<!-- AI-AGENT-SUMMARY
name: opendataloader-pdf
category: PDF data extraction, PDF accessibility automation
license: Apache-2.0
solves: [PDF to structured data for RAG/LLM pipelines, accelerate PDF accessibility remediation โ layout analysis + auto-tagging to Tagged PDF as foundation for PDF/UA (first open-source end-to-end)]
input: PDF files (digital, scanned, tagged)
output: Markdown, JSON (with bounding boxes), HTML, Tagged PDF, PDF/UA (enterprise)
sdk: Python, Node.js, Java
requirements: Java 11+
pricing: open-source core (data extraction, layout analysis, auto-tagging to Tagged PDF), enterprise add-on (PDF/UA export, accessibility studio)
extraction-benchmark: #1 overall extraction accuracy (0.907) in hybrid mode, 0.928 table extraction accuracy, 0.015s/page local mode
accessibility-validation: PDF Association collaboration, Well-Tagged PDF specification, veraPDF automated validation
key-differentiators: [benchmark #1 PDF parser, deterministic output, bounding boxes for every element, XY-Cut++ reading order, AI safety filters, hybrid AI mode, first open-source PDF auto-tagging to Tagged PDF, PDF Association + Dual Lab (veraPDF) collaboration, Well-Tagged PDF spec compliance]
-->
# OpenDataLoader PDF
**PDF Parser for AI-ready data. Automate PDF accessibility. Open-source.**
[](https://github.com/opendataloader-project/opendataloader-pdf/blob/main/LICENSE)
[](https://pypi.org/project/opendataloader-pdf/)
[](https://www.npmjs.com/package/@opendataloader/pdf)
[](https://search.maven.org/artifact/org.opendataloader/opendataloader-pdf-core)
[](https://github.com/opendataloader-project/opendataloader-pdf#java)
<a href="https://trendshift.io/repositories/21917" target="_blank"><img src="https://trendshift.io/api/badge/repositories/21917" alt="opendataloader-project%2Fopendataloader-pdf | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
๐ **PDF parser for AI data extraction** โ Extract Markdown, JSON (with bounding boxes), and HTML from any PDF. #1 in benchmarks (0.907 overall). Deterministic local mode + AI hybrid mode for complex pages.
- **How accurate is it?** โ #1 in benchmarks: 0.907 overall, 0.928 table accuracy across 200 real-world PDFs including multi-column and scientific papers. Deterministic local mode + AI hybrid mode for complex pages ([benchmarks](#extraction-benchmarks))
- **Scanned PDFs and OCR?** โ Yes. Built-in OCR (80+ languages) in hybrid mode. Works with poor-quality scans at 300 DPI+ ([hybrid mode](#hybrid-mode-1-accuracy-for-complex-pdfs))
- **Tables, formulas, images, charts?** โ Yes. Complex/borderless tables, LaTeX formulas, and AI-generated picture/chart descriptions all via hybrid mode ([hybrid mode](#hybrid-mode-1-accuracy-for-complex-pdfs))
- **How do I use this for RAG?** โ `pip install opendataloader-pdf`, convert in 3 lines. Outputs structured Markdown for chunking, JSON with bounding boxes for source citations, and HTML. LangChain integration available. Python, Node.js, Java SDKs ([quick start](#get-started-in-30-seconds) | [LangChain](#langchain-integration))
โฟ **PDF accessibility automation** โ Auto-tag untagged PDFs into screen-reader-ready Tagged PDFs at scale. First open-source tool to generate Tagged PDFs end-to-end.
- **What's the problem?** โ Accessibility regulations are now enforced worldwide. Manual PDF remediation costs $50โ200 per document and doesn't scale ([regulations](#pdf-accessibility--pdfua-conversion))
- **What's free?** โ Layout analysis + auto-tagging (Apache 2.0). Untagged PDF in โ Tagged PDF out. No proprietary SDK dependency ([auto-tagging](#auto-tagging))
- **What about PDF/UA compliance?** โ Converting Tagged PDF to PDF/UA-1 or PDF/UA-2 is an enterprise add-on. Auto-tagging generates the Tagged PDF; PDF/UA export is the final step ([pipeline](#accessibility-pipeline))
- **Why trust this?** โ Built in collaboration with [Dual Lab](https://duallab.com) ([veraPDF](https://verapdf.org) developers) based on [PDF Association](https://pdfa.org) specifications, best practice guides and expertise of the [PDF Community](https://pdfa.org/community/). Auto-tagging follows the [Well-Tagged PDF specification](https://pdfa.org/wtpdf/), validated with veraPDF ([collaboration](https://opendataloader.org/docs/tagged-pdf-collaboration))
## Get Started in 30 Seconds
**Requires**: Java 11+ and Python 3.11+ ([Node.js](https://opendataloader.org/docs/quick-start-nodejs) | [Java](https://opendataloader.org/docs/quick-start-java) also available)
> Before you start: run `java -version`. If not found, install JDK 11+ from [Adoptium](https://adoptium.net/).
