iloveepub로컬 처리 — 파일은 절대 기기를 벗어나지 않습니다
내 EPUB 고치기
복구 작업 전부를 도구 하나로: 열리지 않는 책, 목차가 깨진 채 열리는 책, 서재에서 이름 없이 뜨는 책을 넣어 보세요. 먼저 건강 검사가 실행되고, 구조적 손상은 자동으로 복구되며, 표지와 메타데이터의 빈틈은 직접 채워 넣습니다 — 그리고 복구된 책은 다운로드 전에 다시 검사됩니다. 모두 브라우저 안에서 이루어지고, 파일은 기기를 떠나지 않습니다.
01EPUB 선택
The books that reach this page
A download that Apple Books and Calibre both refuse, often because a zip tool compressed the mimetype entry or moved it off the first position. A merge or a hand edit whose table of contents still points at files that were removed. A shelf that shows the filename because dc:title was never written. You do not yet know which of those it is. The health check is the point of the page.
Come here before the editor. An editor needs a package it can parse. This tool tells you whether the package can be parsed, repairs the structural layer, and runs the same check on the result before the download appears.
The files a repair is allowed to rewrite
META-INF/container.xml is rebuilt when it is missing or malformed. The archive is repacked so mimetype is the first record, stored rather than deflated, containing exactly application/epub+zip. Manifest items and spine entries that point at missing files are removed. A missing EPUB 3 nav, or a missing EPUB 2 NCX, is rebuilt from the chapters that exist. Dead links in a TOC that is already there are stripped.
Chapter HTML, images, and CSS are copied byte for byte. They are not re-encoded and not tidied. Title, author, and cover change only when you type or upload them. Those writes are dc:title, dc:creator, meta name=cover, and the cover-image property. The same checklist then runs on the bytes you are about to download. The cap is 120 MB, with a warning past 100 MB because the repair holds the archive in memory.
Damage this page will name and not pretend to fix
A file that is not a valid ZIP, a book with no readable chapters, and a DRM-protected file are reported as beyond repair. A truncated download has to be fetched again. DRM is not removed. Broken prose inside an otherwise valid chapter — unclosed tags in one story, a stylesheet that hides the text — is not rewritten. The health check does not tidy HTML.
Fonts are not subsetted and images are not recompressed. If the report says the structure is healthy and a reader still misbehaves, the problem is in the chapter markup, which is an editor's job.
Calibre Polish fixes more, and changes more
Calibre's Polish books and Modify EPUB can subset fonts, recompress images, and flatten CSS while they repair a book. That extra cleanup is useful. It is also how an innocent repair changes the typography. The editor's check is closer to EpubCheck, and it leaves you to try the file yourself.
This page stays in the package layer: container, mimetype, manifest, spine, nav, cover markers, Dublin Core. It then repeats its own checklist before download. Use Polish when you want the font and image cleanup. Use this when the chapters should stay byte for byte. If the report says the archive is not a ZIP, Calibre cannot resurrect it either.
고장 난 EPUB 고치는 방법
고장 난 책 넣기
아무 .epub 파일이나 선택하세요 — 열리지 않는 책, 목차가 깨진 책, 서재에서 이름 없이 뜨는 책이면 됩니다. 아무것도 업로드되지 않습니다.
진단 리포트 확인
진단이 모든 검사 항목 — 컨테이너, mimetype, 참조, 목차, 표지, 메타데이터 — 을 정상, 복구 가능, 복구 불가로 표시해 보여줍니다.
복구 적용
구조 복구는 자동으로 실행됩니다. 표지나 제목·저자 메타데이터가 없으면 같은 단계에서 채워 넣으세요 — 둘 다 선택 사항입니다.
검증된 책 다운로드
복구된 EPUB은 다운로드 전에 재검사됩니다. 눈가림 성공 메시지가 아니라 복구 리포트와 전체 통과 확인을 받습니다.
자주 묻는 질문
A zip tool re-saved my EPUB and now nothing opens it. What broke?
Often the mimetype entry. Readers expect it first in the archive, stored rather than deflated, containing exactly application/epub+zip. A general zip app breaks that. The repair writes the entry back and runs the health check again on the result.
Which files does a repair rewrite?
container.xml when it is missing or malformed, the mimetype entry, the OPF when dead manifest or spine references are removed, and nav or NCX when they are missing or contain dead links. Chapter HTML, images, and CSS are copied through. Title, author, and cover change only if you supply them.
Calibre's Polish books also claims to fix EPUBs. Why use this?
Polish can subset fonts, recompress images, and flatten CSS while it fixes the package. That changes more than the structure. This page does not do those transforms. It repairs the package and shows the same checklist again before download.
The report says the book cannot be fixed. Should I try Calibre?
If the file is not a valid ZIP, Calibre cannot read it either. Download it again. If it is DRM, neither tool should strip it. If the package is fine and one chapter is broken HTML, Calibre's editor or Sigil can edit that chapter. This check does not rewrite prose markup.
Does the second check send the book anywhere?
No. The re-check is the same in-browser inspection as the first, on the bytes about to be downloaded. A cover you chose to skip stays listed. It is not described as a failure of the structural repair.