PDF export

WeKan exports a board, list, swimlane or single card as an A4 PDF. The related Excel export offers the same card sections. Both formats build their content from the layout in One Card Layout.

Implementation

• models/exportPDF.js authenticates and scopes the HTTP routes.
• models/server/ExporterCardPDF.js loads board/card data, resolves users and custom fields and reads attachment images.
• models/lib/cardExportDocument.js maps those records into the shared card document. Excel calls exactly the same adapter, so people, dates, checklist items, comments, attachment details, sizes, voting and poker cannot be mapped differently by the two formats.
• models/lib/cardDocument.js decides which blocks exist and omits empty or unselected sections.
• models/lib/pdfDocument.js renders those blocks, paginates them, writes the fallback PDF objects and embeds images.
• models/server/buildUnicodePdf.js uses PDFKit to subset and embed the bundled GNU Unifont fonts for the normal export path.

The board, swimlane, list and card hamburger menus do not have four PDF templates or a second set for Excel. They all include exportScopeBody from client/components/boards/exportScope.jade; one table in exportScope.js defines both download formats, one URL builder adds the selected scope, and one checkbox selection is sent to both routes.

On the server, scope changes only the hierarchy surrounding the cards. Every card at every scope is converted by buildExportCardDocument, then PDF renders the returned medium-independent blocks as pages. PDF-specific code is limited to page geometry, wrapping, pagination, embedded fonts and image encoding; the meaning and ordering of card fields are shared with Excel.

The distributable includes GNU Unifont 17.0.05 and Unifont Upper under the SIL Open Font License 1.1 in private/fonts/unifont. PDFKit subsets and embeds both fonts, so readers do not have to install them. The main font covers the Basic Multilingual Plane and the upper font supplies glyphs from supplementary Unicode planes. This gives every WeKan language a visible glyph and also covers supplementary characters such as emoji. GNU Unifont is deliberately a last-resort coverage font: complex-script shaping and color emoji can be less polished than a platform's script-specific fonts, but text remains present, searchable and portable.

If loading or rendering the embedded fonts fails, the dependency-free writer in models/lib/pdfDocument.js is retained as a safe fallback. It uses base-14 Courier with WinAnsi encoding, preserves Western European text, transliterates some other Latin characters, and replaces unsupported scripts with ?. Markdown headings, lists, emphasis, quotes and code are rendered as document structure instead of printing their Markdown punctuation.

JPEG attachments are embedded using their original /DCTDecode stream. PNG scanlines are decoded, PNG filters are removed, transparency is composited onto white, and the RGB pixels are embedded with /FlateDecode. Images are scaled down without being enlarged. Up to three previews share an A4 PDF row, and only the filename appears below each image. Excel can fit six previews across its six worksheet columns. The whole PDF preview row moves to the next page when it does not fit. A corrupt, unavailable or unsupported image cannot make the PDF export fail.

Before the previews, the attachment detail table includes every attachment and the same six fields as Excel: row number, filename, human-readable size, media type, upload date/time and uploader. Image details therefore remain complete without repeating size or other metadata in the preview caption.

Metadata uses three columns like the printable Excel card. A translated label and its value wrap onto additional lines inside that column instead of being shortened with an ellipsis, so complete date and time values remain visible.

Board PDFs are ordered as board name, members, created and modified metadata, then each swimlane, each list within that swimlane, and each card within that list. Smaller exports begin at their selected level: swimlane then lists and cards, list then cards, or the single card. The only visible swimlane is still named, so the hierarchy does not change merely because a board currently has one. Detailed Excel exports use the same order.

Because every scope uses the shared card document, PDF carries the same complete opened-card fields as detailed Excel, including stickers, dependencies, numeric sort position, and every location's place name, address, latitude and longitude. Legacy single-location fields remain visible too.

Current progress

Completed:

• board, list, swimlane and single-card PDF routes;
• the same field selection and card-document layout as Excel;
• one shared hamburger-menu template and one raw-record adapter across board, swimlane, list and card PDF/Excel exports;
• localized labels, user-timezone dates and rendered Markdown;
• the logged-in user's saved language, falling back to the current browser language when no language is saved;
• the date format displayed by the opened card;
• card metadata, custom fields, checklists, subtasks, comments, attachments, voting and planning poker;
• JPEG and PNG attachment previews in card and detailed board PDFs;
• three previews per row with filename-only captions below and atomic pagination;
• the same six-column attachment details table, colored labels and segmented checklist progress as the printable Excel card;
• wrapped metadata columns that preserve complete date/time values;
• binary-safe object offsets and regression tests that inspect the resulting image XObjects and PDF cross-reference table;
• embedded Unicode-plane fonts, their OFL license, font subsetting and tests that verify PDFKit can parse and embed both shipped fonts.

Remaining: add GIF and BMP decoding if parity with Excel's preview formats is needed.

The former TODO item said PDF still listed images only by name. That step is now implemented and guarded by positive and negative tests.