MathJake commited on
Commit
7a2ea7b
·
verified ·
1 Parent(s): 8de2de0

Upload 3 files

Browse files
Files changed (3) hide show
  1. Llm_local.py +126 -0
  2. Rag_milvus.ipynb +571 -0
  3. chatbox_v1.py +60 -0
Llm_local.py ADDED
@@ -0,0 +1,126 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import ollama
2
+ import time
3
+ import streamlit as st
4
+ import fitz
5
+ from docx import Document
6
+ from io import BytesIO
7
+ from fpdf import FPDF
8
+
9
+ def get_response_from_mistral(query, context):
10
+ prompt_text = f"""
11
+ "Tú eres un asistente para tareas de respuesta a preguntas. "
12
+ "Usa los siguientes fragmentos de contexto recuperado para responder la pregunta. "
13
+ "Si el contexto está vacío o no contiene información relevante, responde: 'Disculpa, no tengo información para responder esa pregunta'. "
14
+ "Si el contexto es válido, responde la pregunta usando un mínimo de 2 oraciones y un máximo de 4, manteniendo la respuesta clara y concisa. "
15
+ "No inventes ni asumas nada que no esté explícitamente en el contexto."
16
+ "\n\n"
17
+
18
+ Usa solo este contexto:
19
+ {context}
20
+ **IMPORTANTE***
21
+ "Ojo siempre que tu contexto es vacio, tu respuesta debe ser : Disculpa, no tengo información para responder esa pregunta"
22
+ **
23
+ Y Responde esta pregunta:
24
+ {query}
25
+ """
26
+
27
+ respuesta = ollama.chat(
28
+ model="mistral",
29
+ messages=[
30
+ {"role": "system", "content": "Eres un asistente especializado en análisis de datos."},
31
+ {"role": "user", "content": prompt_text}
32
+ ]
33
+ )
34
+
35
+ respuesta_texto = respuesta["message"]["content"]
36
+
37
+ for word in respuesta_texto.split():
38
+ yield word + " "
39
+ time.sleep(0.05)
40
+
41
+ def generarPages():
42
+ with st.sidebar:
43
+ st.page_link("chatbox_v1.py", label="Inicio", icon="🏠")
44
+ st.page_link("pages/resumen_word.py", label="Informe de PDF y Word", icon="📄")
45
+ st.page_link("pages/insertardocumentos.py", label="Documentos a vector", icon="🛢️")
46
+
47
+ def informes_mistral(context):
48
+ prompt_text = f"""
49
+ **Atención**: No generes una historia o narrativa, tu tarea es realizar un análisis detallado y preciso del documento legal. No se requiere creatividad, solo precisión.
50
+ Eres un asistente experto en procesamiento y análisis de documentos. Tu tarea es leer y comprender el contenido proporcionado y generar un informe extenso, detallado y bien estructurado.
51
+ El informe debe incluir las siguientes secciones:
52
+
53
+ 1. **Resumen General**: Proporciona un resumen completo y detallado de todo el contenido del documento. Incluye los aspectos más relevantes, pero sin dejar de lado detalles importantes.
54
+
55
+ 2. **Puntos Clave**: Enumera los puntos más importantes del documento, resaltando las ideas principales y los aspectos críticos que se abordan.
56
+
57
+ 3. **Análisis Crítico**: Realiza un análisis en profundidad sobre el contenido del documento. Comenta sobre su calidad, lógica, coherencia, posibles fallos, aspectos positivos, y cualquier otro elemento que pueda ser relevante.
58
+
59
+ 4. **Recomendaciones**: Proporciona sugerencias o recomendaciones para mejorar el contenido. Si el documento se trata de un informe técnico, análisis de datos, o investigación, incluye sugerencias de cómo se podría mejorar la interpretación de los datos, el análisis o la presentación.
60
+
61
+ 5. **Conclusiones**: Finaliza con una sección de conclusiones que recapitule los puntos clave del análisis y del documento en general, además de una visión global de las implicaciones del contenido.
62
+
63
+ 6: **En caso de**: En caso de que el contenido sea acerca de un decreto legislativo o algo acera de una ley incluye un seccion donde hables lo mas importante de todos los articulos y menciones cuales son los mas relevantes.
64
+
65
+ 7: **Documentos Analizados**: Menciona el nombre de todos los documentos que componen el contenido analizado. Si hay más de uno, asegúrate de listarlos todos y dejar claro que el análisis se basa en todos ellos en conjunto.
66
+
67
+ Siempre deberás comenzar el informe con los nombres de los archivos que componen el contenido.
68
+ Recuerda que siempre debes mantener la estructura que te mande
69
+
70
+ Contenido del documento:
71
+ {context}
72
+
73
+ Utiliza un estilo claro y profesional en todo momento, y asegúrate de que cada sección esté claramente diferenciada. Tu informe debe ser extenso y abarcativo, no debe ser corto ni vago.
74
+ recuerda siempre reponder en español
75
+ """
76
+ respuesta = ollama.chat(
77
+ model="mistral",
78
+ messages=[
79
+ {"role": "system", "content": "Eres un asistente especializado en análisis detallado de documentos."},
80
+ {"role": "user", "content": prompt_text}
81
+ ]
82
+ )
83
+
84
+ respuesta_texto = respuesta["message"]["content"]
85
+
86
+ for word in respuesta_texto.split():
87
+ yield word + " "
88
+ time.sleep(0.05)
89
+
90
+ def extraer_texto(archivo):
91
+ if archivo.name.endswith(".pdf"):
92
+ texto = ""
93
+ with fitz.open(stream=archivo.read(), filetype="pdf") as doc:
94
+ for page in doc:
95
+ texto += page.get_text()
96
+ return texto
97
+ elif archivo.name.endswith(".txt"):
98
+ return archivo.read().decode("utf-8")
99
+ else:
100
+ return ""
101
+
102
+ def extraer_texto_word(file):
103
+ texto = ""
104
+ doc = Document(file)
105
+ for para in doc.paragraphs:
106
+ texto += para.text + "\n"
107
+ return texto
108
+
109
+ def generar_docx(texto):
110
+ doc = Document()
111
+ doc.add_heading("Resumen generado por IA", 0)
112
+ for parrafo in texto.split("\n"):
113
+ doc.add_paragraph(parrafo)
114
+ buffer = BytesIO()
115
+ doc.save(buffer)
116
+ buffer.seek(0)
117
+ return buffer
118
+
119
+ def generar_pdf(texto):
120
+ pdf = FPDF()
121
+ pdf.add_page()
122
+ pdf.set_auto_page_break(auto=True, margin=15)
123
+ pdf.set_font("Arial", size=12)
124
+ for linea in texto.split("\n"):
125
+ pdf.multi_cell(0, 10, linea)
126
+ return bytes(pdf.output(dest='S').encode('latin-1'))
Rag_milvus.ipynb ADDED
@@ -0,0 +1,571 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cells": [
3
+ {
4
+ "cell_type": "markdown",
5
+ "id": "bf597549",
6
+ "metadata": {},
7
+ "source": [
8
+ "PRIMERO PREPARAMOS TODAS NUESTRAS FUNCIONES PARA PODER SER CONVOCADAS LUEGO."
