From 017b7f1ca04d5be5397a9479eebd266d0b151741 Mon Sep 17 00:00:00 2001
From: Christopher Gondek <gondek.christopher@gmail.com>
Date: Mon, 1 Sep 2025 16:06:29 +0200
Subject: [PATCH] chore: type checking; input validation for web interface

---
 modules/interfaces/interface_web_model.py   | 24 ++++++++++-----------
 modules/interfaces/interface_web_objects.py | 12 +++++++++--
 2 files changed, 22 insertions(+), 14 deletions(-)

diff --git a/modules/interfaces/interface_web_model.py b/modules/interfaces/interface_web_model.py
index 86f19e08..bb4a82e0 100644
--- a/modules/interfaces/interface_web_model.py
+++ b/modules/interfaces/interface_web_model.py
@@ -2,7 +2,7 @@
 
 from abc import ABC, abstractmethod
 from modules.interfaces.interfaceChatModel import ActionDocument, ActionResult
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, Field, HttpUrl
 from typing import List
 
 
@@ -12,21 +12,21 @@ from typing import List
 
 
 class WebSearchRequest(BaseModel):
-    query: str
-    max_results: int
+    query: str = Field(min_length=1, max_length=400)
+    max_results: int = Field(ge=1, le=20)
 
 
 class WebSearchResultItem(BaseModel):
     """Individual search result"""
 
     title: str
-    url: str
+    url: HttpUrl
 
 
 class WebSearchDocumentData(BaseModel):
     """Complete search results document"""
 
-    query: str
+    query: str = Field(min_length=1, max_length=400)
     results: List[WebSearchResultItem]
     total_count: int
 
@@ -50,20 +50,20 @@ class WebSearchBase(ABC):
 
 
 class WebCrawlRequest(BaseModel):
-    urls: List[str]
+    urls: List[HttpUrl]
 
 
 class WebCrawlResultItem(BaseModel):
     """Individual crawl result"""
 
-    url: str
+    url: HttpUrl
     content: str
 
 
 class WebCrawlDocumentData(BaseModel):
     """Complete crawl results document"""
 
-    urls: List[str]
+    urls: List[HttpUrl]
     results: List[WebCrawlResultItem]
     total_count: int
 
@@ -89,21 +89,21 @@ class WebCrawlBase(ABC):
 
 
 class WebScrapeRequest(BaseModel):
-    query: str
-    max_results: int
+    query: str = Field(min_length=1, max_length=400)
+    max_results: int = Field(ge=1, le=20)
 
 
 class WebScrapeResultItem(BaseModel):
     """Individual scrape result"""
 
-    url: str
+    url: HttpUrl
     content: str
 
 
 class WebScrapeDocumentData(BaseModel):
     """Complete scrape results document"""
 
-    query: str
+    query: str = Field(min_length=1, max_length=400)
     results: List[WebScrapeResultItem]
     total_count: int
 
diff --git a/modules/interfaces/interface_web_objects.py b/modules/interfaces/interface_web_objects.py
index 0ea43bd7..f348f0bd 100644
--- a/modules/interfaces/interface_web_objects.py
+++ b/modules/interfaces/interface_web_objects.py
@@ -1,3 +1,4 @@
+from typing import Optional
 from modules.interfaces.interface_web_model import (
     WebCrawlActionResult,
     WebSearchActionResult,
@@ -11,9 +12,16 @@ from dataclasses import dataclass
 from modules.connectors.connector_tavily import ConnectorTavily
 
 
-@dataclass
+@dataclass(slots=True)
 class WebInterface:
-    connector_tavily: ConnectorTavily = None
+    connector_tavily: ConnectorTavily
+
+    def __post_init__(self) -> None:
+        if self.connector_tavily is None:
+            raise TypeError(
+                "connector_tavily must be provided. "
+                "Use `await WebInterface.create()` or pass a ConnectorTavily."
+            )
 
     @classmethod
     async def create(cls) -> "WebInterface":