Chapter 2.1 - Multi-Format Parsing

[!info] Extracting text from PDF, DOCX, and HTML.

To build a robust Knowledge Base, your RAG system must handle heterogeneous documents. We use a routing function that checks the file extension and applies the optimal library:

def extract_text_from_file(filepath):
    ext = os.path.splitext(filepath)[1].lower()
    if ext == ".txt":
        with open(filepath, "r", encoding="utf-8") as f:
            return f.read()
    elif ext == ".pdf":
        text = ""
        doc = fitz.open(filepath) # PyMuPDF
        for page in doc:
            text += page.get_text() + "\n"
        return text
    elif ext == ".docx":
        doc = docx.Document(filepath) # python-docx
        return "\n".join([para.text for para in doc.paragraphs])
    elif ext == ".html":
        from bs4 import BeautifulSoup
        with open(filepath, "r", encoding="utf-8") as f:
            soup = BeautifulSoup(f.read(), "html.parser")
            return soup.get_text(separator="\n", strip=True)

[!tip] Why PyMuPDF (fitz) Outperforms standard PDF Parsers PyMuPDF operates directly on native PDF C-bindings, extracting multi-column text blocks up to 10x faster than pure-Python parsers while maintaining accurate reading orders.