9
+ ]
10
+ },
11
+ {
12
+ "cell_type": "markdown",
13
+ "id": "7968949c",
14
+ "metadata": {},
15
+ "source": [
16
+ "Instalamos "
17
+ ]
18
+ },
19
+ {
20
+ "cell_type": "code",
21
+ "execution_count": 1,
22
+ "id": "9a192af6",
23
+ "metadata": {},
24
+ "outputs": [
25
+ {
26
+ "name": "stdout",
27
+ "output_type": "stream",
28
+ "text": [
29
+ "Collecting qdrant-client\n",
30
+ " Obtaining dependency information for qdrant-client from https://files.pythonhosted.org/packages/e4/52/f49b0aa96253010f57cf80315edecec4f469e7a39c1ed92bf727fa290e57/qdrant_client-1.14.2-py3-none-any.whl.metadata\n",
31
+ " Downloading qdrant_client-1.14.2-py3-none-any.whl.metadata (10 kB)\n",
32
+ "Requirement already satisfied: transformers in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (4.51.3)\n",
33
+ "Requirement already satisfied: torch in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (2.7.0)\n",
34
+ "Requirement already satisfied: langchain in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (0.3.24)\n",
35
+ "Collecting pymupdf\n",
36
+ " Obtaining dependency information for pymupdf from https://files.pythonhosted.org/packages/71/c2/a9059607f80dcaf2392f991748cfc53456820392c0220cff02572653512a/pymupdf-1.25.5-cp39-abi3-win_amd64.whl.metadata\n",
37
+ " Downloading pymupdf-1.25.5-cp39-abi3-win_amd64.whl.metadata (3.4 kB)\n",
38
+ "Collecting grpcio>=1.41.0 (from qdrant-client)\n",
39
+ " Obtaining dependency information for grpcio>=1.41.0 from https://files.pythonhosted.org/packages/ee/3f/cf92e7e62ccb8dbdf977499547dfc27133124d6467d3a7d23775bcecb0f9/grpcio-1.71.0-cp311-cp311-win_amd64.whl.metadata\n",
40
+ " Using cached grpcio-1.71.0-cp311-cp311-win_amd64.whl.metadata (4.0 kB)\n",
41
+ "Requirement already satisfied: httpx[http2]>=0.20.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from qdrant-client) (0.28.1)\n",
42
+ "Requirement already satisfied: numpy>=1.21 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from qdrant-client) (2.2.5)\n",
43
+ "Collecting portalocker<3.0.0,>=2.7.0 (from qdrant-client)\n",
44
+ " Obtaining dependency information for portalocker<3.0.0,>=2.7.0 from https://files.pythonhosted.org/packages/9b/fb/a70a4214956182e0d7a9099ab17d50bfcba1056188e9b14f35b9e2b62a0d/portalocker-2.10.1-py3-none-any.whl.metadata\n",
45
+ " Using cached portalocker-2.10.1-py3-none-any.whl.metadata (8.5 kB)\n",
46
+ "Collecting protobuf>=3.20.0 (from qdrant-client)\n",
47
+ " Obtaining dependency information for protobuf>=3.20.0 from https://files.pythonhosted.org/packages/97/e9/7b9f1b259d509aef2b833c29a1f3c39185e2bf21c9c1be1cd11c22cb2149/protobuf-6.30.2-cp310-abi3-win_amd64.whl.metadata\n",
48
+ " Downloading protobuf-6.30.2-cp310-abi3-win_amd64.whl.metadata (593 bytes)\n",
49
+ "Requirement already satisfied: pydantic!=2.0.*,!=2.1.*,!=2.2.0,>=1.10.8 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from qdrant-client) (2.11.4)\n",
50
+ "Requirement already satisfied: urllib3<3,>=1.26.14 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from qdrant-client) (2.4.0)\n",
51
+ "Requirement already satisfied: filelock in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from transformers) (3.18.0)\n",
52
+ "Requirement already satisfied: huggingface-hub<1.0,>=0.30.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from transformers) (0.30.2)\n",
53
+ "Requirement already satisfied: packaging>=20.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from transformers) (24.2)\n",
54
+ "Requirement already satisfied: pyyaml>=5.1 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from transformers) (6.0.2)\n",
55
+ "Requirement already satisfied: regex!=2019.12.17 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from transformers) (2024.11.6)\n",
56
+ "Requirement already satisfied: requests in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from transformers) (2.32.3)\n",
57
+ "Requirement already satisfied: tokenizers<0.22,>=0.21 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from transformers) (0.21.1)\n",
58
+ "Requirement already satisfied: safetensors>=0.4.3 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from transformers) (0.5.3)\n",
59
+ "Requirement already satisfied: tqdm>=4.27 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from transformers) (4.67.1)\n",
60
+ "Requirement already satisfied: typing-extensions>=4.10.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from torch) (4.13.2)\n",
61
+ "Requirement already satisfied: sympy>=1.13.3 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from torch) (1.14.0)\n",
62
+ "Requirement already satisfied: networkx in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from torch) (3.4.2)\n",
63
+ "Requirement already satisfied: jinja2 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from torch) (3.1.6)\n",
64
+ "Requirement already satisfied: fsspec in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from torch) (2025.3.2)\n",
65
+ "Requirement already satisfied: langchain-core<1.0.0,>=0.3.55 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from langchain) (0.3.56)\n",
66
+ "Requirement already satisfied: langchain-text-splitters<1.0.0,>=0.3.8 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from langchain) (0.3.8)\n",
67
