calibre Full Text Search Not Working: How to Fix It

When calibre full-text search is not finding words that you can clearly see inside a book, the problem usually falls into one of five categories: full-text indexing is disabled, indexing has not finished, the wrong search box is being used, calibre cannot extract usable text from the selected format, or the book's index entry is stale or damaged. Scanned PDFs introduce another common cause because a page can look readable while containing only an image, not searchable text. The steps below isolate these possibilities without deleting your library, reinstalling calibre, or changing unrelated device and conversion settings.

1. Confirm the Symptom With a Small Safe Test

Begin with one book and one distinctive word. A controlled test is faster and safer than repeatedly searching a large library while changing multiple settings.

1.1 Verify that the word is real searchable text

Open the affected book in calibre's E-book viewer. Find a distinctive word that appears in the main text, preferably an unusual name or technical term. Avoid words shown only on the cover, in an illustration, or as part of a scanned page.

Press Ctrl+F on Windows or Linux, or use the viewer's search control on any supported operating system. Search for the chosen word inside the open book. If the viewer finds it, calibre can see searchable text in that format. If it does not, the problem is probably the source file rather than the library-wide index.

Success looks like this: the viewer highlights the expected word in the book. Once that happens, stop changing conversion or OCR-related settings and proceed to test the full-text index.

1.2 Use the FT search instead of the normal library search

The main calibre search bar normally searches metadata such as title, author, series, tags, comments, identifiers, and available formats. It does not automatically search every sentence inside every book.

Click the FT button at the left side of the library search bar to open calibre's full-text search interface. Enter the same distinctive word there. Depending on your interface layout, you can also open this feature through its assigned keyboard shortcut.

Success looks like this: the full-text results identify the test book as containing the word. Full-text search generally identifies matching books and a relevant occurrence. To find every occurrence within one book, open that book and use the viewer's own search.

1.3 Add a simple EPUB as a control book

If the result remains unclear, create or obtain a small, lawful, DRM-free EPUB containing a unique sentence, such as “Cobalt lantern verification phrase.” Add it to a temporary calibre library or to your regular library if appropriate. Enable indexing, wait for the new book to be processed, and search for “cobalt lantern” through the FT interface.

If the control EPUB works but the original book does not, calibre's search feature is operating and the problem is specific to the original file or its index entry. Stop changing global settings and focus on that book.

2. Check Full-Text Index Availability and Library State

Full-text indexing is optional and is managed per library. Enabling it in one calibre library does not prove that it is enabled in another library.

2.1 Enable indexing for the current library

Open the library that contains the missing book, click FT, and inspect the indexing status. If calibre offers an option to enable indexing, use it. Keep calibre open while the initial indexing process begins.

Users with multiple libraries should verify the library name or location before making changes. A book displayed in one library will not be found by the index belonging to a different library.

Success looks like this: the FT window reports that indexing is enabled and begins showing progress or a completed status. If it is already enabled, do not toggle it repeatedly. Continue to the progress check.

2.2 Wait for indexing to finish

The first index build can take time because calibre must extract text from eligible book files. The duration depends on the number and size of books, their formats, storage speed, and available processor resources. Large PDFs and libraries stored on slow disks can extend the process.

Do not assume that an empty result means failure while the initial index is incomplete. Check the FT indexing status and calibre's Jobs area. Leave the computer awake and calibre running long enough for the test book to be processed.

For a very large library, test a recently re-indexed individual book rather than waiting to draw conclusions from the entire collection. Avoid starting mass conversions or other resource-intensive jobs during this test.

Success looks like this: the pending count falls, indexing reaches a completed or substantially complete state, and the known phrase becomes searchable. Once it does, stop troubleshooting. Slow initial indexing is not evidence of a broken library.

2.3 Clear library restrictions that hide the expected result

Check whether a Virtual library, saved search, or additional restriction is active. Such restrictions can make calibre behave as though only part of the library is available. Return to the full library view and clear active filters before repeating the FT search.

Also confirm that you are searching the library copy, not expecting calibre to index books that exist only on a connected e-reader. The library and device views represent different collections. Add an authorized copy to the calibre library if you need it included in the library's full-text index.

3. Check Whether the Book Contains Extractable Text

A visible page is not always a text page. Full-text indexing depends on text that calibre can extract from a supported, readable book format.

3.1 Test the format in the viewer

Text-based EPUB, AZW3, MOBI, TXT, HTML, DOCX, and many PDFs can be searchable when their contents are available to calibre. Actual results depend on the individual file. A malformed container, unusual encoding, damaged document, or image-only page can prevent expected words from entering the index.

Open the exact format that you expect calibre to index. If a book record contains several formats, the word may exist in one format but not another. For example, an updated EPUB may contain a new chapter while an older PDF attached to the same record does not.