```bash
pip install -U opendataloader-pdf
```
```python
import opendataloader_pdf
# Batch all files in one call โ each convert() spawns a JVM process, so repeated calls are slow
opendataloader_pdf.convert(
input_path=["file1.pdf", "file2.pdf", "folder/"],
output_dir="output/",
format="markdown,json"
)
```
*Annotated PDF output โ each element (heading, paragraph, table, image) detected with bounding boxes and semantic type.*
## What Problems Does This Solve?
| Problem | Solution | Status |
|---------|----------|--------|
| **PDF structure lost during parsing** โ wrong reading order, broken tables, no element coordinates | Deterministic local PDF to Markdown/JSON with bounding boxes, XY-Cut++ reading order | Shipped |
| **Complex tables, scanned PDFs, formulas, charts** need AI-level understanding | Hybrid mode routes complex pages to AI backend (#1 in benchmarks) | Shipped |
| **Manual PDF remediation cost** โ Accessibility regulations (EAA, ADA, Section 508) demand Tagged PDFs. Manual remediation costs $50โ200/doc | Auto-tag untagged PDFs into Tagged PDFs (free, Apache 2.0). Foundation for PDF/UA workflows; full PDF/UA-1/2 export is an enterprise add-on | Auto-tag: Shipped. PDF/UA export: Enterprise |
## Capability Matrix
| Capability | Supported | Tier |
|------------|-----------|------|
| **Data extraction** | | |
| Extract text with correct reading order | Yes | Free |
| Bounding boxes for every element | Yes | Free |
| Table extraction (simple borders) | Yes | Free |
| Table extraction (complex/borderless) | Yes | Free (Hybrid) |
| Heading hierarchy detection | Yes | Free |
| List detection (numbered, bulleted, nested) | Yes | Free |
| Image extraction with coordinates | Yes | Free |
| AI chart/image description | Yes | Free (Hybrid) |
| OCR for scanned PDFs | Yes | Free (Hybrid) |
| Formula extraction (LaTeX) | Yes | Free (Hybrid) |
| Tagged PDF structure extraction | Yes | Free |
| AI safety (prompt injection filtering) | Yes | Free |
| Header/footer/watermark filtering | Yes | Free |
| **Accessibility** | | |
| Auto-tagging โ Tagged PDF for untagged PDFs | Yes | Free (Apache 2.0) |
| PDF/UA-1, PDF/UA-2 export | ๐ผ Available | Enterprise |
| Accessibility studio (visual editor) | ๐ผ Available | Enterprise |
| **Limitations** | | |
| Process Word/Excel/PPT | No | โ |
| GPU required | No | โ |
## Extraction Benchmarks
**opendataloader-pdf [hybrid] ranks #1 overall (0.907)** across reading order, table, and heading extraction accuracy.
| Engine | Overall | Reading Order | Table | Heading | Speed (s/page) | License |
|--------|---------|---------------|-------|---------|----------------|---------|
| **opendataloader [hybrid]** | **0.907** | **0.934** | **0.928** | 0.821 | 0.463 | Apache-2.0 |
| nutrient | 0.885 | 0.925 | 0.708 | 0.819 | **0.008** | Commercial |
| docling | 0.882 | 0.898 | 0.887 | **0.824** | 0.762 | MIT |
| marker | 0.861 | 0.890 | 0.808 | 0.796 | 53.932 | GPL-3.0 |
| unstructured [hi_res] | 0.841 | 0.904 | 0.588 | 0.749 | 3.008 | Apache-2.0 |
| edgeparse | 0.837 | 0.894 | 0.717 | 0.706 | 0.036 | Apache-2.0 |
| opendataloader | 0.831 | 0.902 | 0.489 | 0.739 | 0.015 | Apache-2.0 |
| mineru | 0.831 | 0.857 | 0.873 | 0.743 | 5.962 | AGPL-3.0 |
| pymupdf4llm | 0.732 | 0.885 | 0.401 | 0.412 | 0.091 | AGPL-3.0 |
| unstructured | 0.686 | 0.882 | 0.000 | 0.388 | 0.077 | Apache-2.0 |
| markitdown | 0.589 | 0.844 | 0.273 | 0.000 | 0.114 | MIT |
| liteparse | 0.576 | 0.866 | 0.000 | 0.000 | 1.061 | Apache-2.0 |
> Scores normalized to [0, 1]. Higher is better for accuracy; lower is better for speed. **Bold** = best. [Full benchmark details](https://github.com/opendataloader-project/opendataloader-bench)
[](https://github.com/opendataloader-project/opendataloader-bench)
[](https://github.com/opendataloader-project/opendataloader-bench)