+ "Requirement already satisfied: langsmith<0.4,>=0.1.17 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from langchain) (0.3.39)\n",
68
+ "Requirement already satisfied: SQLAlchemy<3,>=1.4 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from langchain) (2.0.40)\n",
69
+ "Requirement already satisfied: anyio in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from httpx[http2]>=0.20.0->qdrant-client) (4.9.0)\n",
70
+ "Requirement already satisfied: certifi in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from httpx[http2]>=0.20.0->qdrant-client) (2025.4.26)\n",
71
+ "Requirement already satisfied: httpcore==1.* in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from httpx[http2]>=0.20.0->qdrant-client) (1.0.9)\n",
72
+ "Requirement already satisfied: idna in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from httpx[http2]>=0.20.0->qdrant-client) (3.10)\n",
73
+ "Collecting h2<5,>=3 (from httpx[http2]>=0.20.0->qdrant-client)\n",
74
+ " Obtaining dependency information for h2<5,>=3 from https://files.pythonhosted.org/packages/d0/9e/984486f2d0a0bd2b024bf4bc1c62688fcafa9e61991f041fb0e2def4a982/h2-4.2.0-py3-none-any.whl.metadata\n",
75
+ " Using cached h2-4.2.0-py3-none-any.whl.metadata (5.1 kB)\n",
76
+ "Requirement already satisfied: h11>=0.16 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from httpcore==1.*->httpx[http2]>=0.20.0->qdrant-client) (0.16.0)\n",
77
+ "Requirement already satisfied: tenacity!=8.4.0,<10.0.0,>=8.1.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from langchain-core<1.0.0,>=0.3.55->langchain) (9.1.2)\n",
78
+ "Requirement already satisfied: jsonpatch<2.0,>=1.33 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from langchain-core<1.0.0,>=0.3.55->langchain) (1.33)\n",
79
+ "Requirement already satisfied: orjson<4.0.0,>=3.9.14 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from langsmith<0.4,>=0.1.17->langchain) (3.10.18)\n",
80
+ "Requirement already satisfied: requests-toolbelt<2.0.0,>=1.0.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from langsmith<0.4,>=0.1.17->langchain) (1.0.0)\n",
81
+ "Requirement already satisfied: zstandard<0.24.0,>=0.23.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from langsmith<0.4,>=0.1.17->langchain) (0.23.0)\n",
82
+ "Requirement already satisfied: pywin32>=226 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from portalocker<3.0.0,>=2.7.0->qdrant-client) (310)\n",
83
+ "Requirement already satisfied: annotated-types>=0.6.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from pydantic!=2.0.*,!=2.1.*,!=2.2.0,>=1.10.8->qdrant-client) (0.7.0)\n",
84
+ "Requirement already satisfied: pydantic-core==2.33.2 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from pydantic!=2.0.*,!=2.1.*,!=2.2.0,>=1.10.8->qdrant-client) (2.33.2)\n",
85
+ "Requirement already satisfied: typing-inspection>=0.4.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from pydantic!=2.0.*,!=2.1.*,!=2.2.0,>=1.10.8->qdrant-client) (0.4.0)\n",
86
+ "Requirement already satisfied: charset-normalizer<4,>=2 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from requests->transformers) (3.4.1)\n",
87
+ "Requirement already satisfied: greenlet>=1 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from SQLAlchemy<3,>=1.4->langchain) (3.2.1)\n",
88
+ "Requirement already satisfied: mpmath<1.4,>=1.1.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from sympy>=1.13.3->torch) (1.3.0)\n",
89
+ "Requirement already satisfied: colorama in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from tqdm>=4.27->transformers) (0.4.6)\n",
90
+ "Requirement already satisfied: MarkupSafe>=2.0 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from jinja2->torch) (3.0.2)\n",
91
+ "Collecting hyperframe<7,>=6.1 (from h2<5,>=3->httpx[http2]>=0.20.0->qdrant-client)\n",
92
+ " Obtaining dependency information for hyperframe<7,>=6.1 from https://files.pythonhosted.org/packages/48/30/47d0bf6072f7252e6521f3447ccfa40b421b6824517f82854703d0f5a98b/hyperframe-6.1.0-py3-none-any.whl.metadata\n",
93
+ " Using cached hyperframe-6.1.0-py3-none-any.whl.metadata (4.3 kB)\n",
94
+ "Collecting hpack<5,>=4.1 (from h2<5,>=3->httpx[http2]>=0.20.0->qdrant-client)\n",
95
+ " Obtaining dependency information for hpack<5,>=4.1 from https://files.pythonhosted.org/packages/07/c6/80c95b1b2b94682a72cbdbfb85b81ae2daffa4291fbfa1b1464502ede10d/hpack-4.1.0-py3-none-any.whl.metadata\n",
96
+ " Using cached hpack-4.1.0-py3-none-any.whl.metadata (4.6 kB)\n",
97
+ "Requirement already satisfied: jsonpointer>=1.9 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from jsonpatch<2.0,>=1.33->langchain-core<1.0.0,>=0.3.55->langchain) (3.0.0)\n",
98
+ "Requirement already satisfied: sniffio>=1.1 in c:\\users\\adm\\documents\\rag_milvus\\.venv\\lib\\site-packages (from anyio->httpx[http2]>=0.20.0->qdrant-client) (1.3.1)\n",
99
+ "Downloading qdrant_client-1.14.2-py3-none-any.whl (327 kB)\n",
100
+ " ---------------------------------------- 0.0/327.7 kB ? eta -:--:--\n",
101
+ " -------------- ------------------------- 122.9/327.7 kB 3.6 MB/s eta 0:00:01\n",
102
+ " ---------------------------------------- 327.7/327.7 kB 5.1 MB/s eta 0:00:00\n",
103
+ "Downloading pymupdf-1.25.5-cp39-abi3-win_amd64.whl (16.6 MB)\n",
104
