3 files changed, 122 insertions, 0 deletions

diff --git a/src/ollama_web/routes/chat.py b/src/ollama_web/routes/chat.py
new file mode 100644
index 0000000..2612279
--- /dev/null
+++ b/src/ollama_web/routes/chat.py
@@ -0,0 +1,72 @@
+"""SSE streaming chat endpoint."""
+
+from __future__ import annotations
+
+import json
+from collections.abc import AsyncIterator
+from typing import Any
+
+from ollama import Message
+from starlette.requests import Request
+from starlette.responses import StreamingResponse
+
+
+def _parse_messages(raw: list[dict[str, Any]]) -> list[Message]:
+    """Convert raw JSON message dicts into ``ollama.Message`` objects."""
+    out: list[Message] = []
+    for m in raw:
+        role = m.get("role", "user")
+        content = m.get("content", "")
+        name = m.get("name")
+        images = m.get("images")
+        kwargs: dict[str, Any] = {"role": role, "content": content}
+        if name:
+            kwargs["name"] = name
+        if images:
+            kwargs["images"] = images
+        out.append(Message(**kwargs))
+    return out
+
+
+async def _event_stream(request: Request, payload: dict[str, Any]) -> AsyncIterator[bytes]:
+    """Yield SSE events from the tool-augmented chat loop."""
+    model = payload.get("model") or request.app.state.settings.default_model
+    raw_messages = payload.get("messages", [])
+    think = payload.get("think")
+
+    messages = _parse_messages(raw_messages)
+    host = request.app.state.settings.ollama_host
+
+    from ollama_web import llm  # local import to avoid circular at module load
+
+    async for event in llm.astream_chat_with_tools(messages, model, think=think, host=host):
+        data = json.dumps(event, ensure_ascii=False)
+        yield f"data: {data}\n\n".encode()
+
+
+async def chat(request: Request) -> StreamingResponse:
+    """POST /api/chat: stream chat events as Server-Sent Events."""
+    import logging
+
+    logger = logging.getLogger("ollama_web.chat")
+    try:
+        body = await request.body()
+        payload = json.loads(body.decode("utf-8"))
+    except Exception as exc:
+        logger.error("failed to parse request json: %s", exc)
+        data = json.dumps({"type": "error", "message": f"Invalid JSON: {exc}"}, ensure_ascii=False)
+        return StreamingResponse(
+            _single_event(data),
+            media_type="text/event-stream",
+        )
+    msg_count = len(payload.get("messages", []))
+    logger.info("chat request model=%s messages=%d", payload.get("model"), msg_count)
+    return StreamingResponse(
+        _event_stream(request, payload),
+        media_type="text/event-stream",
+    )
+
+
+async def _single_event(data: str) -> AsyncIterator[bytes]:
+    """Yield a single SSE event and finish."""
+    yield f"data: {data}\n\n".encode()
diff --git a/src/ollama_web/routes/models.py b/src/ollama_web/routes/models.py
new file mode 100644
index 0000000..477a44e
--- /dev/null
+++ b/src/ollama_web/routes/models.py
@@ -0,0 +1,23 @@
+"""API routes for model listing and health checks."""
+
+from __future__ import annotations
+
+from starlette.requests import Request
+from starlette.responses import JSONResponse
+
+from ollama_web import llm
+
+
+async def get_models(request: Request) -> JSONResponse:
+    """Return available Ollama models as JSON."""
+    host = request.app.state.settings.ollama_host
+    models = llm.list_models(host=host)
+    return JSONResponse({"models": models})
+
+
+async def health(request: Request) -> JSONResponse:
+    """Report Ollama server reachability."""
+    host = request.app.state.settings.ollama_host
+    models = llm.list_models(host=host)
+    reachable = models is not None
+    return JSONResponse({"reachable": reachable, "models": models})
\ No newline at end of file
diff --git a/src/ollama_web/routes/pages.py b/src/ollama_web/routes/pages.py
new file mode 100644
index 0000000..8e1d084
--- /dev/null
+++ b/src/ollama_web/routes/pages.py
@@ -0,0 +1,27 @@
+"""Page routes rendering Jinja2 templates."""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+from starlette.requests import Request
+from starlette.responses import Response
+from starlette.templating import Jinja2Templates
+
+from ollama_web import llm
+
+_TEMPLATES_DIR = Path(__file__).resolve().parent.parent / "templates"
+templates = Jinja2Templates(directory=str(_TEMPLATES_DIR))
+
+
+async def index(request: Request) -> Response:
+    """Render the chat page with the available model list."""
+    host = request.app.state.settings.ollama_host
+    models = llm.list_models(host=host)
+    default_model = request.app.state.settings.default_model
+    return templates.TemplateResponse(
+        request,
+        "index.html",
+        {"models": models, "default_model": default_model},
+        media_type="text/html; charset=utf-8",
+    )