Carregando Diferentes Formatos de Documentos
Explore métodos para ingerir dados de várias fontes, como PDFs, páginas da Web, bancos de dados e tipos personalizados de arquivos.
Carregando Diferentes Formatos de Documentos é uma aula grátis de LLM Apps in Production (RAG + Vector DB + Caching) no CoddyKit. Esta é a aula 1 de 4. Você pode ler a aula completa abaixo gratuitamente — depois pratica ao vivo no navegador com um editor de código integrado e um tutor de IA 24/7. Faz parte do caminho de aprendizado de LLM Apps in Production (RAG + Vector DB + Caching), e seu progresso é sincronizado entre a web e o app CoddyKit. O curso de LLM Apps in Production (RAG + Vector DB + Caching) inclui 4 aulas no total.
Partes desta aula ainda não foram traduzidas e aparecem em inglês.
Ingesting Diverse Document Types
Welcome! In RAG, your LLM needs information from various sources. This lesson explores how to load data from different document formats into your application.
The goal is to get raw text from places like web pages, PDFs, and databases, preparing it for the next steps in your RAG pipeline.
Loading Web Pages (HTML)
Web pages are a common source of information. To ingest them, you typically:
- Fetch the HTML: Use an HTTP client to download the page content from a URL.
- Parse the HTML: Extract the main text and discard navigation, ads, and other irrelevant elements.
Libraries like requests for fetching and BeautifulSoup for parsing are very popular in Python.
Web Page Loading Example
This Python snippet demonstrates fetching a simple web page and extracting its title. Imagine doing this for many pages to build your knowledge base!
import requests
from bs4 import BeautifulSoup
def main():
url = "http://quotes.toscrape.com/"
try:
response = requests.get(url)
response.raise_for_status() # Raise HTTPError for bad responses
soup = BeautifulSoup(response.text, 'html.parser')
title = soup.find('title').get_text()
print(f"Page Title: {title}")
except requests.exceptions.RequestException as e:
print(f"Error fetching URL: {e}")
if __name__ == "__main__":
main()Extracting Text from PDFs
PDFs (Portable Document Format) are widely used for reports and documents. Extracting text from PDFs can be tricky due to their complex structure, which combines text, images, and formatting.
Fortunately, programming libraries exist to help. They can read the PDF structure and pull out the textual content, often page by page.
PDF Text Extraction Example
This conceptual Python code shows how you might extract text from the first page of a PDF using a library like pypdf. For a real run, you'd need a sample.pdf file.
from pypdf import PdfReader
def main():
# In a real scenario, 'sample.pdf' would exist
# For this example, we'll simulate the output
pdf_file_path = "sample.pdf"
print(f"Attempting to read from: {pdf_file_path}")
print("\n--- Simulated PDF Content ---")
print("This is some text from the first page of a sample PDF document.")
print("It contains important information for our RAG system.")
print("-----------------------------")
# Actual code might look like this:
# reader = PdfReader(pdf_file_path)
# page = reader.pages[0]
# text = page.extract_text()
# print(text)
if __name__ == "__main__":
main()Loading Data from Databases
Databases, both SQL (like PostgreSQL, MySQL) and NoSQL (like MongoDB, Cassandra), store structured data that can be valuable for RAG.
To ingest from databases:
- Connect: Establish a connection using database drivers.
- Query: Write queries (e.g., SQL statements) to retrieve relevant data.
- Process: Extract text fields from the query results.