+ " ---------------------------------------- 0.0/16.6 MB ? eta -:--:--\n",
105
+ " - -------------------------------------- 0.5/16.6 MB 14.4 MB/s eta 0:00:02\n",
106
+ " -- ------------------------------------- 1.0/16.6 MB 12.8 MB/s eta 0:00:02\n",
107
+ " ---- ----------------------------------- 1.8/16.6 MB 14.0 MB/s eta 0:00:02\n",
108
+ " ------ --------------------------------- 2.5/16.6 MB 14.6 MB/s eta 0:00:01\n",
109
+ " ------- -------------------------------- 3.3/16.6 MB 15.0 MB/s eta 0:00:01\n",
110
+ " --------- ------------------------------ 4.0/16.6 MB 14.9 MB/s eta 0:00:01\n",
111
+ " ----------- ---------------------------- 4.7/16.6 MB 15.1 MB/s eta 0:00:01\n",
112
+ " ------------- -------------------------- 5.4/16.6 MB 15.1 MB/s eta 0:00:01\n",
113
+ " -------------- ------------------------- 6.2/16.6 MB 15.1 MB/s eta 0:00:01\n",
114
+ " ---------------- ----------------------- 6.9/16.6 MB 15.2 MB/s eta 0:00:01\n",
115
+ " ------------------ --------------------- 7.6/16.6 MB 15.2 MB/s eta 0:00:01\n",
116
+ " -------------------- ------------------- 8.3/16.6 MB 15.6 MB/s eta 0:00:01\n",
117
+ " --------------------- ------------------ 9.1/16.6 MB 15.7 MB/s eta 0:00:01\n",
118
+ " ----------------------- ---------------- 9.8/16.6 MB 15.6 MB/s eta 0:00:01\n",
119
+ " ------------------------- -------------- 10.4/16.6 MB 15.6 MB/s eta 0:00:01\n",
120
+ " -------------------------- ------------- 11.0/16.6 MB 15.6 MB/s eta 0:00:01\n",
121
+ " ---------------------------- ----------- 11.8/16.6 MB 15.6 MB/s eta 0:00:01\n",
122
+ " ------------------------------ --------- 12.5/16.6 MB 15.6 MB/s eta 0:00:01\n",
123
+ " ------------------------------- -------- 13.2/16.6 MB 15.6 MB/s eta 0:00:01\n",
124
+ " --------------------------------- ------ 13.9/16.6 MB 15.6 MB/s eta 0:00:01\n",
125
+ " ----------------------------------- ---- 14.6/16.6 MB 15.2 MB/s eta 0:00:01\n",
126
+ " ------------------------------------ --- 15.3/16.6 MB 15.2 MB/s eta 0:00:01\n",
127
+ " -------------------------------------- - 16.1/16.6 MB 16.0 MB/s eta 0:00:01\n",
128
+ " --------------------------------------- 16.6/16.6 MB 15.6 MB/s eta 0:00:01\n",
129
+ " ---------------------------------------- 16.6/16.6 MB 15.2 MB/s eta 0:00:00\n",
130
+ "Using cached grpcio-1.71.0-cp311-cp311-win_amd64.whl (4.3 MB)\n",
131
+ "Using cached portalocker-2.10.1-py3-none-any.whl (18 kB)\n",
132
+ "Downloading protobuf-6.30.2-cp310-abi3-win_amd64.whl (431 kB)\n",
133
+ " ---------------------------------------- 0.0/431.0 kB ? eta -:--:--\n",
134
+ " --------------------------------------- 431.0/431.0 kB 26.3 MB/s eta 0:00:00\n",
135
+ "Using cached h2-4.2.0-py3-none-any.whl (60 kB)\n",
136
+ "Using cached hpack-4.1.0-py3-none-any.whl (34 kB)\n",
137
+ "Using cached hyperframe-6.1.0-py3-none-any.whl (13 kB)\n",
138
+ "Installing collected packages: pymupdf, protobuf, portalocker, hyperframe, hpack, grpcio, h2, qdrant-client\n",
139
+ "Successfully installed grpcio-1.71.0 h2-4.2.0 hpack-4.1.0 hyperframe-6.1.0 portalocker-2.10.1 protobuf-6.30.2 pymupdf-1.25.5 qdrant-client-1.14.2\n",
140
+ "Note: you may need to restart the kernel to use updated packages.\n"
141
+ ]
142
+ },
143
+ {
144
+ "name": "stderr",
145
+ "output_type": "stream",
146
+ "text": [
147
+ "\n",
148
+ "[notice] A new release of pip is available: 23.2.1 -> 25.1\n",
149
+ "[notice] To update, run: python.exe -m pip install --upgrade pip\n"
150
+ ]
151
+ }
152
+ ],
153
+ "source": [
154
+ "%pip install qdrant-client transformers torch langchain pymupdf"
155
+ ]
156
+ },
157
+ {
158
+ "cell_type": "markdown",
159
+ "id": "a4833977",
160
+ "metadata": {},
161
+ "source": [
162
+ "Importamos librerias"
163
+ ]
164
+ },
165
+ {
166
+ "cell_type": "code",
167
+ "execution_count": 19,
168
+ "id": "1684b4de",
169
+ "metadata": {},
170
+ "outputs": [],
171
+ "source": [
172
+ "from sentence_transformers import SentenceTransformer\n",
173
+ "from langchain.text_splitter import RecursiveCharacterTextSplitter\n",
174
+ "from langchain.schema import Document\n",
175
+ "from qdrant_client import QdrantClient\n",
176
+ "from qdrant_client.models import PointStruct, Distance, VectorParams\n",
177
+ "import fitz # PyMuPDF"
178
+ ]
179
+ },
180
+ {
181
+ "cell_type": "markdown",
182
+ "id": "e612e674",
183
+ "metadata": {},
184
+ "source": [
185
+ "Definimos funciones\n",
186
+ "1) Cargar los pdf por sus bloques de paginas"
187
+ ]
188
+ },
189
+ {
190
+ "cell_type": "code",
191
+ "execution_count": 20,
192
+ "id": "5a594ed8",
193
+ "metadata": {},
194
+ "outputs": [],
195
+ "source": [
196
+ "def pdfachunk(path, chunk_size_pages=20):\n",
197
+ " doc = fitz.open(path)\n",
198
+ " chunks = []\n",
199
+ " for i in range(0, len(doc), chunk_size_pages):\n",
200
+ " text = \"\"\n",
201
+ " for page_num in range(i, min(i + chunk_size_pages, len(doc))):\n",
202
+ " text += doc[page_num].get_text()\n",
203
+ " chunks.append(text)\n",
204
+ " doc.close()\n",
205
+ " return chunks"
206
+ ]
207
+ },
208
+ {
209
+ "cell_type": "markdown",
210
+ "id": "59d048b7",
211
+ "metadata": {},
212
+ "source": [
213
+ "2) Dividir texto en chunks más pequeños con solapamiento"
214
+ ]
215
+ },
216
+ {
217
+ "cell_type": "code",
218
+ "execution_count": 21,
219
