feat: adicionando opção de enviar o PDF próprio para gerar documento para testes

gerar_documento/serializer.py

@@ -14,7 +14,7 @@ class FileInfoSerializer(serializers.Serializer):
     link_arquivo = serializers.URLField()
 
 
-class ResumoCursorCompeltoSerializer(ResumoCursorSerializer):
+class GerarDocumentoSerializer(ResumoCursorSerializer):
     system_prompt = None
 
     files = serializers.ListField(child=FileInfoSerializer(), required=True)
@@ -45,3 +45,30 @@ class ResumoCursorCompeltoSerializer(ResumoCursorSerializer):
     gpt_temperature = serializers.FloatField(default=0)
     id_modelo_do_usuario = serializers.IntegerField(required=False)
     should_have_contextual_chunks = serializers.BooleanField(default=False)
+
+
+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=5)
+    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=200)
+    num_k_rerank = serializers.IntegerField(default=5)
+    model_cohere_rerank = serializers.CharField(
+        required=False, default="rerank-english-v2.0"
+    )
+    more_initial_chunks_for_reranking = serializers.IntegerField(default=20)
+    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)

gerar_documento/urls.py

@@ -1,11 +1,19 @@
 from django.urls import path
 
-from gerar_documento.views import ResumoSimplesCursorCompletoView
+from gerar_documento.views import (
+    GerarDocumentoView,
+    GerarDocumentoComPDFProprioView,
+)
 
 urlpatterns = [
     path(
         "gerar-documento",
-        ResumoSimplesCursorCompletoView.as_view(),
+        GerarDocumentoView.as_view(),
         name="summary-cursor-completo-pdf",
     ),
+    path(
+        "gerar-documento/pdf",
+        GerarDocumentoComPDFProprioView.as_view(),
+        name="generate-document-with-own-pdf",
+    ),
 ]

gerar_documento/views.py

@@ -12,19 +12,20 @@ from _utils.resumo_completo_cursor import (
 )
 from _utils.gerar_relatorio_modelo_usuario.prompts import prompt_auxiliar_inicio
 from .serializer import (
-    ResumoCursorCompeltoSerializer,
+    GerarDocumentoComPDFProprioSerializer,
+    GerarDocumentoSerializer,
 )
 
 
-class ResumoSimplesCursorCompletoView(AsyncAPIView):
+class GerarDocumentoView(AsyncAPIView):
     # parser_classes = [MultiPartParser]
 
     @extend_schema(
-        request=ResumoCursorCompeltoSerializer,
+        request=GerarDocumentoSerializer,
     )
     async def post(self, request):
         print(f"\n\nDATA E HORA DA REQUISIÇÃO: {datetime.now()}")
-        serializer = ResumoCursorCompeltoSerializer(data=request.data)
+        serializer = GerarDocumentoSerializer(data=request.data)
         if serializer.is_valid(raise_exception=True):
             data = serializer.validated_data
             print("\n\ndata: ", data)
@@ -46,3 +47,28 @@ class ResumoSimplesCursorCompletoView(AsyncAPIView):
             # remove_pdf_temp_files(listaPDFs)
 
             return Response({"resposta": resposta_llm})
+
+
+class GerarDocumentoComPDFProprioView(AsyncAPIView):
+    parser_classes = [MultiPartParser]
+
+    @extend_schema(
+        request=GerarDocumentoComPDFProprioSerializer,
+    )
+    async def post(self, request):
+        print(f"\n\nDATA E HORA DA REQUISIÇÃO: {datetime.now()}")
+        serializer = GerarDocumentoComPDFProprioSerializer(data=request.data)
+        if serializer.is_valid(raise_exception=True):
+            data = serializer.validated_data
+            print("\n\ndata: ", data)
+
+            listaPDFs = handle_pdf_files_from_serializer(data["files"])
+
+            resposta_llm = await get_llm_summary_answer_by_cursor_complete(
+                data, listaPDFs
+            )
+            print("\n\nresposta_llm: ", resposta_llm)
+
+            remove_pdf_temp_files(listaPDFs)
+
+            return Response({"resposta": resposta_llm})

ragas_api/serializer.py

@@ -1,14 +1,14 @@
 from rest_framework import serializers
-from gerar_documento.serializer import ResumoCursorCompeltoSerializer
+from gerar_documento.serializer import GerarDocumentoSerializer
 
 
-class RagasSerializer(ResumoCursorCompeltoSerializer):
+class RagasSerializer(GerarDocumentoSerializer):
     files = serializers.ListField(child=serializers.FileField(), required=True)
     id_modelo_do_usuario = serializers.IntegerField(required=False)
     hf_embedding = serializers.CharField(required=False, default="all-MiniLM-L6-v2")
 
 
-class RagasFromTextSerializer(ResumoCursorCompeltoSerializer):
+class RagasFromTextSerializer(GerarDocumentoSerializer):
     files = None
     id_modelo_do_usuario = serializers.IntegerField(required=False, default=9)
     user_message = serializers.CharField(required=True)