Add fetch_url tool for link ingestion

custom_components/groqd/conversation.py

@@ -4,6 +4,8 @@ from __future__ import annotations
 
 from collections.abc import Callable
 import json
+import re
+from html import unescape
 from typing import Any, Literal
 
 from groq._types import NOT_GIVEN
@@ -142,6 +144,42 @@ def _searxng_tool() -> ChatCompletionToolParam:
     return ChatCompletionToolParam(type="function", function=tool_spec)
 
 
+def _fetch_tool() -> ChatCompletionToolParam:
+    tool_spec = FunctionDefinition(
+        name="fetch_url",
+        parameters={
+            "type": "object",
+            "properties": {
+                "url": {"type": "string"},
+                "max_chars": {"type": "integer", "default": 4000},
+            },
+            "required": ["url"],
+        },
+        description="Fetch a URL and return cleaned text content.",
+    )
+    return ChatCompletionToolParam(type="function", function=tool_spec)
+
+
+def _strip_html(text: str) -> str:
+    # Cheap HTML-to-text for summarization; avoids extra deps.
+    text = re.sub(r"(?s)<script.*?>.*?</script>", " ", text)
+    text = re.sub(r"(?s)<style.*?>.*?</style>", " ", text)
+    text = re.sub(r"<[^>]+>", " ", text)
+    text = re.sub(r"\\s+", " ", text)
+    return unescape(text).strip()
+
+
+async def _run_fetch(hass: HomeAssistant, url: str, max_chars: int) -> dict[str, Any]:
+    session = async_get_clientsession(hass)
+    async with session.get(url, timeout=20) as resp:
+        resp.raise_for_status()
+        content_type = resp.headers.get("content-type", "")
+        text = await resp.text()
+    cleaned = _strip_html(text)
+    if max_chars > 0:
+        cleaned = cleaned[:max_chars]
+    return {"url": url, "content_type": content_type, "text": cleaned}
+
 async def _run_searxng(
     hass: HomeAssistant,
     options: dict[str, Any],
@@ -258,6 +296,9 @@ class GroqdConversationEntity(
             if tools is None:
                 tools = []
             tools.append(_searxng_tool())
+        if tools is None:
+            tools = []
+        tools.append(_fetch_tool())
 
         memory_scope = options.get(CONF_MEMORY_SCOPE, DEFAULT_MEMORY_SCOPE)
         memory_key = None
@@ -448,6 +489,17 @@ class GroqdConversationEntity(
                         tool_response = await _run_searxng(self.hass, options, tool_args)
                     except Exception as err:
                         tool_response = {"error": type(err).__name__, "error_text": str(err)}
+                elif tool_name == "fetch_url":
+                    url = tool_args.get("url", "")
+                    max_chars = tool_args.get("max_chars", 4000)
+                    try:
+                        max_chars = int(max_chars)
+                    except (TypeError, ValueError):
+                        max_chars = 4000
+                    try:
+                        tool_response = await _run_fetch(self.hass, url, max_chars)
+                    except Exception as err:
+                        tool_response = {"error": type(err).__name__, "error_text": str(err)}
                 elif llm_api:
                     tool_input = llm.ToolInput(
                         tool_name=tool_name,