# Import and export Use a project archive to move a project between Fablewright instances. Use Word, EPUB, or print to hand over the live manuscript for reading. Use an instance backup to preserve accounts and the whole server. These are different copies. A project archive includes deleted work and earlier drafts, so do not send one to a reviewer who only needs the current manuscript. ## Choose an export Open Export from a library card or the project menu. Owners and co-authors can export. Readers and guests cannot use the export routes. Word, EPUB, and browser print can export the whole project or one book, chapter, or scene. The project archive always covers the whole project, including all its live books. There is no partial archive. Wait for saves to finish before exporting. An export reads the server, not unsaved words in an editor or another tab. | Format | Use it for | What it leaves out | | ------------------------ | ------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------ | | Word (`.docx`) | A submission manuscript or further editing in a word processor. | Notes, note images, planning, feedback, deleted work, and history. | | EPUB (`.epub`) | Reading on an ebook reader. | Notes, note images, planning, feedback, deleted work, and history. | | Project archive (`.zip`) | A whole-project copy, migration, or long-term copy outside the instance. | Accounts, access permissions, writer goals and daily counts, and the original import report and source file. | | Browser print or PDF | Reading the selected project, book, chapter, or scene. | Project data outside the printed manuscript. | ### Export one book, chapter, or scene Open Read-through and choose the part you want. Use Read alone on a scene, Read the chapter in the outline, or open a book. The scope bar names what is on screen. The Export control offers Word and EPUB for that scope, and Print saves the same scope as a PDF. A scoped Word or EPUB contains live words only. Like a whole-project Word or EPUB export, it leaves out deleted work, recovery history, notes, planning, feedback, and note images. The document title and file name identify the selected part. A book export uses the book title. A chapter or scene export combines its parent titles. ### Word The submission preset uses Times New Roman at 12 points, double spacing, half-inch first-line indents, and one-inch margins. Each chapter starts on a new page. A `#` separates scenes. The running header uses the project's author, title, and page number. Set the pen name in Details on the library card. Your account name does not become the manuscript author. If the author is empty, the header uses `Author`. Front matter comes before the story and back matter after it. Inspect the result in the word processor you intend to use. The tests cover document structure and Fablewright's reader. They do not prove every version of Word renders identically. ### EPUB EPUB includes a table of contents and the live manuscript in reading order. The project author and description supply book metadata when present. The reader controls typography, so the app's body-font choice does not prescribe the reading device's font. This is a reading copy, not an editable project archive. Fablewright does not import EPUB. For print-ready typesetting or PDF/X, use another tool after export. ## Inside a project archive The archive is an ordinary zip. You can open it without a running Fablewright instance. | Path | Contents | | ----------------- | ---------------------------------------------------------------------------------------------------------------------- | | `manuscript.json` | Project details, structure, prose, notes, templates, plot data, links, feedback, trash, and retained recovery history. | | `books/` | Live manuscript as Markdown, with folders for books, parts, chapters, scenes, and front and back matter. | | `notes/` | Live notes as Markdown. | | `images/` | Available note display images and any preserved originals, including images on trashed notes. | | `README.md` | A short explanation of the archive and its images. | `manuscript.json` is the structured project copy. Markdown is a reading copy for other tools and version control. Editing Markdown inside the zip does not change what Fablewright imports: archive import reads the JSON. The current archive format is version 1. It keeps front and back matter placement. A newer format is refused by name. Files without a supported format marker do not import. Import creates new identifiers and gives ownership to the account that imports. Deleted items stay in Recovery. Scene versions, note and card versions, project versions, and feedback travel with the project. Memberships, invitations, and secret links do not. The archive does not include the original document you imported or its import report. Download those separately from the library card if you need them. Instance backups include them. ### Pictures and standalone JSON A display image travels as WebP. A preserved original travels beside it under an `.orig` extension. Import reads the original when available, or the display copy otherwise. A standalone `manuscript.json` holds image descriptions and names, but no picture bytes. Importing it does not reuse images already on the destination instance. Use the complete zip when notes have pictures. An export cannot recover