Try selecting and copying the relevant sentence in the viewer or another trusted document reader. If text selection produces sensible words, the file probably has a text layer. If selection is impossible or yields random characters, extraction may fail or produce different text from what appears visually.

3.2 Recognize scanned PDF and OCR limitations

A scanned PDF often consists of photographs of pages. Human readers can recognize the letters, but full-text software sees images unless the PDF also includes an OCR text layer. calibre's index cannot reliably find a word that does not exist as machine-readable text.

If the PDF is image-only, use a lawful OCR tool or request an accessible text-based edition from the document provider. OCR quality matters. It can confuse characters such as lowercase “l,” uppercase “I,” and the number “1,” split hyphenated words, or misread accented characters. Search for a short neighboring word to determine whether the OCR text differs from the printed appearance.

Do not use this troubleshooting process to bypass DRM or other access controls. If a protected retailer file cannot be read normally by calibre, obtain a compatible, authorized edition from the provider.

Success looks like this: the word can be found by searching inside the opened document. If the viewer cannot find it, rebuilding the library index is unlikely to help until the source text issue is corrected.

3.3 Convert only when conversion solves a demonstrated source problem

Conversion is not a mandatory full-text-search repair. Do not convert a healthy EPUB merely because an index is delayed. Consider conversion only if calibre can legally open the source, the format behaves poorly, and a test conversion produces a clean text-based EPUB or AZW3.

After conversion, open the new format and search inside it before rebuilding anything. If the expected phrase appears in the converted file, re-index that book. If it remains absent, inspect the source quality rather than repeating conversions with random settings.

4. Re-Index the Affected Book Before Rebuilding Everything

If the viewer finds the word but the FT search does not, the individual book's index entry may be missing or stale.

4.1 Re-index one book

Select the affected book. In calibre's Book details panel, right-click and choose Re-index this book for full text searching. Wait for the indexing job to complete, then repeat the same distinctive-word search in the FT interface.

This targeted action is appropriate after replacing a format, editing a book, correcting a damaged file, or confirming that searchable text exists but is absent from library-wide results.

Success looks like this: the book appears in FT results after its individual re-index finishes. Stop there. There is no benefit in rebuilding the entire library index when one record was the only problem.

4.2 Rebuild the broader index only when the pattern justifies it

A full re-index is reasonable when many known-good EPUBs fail, the index status is inconsistent, or the library was moved or restored and widespread results are missing. Use the FT interface's available index controls. Advanced users can also use calibredb fts_index to inspect status or request re-indexing, but the graphical interface is preferable for most readers.

Back up the calibre library before performing broad maintenance. A full re-index can consume substantial time and computing resources in a large collection, so schedule it when the computer can remain on and the library folder will stay available.

5. Check Permissions, Storage, Security Software, and Sync Tools

Full-text indexing requires reliable access to the library database, book files, index data, and temporary working locations. Operating system or storage interference can prevent indexing even when the books themselves open.

5.1 Confirm write access to the library

Make sure your Windows, macOS, or Linux account can read and write the entire calibre library folder. Warning signs include metadata edits that do not persist, failures when adding or removing formats, database errors, or repeated indexing work after every restart.

Avoid using system-protected application directories for a personal library. On Linux, check ownership and permissions rather than launching calibre as root. On Windows and macOS, verify that the library is not read-only and that privacy or controlled-folder settings are not blocking calibre.

5.2 Test outside cloud synchronization

Cloud-sync applications can lock, delay, replace, or partially synchronize files while calibre is updating its database and index. Pause synchronization during a controlled test, or copy a few books into a new local temporary library stored on an internal drive.

Do not open and modify the same live calibre library from multiple computers at once through a synchronized folder. If the local test works, the failure likely involves storage availability or synchronization behavior rather than the book format.

5.3 Review antivirus and endpoint protection events

Security software may block a calibre process, quarantine a temporary file, or interfere with repeated content extraction. Check the product's event history for actions involving calibre or the library path. If organizational security policies control the computer, ask the administrator to review the event rather than disabling protection.

Create exclusions only when you understand the security implications and trust the library's files. A safer first test is a small local library containing a known EPUB.

5.4 Separate local FT problems from server and device limitations

USB mode, device detection, email delivery, and send-to-device settings do not normally control the desktop library's full-text index. Likewise, metadata downloads change fields such as title, author, tags, comments, and covers, not the prose inside a book.

The calibre Content server's ordinary catalog search may expose library metadata search behavior rather than duplicating every desktop FT workflow. First prove that FT search works locally in the calibre desktop application. Only then investigate browser caching, server permissions, library visibility, network access, or interface-specific limitations.

6. Use Logs and Debug Tools When the Index Still Fails

Logs are useful when indexing stalls, reports an error, or repeatedly skips known-good text books. They are less useful when the real issue is simply an image-only PDF.

