* langchain example using langchain expression language

examples/langchain_lcel.py

@@ -0,0 +1,47 @@
+from langchain import chat_models, prompts, schema
+
+from llm_guard.input_scanners import Toxicity
+
+
+class LLMGuardPromptInvalidException(Exception):
+    """Exception to raise when llm-guard marks prompt invalid."""
+
+
+class LLMGuardOutputInvalidException(Exception):
+    """Exception to raise when llm-guard marks result invalid."""
+
+
+def scan_input_toxicity_scanner(params) -> str:
+    text = params.get("text", "")
+    threshold = float(params.get("threshold", 0.7))
+    scanner = Toxicity(threshold)
+    sanitized_input, is_valid, risk_score = scanner.scan(text)
+    print(risk_score)
+    if is_valid:
+        return sanitized_input
+    else:
+        raise LLMGuardPromptInvalidException(
+            f"The input text '{text}' was determined as toxic with risk score {risk_score}"
+        )
+
+
+chain = (
+    prompts.ChatPromptTemplate.from_template("Reverse the following string: {text}")
+    | chat_models.ChatOpenAI()
+    | schema.output_parser.StrOutputParser()
+)
+
+guard_chain = scan_input_toxicity_scanner | schema.output_parser.StrOutputParser()
+
+overall_chain = {"text": guard_chain} | chain
+
+
+try:
+    input_text = "Hello, world!"
+    threshold = 0.0
+    output = overall_chain.invoke({"text": input_text, "threshold": threshold})
+    print(output)
+except LLMGuardPromptInvalidException as e:
+    print(f"Prompt invalid: {str(e)}")
+except Exception as e:
+    print(f"An error occurred: {str(e)}")