## Which Mode Should I Use?
| Your Document | Mode | Install | Server Command | Client Command |
|---------------|------|---------|----------------|----------------|
| Standard digital PDF | Fast (default) | `pip install opendataloader-pdf` | None needed | `opendataloader-pdf file1.pdf file2.pdf folder/` |
| Complex or nested tables | **Hybrid** | `pip install "opendataloader-pdf[hybrid]"` | `opendataloader-pdf-hybrid --port 5002` | `opendataloader-pdf --hybrid docling-fast file1.pdf file2.pdf folder/` |
| Scanned / image-based PDF | Hybrid + OCR | `pip install "opendataloader-pdf[hybrid]"` | `opendataloader-pdf-hybrid --port 5002 --force-ocr` | `opendataloader-pdf --hybrid docling-fast file1.pdf file2.pdf folder/` |
| Non-English scanned PDF | Hybrid + OCR | `pip install "opendataloader-pdf[hybrid]"` | `opendataloader-pdf-hybrid --port 5002 --force-ocr --ocr-engine <engine> --ocr-lang <codes>` โ see [Choosing an OCR Engine](#choosing-an-ocr-engine) | `opendataloader-pdf --hybrid docling-fast file1.pdf file2.pdf folder/` |
| Mathematical formulas | Hybrid + formula | `pip install "opendataloader-pdf[hybrid]"` | `opendataloader-pdf-hybrid --enrich-formula` | `opendataloader-pdf --hybrid docling-fast --hybrid-mode full file1.pdf file2.pdf folder/` |
| Charts needing description | Hybrid + picture | `pip install "opendataloader-pdf[hybrid]"` | `opendataloader-pdf-hybrid --enrich-picture-description` | `opendataloader-pdf --hybrid docling-fast --hybrid-mode full file1.pdf file2.pdf folder/` |
| Nested sections for RAG chunking | Hybrid + heading levels | `pip install "opendataloader-pdf[hybrid]"` | `opendataloader-pdf-hybrid --port 5002 --heading-hierarchy` | `opendataloader-pdf --hybrid docling-fast file1.pdf file2.pdf folder/` |
| Untagged PDFs needing accessibility | Auto-tagging โ Tagged PDF | `pip install opendataloader-pdf` | None needed | `opendataloader-pdf --format tagged-pdf file1.pdf file2.pdf folder/` |
## Quick Start
### Python
```bash
pip install -U opendataloader-pdf
```
```python
import opendataloader_pdf
# Batch all files in one call โ each convert() spawns a JVM process, so repeated calls are slow
opendataloader_pdf.convert(
input_path=["file1.pdf", "file2.pdf", "folder/"],
output_dir="output/",
format="markdown,json"
)
```
### Node.js
```bash
npm install @opendataloader/pdf
```
```typescript
import { convert } from '@opendataloader/pdf';
await convert(['file1.pdf', 'file2.pdf', 'folder/'], {
outputDir: 'output/',
format: 'markdown,json'
});
```
### Java
```xml
<dependency>
<groupId>org.opendataloader</groupId>
<artifactId>opendataloader-pdf-core</artifactId>
</dependency>
```
[Python Quick Start](https://opendataloader.org/docs/quick-start-python) | [Node.js Quick Start](https://opendataloader.org/docs/quick-start-nodejs) | [Java Quick Start](https://opendataloader.org/docs/quick-start-java)
## Hybrid Mode: #1 Accuracy for Complex PDFs
Hybrid mode combines fast local Java processing with AI backends. Simple pages stay local (0.02s); complex pages route to AI for +90% table accuracy.
> **Don't combine with `--use-struct-tree` on tagged PDFs.** `--use-struct-tree` takes precedence, so the hybrid backend is **not** called (a warning is logged). If you want the hybrid backend, drop `--use-struct-tree`.
```bash
pip install -U "opendataloader-pdf[hybrid]"
```
**Terminal 1** โ Start the backend server:
```bash
opendataloader-pdf-hybrid --port 5002
```
**Terminal 2** โ Process PDFs:
```bash
# Batch all files in one call โ each invocation spawns a JVM process, so repeated calls are slow
opendataloader-pdf --hybrid docling-fast file1.pdf file2.pdf folder/
```
**Python:**
```python
# Batch all files in one call โ each convert() spawns a JVM process, so repeated calls are slow
opendataloader_pdf.convert(
input_path=["file1.pdf", "file2.pdf", "folder/"],
output_dir="output/",
hybrid="docling-fast"
)
```
### OCR for Scanned PDFs
Start the backend with `--force-ocr` for image-based PDFs with no selectable text:
```bash
opendataloader-pdf-hybrid --port 5002 --force-ocr
```
For non-English documents, specify the language:
```bash
opendataloader-pdf-hybrid --port 5002 --force-ocr --ocr-lang "ko,en"
```
The language codes above are for the default engine (EasyOCR): `en`, `ko`, `ja`, `ch_sim`, `ch_tra`, `de`, `fr`, `ar`, and more.
#### Choosing an OCR Engine
OCR quality varies with the engine, the language, and the document itself (scan resolution, fonts, layout). No engine is best for every document, so run a few representative pages through several engines and compare the output before processing a whole collection.
Select the engine with `--ocr-engine`. Each engine uses its own language code system for `--ocr-lang`:
| Engine | `--ocr-engine` | `--ocr-lang` example | Platform / extra install |
|--------|----------------|----------------------|--------------------------|
| EasyOCR (default) | `easyocr` | `ko,en` (ISO 639-1) | Included in `opendataloader-pdf[hybrid]` |
| RapidOCR | `rapidocr` | `korean` (RapidOCR codes: `en`, `ch`, `japan`, `korean`, ...) | `pip install rapidocr onnxruntime` |
| Tesseract (CLI) | `tesseract` | `kor,eng` (ISO 639-2) | Tesseract binary + language data (e.g. `tesseract-ocr-kor`) |
| Tesseract (Python binding) | `tesserocr` | `kor,eng` (ISO 639-2) | `pip install tesserocr` + libtesseract |
| Apple Vision | `ocrmac` | `ko-KR,en-US` (BCP-47) | macOS only, `pip install ocrmac` |
| NVIDIA Nemotron OCR | `nemotron-ocr` | `multilingual` or `english` | Linux x86_64 + CUDA, `pip install "docling[feat-ocr-nemotron]"` |
| Automatic | `auto` | Not configurable | Docling picks an available engine |
To compare engines on a sample, start one server per engine and send the same file to each.
