CrewAI

import weave
from crewai import Agent, Task, Crew, LLM, Process

# Initialiser Weave avec le nom de votre projet
weave.init(project_name="crewai_demo")

# Créer un LLM avec une température de 0 pour obtenir des sorties déterministes
llm = LLM(model="gpt-4o-mini", temperature=0)

# Créer les agents
researcher = Agent(
    role='Research Analyst',
    goal='Find and analyze the best investment opportunities',
    backstory='Expert in financial analysis and market research',
    llm=llm,
    verbose=True,
    allow_delegation=False,
)

writer = Agent(
    role='Report Writer',
    goal='Write clear and concise investment reports',
    backstory='Experienced in creating detailed financial reports',
    llm=llm,
    verbose=True,
    allow_delegation=False,
)

# Créer les tâches
research_task = Task(
    description='Deep research on the {topic}',
    expected_output='Comprehensive market data including key players, market size, and growth trends.',
    agent=researcher
)

writing_task = Task(
    description='Write a detailed report based on the research',
    expected_output='The report should be easy to read and understand. Use bullet points where applicable.',
    agent=writer
)

# Créer un crew
crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, writing_task],
    verbose=True,
    process=Process.sequential,
)

# Lancer le crew
result = crew.kickoff(inputs={"topic": "AI in material science"})
print(result)

# .... imports existants ....
from crewai_tools import SerperDevTool

# On fournit l’outil à l’agent.
researcher = Agent(
    role='Research Analyst',
    goal='Find and analyze the best investment opportunities',
    backstory='Expert in financial analysis and market research',
    llm=llm,
    verbose=True,
    allow_delegation=False,
    tools=[SerperDevTool()],
)

# .... code existant ....

import weave
# Initialiser Weave avec le nom de votre projet
weave.init("crewai_demo")

from crewai.flow.flow import Flow, listen, router, start
from litellm import completion


class CustomerFeedbackFlow(Flow):
    model = "gpt-4o-mini"

    @start()
    def fetch_feedback(self):
        print("Fetching customer feedback")
        # Dans un cas réel, ce code pourrait être remplacé par un appel d’API.
        # Dans cet exemple, nous simulons le feedback d’un client.
        feedback = (
            "I had a terrible experience with the product. "
            "It broke after one use and customer service was unhelpful."
        )
        self.state["feedback"] = feedback
        return feedback

    @router(fetch_feedback)
    def analyze_feedback(self, feedback):
        # Utiliser le modèle de langage pour analyser le sentiment
        prompt = (
            f"Analyze the sentiment of this customer feedback and "
            "return only 'positive' or 'negative':\n\n"
            f"Feedback: {feedback}"
        )
        response = completion(
            model=self.model,
            messages=[{"role": "user", "content": prompt}],
        )
        sentiment = response["choices"][0]["message"]["content"].strip().lower()
        # Si la réponse est ambiguë, utiliser « negative » par défaut
        if sentiment not in ["positive", "negative"]:
            sentiment = "negative"
        return sentiment

    @listen("positive")
    def handle_positive_feedback(self):
        # Générer un message de remerciement en réponse à un feedback positif
        prompt = "Generate a thank you message for a customer who provided positive feedback."
        response = completion(
            model=self.model,
            messages=[{"role": "user", "content": prompt}],
        )
        thank_you_message = response["choices"][0]["message"]["content"].strip()
        self.state["response"] = thank_you_message
        return thank_you_message

    @listen("negative")
    def handle_negative_feedback(self):
        # En réponse à un feedback négatif, générer un message d’excuses promettant d’améliorer le service
        prompt = (
            "Generate an apology message to a customer who provided negative feedback and offer assistance or a solution."
        )
        response = completion(
            model=self.model,
            messages=[{"role": "user", "content": prompt}],
        )
        apology_message = response["choices"][0]["message"]["content"].strip()
        self.state["response"] = apology_message
        return apology_message

# Instancier et lancer le flow
flow = CustomerFeedbackFlow()
result = flow.kickoff()
print(result)

# .... imports existants et initialisation de Weave ....

# Décorez votre fonction de garde-fou avec `@weave.op()`
@weave.op(name="guardrail-validate_blog_content")
def validate_blog_content(result: TaskOutput) -> Tuple[bool, Any]:
    # Obtenir le résultat brut sous forme de chaîne
    result = result.raw

    """Validate blog content meets requirements."""
    try:
        # Vérifier le nombre de mots
        word_count = len(result.split())

        if word_count > 200:
            return (False, {
                "error": "Blog content exceeds 200 words",
                "code": "WORD_COUNT_ERROR",
                "context": {"word_count": word_count}
            })

        # Ajoutez ici une logique de validation supplémentaire
        return (True, result.strip())
    except Exception as e:
        return (False, {
            "error": "Unexpected error during validation",
            "code": "SYSTEM_ERROR"
        })


# .... agents existants et tâche de l’analyste de recherche ....

writing_task = Task(
    description='Write a detailed report based on the research under 200 words',
    expected_output='The report should be easy to read and understand. Use bullet points where applicable.',
    agent=writer,
    guardrail=validate_blog_content,
)

# .... code existant pour exécuter le crew ....