an image file already missing from the server. Inspect important pictures after a trial import. An image description in JSON is not evidence that the archive contains the image itself. ## Import a manuscript Choose Import on the library page. You need a signed-in account. Import always creates a new project and never merges into or replaces an existing one. Accepted inputs are: | File | Behavior | | -------------------------------- | ---------------------------------------------------------------------------------------------------------------- | | Fablewright `.zip` | Imports directly from its `manuscript.json` and available images. An arbitrary zip of Markdown is not supported. | | Fablewright `.json` | Imports directly as a structured project without image files. | | Scrivener `.scriv` folder or ZIP | Shows manuscript structure, counts, excerpts, and conversion warnings before import. Research becomes notes. | | Word `.docx` | Shows chapters, scenes, word counts, excerpts, and conversion warnings before import. | | `.md` or `.markdown` | Shows the structure derived from headings before import. | | `.txt` | Uses the same preview and parser as Markdown. Markdown-like punctuation can become formatting. | Word and text imports create one book. Archives can restore several books. A title entered for a Word or text import names the new project. An archive keeps its project details. For a Word or text file: 1. Keep a separate copy of the source document. 2. Choose the file in Import. 3. Read the preview, including conversion warnings and chapter boundaries. 4. If the structure is wrong, cancel and adjust the source file. 5. Approve the import. 6. Open the new project and inspect the beginning, a middle chapter, and the ending. 7. Download its Report and Source file from the library card when available. The report records proposed and imported counts, warnings, the source name, and its digest. A digest is a value used to identify file contents. The instance also tries to keep the original file. A storage failure can leave an imported project without a report or source download. Keep your own original until you accept the result. If the response fails after submission, refresh the library before trying again. The project can already exist even when the browser did not receive confirmation. A second import creates another copy. ## Prepare Word documents Use paragraph styles rather than appearance alone. A large bold line is not necessarily a heading. Fablewright reads the structure this way: - Heading 1 starts a chapter. - Heading 2 starts a scene within a chapter. - A recognized scene-break paragraph starts another scene. - Higher-level headings remain headings inside scene prose. - Text before the first structural heading becomes a chapter named Front matter. - A document with no structural headings becomes one chapter and scene. Scene-break paragraphs include `***`, `* * *`, `---`, `- - -`, `#`, and three bullets. An ordinary page break does not by itself define a chapter. A chapter without prose gets an empty scene. The chapter named Front matter is not automatically assigned front-matter placement. After import, use Move to front matter if that text belongs before the numbered chapters. External files do not reconstruct a series hierarchy or planning notes. ### What Word conversion changes | Source content | Result | | --------------------------------------------------------- | --------------------------------------------------------------------------- | | Ordinary text, bold, italic, underline, and strikethrough | Kept as supported prose and marks. | | Lists | List text becomes plain paragraphs. Numbering and nesting are lost. | | Tables | Table text becomes paragraphs. Cell layout is lost. | | Images | Omitted from scene prose. | | Footnotes and endnotes | Omitted. | | Comments | Omitted. They do not become Fablewright feedback. | | Tracked changes | Insertions are kept and deletions are dropped. The revision record is lost. | | Unrecognized paragraph styles | Converted to ordinary paragraphs. | Read the warnings before accepting. If tracked changes matter, resolve them in the source document first. A Word round trip is not a lossless project round trip. Keep the project archive too. ## Prepare Markdown and plain text Use UTF-8 text. A level-one heading starts a chapter and a level-two heading starts a scene. Level-three and deeper headings become level-three headings within the prose. For example: ```markdown # Chapter one ## Arrival The first scene goes here. ## Departure The second scene goes here. # Chapter two Text directly under a chapter heading becomes its first scene. ``` Text before the first heading becomes a chapter named Front matter. With no headings, the file becomes one chapter and scene. A horizontal rule in Markdown does not split scenes as a Word scene-break paragraph does. An opening YAML metadata block is dropped. Pipe tables lose their table layout, and nested lists flatten to one level. The parser handles the app's supported prose forms, not every Markdown extension or embedded HTML feature. Inspect the preview and the imported prose. Plain text uses that same parser. If you need punctuation such as heading marks or asterisks to remain literal, escape it as Markdown or use a prepared Word document. ## Import a Scrivener project Close the