**Terminal 1** โ EasyOCR server:
```bash
opendataloader-pdf-hybrid --port 5002 --force-ocr --ocr-engine easyocr --ocr-lang "ko,en"
```
**Terminal 2** โ Tesseract server:
```bash
opendataloader-pdf-hybrid --port 5003 --force-ocr --ocr-engine tesseract --ocr-lang "kor,eng"
```
**Terminal 3** โ Send the same sample to both:
```bash
opendataloader-pdf sample.pdf --hybrid docling-fast --hybrid-mode full --hybrid-url http://localhost:5002 -o out-easyocr
opendataloader-pdf sample.pdf --hybrid docling-fast --hybrid-mode full --hybrid-url http://localhost:5003 -o out-tesseract
```
### Formula Extraction (LaTeX)
Extract mathematical formulas as LaTeX from scientific PDFs:
```bash
# Server: enable formula enrichment
opendataloader-pdf-hybrid --enrich-formula
# Batch all files in one call โ each invocation spawns a JVM process, so repeated calls are slow
opendataloader-pdf --hybrid docling-fast --hybrid-mode full file1.pdf file2.pdf folder/
```
Output in JSON:
```json
{
"type": "formula",
"page number": 1,
"bounding box": [226.2, 144.7, 377.1, 168.7],
"content": "\\frac{f(x+h) - f(x)}{h}"
}
```
> **Note**: Formula and picture description enrichments require `--hybrid-mode full` on the client side.
### Chart & Image Description
Generate AI descriptions for charts and images โ useful for RAG search and accessibility alt text:
```bash
# Server
opendataloader-pdf-hybrid --enrich-picture-description
# Batch all files in one call โ each invocation spawns a JVM process, so repeated calls are slow
opendataloader-pdf --hybrid docling-fast --hybrid-mode full file1.pdf file2.pdf folder/
```
Output in JSON:
```json
{
"type": "image",
"pdfua_tag": "Figure",
"id": 285,
"page number": 1,
"bounding box": [72.0, 400.0, 540.0, 650.0],
"alt": "A bar chart showing waste generation by region from 2016 to 2030...",
"alt_source": "ai-generated"
}
```
`alt_source` says where the text came from: `original` (the PDF's own /Alt), `ai-generated`,
or `missing` โ and when it is `missing`, there is no `alt` field at all.
With `--enrich-picture-description`, every picture goes to the model whatever its size. The
server reports the tally per request, so a picture that came back without a description shows
up in the log rather than only in the output:
```
picture_description pictures=4 described=4
```
Cost follows the number of pictures, not their size: a 40ร20 pt logo costs about what a
full-page chart does, and on CPU one picture can take over a minute. A document full of icons
is therefore minutes of work per page, and conversions are serialized, so other requests wait.
`--picture-area-threshold 0.05` spends the model only on pictures covering at least 5% of
their page:
```
picture_description pictures=4 described=1 skipped=3 threshold=0.05
```
Skips are logged at INFO when you set a threshold, and at WARNING when you did not โ in that
case nothing was asked to drop them, so the cause is worth a look.
> Uses SmolVLM (256M), a lightweight vision model. Custom prompts supported via `--picture-description-prompt`.
> Since the picture count comes from the uploaded file, a server reachable by untrusted callers
> should set `--picture-area-threshold`, and `--max-file-size` to cap the conversion work one
> file can ask for. That limit is applied once the request body has been received, so it bounds
> conversion cost rather than upload bandwidth or temporary disk.
### Heading Hierarchy
The layout model labels a region as a section header without a depth, so by default every
heading comes back at level 1 and subsections sit at the same depth as the document title.
Flat headings make it hard to tell a section title from a document title when chunking for RAG.
`--heading-hierarchy` infers the depth โ from the PDF outline first, then section numbering
(`1.` โ `1.1` โ `1.1.1`), then visual style:
```bash
# Server
opendataloader-pdf-hybrid --port 5002 --heading-hierarchy
# Client โ no extra flag needed
opendataloader-pdf --hybrid docling-fast file1.pdf file2.pdf folder/
```
```diff
# 3. Methodology
- # 3.1. Multi-Object Rectification Network
+ ## 3.1. Multi-Object Rectification Network
- # Abstract
+ ## Abstract
```
Numbering covers numbered sections; style is what pulls an unnumbered `Abstract` or
`References` out from under the document title.
Off by default, so existing output is unchanged. Levels are capped at H6.
