等文本推理任务。"
      />
      <Form form={form} layout="vertical" onFinish={onFinish}>
        <Form.Item
          label="系统 API BaseURL（前端请求后端）"
          name="system_api_base_url"
-          extra="默认 /api；若前后端分离部署，可填如 http://127.0.0.1:8000/api"
+          extra="默认 /api；前后端分离时填 http://127.0.0.1:8000/api"
        >
-          <Input placeholder="/api 或 http://127.0.0.1:8000/api" />
+          <Input placeholder="/api" />
        </Form.Item>
-        <Form.Item
-          label="默认模型提供商"
-          name="llm_provider"
-          rules={[{ required: true, message: "请选择提供商" }]}
-        >
-          <Select
-            options={[
-              { label: "OpenAI", value: "openai" },
-              { label: "Anthropic", value: "anthropic" },
-              { label: "DeepSeek", value: "deepseek" },
-              { label: "自定义(OpenAI兼容)", value: "custom_openai" },
-            ]}
-          />
+
+        <Divider orientation="left">OCR 视觉模型（截图识别交易）</Divider>
+
+        <Form.Item label="OCR 提供商" name="ocr_provider" rules={[{ required: true }]}>
+          <Select options={PROVIDER_OPTIONS} />
        </Form.Item>
+        <Form.Item label="OCR 模型名" name="ocr_model" extra="留空则使用该提供商的默认模型">
+          <Input placeholder="如 gpt-4o / qwen-vl-max / ..." />
+        </Form.Item>
+
+        <Divider orientation="left">推理模型（报告生成等文本推理）</Divider>
+
+        <Form.Item label="推理提供商" name="inference_provider" rules={[{ required: true }]}>
+          <Select options={PROVIDER_OPTIONS} />
+        </Form.Item>
+        <Form.Item label="推理模型名" name="inference_model" extra="留空则使用该提供商的默认模型">
+          <Input placeholder="如 gpt-4o-mini / deepseek-chat / ..." />
+        </Form.Item>
+
+        <Divider orientation="left">API Key 与厂商配置</Divider>
+
        <Form.Item label="OpenAI API Key" name="openai_api_key">
          <Input.Password placeholder="sk-..." />
        </Form.Item>
@@ -113,29 +124,33 @@ export default function Settings() {
        >
          <Input placeholder="https://api.xxx.com/v1" />
        </Form.Item>
-        <Form.Item label="自定义厂商 Model" name="custom_openai_model">
+        <Form.Item label="自定义厂商默认模型" name="custom_openai_model">
          <Input placeholder="gpt-4o-mini / qwen-vl-plus / ..." />
        </Form.Item>
        <Form.Item label="自定义厂商 API Key" name="custom_openai_api_key">
          <Input.Password placeholder="sk-..." />
        </Form.Item>
+
        <Space>
-          <Button type="primary" htmlType="submit" loading={saving}>
-            保存设置
-          </Button>
+          <Button type="primary" htmlType="submit" loading={saving}>保存设置</Button>
          <Button onClick={loadSettings}>刷新</Button>
        </Space>
      </Form>

      {runtime && (
-        <Card title="当前状态" size="small" style={{ marginTop: 16 }}>
-          <div>系统 API BaseURL: {getApiBaseUrl()}</div>
-          <div>当前提供商: {runtime.llm_provider}</div>
-          <div>OpenAI Key: {runtime.has_keys.openai ? "已配置" : "未配置"}</div>
-          <div>Anthropic Key: {runtime.has_keys.anthropic ? "已配置" : "未配置"}</div>
-          <div>DeepSeek Key: {runtime.has_keys.deepseek ? "已配置" : "未配置"}</div>
-          <div>自定义厂商 Key: {runtime.has_keys.custom_openai ? "已配置" : "未配置"}</div>
-          <div>自定义厂商 BaseURL: {runtime.base_urls.custom_openai || "-"}</div>
+        <Card title="当前生效配置" size="small" style={{ marginTop: 16 }}>
+          <Descriptions column={2} size="small" bordered>
+            <Descriptions.Item label="系统 API BaseURL">{getApiBaseUrl()}</Descriptions.Item>
+            <Descriptions.Item label="自定义厂商 BaseURL">{runtime.base_urls.custom_openai || "-"}</Descriptions.Item>
+            <Descriptions.Item label="OCR 提供商">{runtime.ocr_provider}</Descriptions.Item>
+            <Descriptions.Item label="OCR 模型">{runtime.ocr_model}</Descriptions.Item>
+            <Descriptions.Item label="推理提供商">{runtime.inference_provider}</Descriptions.Item>
+            <Descriptions.Item label="推理模型">{runtime.inference_model}</Descriptions.Item>
+            <Descriptions.Item label="OpenAI Key">{runtime.has_keys.openai ? "已配置" : "未配置"}</Descriptions.Item>
+            <Descriptions.Item label="Anthropic Key">{runtime.has_keys.anthropic ? "已配置" : "未配置"}</Descriptions.Item>
+            <Descriptions.Item label="DeepSeek Key">{runtime.has_keys.deepseek ? "已配置" : "未配置"}</Descriptions.Item>
+            <Descriptions.Item label="自定义厂商 Key">{runtime.has_keys.custom_openai ? "已配置" : "未配置"}</Descriptions.Item>
+          </Descriptions>
        </Card>
      )}
    </Card>
--- a/frontend/src/services/api.ts
+++ b/frontend/src/services/api.ts
@@ -56,6 +56,13 @@ export interface ScreenshotItem {
  filename: string;
  file_path: string;
  status: string;
+  progress_step: string | null;
+  progress_percent: number;
+  progress_detail: string | null;
+  started_at: string | null;
+  finished_at: string | null;
+  duration_ms: number | null;
+  error_message: string | null;
  created_at: string;
 }

@@ -72,16 +79,24 @@ export interface FlowGraph {
  edges: Array<{ source: string; target: string; amount: number; count?: number }>;
 }

+export type ProviderKey = "openai" | "anthropic" | "deepseek" | "custom_openai";
+
 export interface RuntimeSettings {
-  llm_provider: "openai" | "anthropic" | "deepseek" | "custom_openai";
-  providers: Array<"openai" | "anthropic" | "deepseek" | "custom_openai">;
-  models: Record<string, string>;
+  ocr_provider: ProviderKey;
+  ocr_model: string;
+  inference_provider: ProviderKey;
+  inference_model: string;
+  providers: ProviderKey[];
+  provider_defaults: Record<string, string>;
  base_urls: Record<string, string>;
  has_keys: Record<string, boolean>;
 }

 export interface RuntimeSettingsUpdate {
-  llm_provider?: "openai" | "anthropic" | "deepseek" | "custom_openai";
+  ocr_provider?: ProviderKey;
+  ocr_model?: string;
+  inference_provider?: ProviderKey;
+  inference_model?: string;
  openai_api_key?: string;
  anthropic_api_key?: string;
  deepseek_api_key?: string;