langchain-jostack-mdparse

A LangChain Document Loader for jostack-mdparse.

Installation

pip install langchain-jostack-mdparse

Quick Start

from langchain_jostack_mdparse import MarkdownExtractLoader

# Load a single file
loader = MarkdownExtractLoader("README.md")
docs = loader.load()

# Load with options
loader = MarkdownExtractLoader(
    "docs/guide.md",
    heading_level="1,2",
    strip_html=True,
    split_sections=True,
)
docs = loader.load()

# Load multiple files
loader = MarkdownExtractLoader(["a.md", "b.md"])
docs = loader.load()

Parameters Reference

| format | str | "text" | Output format. Values: json, text, html. Default: json | | quiet | bool | False | Suppress console logging output | | heading_level | str | None | Filter by heading levels (comma-separated, e.g. '1,2,3'). Default: all levels | | sections | str | None | Extract only sections matching the given heading text (comma-separated). Case-insensitive substring match | | include_frontmatter | bool | True | Include YAML/TOML frontmatter in output. Default: true | | strip_html | bool | False | Strip inline HTML tags from Markdown content | | include_code_blocks | bool | True | Include fenced code blocks in output. Default: true | | include_toc | bool | False | Add a generated table of contents to the output | | flatten_lists | bool | False | Flatten nested lists into a single level | | section_separator | str | None | Separator between sections in text output. Use %section-title% for heading text. Default: none | | normalize_links | bool | False | Convert relative links to absolute using the source file path | | detect_tables | bool | False | Detect and extract Markdown tables as structured data |

Development

pip install -e .
make test
make lint

License

Apache-2.0