### Hancom Data Loader Integration โ Coming Soon
Enterprise-grade AI document analysis via [Hancom Data Loader](https://sdk.hancom.com/en/services/1?utm_source=github&utm_medium=readme&utm_campaign=opendataloader-pdf) โ customer-customized models trained on your domain-specific documents. 30+ element types (tables, charts, formulas, captions, footnotes, etc.), VLM-based image/chart understanding, complex table extraction (merged cells, nested tables), SLA-backed OCR for scanned documents, and native HWP/HWPX support. Supports PDF, DOCX, XLSX, PPTX, HWP, PNG, JPG. [Live demo](https://livedemo.sdk.hancom.com/en/dataloader?utm_source=github&utm_medium=readme&utm_campaign=opendataloader-pdf)
[Hybrid Mode Guide](https://opendataloader.org/docs/hybrid-mode)
## Output Formats
| Format | Use Case |
|--------|----------|
| **JSON** | Structured data with bounding boxes, semantic types |
| **Markdown** | Clean text for LLM context, RAG chunks |
| **HTML** | Web display with styling |
| **Annotated PDF** | Visual debugging โ see detected structures ([sample](https://opendataloader.org/demo/samples/01030000000000)) |
| **Text** | Plain text extraction |
Combine formats: `format="json,markdown"`
### JSON Output Example
```json
{
"type": "heading",
"id": 42,
"level": "Title",
"page number": 1,
"bounding box": [72.0, 700.0, 540.0, 730.0],
"heading level": 1,
"font": "Helvetica-Bold",
"font size": 24.0,
"text color": "[0.0]",
"content": "Introduction"
}
```
| Field | Description |
|-------|-------------|
| `type` | Element type: heading, paragraph, table, list, image, caption, formula |
| `id` | Unique identifier for cross-referencing |
| `page number` | 1-indexed page reference |
| `bounding box` | `[left, bottom, right, top]` in PDF points (72pt = 1 inch) |
| `heading level` | Heading depth (1+) |
| `content` | Extracted text |
[Full JSON Schema](https://opendataloader.org/docs/reference/json-schema)
## Advanced Features
### Tagged PDF Support
When a PDF has structure tags, OpenDataLoader extracts the **exact layout** the author intended โ no guessing, no heuristics. Headings, lists, tables, and reading order are preserved from the source.
> **Output quality depends on tag quality.** Not all tagged PDFs are well-tagged. For PDFs with sparse or incorrect tags, the default heuristic mode or `--hybrid docling-fast` often produces better results.
> **`--use-struct-tree` takes precedence over `--hybrid`.** If both are set on a tagged PDF, the structure tree is used and the hybrid backend is **not** called (a well-tagged PDF already carries reading order and structure). Drop `--use-struct-tree` if you want the hybrid backend instead.
```python
# Batch all files in one call โ each convert() spawns a JVM process, so repeated calls are slow
opendataloader_pdf.convert(
input_path=["file1.pdf", "file2.pdf", "folder/"],
output_dir="output/",
use_struct_tree=True # Use native PDF structure tags
)
```
Most PDF parsers ignore structure tags entirely. [Learn more](https://opendataloader.org/docs/tagged-pdf)
### AI Safety: Prompt Injection Protection
PDFs can contain hidden prompt injection attacks. OpenDataLoader automatically filters:
- Hidden text (transparent, zero-size fonts)
- Off-page content
- Suspicious invisible layers
To sanitize sensitive data (emails, URLs, phone numbers โ placeholders), enable it explicitly:
```bash
# Batch all files in one call โ each invocation spawns a JVM process, so repeated calls are slow
opendataloader-pdf file1.pdf file2.pdf folder/ --sanitize
```
#### Custom Sanitization Rules
To use your own rules, give a JSON rules file with `--sanitize-rules` (Python: `sanitize_rules`, Node.js: `sanitizeRules`). This option also turns on `--sanitize`. Only the rules of the file run, in place of the built-in rules.
```json
{
"rules": [
{ "id": "kr-mobile", "pattern": "(?<!\\d)01[016789]-\\d{3,4}-\\d{4}(?!\\d)", "replacement": "010-0000-0000" }
]
}
```
```bash
opendataloader-pdf file1.pdf folder/ --sanitize-rules rules.json
```
To keep the built-in rules, start from [`examples/sanitize-rules/default-rules.json`](examples/sanitize-rules/default-rules.json). It has the built-in rules of `--sanitize`. Copy it, then add, change, or delete rules. If a later release changes the built-in rules, compare your copy with the new template.
Each rule has three keys:
| Key | Value |
|-----|-------|
| `id` | 1 to 64 characters from `a-z`, `0-9`, and `-`. Each ID only once |
| `pattern` | A Java regular expression, 1 to 5,000 characters |
| `replacement` | Literal text, 1 to 200 characters, with no control characters (`$1` is not a group reference) |
The run stops before any file is processed if the rules file has one of these problems:
- An unknown or missing key, or a value of the wrong type.
- A rule ID that is repeated or has a wrong format.
- A pattern that is empty or does not compile, or a replacement that is empty or has a control character.
- A value over a limit: a file larger than 1 MiB (1,048,576 bytes), more than 200 rules, or a pattern or replacement that is too long.
- No rules.
The error gives the reason, and the rule or the key.
How the rules work:
- All rules run on the original text of each line. A match cannot cross a line, and a rule does not see the placeholder of another rule.
- If two matches overlap, the match that starts first is kept. If they start at the same position, the longer match is kept. If they also have the same length, the match of the earlier rule in the file is kept.
- Rules apply to the extracted text. They do not apply to image descriptions, formulas, or document metadata such as the title.
> **Warning:** A pattern runs on every text line, with no time limit. Some patterns are very slow, also on a short line. For example, nested repetition with limits such as `(a{1,20}){1,20}$` can take tens of seconds on a line of 31 characters, and the time grows fast with the length of the line. Some patterns overflow the stack on a long line, for example `(a|b)*c`. Then that file fails with an error that names the rule, and the run continues with the next file. Use only rules files that you trust. Test each new pattern on lines as long as your real text.
>
> `\w` matches only ASCII letters, digits, and `_`. On Java 19 and later, `\b` uses the same ASCII rule. On Java 11 to 18, `\b` also counts other letters, such as Hangul, as word characters. Use `(?<!\d)` and `(?!\d)` around numbers.
