MinerU Reader
pip install llama-index-readers-mineru
This reader uses the MinerU document parsing API to extract high-quality Markdown from PDF, Doc/Docx, PPT/PPTx, images, and Excel files. It supports two parsing modes:
| Feature | Flash (default) | Precision |
|---|---|---|
| Auth | No token required | Token required |
| Speed | Blazing fast | Standard |
| File size | Max 10 MB | Max 200 MB |
| Page limit | Max 20 pages | Max 600 pages |
| Formula / Table | Disabled | Configurable |
| Output in this Reader | Markdown only | Markdown only |
Note: MinerU Python SDK precision mode supports extra output formats (images/JSON/docx/html/latex), but
MinerUReaderin this integration currently returns onlyresult.markdownasDocument.text.
Usage
Flash Mode (default, no token needed)
from llama_index.readers.mineru import MinerUReader
reader = MinerUReader()
# Parse a single PDF from URL
documents = reader.load_data(
"https://cdn-mineru.openxlab.org.cn/demo/example.pdf"
)
print(documents[0].text)
# Parse a local file
documents = reader.load_data("/path/to/local.pdf")
# Parse multiple files at once
documents = reader.load_data(
[
"https://cdn-mineru.openxlab.org.cn/demo/example.pdf",
"/path/to/local.pdf",
]
)
Precision Mode (token required)
Get your free token from MinerU API Management.
from llama_index.readers.mineru import MinerUReader
reader = MinerUReader(
mode="precision",
token="your-api-token", # or set MINERU_TOKEN env var
ocr=True,
formula=True,
table=True,
language="en",
pages="1-20",
)
documents = reader.load_data("/path/to/scanned_paper.pdf")
Mixed Sources (local path + URL)
You can parse local files and remote URLs in one call:
from llama_index.readers.mineru import MinerUReader
reader = MinerUReader()
documents = reader.load_data(
[
"/path/to/local_a.pdf",
"/path/to/local_b.docx",
"https://cdn-mineru.openxlab.org.cn/demo/example.pdf",
]
)
for doc in documents:
print(doc.metadata["source"], "-", doc.text[:100])
Attach custom metadata with extra_info
Use extra_info when you want to merge custom metadata fields into every
returned Document.metadata (for example project/tenant/tag). It does not
change parsing behavior or output format.
from llama_index.readers.mineru import MinerUReader
reader = MinerUReader()
documents = reader.load_data(
"/path/to/paper.pdf",
extra_info={
"project": "paper-rag",
"tenant": "team-a",
"source_type": "research_pdf",
},
)
print(documents[0].metadata["project"]) # paper-rag
print(documents[0].metadata["source_type"]) # research_pdf
print(documents[0].text[:120]) # still Markdown text
Per-Page Splitting
When split_pages=True, each PDF page becomes a separate Document — ideal for RAG pipelines that need page-level granularity.
reader = MinerUReader(split_pages=True, pages="1-5")
documents = reader.load_data("/path/to/paper.pdf")
for doc in documents:
print(f"Page {doc.metadata['page']}: {doc.text[:100]}...")
Page Range + Split Pages (PDF only)
Use pages together with split_pages=True to parse only selected PDF pages.
In this mode, each selected page becomes one Document.
from llama_index.readers.mineru import MinerUReader
reader = MinerUReader(
mode="precision",
token="your-api-token", # or set MINERU_TOKEN
pages="2-4",
split_pages=True,
language="en",
)
documents = reader.load_data("/path/to/paper.pdf")
for doc in documents:
print(doc.metadata.get("page"), doc.metadata.get("source"))
Use with LlamaIndex Pipeline
from llama_index.core import VectorStoreIndex
from llama_index.readers.mineru import MinerUReader
reader = MinerUReader()
documents = reader.load_data("/path/to/paper.pdf")
index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine()
response = query_engine.query("Summarize the key findings")
print(response)
Parameters
Reader initialization (MinerUReader(...))
| Parameter | Type | Default | Description |
|---|---|---|---|
mode |
str |
"flash" |
Parsing mode: "flash" or "precision" |
token |
str | None |
None |
MinerU API token (precision mode). Falls back to MINERU_TOKEN env var. Apply here: https://mineru.net/apiManage/token |
language |
str |
"ch" |
Document language code |
pages |
str | None |
None |
Page range, e.g. "1-10" |
timeout |
int |
600 |
Max seconds to wait for task completion |
split_pages |
bool |
False |
Split PDF into per-page Documents |
ocr |
bool |
False |
Enable OCR (precision mode only) |
formula |
bool |
True |
Enable formula recognition (precision mode only) |
table |
bool |
True |
Enable table recognition (precision mode only) |
load_data(...) arguments
| Parameter | Type | Default | Description |
|---|---|---|---|
sources |
str | Path | list[str | Path] |
— | Single file path/URL, or a list of file paths/URLs |
extra_info |
dict | None |
None |
Custom metadata merged into each returned Document.metadata |
Links
Metadata
Release files for llama-index-readers-mineru 0.1.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| llama_index_readers_mineru-0.1.1.tar.gz | 8.0 kB | Without provenance | Details for llama_index_readers_mineru-0.1.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| llama_index_readers_mineru-0.1.1-py3-none-any.whl | Python 3 | none | any | Details for llama_index_readers_mineru-0.1.1-py3-none-any.whl |
Total release size: 15.8 kB
Release files / llama_index_readers_mineru-0.1.1.tar.gz
| Download URL | llama_index_readers_mineru-0.1.1.tar.gz |
|---|---|
| Size | 8.0 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
4f60f8d4111a136544144519bc5aa5627fbbb74931646481f9850660dfb238dd
|
|
BLAKE2b-256 checksum How to use checksums |
56e6850fc2426693ece5203027c3c555356ee713c7aa3af075d84d400c586ed3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.13
|
Release files / llama_index_readers_mineru-0.1.1-py3-none-any.whl
| Download URL | llama_index_readers_mineru-0.1.1-py3-none-any.whl |
|---|---|
| Size | 7.8 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
feee82a23296905669503f4dd9d7df36bf07b9e847724a3c839cb95589f808aa
|
|
BLAKE2b-256 checksum How to use checksums |
5068a5f3c37480eaf0730fd64595b44522e2d1745defd37a915580cc294e2ec7
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.12.13
|
Security for llama-index-readers-mineru 0.1.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.