+ "id": "bffac6eb",
220
+ "metadata": {},
221
+ "outputs": [],
222
+ "source": [
223
+ "def split_chunks(raw_chunks, chunk_size=1024, chunk_overlap=100):\n",
224
+ " docs = [Document(page_content=chunk) for chunk in raw_chunks]\n",
225
+ " splitter = RecursiveCharacterTextSplitter(\n",
226
+ " chunk_size=chunk_size,\n",
227
+ " chunk_overlap=chunk_overlap,\n",
228
+ " separators=[\"\\n\\n\", \"\\n\", \".\", \" \"]\n",
229
+ " )\n",
230
+ " return splitter.split_documents(docs)"
231
+ ]
232
+ },
233
+ {
234
+ "cell_type": "markdown",
235
+ "id": "8664bf6f",
236
+ "metadata": {},
237
+ "source": [
238
+ "3) Generar embeddings en batch"
239
+ ]
240
+ },
241
+ {
242
+ "cell_type": "code",
243
+ "execution_count": 22,
244
+ "id": "35a4df0b",
245
+ "metadata": {},
246
+ "outputs": [],
247
+ "source": [
248
+ "def generaremben(model, texts):\n",
249
+ " texts = [t for t in texts if t.strip()] # filtra vacíos\n",
250
+ " if not texts:\n",
251
+ " raise ValueError(\"No hay textos válidos para generar embeddings.\")\n",
252
+ " return model.encode(texts, batch_size=16, show_progress_bar=True)\n"
253
+ ]
254
+ },
255
+ {
256
+ "cell_type": "markdown",
257
+ "id": "c28a5724",
258
+ "metadata": {},
259
+ "source": [
260
+ "4) Insertar los docemtos en QDRANT localemente"
261
+ ]
262
+ },
263
+ {
264
+ "cell_type": "code",
265
+ "execution_count": 23,
266
+ "id": "3c61ddca",
267
+ "metadata": {},
268
+ "outputs": [],
269
+ "source": [
270
+ "def insertarenqdra(embeddings, texts, collection_name=\"pdf_chunks\"):\n",
271
+ " client = QdrantClient(path=\"./qdrant_data\") # persistente\n",
272
+ "\n",
273
+ " dim = len(embeddings[0])\n",
274
+ " client.recreate_collection(\n",
275
+ " collection_name=collection_name,\n",
276
+ " vectors_config=VectorParams(size=dim, distance=Distance.COSINE)\n",
277
+ " )\n",
278
+ "\n",
279
+ " points = [\n",
280
+ " PointStruct(id=i, vector=embeddings[i].tolist(), payload={\"text\": texts[i]})\n",
281
+ " for i in range(len(embeddings))\n",
282
+ " ]\n",
283
+ "\n",
284
+ " client.upsert(collection_name=collection_name, points=points)\n",
285
+ " print(f\"✅ Insertados {len(points)} vectores en Qdrant.\")"
286
+ ]
287
+ },
288
+ {
289
+ "cell_type": "markdown",
290
+ "id": "566e06c8",
291
+ "metadata": {},
292
+ "source": [
293
+ "5) Funcion modelo para no cargarlo siempre"
294
+ ]
295
+ },
296
+ {
297
+ "cell_type": "code",
298
+ "execution_count": 6,
299
+ "id": "eec86477",
300
+ "metadata": {},
301
+ "outputs": [],
302
+ "source": [
303
+ "def load_nv_model():\n",
304
+ " return AutoModel.from_pretrained(\"nvidia/NV-Embed-v2\", trust_remote_code=True)"
305
+ ]
306
+ },
307
+ {
308
+ "cell_type": "markdown",
309
+ "id": "08510d36",
310
+ "metadata": {},
311
+ "source": [
312
+ "Probamos"
313
+ ]
314
+ },
315
+ {
316
+ "cell_type": "code",
317
+ "execution_count": 24,
318
+ "id": "2735d1a1",
319
+ "metadata": {},
320
+ "outputs": [],
321
+ "source": [
322
+ "pdf_path=\"./DOCS/Decreto-Supremo-N_-018-2019-JUS.pdf\" "
323
+ ]
324
+ },
325
+ {
326
+ "cell_type": "code",
327
+ "execution_count": 25,
328
+ "id": "6ede8122",
329
+ "metadata": {},
330
+ "outputs": [],
331
+ "source": [
332
+ "pdf_chunks = pdfachunk(pdf_path)"
333
+ ]
334
+ },
335
+ {
336
+ "cell_type": "code",
337
+ "execution_count": 26,
338
+ "id": "8f33af13",
339
+ "metadata": {},
340
+ "outputs": [],
341
+ "source": [
342
+ "split_docs = split_chunks(pdf_chunks)"
343
+ ]
344
+ },
345
+ {
346
+ "cell_type": "code",
347
+ "execution_count": 27,
348
+ "id": "b0fb3761",
349
+ "metadata": {},
350
+ "outputs": [],
351
+ "source": [
352
+ "texts = [doc.page_content for doc in split_docs]"
353
+ ]
354
+ },
355
+ {
356
+ "cell_type": "markdown",
357
+ "id": "85f2ee3f",
358
+ "metadata": {},
359
+ "source": [
360
+ "Definimos nuestro modelo de embbending"
361
+ ]
362
+ },
363
+ {
364
+ "cell_type": "markdown",
365
+ "id": "5eb18c36",
366
+ "metadata": {},
367
+ "source": [
368
+ "NECESITAMOS DATASETS Y EINOPS"
369
+ ]
370
+ },
371
+ {
372
+ "cell_type": "code",
373
+ "execution_count": null,
374
+ "id": "74262eaa",
375
+ "metadata": {},
376
+ "outputs": [],
377
+ "source": [
378
+ "%pip install datasets einops"
379
+ ]
380
+ },
381
+ {
382
+ "cell_type": "code",
383
+ "execution_count": 28,
384
+ "id": "93bbbcde",
385
+ "metadata": {},
386
+ "outputs": [
387
+ {
388
+ "name": "stderr",
389
+ "output_type": "stream",
390
+ "text": [
391
+ "Xet Storage is enabled for this repo, but the 'hf_xet' package is not installed. Falling back to regular HTTP download. For better performance, install the package with: `pip install huggingface_hub[hf_xet]` or `pip install hf_xet`\n"
392
+ ]
393
+ }
394
+ ],
395
+ "source": [
396
+ "model = SentenceTransformer(\"all-MiniLM-L6-v2\")"
397
+ ]
398
+ },
399
+ {
400
+ "cell_type": "code",
401
+ "execution_count": 13,
402
+ "id": "f9f4d5bd",
403
+ "metadata": {},
404
+ "outputs": [
405
+ {
406
+ "name": "stdout",
407
+ "output_type": "stream",
408
+ "text": [
409
+ "<class 'transformers_modules.nvidia.NV-Embed-v2.c50d55f43bde7e6a18e0eaa15a62fd63a930f1a1.modeling_nvembed.NVEmbedModel'>\n"
410
+ ]
411
+ }
412
+ ],
413
+ "source": [
414
+ "print(type(model))"
415
+ ]
416
+ },
417
+ {
418
+ "cell_type": "code",
419
+ "execution_count": 29,
420