[AI Safety Guide](https://opendataloader.org/docs/ai-safety)
### LangChain Integration
```bash
pip install -U langchain-opendataloader-pdf
```
```python
from langchain_opendataloader_pdf import OpenDataLoaderPDFLoader
loader = OpenDataLoaderPDFLoader(
file_path=["file1.pdf", "file2.pdf", "folder/"],
format="text"
)
documents = loader.load()
```
[LangChain Docs](https://docs.langchain.com/oss/python/integrations/document_loaders/opendataloader_pdf) | [GitHub](https://github.com/opendataloader-project/langchain-opendataloader-pdf) | [PyPI](https://pypi.org/project/langchain-opendataloader-pdf/)
### Advanced Options
```python
# Batch all files in one call โ each convert() spawns a JVM process, so repeated calls are slow
opendataloader_pdf.convert(
input_path=["file1.pdf", "file2.pdf", "folder/"],
output_dir="output/",
format="json,markdown,pdf",
image_output="embedded", # "off", "embedded" (Base64), or "external" (default)
image_format="jpeg", # "png" or "jpeg"
use_struct_tree=True, # Use native PDF structure
)
```
[Full CLI Options Reference](https://opendataloader.org/docs/reference/cli-options)
## PDF Accessibility & PDF/UA Conversion
**Problem**: Millions of existing PDFs lack structure tags, failing accessibility regulations (EAA, ADA/Section 508, Korea Digital Inclusion Act). Manual remediation costs $50โ200 per document and doesn't scale.
**OpenDataLoader's approach**: Built in collaboration with [PDF Association](https://pdfa.org) and [Dual Lab](https://duallab.com) (developers of [veraPDF](https://verapdf.org), the industry-reference open-source PDF/A and PDF/UA validator). Auto-tagging follows the [Well-Tagged PDF specification](https://pdfa.org/resource/well-tagged-pdf/) and is validated programmatically using veraPDF โ automated conformance checks against PDF accessibility standards, not manual review. No existing open-source tool generates Tagged PDFs end-to-end โ most rely on proprietary SDKs for the tag-writing step. OpenDataLoader does it all under Apache 2.0. ([collaboration details](https://opendataloader.org/docs/tagged-pdf-collaboration))
| Regulation | Deadline | Requirement |
|------------|----------|-------------|
| **European Accessibility Act (EAA)** | June 28, 2025 | Accessible digital products across the EU |
| **ADA & Section 508** | In effect | U.S. federal agencies and public accommodations |
| **Digital Inclusion Act** | In effect | South Korea digital service accessibility |
### Standards & Validation
| Aspect | Detail |
|--------|--------|
| **Specification** | [Well-Tagged PDF](https://pdfa.org/resource/well-tagged-pdf/) by PDF Association |
| **Validation** | [veraPDF](https://verapdf.org) โ industry-reference open-source PDF/A & PDF/UA validator |
| **Collaboration** | PDF Association + [Dual Lab](https://duallab.com) (veraPDF developers) co-develop tagging and validation |
| **License** | Auto-tagging โ Tagged PDF: Apache 2.0 (free). PDF/UA export: Enterprise |
### Accessibility Pipeline
| Step | Feature | Status | Tier |
|------|---------|--------|------|
| 1. **Audit** | Read existing PDF tags, detect untagged PDFs | Shipped | Free |
| 2. **Auto-tag โ Tagged PDF** | Generate structure tags for untagged PDFs | Shipped | Free (Apache 2.0) |
| 3. **Export PDF/UA** | Convert to PDF/UA-1 or PDF/UA-2 compliant files | ๐ผ Available | Enterprise |
| 4. **Visual editing** | Accessibility studio โ review and fix tags | ๐ผ Available | Enterprise |
> **๐ผ Enterprise features** are available on request. [Contact us](https://opendataloader.org/contact) to get started.
### Auto-Tagging
Generate Tagged PDFs from untagged PDFs โ output is a screen-reader-ready PDF with structure tags (headings, paragraphs, lists, tables, reading order).
```python
import opendataloader_pdf
# Untagged PDF in โ Tagged PDF out
opendataloader_pdf.convert(
input_path=["file1.pdf", "file2.pdf", "folder/"],
output_dir="output/",
format="tagged-pdf"
)
```
```bash
# CLI
opendataloader-pdf --format tagged-pdf file1.pdf file2.pdf folder/
```
Combine with other formats: `format="json,tagged-pdf"`.
### End-to-End Compliance Workflow
```
Existing PDFs (untagged)
โ
โผ
โโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโโ
โ 1. Audit โโโโ>โ 2. Auto-Tag โโโโ>โ 3. Export โโโโ>โ 4. Studio โ
โ (check tags) โ โ (โ Tagged PDF) โ โ (PDF/UA) โ โ (visual editor) โ
โโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโโ
โ โ โ โ
โผ โผ โผ โผ
use_struct_tree format="tagged-pdf" PDF/UA export Accessibility Studio
(Available now) (Available, Apache 2.0) (Enterprise) (Enterprise)
```
[PDF Accessibility Guide](https://opendataloader.org/docs/accessibility-compliance)
## Roadmap
| Feature | Timeline | Tier |
|---------|----------|------|
| **[Hancom Data Loader](https://sdk.hancom.com/en/services/1?utm_source=github&utm_medium=readme&utm_campaign=opendataloader-pdf)** โ Enterprise AI document analysis, customer-customized models, VLM-based chart/image understanding, production-grade OCR | Q2-Q3 2026 | Planned |
| **Structure validation** โ Verify PDF tag trees | Q3 2026 | Planned |
[Full Roadmap](https://opendataloader.org/docs/upcoming-roadmap)
## Frequently Asked Questions
### What is the best PDF parser for RAG?
For RAG pipelines, you need a parser that preserves document structure, maintains correct reading order, and provides element coordinates for citations. OpenDataLoader is designed specifically for this โ it outputs structured JSON with bounding boxes, handles multi-column layouts with XY-Cut++, and runs locally without GPU. In hybrid mode, it ranks #1 overall (0.907) in benchmarks.
