HWPX Text Extractor — Korean Hangul documents to plain text

What does HWPX Text Extractor do?

HWPX Text Extractor pulls clean plain text out of Korean HWPX documents (the XML-based format of Hangul Word Processor 2014+). HWPX is the standard document format across Korean government agencies, public institutions, and enterprises — grant announcements, procurement notices, official forms, and reports are published in it every day. Most text-processing pipelines cannot read it.

Give this Actor a list of file URLs and it returns the extracted text of each document as structured JSON — ready for search indexing, LLM pipelines, translation, or archiving. Running on the Apify platform gives you API access, scheduling, and integrations out of the box.

Why use HWPX Text Extractor?

• Korean public-sector data is locked in HWPX. Government grant portals, procurement systems, and municipal sites attach HWPX files that generic parsers (and most OCR pipelines) skip.
• No Hangul software required. No HWP viewer, no Windows, no manual conversion.
• Honest error reporting. Legacy binary .hwp (HWP 5.x, OLE2) files are detected and named explicitly — you get "this is a legacy binary HWP, re-save as HWPX" instead of a silent failure or garbage output.
• Batch-friendly. One run processes a whole list of URLs; each file becomes one dataset item, and one broken file never kills the batch.

How to use it

1. Open the Actor and paste one or more direct HWPX file URLs into HWPX file URLs.
2. Click Start. The Actor downloads each file and extracts its text.
3. Open the Output tab (or the dataset API) and download results as JSON, CSV, or Excel.

Input

{
    "fileUrls": [
        "https://example.go.kr/files/announcement.hwpx",
        "https://example.go.kr/files/form.hwpx"
    ]
}

URLs served with a different extension work too — the Actor checks the actual file bytes, not the filename.

Output

One dataset item per URL:

{
    "url": "https://example.go.kr/files/announcement.hwpx",
    "filename": "announcement.hwpx",
    "ok": true,
    "text": "전북특별자치도 공고 제2026-1353호 ...",
    "textLength": 13688,
    "sectionCount": 1,
    "error": null
}

Failed files stay in the dataset with ok: false and a human-readable error, so you always know exactly what was skipped and why:

error	meaning
legacy binary HWP 5.x file (OLE2) ...	The file is old-format .hwp, not HWPX. Re-save it in Hangul 2014+
not a zip-based HWPX file	The URL did not return an HWPX document
no Contents/section*.xml inside	Zip container without HWPX structure
HTTP errors (404, 403, timeout)	The file could not be downloaded

Pricing

The Actor charges per successfully parsed document — failed downloads and unreadable files are free. Parsing is fast (a typical 40-page government announcement takes under a second), so platform usage costs stay minimal.

Limitations & tips

• HWPX only. Legacy binary .hwp (HWP 5.x) is intentionally out of scope — it is a completely different, undocumented OLE2 format. The Actor detects it and tells you.
• Text comes out in reading order, paragraph by paragraph. Tables are flattened into text lines; cell structure is not preserved in this version.
• Files over 50 MB are rejected to protect your usage costs.

FAQ & support

Is scraping the files legal? This Actor does not crawl any site — it only downloads the URLs you provide. Make sure you have the right to access those files.

Found a bug or need a feature (table structure, HWP 5.x support, OCR fallback)? Open an issue on the Actor's Issues tab — the roadmap is driven by real requests.