Skip to content

Library and IIIF viewer ​

The Library contains sources added to your personal catalogue, their metadata and workspace links. Search is available in the Dashboard. IIIF is the protocol used to describe and display many supported digital copies: a manifest provides metadata, page order and image references.

Catalogue structure ​

ElementPurpose
WorkBibliographic record with title, author, date, language and identifiers
DigitisationDigital representation linked to the record, such as an IIIF manifest or PDF
Local versionImages from a digitisation stored in the file repository at a particular resolution

The catalogue does not automatically merge works by title. Manifest identity prevents the same source from being added twice. A digitisation may have several local versions, including versions created by resizing images rather than downloading them.

Organisation ​

Every work looks the same in search results, in the catalogue, on its page, and in the transcription Studio: author, year, place, and printer first, then the title in italics, cut after two lines (one in headers) and readable in full on hover. At the bottom, in small print, library, pages, and status. Early printed books are recognised by who wrote and printed them, not by titles half a page long.

The right-hand column holds the shelves, fixed ways of looking at the catalogue: All, Recent (added or opened in the last 30 days), To download (not all on the computer yet), Being transcribed (with a transcription not yet fully verified), Not linked to a workspace, Archived. Archived works appear only on their own shelf. Below are the collections and the saved views. Every entry shows how many works it holds. The "+" next to each section heading opens the name field where the new entry will appear: Enter saves, Esc or a click elsewhere cancels. Deleting a collection does not touch its works.

Above the list, on its own line, the search looks at every piece of data about the work — title, author, printer, place, notes, identifier — and below it the quick filters narrow the chosen shelf or collection by type, century (from the first year in the date), language, library, download state and workspace. Each value shows how many works it would have with the other filters already applied. The list sorts by title, author, year, recently added or recently opened. A saved view remembers the chosen filters: its "+" is enabled only when at least one filter is on, and says why on hover. The view stays a filter and updates by itself as new works come in.

The list shows as rows, covers or a table. Covers are all the same size: when space runs out the title gets shorter, and two links are shown; the others are counted ("+3") and listed on hover. The table has a column for author, title, year, place and printer, library, pages and state (work and download); author, title and year sort by clicking the header. With Group the list splits by century, author, library or collection, with the number of works next to each group; works without the value go into a group at the end, and a work in several collections appears under each. View and grouping stay as you leave them.

List and grid views show provenance, declared page count, local resolutions, disk usage and links. An unavailable count is not equivalent to zero. Collections group works without moving or duplicating them.

A work can belong to several workspaces and collections. Removing a link does not delete the record or its files. At the bottom left of a row are the commands to link it to a workspace and add it to a collection; at the top right, on the first line, the other commands as icons named on hover, in three groups: create a transcription | download, verify, reduce images, free space | archive, delete. All of them appear on hover or when the row has focus. In the cover grid and in the table the same commands sit in a menu. The small line also tells how far the work has gone: being transcribed (a transcription not yet fully verified), transcribed, translated.

Several works at once ​

A click opens the work. With Ctrl (⌘ on the Mac) a click adds it to the choice or removes it; with Shift every work between the last choice and the clicked one is chosen, in list order; the check mark on the left of the row does the same as Ctrl. While the choice is not empty, a bar above the list offers the commands that apply to all of them: collection (a new one too), workspace, download, archive or return to the catalogue. Archiving several works at once does not offer to free space, as it does for a single work. Esc or the cross clears the choice, which also clears when you change shelf or collection.

A row can be dragged onto a collection in the shelves column: the row goes in, or the whole choice if the row is part of it.

Work details ​

The detail view places the viewer alongside a panel with four sections:

  • Work: bibliographic metadata, additional fields and technical references.
  • Digitisations: registered copies, downloads and local versions.
  • Organisation: workspace and collection links.
  • Notes: notes about the work, saved automatically.

Title, author, date and language can be edited manually. Overrides are stored separately from the original values and can be removed. Resynchronise with the library retrieves fresh metadata and clears manual overrides only for the fields the library declares in that reading: a correction on a field the library does not provide is kept, along with notes and downloaded files.

Reading pages ​

The viewer provides thumbnails, page-number navigation and zoom, and remembers the reading position. It uses local images when available. Online reading loads a page image first and may request higher-detail tiles as you zoom in. Enlarging a local image does not increase its resolution.