### What is the best open-source PDF parser?
OpenDataLoader PDF is the only open-source parser that combines: rule-based deterministic extraction (no GPU), bounding boxes for every element, XY-Cut++ reading order, built-in AI safety filters, native Tagged PDF support, and hybrid AI mode for complex documents. It ranks #1 in overall accuracy (0.907) while running locally on CPU.
### How do I extract tables from PDF for LLM?
OpenDataLoader detects tables using border analysis and text clustering, preserving row/column structure. For complex tables, enable hybrid mode for +90% accuracy improvement (0.489 to 0.928 TEDS score):
```python
# Batch all files in one call โ each convert() spawns a JVM process, so repeated calls are slow
opendataloader_pdf.convert(
input_path=["file1.pdf", "file2.pdf", "folder/"],
output_dir="output/",
format="json",
hybrid="docling-fast" # For complex tables
)
```
### How does it compare to docling, marker, or pymupdf4llm?
OpenDataLoader [hybrid] ranks #1 overall (0.907) across reading order, table, and heading accuracy. Key differences: docling (0.882) is strong but lacks bounding boxes and AI safety filters. marker (0.861) requires GPU and is 1000x slower (53.932s/page). pymupdf4llm (0.732) is fast but has poor table (0.401) and heading (0.412) accuracy. OpenDataLoader is the only parser that combines deterministic local extraction, bounding boxes for every element, and built-in prompt injection protection. See [full benchmark](https://github.com/opendataloader-project/opendataloader-bench).
### Can I use this without sending data to the cloud?
Yes. OpenDataLoader runs 100% locally. No API calls, no data transmission โ your documents never leave your environment. The hybrid mode backend also runs locally on your machine. Ideal for legal, healthcare, and financial documents.
### Does it support OCR for scanned PDFs?
Yes, via hybrid mode. Install with `pip install "opendataloader-pdf[hybrid]"`, start the backend with `--force-ocr`, then process as usual. Supports multiple languages including Korean, Japanese, Chinese, Arabic, and more via `--ocr-lang`, with a choice of OCR engines via `--ocr-engine`. Quality varies by engine, language, and document โ see [Choosing an OCR Engine](#choosing-an-ocr-engine).
### Does it work with Korean, Japanese, or Chinese documents?
Yes. For digital PDFs, text extraction works out of the box. For scanned PDFs, use hybrid mode with `--force-ocr --ocr-lang "ko,en"` (or `ja`, `ch_sim`, `ch_tra`), and compare OCR engines on a sample of your documents โ see [Choosing an OCR Engine](#choosing-an-ocr-engine). Coming soon: [Hancom Data Loader](https://sdk.hancom.com/en/services/1?utm_source=github&utm_medium=readme&utm_campaign=opendataloader-pdf) integration โ enterprise-grade AI document analysis with built-in production-grade OCR and customer-customized models optimized for your specific document types and workflows.
### How fast is it?
Local mode processes 60+ pages per second on CPU (0.02s/page). Hybrid mode processes 2+ pages per second (0.46s/page) with significantly higher accuracy for complex documents. No GPU required. Benchmarked on Apple M4. [Full benchmark details](https://github.com/opendataloader-project/opendataloader-bench). With multi-process batch processing, throughput exceeds 100 pages per second on 8+ core machines.
### Does it handle multi-column layouts?
Yes. OpenDataLoader uses XY-Cut++ reading order analysis to correctly sequence text across multi-column pages, sidebars, and mixed layouts. This works in both local and hybrid modes without any configuration.
### What is hybrid mode?
Hybrid mode combines fast local Java processing with an AI backend. Simple pages are processed locally (0.02s/page); complex pages (tables, scanned content, formulas, charts) are automatically routed to the AI backend for higher accuracy. The backend runs locally on your machine โ no cloud required. See [Which Mode Should I Use?](#which-mode-should-i-use) and [Hybrid Mode Guide](https://opendataloader.org/docs/hybrid-mode).
### Does it work with LangChain?
Yes. Install `langchain-opendataloader-pdf` for an official LangChain document loader integration. See [LangChain docs](https://docs.langchain.com/oss/python/integrations/document_loaders/opendataloader_pdf).
### How do I chunk PDFs for RAG?
OpenDataLoader outputs structured Markdown with headings, tables, and lists preserved โ ideal input for semantic chunking. Each element in JSON output includes `type`, `heading level`, and `page number`, so you can split by section or page boundary. For most RAG pipelines: parse with `format="markdown"` for text chunks, or `format="json"` when you need element-level control. Pair with LangChain's `RecursiveCharacterTextSplitter` or your own heading-based splitter for best results.
### How do I cite PDF sources in RAG answers?
Every element in JSON output includes a `bounding box` (`[left, bottom, right, top]` in PDF points) and `page number`. When your RAG pipeline returns an answer, map the source chunk back to its bounding box to highlight the exact location in the original PDF. This enables "click to source" UX โ users see which paragraph, table, or figure the answer came from. No other open-source parser provides bounding boxes for every element by default.