+ "id": "e594d6af",
421
+ "metadata": {},
422
+ "outputs": [
423
+ {
424
+ "name": "stderr",
425
+ "output_type": "stream",
426
+ "text": [
427
+ "Batches: 100%|██████████| 12/12 [00:03<00:00, 3.29it/s]\n"
428
+ ]
429
+ }
430
+ ],
431
+ "source": [
432
+ "embeddings = generaremben(model, texts)"
433
+ ]
434
+ },
435
+ {
436
+ "cell_type": "code",
437
+ "execution_count": 30,
438
+ "id": "beba3991",
439
+ "metadata": {},
440
+ "outputs": [
441
+ {
442
+ "name": "stderr",
443
+ "output_type": "stream",
444
+ "text": [
445
+ "C:\\Users\\adm\\AppData\\Local\\Temp\\ipykernel_26272\\752761030.py:5: DeprecationWarning: `recreate_collection` method is deprecated and will be removed in the future. Use `collection_exists` to check collection existence and `create_collection` instead.\n",
446
+ " client.recreate_collection(\n"
447
+ ]
448
+ },
449
+ {
450
+ "name": "stdout",
451
+ "output_type": "stream",
452
+ "text": [
453
+ "✅ Insertados 181 vectores en Qdrant.\n"
454
+ ]
455
+ }
456
+ ],
457
+ "source": [
458
+ "insertarenqdra(embeddings, texts, collection_name=\"jus_decreto_018\")"
459
+ ]
460
+ },
461
+ {
462
+ "cell_type": "markdown",
463
+ "id": "2c241dc6",
464
+ "metadata": {},
465
+ "source": [
466
+ "Funcion para consultar con qdrant"
467
+ ]
468
+ },
469
+ {
470
+ "cell_type": "code",
471
+ "execution_count": 31,
472
+ "id": "86beaf73",
473
+ "metadata": {},
474
+ "outputs": [],
475
+ "source": [
476
+ "from qdrant_client import QdrantClient\n",
477
+ "import numpy as np"
478
+ ]
479
+ },
480
+ {
481
+ "cell_type": "code",
482
+ "execution_count": null,
483
+ "id": "b81de586",
484
+ "metadata": {},
485
+ "outputs": [],
486
+ "source": [
487
+ "def query_qdrant(query, model, collection_name, top_k=5):\n",
488
+ " # Generar embedding de la consulta\n",
489
+ " query_embedding = model.encode([query])[0]\n",
490
+ " \n",
491
+ " # Conexión al cliente Qdrant\n",
492
+ " client = QdrantClient(path=\"./qdrant_data\")\n",
493
+ "\n",
494
+ " # Realizar búsqueda en la colección\n",
495
+ " results = client.search(\n",
496
+ " collection_name=collection_name,\n",
497
+ " query_vector=query_embedding.tolist(),\n",
498
+ " limit=top_k, # Limitar a los primeros K resultados más similares\n",
499
+ " with_payload=True # Incluir el texto en los resultados\n",
500
+ " )\n",
501
+ "\n",
502
+ " return results"
503
+ ]
504
+ },
505
+ {
506
+ "cell_type": "code",
507
+ "execution_count": 36,
508
+ "id": "9d448736",
509
+ "metadata": {},
510
+ "outputs": [
511
+ {
512
+ "ename": "RuntimeError",
513
+ "evalue": "Storage folder ./qdrant_data is already accessed by another instance of Qdrant client. If you require concurrent access, use Qdrant server instead.",
514
+ "output_type": "error",
515
+ "traceback": [
516
+ "\u001b[31m---------------------------------------------------------------------------\u001b[39m",
517
+ "\u001b[31merror\u001b[39m Traceback (most recent call last)",
518
+ "\u001b[36mFile \u001b[39m\u001b[32mc:\\Users\\adm\\Documents\\rag_MILVUS\\.venv\\Lib\\site-packages\\portalocker\\portalocker.py:49\u001b[39m, in \u001b[36mlock\u001b[39m\u001b[34m(file_, flags)\u001b[39m\n\u001b[32m 48\u001b[39m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[32m---> \u001b[39m\u001b[32m49\u001b[39m \u001b[43mwin32file\u001b[49m\u001b[43m.\u001b[49m\u001b[43mLockFileEx\u001b[49m\u001b[43m(\u001b[49m\u001b[43mos_fh\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mmode\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[32;43m0\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43m-\u001b[49m\u001b[32;43m0x10000\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43m__overlapped\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m 50\u001b[39m \u001b[38;5;28;01mexcept\u001b[39;00m pywintypes.error \u001b[38;5;28;01mas\u001b[39;00m exc_value:\n\u001b[32m 51\u001b[39m \u001b[38;5;66;03m# error: (33, 'LockFileEx', 'The process cannot access the file\u001b[39;00m\n\u001b[32m 52\u001b[39m \u001b[38;5;66;03m# because another process has locked a portion of the file.')\u001b[39;00m\n",
519
+ "\u001b[31merror\u001b[39m: (33, 'LockFileEx', 'El proceso no tiene acceso al archivo porque otro proceso tiene bloqueada una parte del archivo.')",
520
+ "\nThe above exception was the direct cause of the following exception:\n",
521
+ "\u001b[31mAlreadyLocked\u001b[39m Traceback (most recent call last)",
522
+ "\u001b[36mFile \u001b[39m\u001b[32mc:\\Users\\adm\\Documents\\rag_MILVUS\\.venv\\Lib\\site-packages\\qdrant_client\\local\\qdrant_local.py:133\u001b[39m, in \u001b[36mQdrantLocal._load\u001b[39m\u001b[34m(self)\u001b[39m\n\u001b[32m 132\u001b[39m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[32m--> \u001b[39m\u001b[32m133\u001b[39m \u001b[43mportalocker\u001b[49m\u001b[43m.\u001b[49m\u001b[43mlock\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m 134\u001b[39m \u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43m_flock_file\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 135\u001b[39m \u001b[43m \u001b[49m\u001b[43mportalocker\u001b[49m\u001b[43m.\u001b[49m\u001b[43mLockFlags\u001b[49m\u001b[43m.\u001b[49m\u001b[43mEXCLUSIVE\u001b[49m\u001b[43m \u001b[49m\u001b[43m|\u001b[49m\u001b[43m \u001b[49m\u001b[43mportalocker\u001b[49m\u001b[43m.\u001b[49m\u001b[43mLockFlags\u001b[49m\u001b[43m.\u001b[49m\u001b[43mNON_BLOCKING\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 136\u001b[39m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m 137\u001b[39m \u001b[38;5;28;01mexcept\u001b[39;00m portalocker.exceptions.LockException:\n",
