gerar_documento/serializer.py

from rest_framework import serializers
from _antigos.resumos.serializer import ResumoCursorSerializer
from _utils.gerar_relatorio_modelo_usuario.prompts import (
    prompt_gerar_documento,
    prompt_auxiliar_padrao,
)

user_message = "What are the main points of this document?"


class FileInfoSerializer(serializers.Serializer):
    unique_id = serializers.CharField(max_length=255)
    tipo_arquivo = serializers.CharField(max_length=255)
    link_arquivo = serializers.URLField()


class GerarDocumentoSerializer(ResumoCursorSerializer):
    system_prompt = None

    files = serializers.ListField(child=FileInfoSerializer(), required=True)
    bubble_editor_version = serializers.CharField(
        required=False, default="version-test"
    )  # Será o valor utilizado dentro da URL da requisição pro Bubble

    # prompt_auxiliar = serializers.CharField(
    #     required=False, default=prompt_auxiliar_padrao
    # )
    prompt_gerar_documento = serializers.CharField(
        required=False, default=prompt_gerar_documento
    )
    user_message = serializers.CharField(required=False, default=user_message)
    num_chunks_retrieval = serializers.IntegerField(default=20)
    embedding_weight = serializers.FloatField(default=0.5)
    bm25_weight = serializers.FloatField(default=0.5)
    context_window = serializers.IntegerField(default=3)
    chunk_overlap = serializers.IntegerField(default=800)
    num_k_rerank = serializers.IntegerField(default=20)
    model_cohere_rerank = serializers.CharField(
        required=False, default="rerank-english-v2.0"
    )
    more_initial_chunks_for_reranking = serializers.IntegerField(default=100)
    claude_context_model = serializers.CharField(
        required=False, default="claude-3-haiku-20240307"
    )
    gpt_temperature = serializers.FloatField(default=0)
    id_modelo_do_usuario = serializers.IntegerField(required=False)
    should_have_contextual_chunks = serializers.BooleanField(default=False)  # type: ignore
    should_use_llama_parse = serializers.BooleanField(required=False, default=False)  # type: ignore
    llm_ultimas_requests = serializers.CharField(
        required=False, default="gemini-2.0-flash"
    )


class GerarDocumentoComPDFProprioSerializer(ResumoCursorSerializer):
    system_prompt = None
    # prompt_auxiliar = serializers.CharField(
    #     required=False, default=prompt_auxiliar_padrao
    # )
    prompt_gerar_documento = serializers.CharField(
        required=False, default=prompt_gerar_documento
    )
    user_message = serializers.CharField(required=False, default=user_message)
    num_chunks_retrieval = serializers.IntegerField(default=20)
    embedding_weight = serializers.FloatField(default=0.5)
    bm25_weight = serializers.FloatField(default=0.5)
    context_window = serializers.IntegerField(default=3)
    chunk_overlap = serializers.IntegerField(default=800)
    num_k_rerank = serializers.IntegerField(default=20)
    model_cohere_rerank = serializers.CharField(
        required=False, default="rerank-english-v2.0"
    )
    more_initial_chunks_for_reranking = serializers.IntegerField(default=100)
    claude_context_model = serializers.CharField(
        required=False, default="claude-3-haiku-20240307"
    )
    gpt_temperature = serializers.FloatField(default=0)
    id_modelo_do_usuario = serializers.IntegerField(required=False, default=11)
    should_have_contextual_chunks = serializers.BooleanField(default=False)  # type: ignore
    should_use_llama_parse = serializers.BooleanField(required=False, default=False)  # type: ignore
    llm_ultimas_requests = serializers.CharField(required=False, default="gpt-4o-mini")