project in Scrivener before copying or importing it. Import reads the saved project, not unsaved editor changes. Keep a separate copy of the complete `.scriv` folder. Choose **Choose a Scrivener folder** in Import, or upload a ZIP of the folder. The folder must contain its `.scrivx` file and `Files` directory. You can also ZIP the folder contents directly. Do not upload `.scrivx` alone: that file describes the binder but does not hold the prose. A ZIP must contain only one project. The importer reads the older numbered `Files/Docs` layout and the UUID-based `Files/Data` layout used by Scrivener 3. It creates one book. Review the proposed chapters, scenes, original paths, note count, and conversion notices before approving. You can change the project title. | Scrivener content | Fablewright result | | ----------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------- | | Manuscript binder | Parts, chapters, and scenes in binder order. Deep nesting becomes ordered scenes with path-based titles. | | Text on folders or parent documents | Separate scenes before their children. Empty manuscript chapters receive an empty scene. | | Front Matter or Back Matter folders outside the manuscript | Unnumbered matter pages. The importer recognizes these English folder names. | | RTF text | Paragraphs, Unicode, bold, italic, underline, strikethrough, external links, and one-level lists. | | Synopses and standard statuses | Chapter or scene synopsis and a matching drafting, revising, or complete status. Original status names remain in metadata notes. | | Research and other binder material | Notes grouped by note type. Note summaries keep the original binder path. | | Document notes, footnotes, annotations, inspector comments | Separate notes, linked to the corresponding scene where possible. Comments do not become passage feedback. | | Metadata, keywords, labels, bookmarks, collections, targets | Binder map and project metadata notes. Bookmarks to imported research notes also become scene-to-note links. | | Template folder | Notes plus note types with the direct child templates. | | Research images and embedded PNG or JPEG pictures | Separate note images where validation succeeds. TIFF research images become WebP, with their originals in the source ZIP. | | Trash | Deleted notes in Recovery, with their text and original paths. | | Recognized snapshot files | Scene versions or note versions. Unassociated snapshot files become notes. | The report identifies flattened tables and list nesting, unsupported formatting, and links that became plain text. Internal Scrivener links do not become links to Fablewright scenes. Fonts, colors, page layout, compile instructions, editor preferences, and media playback are not reproduced. Literal compile placeholders stay in the text rather than being evaluated. Unsupported attachments remain in the source ZIP, with descriptive notes in the project. When a snapshot index supplies a matching title and date, the version keeps them. Otherwise, the version uses its filename and project date. A source without a project date uses the Unix epoch as an unknown-date marker. Imported scene versions receive names so automatic history retention does not remove them. The browser packages a folder as a ZIP before upload. The source download returns that ZIP, not a single RTF or `.scrivx` file. Keep it separately: a Fablewright export does not include the original import source. Inspect prose, notes, images, trash, and important versions after importing. For another writing app, export one manuscript as Word or Markdown. Those formats do not transfer its research binder or revision history. Import a representative sample before moving a whole series. ## Limits and recovery One MiB is 1,048,576 bytes. The file limit is 100 MiB, or 104,857,600 bytes. The interface calls this 100 MB. An archived `manuscript.json` can expand to 128 MiB. A proxy or request-body limit can reject a smaller upload before the app reads it. The default `BODY_SIZE_LIMIT=101M` sits one MiB above the file limit, so a full-size file fits with its multipart headers. That ceiling applies to imports. Other routes are capped at 12 MiB when they carry a document, and 2 MiB otherwise. The entry point refuses a larger body before anything reads it. A reverse proxy can also reject the request before the app reads it. Allow more than 100 MiB there too. Raising the request limit does not raise the 100 MiB file limit. Individual expanded Word XML entries also have a 16 MiB limit. A Scrivener ZIP can expand to 200 MiB. Each project document has a 16 MiB limit. A project can contain at most 20,000 ZIP entries and 10,000 binder items. Images use the normal 10 MiB and 64-million-pixel limits. Encrypted ZIPs, symbolic links, unsafe paths, and damaged prose are refused before project creation. Exports have no matching 100 MiB cap. A project with many images or versions can produce an archive too large to import. Keep that archive, but use an instance backup for server recovery. Do not prune irreplaceable history just to force an import under the limit. An archive is private manuscript data. It contains old drafts, trash, and feedback author names as well as the live prose. Store it somewhere you trust. For instance backups, restore procedures, and moving the whole server, use [the operations guide](operations.md).