6.1 Inspect Jobs and conversion details

Open calibre's Jobs area and look for failed or unfinished operations. Double-click a relevant job to view details. Record the book title, format, library path, and exact error text.

If conversion is part of the test, open the conversion job log to see the settings that were actually used. calibre also provides conversion debug output that saves intermediate pipeline stages. This can reveal whether readable text survived the input stage before the output book was created.

6.2 Run calibre in debug mode

For a repeatable failure, restart calibre in debug mode through the Preferences menu option, reproduce the smallest possible test, and close calibre normally so the log can be reviewed. Advanced users can launch the graphical interface from a terminal with calibre-debug -g to send diagnostic output to the terminal.

Look for errors mentioning the affected file, text extraction, database access, permissions, temporary directories, or a third-party plugin. Do not publish a complete log without checking it for usernames, local paths, email addresses, server details, or other private information.

6.3 Temporarily disable relevant third-party plugins

Most plugins have no role in full-text indexing, so disabling every plugin should not be the first step. However, a plugin that changes imported files, replaces formats, modifies library behavior, or integrates with storage could affect the test. Disable only the relevant third-party plugin, restart calibre, re-index the control book, and test again.

If the search begins working, leave that plugin disabled while checking for an update or contacting its maintainer. If nothing changes, re-enable it and continue. Change one variable at a time.

7. Run a Clean Temporary Test Before Reinstalling

Reinstalling calibre usually does not repair a damaged source file, incomplete OCR, a disabled per-library index, or a cloud-locked library. A temporary library gives much clearer evidence.

1. Create a new empty library in a normal local folder.
2. Add one small DRM-free EPUB containing a known unique phrase.
3. Enable full-text indexing for that temporary library.
4. Wait until the book has been indexed.
5. Search for the phrase through the FT interface.
6. Open the EPUB and confirm the same phrase with the viewer search.

If both searches work, the calibre installation is functional. Return to the original library and investigate its index state, storage location, restrictions, or individual files. If the temporary test fails, capture a debug log and check permissions or security software before reinstalling.

Do not delete the original library, its metadata database, or unfamiliar index files as an experiment. Do not move book folders manually within a calibre library. Use calibre's own library tools and make a verified backup before structural changes.

8. Quick Fix Checklist

• Click FT instead of using the normal metadata search bar.
• Confirm indexing is enabled for the currently open library.
• Wait for initial indexing and check the Jobs area for failures.
• Search inside the book first to prove that the word is extractable text.
• Test a small, known-good, DRM-free EPUB with a unique phrase.
• Clear Virtual library and additional search restrictions.
• Re-index the affected book from the Book details panel.
• Use a full re-index only when many known-good books are missing.
• Check scanned PDFs for an OCR text layer.
• Pause cloud synchronization during a controlled local test.
• Verify write permission for the library and temporary folders.
• Review relevant antivirus events without disabling protection blindly.
• Use job details or debug mode if indexing reports repeatable errors.
• Test in a new temporary library before reinstalling or deleting anything.

9. Frequently Asked Questions

9.1 Why does calibre find a title but not a word inside the book?

The normal library search primarily searches metadata, while words inside book contents require the separate FT interface and an enabled full-text index. Finding a title proves that the book's metadata record exists. It does not prove that the book's content has been indexed.

9.2 Which book formats are searchable?

Searchability depends on whether calibre can read and extract actual text from the specific file. Common text-based e-book formats such as EPUB and AZW3 make good tests. MOBI, DOCX, TXT, HTML, and text-based PDFs may also be searchable. An image-only PDF or a damaged file can fail even when its extension is normally supported.

9.3 Why is my scanned PDF missing from the results?

The PDF probably contains page images without a usable OCR layer. Full-text indexing cannot search visual letters in an image as ordinary text. Test by searching inside the opened PDF or selecting and copying a sentence. Obtain or create a lawful OCR-enabled version if necessary.

9.4 How long should calibre indexing take?

There is no universal time because library size, file size, format complexity, storage performance, and computer resources vary. Watch the indexing status rather than assuming a fixed deadline. A small EPUB should provide a quicker diagnostic result than a large scanned or complex PDF.

9.5 Should I rebuild the entire index?

Not initially. Re-index one affected book after proving that its contents are searchable in the viewer. Rebuild the wider index only when multiple known-good books fail or the index reports a broader problem. Stop as soon as the targeted re-index restores correct results.

9.6 Will reinstalling calibre fix full-text search?

Usually not when the cause is a disabled library index, incomplete indexing, an image-only PDF, insufficient permissions, cloud interference, or one stale book entry. Run the temporary-library test first. Reinstallation becomes reasonable only if a clean local test also fails and logs suggest damaged program files rather than a library or source-file problem.