### How do I convert PDF to Markdown for LLM?
```python
import opendataloader_pdf
# Batch all files in one call โ each convert() spawns a JVM process, so repeated calls are slow
opendataloader_pdf.convert(
input_path=["file1.pdf", "file2.pdf", "folder/"],
output_dir="output/",
format="markdown"
)
```
OpenDataLoader preserves heading hierarchy, table structure, and reading order in the Markdown output. For complex documents with borderless tables or scanned pages, use hybrid mode (`hybrid="docling-fast"`) for higher accuracy. The output is clean enough to feed directly into LLM context windows or RAG chunking pipelines.
### Is there an automated PDF accessibility remediation tool?
Yes. OpenDataLoader is the first open-source tool that automates PDF accessibility end-to-end. Built in collaboration with [PDF Association](https://pdfa.org) and [Dual Lab](https://duallab.com) (veraPDF developers), auto-tagging follows the Well-Tagged PDF specification and is validated programmatically using veraPDF. The layout analysis engine detects document structure (headings, tables, lists, reading order) and generates accessibility tags automatically. Auto-tagging converts untagged PDFs into Tagged PDFs under Apache 2.0 โ no proprietary SDK dependency. Use `format="tagged-pdf"` (Python/Node.js) or `--format tagged-pdf` (CLI). For organizations needing full PDF/UA compliance, enterprise add-ons provide PDF/UA export and a visual tag editor. This replaces manual remediation workflows that typically cost $50โ200+ per document.
### Is this really the first open-source PDF auto-tagging tool?
Yes. Existing tools either depend on proprietary SDKs for writing structure tags, only output non-PDF formats (e.g., Docling outputs Markdown/JSON but cannot produce Tagged PDFs), or require manual intervention. OpenDataLoader is the first to do layout analysis โ tag generation โ Tagged PDF output entirely under an open-source license (Apache 2.0), with no proprietary dependency. Auto-tagging follows the PDF Association's Well-Tagged PDF specification and is validated using veraPDF, the industry-reference open-source PDF/A and PDF/UA validator.
### How do I convert existing PDFs to PDF/UA?
OpenDataLoader provides an end-to-end pipeline: audit existing PDFs for tags (`use_struct_tree=True`), auto-tag untagged PDFs into Tagged PDFs (`format="tagged-pdf"`, free under Apache 2.0), and export as PDF/UA-1 or PDF/UA-2 (enterprise add-on). Auto-tagging follows the PDF Association's Well-Tagged PDF specification and is validated using veraPDF. Auto-tagging generates the Tagged PDF; PDF/UA export is the final step. [Contact us](https://opendataloader.org/contact) for enterprise integration.
### How do I make my PDFs accessible for EAA compliance?
The European Accessibility Act requires accessible digital products by June 28, 2025. OpenDataLoader supports the full remediation workflow: audit โ auto-tag โ Tagged PDF โ PDF/UA export. Auto-tagging follows the PDF Association's Well-Tagged PDF specification and is validated using veraPDF, ensuring standards-compliant output. Auto-tagging to Tagged PDF is open-source under Apache 2.0. PDF/UA export and accessibility studio are enterprise add-ons. See our [Accessibility Guide](https://opendataloader.org/docs/accessibility-compliance).
### Is OpenDataLoader PDF free?
The core library is **open-source under Apache 2.0** โ free for commercial use. This includes all extraction features (text, tables, images, OCR, formulas, charts via hybrid mode), AI safety filters, Tagged PDF support, and auto-tagging to Tagged PDF. We are committed to keeping the core accessibility pipeline (layout analysis โ auto-tagging โ Tagged PDF) free and open-source. Enterprise add-ons (PDF/UA export, accessibility studio) are available for organizations needing end-to-end regulatory compliance.
### Why did the license change from MPL 2.0 to Apache 2.0?
MPL 2.0 requires file-level copyleft, which often triggers legal review before enterprise adoption. Apache 2.0 is fully permissive โ no copyleft obligations, easier to integrate into commercial projects. If you are using a pre-2.0 version, it remains under MPL 2.0 and you can continue using it. Upgrading to 2.0+ means your project follows Apache 2.0 terms, which are strictly more permissive โ no additional obligations, no action needed on your side.
## Documentation
- [Quick Start (Python)](https://opendataloader.org/docs/quick-start-python)
- [Quick Start (Node.js)](https://opendataloader.org/docs/quick-start-nodejs)
- [Quick Start (Java)](https://opendataloader.org/docs/quick-start-java)
- [JSON Schema Reference](https://opendataloader.org/docs/reference/json-schema)
- [CLI Options](https://opendataloader.org/docs/reference/cli-options)
- [Hybrid Mode Guide](https://opendataloader.org/docs/hybrid-mode)
- [Tagged PDF Support](https://opendataloader.org/docs/tagged-pdf)
- [AI Safety Features](https://opendataloader.org/docs/ai-safety)
- [PDF Accessibility](https://opendataloader.org/docs/accessibility-compliance)
## Contributing
We welcome contributions! See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
## License
[Apache License 2.0](LICENSE)
> **Note:** Versions prior to 2.0 are licensed under the [Mozilla Public License 2.0](https://www.mozilla.org/MPL/2.0/).
---
**Found this useful?** Give us a star to help others discover OpenDataLoader.