Database Loading Example
Here's a Python example using SQLite, an embedded SQL database. It creates a simple table, inserts data, and then retrieves it. This is a common pattern for database ingestion.
import sqlite3
def main():
# Connect to an in-memory SQLite database
conn = sqlite3.connect(':memory:')
cursor = conn.cursor()
# Create a simple table
cursor.execute('''
CREATE TABLE IF NOT EXISTS documents (
id INTEGER PRIMARY KEY,
title TEXT,
content TEXT
)
''')
# Insert some data
cursor.execute("INSERT INTO documents (title, content) VALUES (?, ?)",
("RAG Overview", "RAG enhances LLMs by retrieving relevant docs."))
cursor.execute("INSERT INTO documents (title, content) VALUES (?, ?)",
("Vector DBs", "Store embeddings for fast similarity search."))
conn.commit()
# Retrieve data
cursor.execute("SELECT title, content FROM documents")
rows = cursor.fetchall()
print("--- Retrieved Documents ---")
for row in rows:
print(f"Title: {row[0]}, Content: {row[1]}")
print("---------------------------")
conn.close()
if __name__ == "__main__":
main()Handling Plain Text & Custom Files
Beyond specific formats, you'll often deal with plain text files (.txt, .md) or custom formats (e.g., CSV, JSON). For these:
- Plain Text: Read directly, paying attention to encoding (UTF-8 is common).
- Custom Formats: Use libraries specific to the format (e.g.,
csv,jsonmodules in Python) to parse and extract text fields.
The key is transforming the data into a usable text string.
Unified Data Loading with Libraries
For complex RAG systems, you don't always need to write custom loaders for every format. Libraries like LlamaIndex and LangChain offer 'Document Loaders' that abstract away much of this complexity.
- They provide connectors for many data sources (web, PDF, databases, cloud storage).
- They often handle basic parsing and text extraction automatically.
These tools simplify the initial ingestion step, letting you focus on retrieval and generation.
Check Your Knowledge
Which of the following is typically a primary challenge when extracting text content from PDF documents for a RAG system?
Recap: Loading Diverse Data
Great job! You've learned the fundamentals of loading diverse document formats for your RAG system.
- We covered fetching and parsing web pages.
- Discussed extracting text from complex PDFs.
- Explored querying databases for structured content.
- Touched upon handling plain text and other custom files.
- Recognized the value of unified data loading libraries.
The next step is to prepare this raw text for effective retrieval!
Perguntas Frequentes
A aula “Carregando Diferentes Formatos de Documentos” é grátis?
Sim — o texto completo de “Carregando Diferentes Formatos de Documentos” é grátis para ler aqui na web. Para praticá-la interativamente (um editor de código integrado e um tutor de IA 24/7) e desbloquear o restante do curso de LLM Apps in Production (RAG + Vector DB + Caching), atualize para CoddyKit PRO. O curso de LLM Apps in Production (RAG + Vector DB + Caching) inclui 4 aulas no total.
O que vou aprender em “Carregando Diferentes Formatos de Documentos”?
Explore métodos para ingerir dados de várias fontes, como PDFs, páginas da Web, bancos de dados e tipos personalizados de arquivos. Você pratica LLM Apps in Production (RAG + Vector DB + Caching) com código prático que executa diretamente no navegador, e um tutor de IA 24/7 responde suas dúvidas enquanto trabalha na aula.
Preciso ter experiência prévia para começar LLM Apps in Production (RAG + Vector DB + Caching)?
Nenhuma experiência prévia é necessária. LLM Apps in Production (RAG + Vector DB + Caching) no CoddyKit é estruturado para alunos iniciantes até avançados, então você pode começar aqui ou desde o início e aprender no seu ritmo. Esta é a aula 1 de 4.
Quanto tempo leva a aula “Carregando Diferentes Formatos de Documentos”?
A maioria das aulas CoddyKit leva cerca de 5–10 minutos. Cada uma é compacta e interativa, então você faz progresso constante e retoma exatamente de onde parou entre web e app.
Posso escrever e executar código nesta aula de LLM Apps in Production (RAG + Vector DB + Caching)?
Sim. Cada aula de LLM Apps in Production (RAG + Vector DB + Caching) inclui um editor de código integrado, então você escreve e executa código real direto no navegador e recebe feedback de IA instantaneamente — nenhuma configuração local necessária.
Todas as aulas deste curso
- Carregando Diferentes Formatos de Documentos
- Estratégias de Divisão de Texto com Consciência de Contexto
- Gerenciamento e Filtragem de Metadados
- Limpando e eliminando duplicidades nos dados de origem