523
+ "\u001b[36mFile \u001b[39m\u001b[32mc:\\Users\\adm\\Documents\\rag_MILVUS\\.venv\\Lib\\site-packages\\portalocker\\portalocker.py:54\u001b[39m, in \u001b[36mlock\u001b[39m\u001b[34m(file_, flags)\u001b[39m\n\u001b[32m 53\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m exc_value.winerror == winerror.ERROR_LOCK_VIOLATION:\n\u001b[32m---> \u001b[39m\u001b[32m54\u001b[39m \u001b[38;5;28;01mraise\u001b[39;00m exceptions.AlreadyLocked(\n\u001b[32m 55\u001b[39m exceptions.LockException.LOCK_FAILED,\n\u001b[32m 56\u001b[39m exc_value.strerror,\n\u001b[32m 57\u001b[39m fh=file_,\n\u001b[32m 58\u001b[39m ) \u001b[38;5;28;01mfrom\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34;01mexc_value\u001b[39;00m\n\u001b[32m 59\u001b[39m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[32m 60\u001b[39m \u001b[38;5;66;03m# Q: Are there exceptions/codes we should be dealing with\u001b[39;00m\n\u001b[32m 61\u001b[39m \u001b[38;5;66;03m# here?\u001b[39;00m\n",
524
+ "\u001b[31mAlreadyLocked\u001b[39m: (1, 'El proceso no tiene acceso al archivo porque otro proceso tiene bloqueada una parte del archivo.')",
525
+ "\nDuring handling of the above exception, another exception occurred:\n",
526
+ "\u001b[31mRuntimeError\u001b[39m Traceback (most recent call last)",
527
+ "\u001b[36mCell\u001b[39m\u001b[36m \u001b[39m\u001b[32mIn[36]\u001b[39m\u001b[32m, line 2\u001b[39m\n\u001b[32m 1\u001b[39m query = \u001b[33m\"\u001b[39m\u001b[33m¿Cuál es el propósito de la Ley 018-2019?\u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m----> \u001b[39m\u001b[32m2\u001b[39m results = \u001b[43mquery_qdrant\u001b[49m\u001b[43m(\u001b[49m\u001b[43mquery\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\u001b[33;43m\"\u001b[39;49m\u001b[33;43mjus_decreto_018\u001b[39;49m\u001b[33;43m\"\u001b[39;49m\u001b[43m)\u001b[49m\n",
528
+ "\u001b[36mCell\u001b[39m\u001b[36m \u001b[39m\u001b[32mIn[34]\u001b[39m\u001b[32m, line 6\u001b[39m, in \u001b[36mquery_qdrant\u001b[39m\u001b[34m(query, model, collection_name, top_k)\u001b[39m\n\u001b[32m 3\u001b[39m query_embedding = model.encode([query])[\u001b[32m0\u001b[39m]\n\u001b[32m 5\u001b[39m \u001b[38;5;66;03m# Conexión al cliente Qdrant\u001b[39;00m\n\u001b[32m----> \u001b[39m\u001b[32m6\u001b[39m client = \u001b[43mQdrantClient\u001b[49m\u001b[43m(\u001b[49m\u001b[43mpath\u001b[49m\u001b[43m=\u001b[49m\u001b[33;43m\"\u001b[39;49m\u001b[33;43m./qdrant_data\u001b[39;49m\u001b[33;43m\"\u001b[39;49m\u001b[43m)\u001b[49m\n\u001b[32m 8\u001b[39m \u001b[38;5;66;03m# Realizar búsqueda en la colección\u001b[39;00m\n\u001b[32m 9\u001b[39m results = client.search(\n\u001b[32m 10\u001b[39m collection_name=collection_name,\n\u001b[32m 11\u001b[39m query_vector=query_embedding.tolist(),\n\u001b[32m 12\u001b[39m limit=top_k, \u001b[38;5;66;03m# Limitar a los primeros K resultados más similares\u001b[39;00m\n\u001b[32m 13\u001b[39m with_payload=\u001b[38;5;28;01mTrue\u001b[39;00m \u001b[38;5;66;03m# Incluir el texto en los resultados\u001b[39;00m\n\u001b[32m 14\u001b[39m )\n",
529
+ "\u001b[36mFile \u001b[39m\u001b[32mc:\\Users\\adm\\Documents\\rag_MILVUS\\.venv\\Lib\\site-packages\\qdrant_client\\qdrant_client.py:133\u001b[39m, in \u001b[36mQdrantClient.__init__\u001b[39m\u001b[34m(self, location, url, port, grpc_port, prefer_grpc, https, api_key, prefix, timeout, host, path, force_disable_check_same_thread, grpc_options, auth_token_provider, cloud_inference, local_inference_batch_size, check_compatibility, **kwargs)\u001b[39m\n\u001b[32m 128\u001b[39m \u001b[38;5;28mself\u001b[39m._client = QdrantLocal(\n\u001b[32m 129\u001b[39m location=location,\n\u001b[32m 130\u001b[39m force_disable_check_same_thread=force_disable_check_same_thread,\n\u001b[32m 131\u001b[39m )\n\u001b[32m 132\u001b[39m \u001b[38;5;28;01melif\u001b[39;00m path \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n\u001b[32m--> \u001b[39m\u001b[32m133\u001b[39m \u001b[38;5;28mself\u001b[39m._client = \u001b[43mQdrantLocal\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m 134\u001b[39m \u001b[43m \u001b[49m\u001b[43mlocation\u001b[49m\u001b[43m=\u001b[49m\u001b[43mpath\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 135\u001b[39m \u001b[43m \u001b[49m\u001b[43mforce_disable_check_same_thread\u001b[49m\u001b[43m=\u001b[49m\u001b[43mforce_disable_check_same_thread\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m 136\u001b[39m \u001b[43m \u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m 137\u001b[39m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[32m 138\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m location \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m \u001b[38;5;129;01mand\u001b[39;00m url \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m:\n",
530
