Microsoft MarkItDown：把任意文件转换为 Markdown 的完整指南

pip install 'markitdown[all]'

markitdown --version

pip install 'markitdown[pdf, docx, pptx]'

markitdown --list-plugins
markitdown --use-plugins path-to-file.pdf

pip install markitdown-ocr

pip install 'markitdown[az-content-understanding]'

git clone git@github.com:microsoft/markitdown.git && cd markitdown && pip install -e 'packages/markitdown[all]'

markitdown path-to-file.pdf > document.md

markitdown path-to-file.pdf -o document.md

cat path-to-file.pdf | markitdown

curl -sL https://example.com/document.pdf | markitdown

markitdown report.docx > report.md

markitdown presentation.pptx > slides.md

markitdown data.xlsx > data.md

markitdown scan.png > scan.md

sudo apt-get install tesseract-ocr
pip install markitdown-ocr

markitdown ./documents/ -o ./output/

import markitdown

md = markitdown.MarkItDown()
result = md.convert("document.pdf")
print(result.text_content)

import markitdown

md = markitdown.MarkItDown()
with open("report.docx", "rb") as f:
    result = md.convert(f)
    print(result.text_content)

import markitdown

md = markitdown.MarkItDown()
result = md.convert("document.pdf")
print(result.metadata)
print(result.text_content)

import markitdown
import glob
import os

md = markitdown.MarkItDown()
files = glob.glob("docs/**/*.pdf", recursive=True)
for filepath in files:
    result = md.convert(filepath)
    output_path = os.path.splitext(filepath)[0] + ".md"
    with open(output_path, "w") as f:
        f.write(result.text_content)
    print(f"Converted: {filepath} -> {output_path}")

import markitdown

md = markitdown.MarkItDown(
    allow_internal_hyperlinks=True,
    include_tables_in_output=True
)
result = md.convert("document.pdf")
print(result.text_content)

import markitdown
import os
from langchain_text_splitters import RecursiveCharacterTextSplitter

def ingest_documents(directory):
    md = markitdown.MarkItDown()
    splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)

    documents = []
    for filename in os.listdir(directory):
        filepath = os.path.join(directory, filename)
        if os.path.isfile(filepath):
            result = md.convert(filepath)
            if result:
                chunks = splitter.split_text(result.text_content)
                for i, chunk in enumerate(chunks):
                    documents.append({
                        "source": filename,
                        "chunk_index": i,
                        "content": chunk
                    })
    return documents

docs = ingest_documents("./knowledge_base")
print(f"Processed {len(docs)} document chunks")

#!/bin/bash
# process_uploads.sh — 每天处理所有上传的文档

MARKDOWN_DIR="/var/markdown"
UPLOAD_DIR="/var/uploads"

mkdir -p "$MARKDOWN_DIR"

for file in "$UPLOAD_DIR"/*.pdf "$UPLOAD_DIR"/*.docx "$UPLOAD_DIR"/*.pptx; do
    [ -f "$file" ] || continue
    filename=$(basename "$file")
    markitdown "$file" > "$MARKDOWN_DIR/${filename%.*}.md"
    echo "Converted: $file"
done

import markitdown

def prepare_document_for_llm(filepath, max_tokens=4000):
    md = markitdown.MarkItDown()
    result = md.convert(filepath)

    if result:
        content = result.text_content[:max_tokens * 4]
        return {
            "status": "success",
            "content": content,
            "tokens_estimated": len(content) // 4,
            "format": "markdown"
        }
    return {"status": "error", "message": "Conversion failed"}