How to use
- Add DOCX, HWP or HWPX files (up to 100).
- Enter text to find and its replacement under Replacement rules. Use Add rule for more lines, or import a two-column CSV (first column find, second column replace) with Load rules from CSV. A first row of
find,replace(or찾을 말,바꿀 말) is skipped as a header. - Press Preview. For each file you see how many replacements will be made and the context of the first few (text to be removed in red, new text in green). Nothing is saved.
- If it looks right, press Replace now. The results are re-opened to check that the text still reads correctly, and then Download results gives you a single file directly or a ZIP for several. File names are unchanged, and HWP stays HWP, HWPX stays HWPX and DOCX stays DOCX.
If you change the rules, options or files, run the preview again before replacing.
Options and rule order
| Option | Behavior |
|---|---|
| Match case | Off by default |
| Whole word | Replaces only when not next to letters or digits. In Korean, a word with a particle attached is excluded |
| Regex | Treats the find text as a regular expression. In the replacement you can use $1, $2 (groups), $& (whole match) and $$ (a dollar sign) |
- Rules run from top to bottom, and a later rule applies to the result of earlier rules. For example, rules
가→나then나→다turn가 나into다 다. - Within one rule, matches are replaced left to right without overlapping. The regex
(\d+)-(\d+)-(\d+)replaced by$3/$2/$1turns2026-10-07into07/10/2026. - In plain mode,
$is just a character.
DOCX: how formatting is preserved
Word splits text into several pieces (runs) whenever formatting changes. If only 홍 of 홍길동 is bold, then 홍, 길 and 동 are separate runs. This tool joins the run text of each paragraph to search, and when a match spans several runs it puts the replacement into the first run and deletes only the matched characters from the others, so the first character’s formatting carries over to the new text. Example: with runs 안녕 홍, 길, 동입니다, the rule 홍길동→김철수 gives 안녕 김철수, an empty run and 입니다.
- When the new text starts or ends with a space,
xml:space="preserve"is added so the space is not lost. - Tracked deletions, field codes, pictures, styles and other parts are untouched. Only the text is rewritten in the original XML, so the other formatting tags stay exactly as they were.
- Replaced in: body (including tables), headers, footers, footnotes and endnotes. Comments and text-box fallback copies are excluded.
HWP and HWPX: what is replaced safely
For Hangul documents the text of each paragraph is read to find matches, then new text is inserted and the old text deleted from the back so positions never shift. Text inside tables is replaced too.
- Skipped: if a field or picture sits in the middle of the matched text, it cannot be replaced safely, so it is skipped and the number is shown. Paragraphs in table cells that contain objects are skipped as well.
- Headers, footers, footnotes, endnotes and text boxes are not changed yet. Check those places in Hangul if you need them.
- A line break in the replacement becomes a space. Password-protected documents are not supported.
- New text inherits the character style of the text it replaces. For important documents, open the result in Hangul once to check.
Limits
- Phrases that span paragraphs (a sentence with a line break) are not found; only text inside a paragraph is replaced.
- Up to 100 MB per file and 200 rules. Regexes run in a separate worker and a file is skipped if it takes over 10 seconds.
- If a result cannot be re-opened and verified, the original is left as it is and an error is shown.
Related: Find and Replace Text, Find & Replace in Excel, HWPX Mail Merge, Fill HWPX Fields, Search Text in Multiple Files.
Your files are processed only in your browser and never sent to a server.