LangChain Classic: Reading Old Code

import warnings; warnings.filterwarnings("ignore")
from standins import RuleModel, KeywordEmbeddings
from langchain_core.prompts import PromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_classic.chains import LLMChain, SimpleSequentialChain, ConversationChain, RetrievalQA
from langchain_classic.memory import ConversationBufferMemory, ConversationBufferWindowMemory
from langchain_community.vectorstores import FAISS

llm = RuleModel(rules={"organic tea": "Chai Bloom", "company name for electric": "VoltRide",
                       "slogan": "VoltRide: Silent rides, bright future.", "upi limit": "Rs 25,000 in the first 24 hours."}, default="OK")
prompt = PromptTemplate.from_template("Suggest one name for a {business} company")
print("LLMChain (old):", LLMChain(llm=llm, prompt=prompt).invoke({"business": "organic tea"}))
print("LCEL     (new):", (prompt | llm | StrOutputParser()).invoke({"business": "organic tea"}))

name_chain = LLMChain(llm=llm, prompt=PromptTemplate.from_template("Give one company name for electric {product}. Only the name."))
slogan_chain = LLMChain(llm=llm, prompt=PromptTemplate.from_template("Write a slogan for the company {name}."))
print("SimpleSequentialChain:", SimpleSequentialChain(chains=[name_chain, slogan_chain]).invoke("scooters"))

class Rememberer(RuleModel):
    def _pick(self, messages):
        t = messages[-1].content
        hist = t.split("Current conversation:")[-1].rsplit("Human:", 1)[0]
        q = t.rsplit("Human:", 1)[-1].lower()
        if "name" in q: return "Your name is Ajeet." if "Ajeet" in hist else "I'm sorry, I don't know your name."
        return "Nice to meet you!"
for mem in [ConversationBufferMemory(), ConversationBufferWindowMemory(k=1)]:
    conv = ConversationChain(llm=Rememberer(), memory=mem)
    conv.invoke({"input": "My name is Ajeet."}); conv.invoke({"input": "I like cricket."})
    print(f"{type(mem).__name__:32s} -> {conv.invoke({'input': 'What is my name?'})['response']}")

retriever = FAISS.from_texts(["Report a lost card within 3 days for zero liability.", "UPI limit for a new payee is Rs 25,000 in the first 24 hours.",
                              "Video KYC is available from 9 AM to 9 PM."], KeywordEmbeddings()).as_retriever(search_kwargs={"k": 1})
out = RetrievalQA.from_chain_type(llm=llm, retriever=retriever, chain_type="stuff", return_source_documents=True).invoke({"query": "What is the UPI limit for a new payee?"})
print("RetrievalQA:", out["result"], "| source:", out["source_documents"][0].page_content)