Okab

Generate searchable PDFs with embedded TrueType and CFF1 fonts in pure Ruby

Features · Installation · Quick start · Limits · Development

Okab is a small Ruby library for generating PDF 1.7 documents. It embeds font subsets through Alhena and writes ToUnicode mappings so text remains searchable and copyable. The same document inputs produce stable output without timestamps. Its name comes from Arabic ʿuqāb, “eagle” (ζ Aquilae).

• TrueType and supported static CFF1 font embedding, including Japanese text when the font contains the glyphs
• Text and wrapped text, vector paths, clipping, transforms, and opacity
• PNG images with alpha and JPEG images embedded without re-encoding
• Optional Zaniah vector-document bridge with searchable glyph runs
• Page links, document outlines, and deterministic PDF output

gem install okab

Okab requires Ruby 3.2 or newer and Alhena 0.3.x. With Bundler, add gem "okab" to your Gemfile.

Supply a TrueType font containing the text you want to render:

require "okab"

font = Okab::Font.load("/path/to/font.ttf")
document = Okab::Document.new(title: "Quarterly report")
page = document.page(width: 595, height: 842) # PDF points
page.text("Quarterly report", x: 48, y: 790, font: font, size: 24)
page.text_block("A searchable report with embedded fonts.",
  x: 48, y: 750, width: 360, font: font, size: 12, line_height: 18)
document.outline("Report", page: page)
document.write("report.pdf")

Coordinates are in PDF points, with the origin at the lower-left corner. Text must be valid UTF-8. Use a font with Japanese glyphs to render Japanese text.

Install Zaniah separately, then load the optional bridge explicitly:

require "okab/zaniah_vector"

vector = Zaniah::Vector.record(width: 595, height: 842) { slide_element }
pdf = Okab::Document.new
page = pdf.page(width: vector.width, height: vector.height)
Okab::ZaniahVector.draw(page, vector)
pdf.write("slide.pdf")

The bridge converts Zaniah's top-left coordinates to PDF's lower-left coordinates. Solid quads and paths remain PDF vector operations, shaped glyph IDs are embedded with ToUnicode mappings, and images remain PDF images. Unsupported paint effects such as multi-stop gradients and shadows are rasterized individually, never as a full-page screenshot. See the adapter notes for the mapping and limits.

Okab generates PDFs; it does not read or edit them. Encryption, signatures, forms, and PDF/A are out of scope.

• CFF2, variable CFF1, and already CID-keyed CFF fonts are unsupported. CFF embedding requires subset: true.
• PNG input must be non-interlaced and 8-bit.
• JPEG input must be 8-bit grayscale or RGB, using baseline, extended-sequential, or progressive encoding. Exif orientation is not applied.

bundle install
bundle exec rake
bundle exec rbs -I sig -r alhena validate

To test real Japanese text extraction, set OKAB_TEST_FONT to a TrueType font with Japanese glyphs and install Poppler's pdftotext utility:

OKAB_TEST_FONT=/path/to/japanese-font.ttf bundle exec rake

Okab is released under the MIT License.