A command in the viewer toolbar opens the page you are looking at in the library's own viewer, where the address shape has been verified — today Gallica and Internet Archive. The link to the whole work sits at the top of the item page, next to the other commands.

The Local file / Online file indicator distinguishes repository files from remote reading. Its tooltip provides the source, cache status and image dimensions. Viewer caching is not a permanent download.

When local-only reading is enabled, missing pages display a notice and the viewer does not request images from the library. This setting applies to the open work and resets when it closes.

Saving pages and managing versions ​

Saving the current page writes the bytes already displayed. It does not request a new image at the configured resolution; the tooltip shows the actual dimensions. Reading a partially downloaded version can fill gaps with pages retrieved during browsing.

Downloading a complete digitisation creates a background job. Requesting a different resolution creates a separate local version. Each version has its own controls for opening, resizing and deletion. Resolution policies are described under Storage and jobs.

The PDF of the work ​

Some libraries, next to the images, offer the same work as a PDF. They declare it in their manifest — on the root or on the sequence, depending on the version of the standard — and that is where Glossa finds it: no addresses built by analogy.

In search results the closed row says “PDF available” only when there is one. The expanded row always states the full status — available, not available, or not verified when the manifest could not be read — and adds the declared pages and the pixel size of the first page, the only hint about scan quality available before downloading. The manifest is read once per work, only for rows on screen, at most two reads at a time, and never for a result the catalogue already declares without a reproduction.

In the work record the PDF is a row of the book section, below the image copies. It states availability and local status; a command checks with the library again — useful when the PDF was published later — and the same check runs on resynchronisation. When the PDF is available it is downloaded from the same row; when it is on disk, the row states its pages and size and offers the commands to display it, open it with the system application or delete it. Deleting it does not touch the images of the same work.

From the same row you choose what to display: the PDF or the images. The viewer always states which of the two copies is on screen, because the pages of the PDF and those of the image sequence do not correspond and are not merged into a single browsing sequence.

Stated limits: a PDF over 256 MB does not open in the built-in viewer and must be read with the system application; for a password-protected or malformed PDF the pages are not counted, and the record says so rather than inventing a number. The page is drawn at a fixed resolution: magnified to the maximum it looks less sharp than IIIF tiles.

Every check has a deadline: if the library is busy or does not answer, the status stays “not verified” and you can try again. No wait is left hanging, and a slow request does not block the others.

Archiving and deletion ​

ActionEffect
ArchiveHides the work from the active catalogue and retains files; freeing space is a separate choice
Free spaceDeletes downloaded images while retaining the record, manifest and thumbnails
Delete a local versionRemoves only the selected version’s files
Delete the workRemoves the record, links, the files of all copies — images and document — and the associated cache

Deletion does not use a recycle bin. Destructive file operations require jobs that could modify those files to finish or be cancelled; pausing them is not sufficient.

Limitations ​

Importing text from a PDF into a translation project remains a separate feature: a transcription cannot yet be started from a document kept in the Library. Advanced page management and multiple-page selection are incomplete. Institutional download restrictions are not enforced automatically; consult the source’s conditions of use.

On the item page, under Digital copies, the collapsible «Technical data» section gathers every address of that copy — IIIF manifest, work page, catalogue record, the open page in the library's viewer, the image of that page, the library site — each marked by its own symbol, copyable and openable in the browser. Long addresses read in full on hover.

The library's manifest can be requested there too: it is not dumped as it comes — its address is right above for that — but read and shown as the library's statement about the work, with page count, description, catalogue entries and rights.

Work data ​

The item page shows every field it knows about, including those the library did not fill in: an empty field states that the information never arrived, and from there you can write it yourself. Each row is corrected with the pencil and keeps the library's original value, which stays readable and restorable.

The essential fields — title, kind of work, author, date, publisher, language — are always in view. The rest is gathered into collapsible groups: content, copy, provenance, rights and notes. The groups remember whether they are open, and that choice applies across the whole Library.

The kind of work is chosen among the expected values, because the catalogue filters rely on them. Fields holding several values — subjects, other contributors, rights, provenance — are written on a single line separated by «·», exactly as they are shown.

Public documentation for the Glossa desktop app