+ "\u001b[36mFile \u001b[39m\u001b[32mc:\\Users\\adm\\Documents\\rag_MILVUS\\.venv\\Lib\\site-packages\\qdrant_client\\local\\qdrant_local.py:66\u001b[39m, in \u001b[36mQdrantLocal.__init__\u001b[39m\u001b[34m(self, location, force_disable_check_same_thread)\u001b[39m\n\u001b[32m 64\u001b[39m \u001b[38;5;28mself\u001b[39m.aliases: \u001b[38;5;28mdict\u001b[39m[\u001b[38;5;28mstr\u001b[39m, \u001b[38;5;28mstr\u001b[39m] = {}\n\u001b[32m 65\u001b[39m \u001b[38;5;28mself\u001b[39m._flock_file: Optional[TextIOWrapper] = \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[32m---> \u001b[39m\u001b[32m66\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43m_load\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[32m 67\u001b[39m \u001b[38;5;28mself\u001b[39m._closed: \u001b[38;5;28mbool\u001b[39m = \u001b[38;5;28;01mFalse\u001b[39;00m\n",
531
+ "\u001b[36mFile \u001b[39m\u001b[32mc:\\Users\\adm\\Documents\\rag_MILVUS\\.venv\\Lib\\site-packages\\qdrant_client\\local\\qdrant_local.py:138\u001b[39m, in \u001b[36mQdrantLocal._load\u001b[39m\u001b[34m(self)\u001b[39m\n\u001b[32m 133\u001b[39m portalocker.lock(\n\u001b[32m 134\u001b[39m \u001b[38;5;28mself\u001b[39m._flock_file,\n\u001b[32m 135\u001b[39m portalocker.LockFlags.EXCLUSIVE | portalocker.LockFlags.NON_BLOCKING,\n\u001b[32m 136\u001b[39m )\n\u001b[32m 137\u001b[39m \u001b[38;5;28;01mexcept\u001b[39;00m portalocker.exceptions.LockException:\n\u001b[32m--> \u001b[39m\u001b[32m138\u001b[39m \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mRuntimeError\u001b[39;00m(\n\u001b[32m 139\u001b[39m \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mStorage folder \u001b[39m\u001b[38;5;132;01m{\u001b[39;00m\u001b[38;5;28mself\u001b[39m.location\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m is already accessed by another instance of Qdrant client.\u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m 140\u001b[39m \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33m If you require concurrent access, use Qdrant server instead.\u001b[39m\u001b[33m\"\u001b[39m\n\u001b[32m 141\u001b[39m )\n",
532
+ "\u001b[31mRuntimeError\u001b[39m: Storage folder ./qdrant_data is already accessed by another instance of Qdrant client. If you require concurrent access, use Qdrant server instead."
533
+ ]
534
+ }
535
+ ],
536
+ "source": [
537
+ "query = \"¿Cuál es el propósito de la Ley 018-2019?\"\n",
538
+ "results = query_qdrant(query, model,\"jus_decreto_018\")"
539
+ ]
540
+ },
541
+ {
542
+ "cell_type": "code",
543
+ "execution_count": null,
544
+ "id": "61d76427",
545
+ "metadata": {},
546
+ "outputs": [],
547
+ "source": []
548
+ }
549
+ ],
550
+ "metadata": {
551
+ "kernelspec": {
552
+ "display_name": ".venv",
553
+ "language": "python",
554
+ "name": "python3"
555
+ },
556
+ "language_info": {
557
+ "codemirror_mode": {
558
+ "name": "ipython",
559
+ "version": 3
560
+ },
561
+ "file_extension": ".py",
562
+ "mimetype": "text/x-python",
563
+ "name": "python",
564
+ "nbconvert_exporter": "python",
565
+ "pygments_lexer": "ipython3",
566
+ "version": "3.11.5"
567
+ }
568
+ },
569
+ "nbformat": 4,
570
+ "nbformat_minor": 5
571
+ }
chatbox_v1.py ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import streamlit as st
2
+ from Rag_milvus import query_qdrant, obtener_colecciones, query_qdrant_sinumbral
3
+ from Llm_local import get_response_from_mistral, generarPages
4
+ from sentence_transformers import SentenceTransformer
5
+
6
+ col1, col2 = st.columns([1, 4])
7
+ with col1:
8
+ st.image("Procuradurialogo.jpg", width=600)
9
+
10
+ with col2:
11
+ st.markdown("""
12
+ <div style='display: flex; align-items: center; height: 100%;'>
13
+ <h1 style='margin: 0; text-align: center;'>ProcurIA</h1>
14
+ </div>
15
+ """, unsafe_allow_html=True)
16
+
17
+ st.sidebar.title("Menú de Funciones")
18
+ generarPages()
19
+ #Inicializamos nuestro historial de chat
20
+ if "messages" not in st.session_state:
21
+ st.session_state.messages = [{"role": "assistant", "content": "Hola!, en que puedo ayudarte?"}]
22
+
23
+ #Definimos modelo
24
+ model = SentenceTransformer("all-MiniLM-L6-v2")
25
+
26
+ #Elegimos una coleccion
27
+ colecciones = obtener_colecciones()
28
+ coleccion_seleccionada = st.sidebar.selectbox("Selecciona una colección", colecciones)
29
+
30
+ # Mostrar el historial de mensajes
31
+ for message in st.session_state.messages:
32
+ with st.chat_message(message["role"]):
33
+ st.markdown(message["content"])
34
+
35
+ # Entrada del usuario
36
+ if prompt := st.chat_input("Escribe tus dudas"):
37
+ st.session_state.messages.append({"role": "user", "content": prompt})
38
+
39
+ with st.chat_message("user"):
40
+ st.markdown(prompt)
41
+
42
+ with st.chat_message("assistant"):
43
+ if coleccion_seleccionada == "Todas las colecciones":
44
+ colecciones_disponibles = obtener_colecciones()
45
+ results = []
46
+ umbral=1
47
+ for coleccion in colecciones_disponibles[1:]:
48
+ coleccion_results = query_qdrant_sinumbral(prompt,model,coleccion)
49
+ results.extend(coleccion_results)
50
+ else:
51
+ umbral=0.56
52
+ results = query_qdrant(prompt, model, coleccion_seleccionada,5,umbral)
53
+
54
+ if not results:
55
+ response = "Disculpa, no tengo información para responder esa pregunta."
56
+ else:
57
+ response = st.write_stream(get_response_from_mistral(prompt, results))
58
+
59
+ st.session_state.messages.append({"role": "assistant", "content": response})
60
+ st.write(results)