Google Fonts

Lang Metadata System

🦜 googlefonts/lang is the GitHub repository that defines the foundation for the *lang metadata system*

This section specifies the main concepts of the system and provides information on how if influences the displaying of the font in the Catalog.

Background reading:
team  METADATA file

Table of contents

• Constructs
• Other definitions
• METADATA.pb
• Catalog (fonts.google.com)

Constructs

• Language
  • Application of some writing system (e.g. English)
• Region
  • Geo area (e.g. United States)
  • Grouped further into region groups (e.g. Asia)
• Script
  • Writing system (e.g. Latin)

Other definitions

Coverage – the set of codepoints covered by a given family; in this context, that set is further reduced to the codepoints that would be available at serving time via the Google Fonts API.

METADATA.pb

In font metadata files, we have several controls that influence how lang metadata manifests further downstream.

• Primary script (primary_script)
  • Sets the script of the default sample text for when no other lang filters are specified
• Primary language (primary_language)
  • Sets the language of the default sample text for when no other lang filters are specified
• Supported languages (languages)
  • Autogenerated; can manually override by specifying in METADATA.pb
  • Language support is determined by whether coverage includes the entire base character set (i.e. exemplar_chars.base) of a language
  • Has the side effect that languages without exemplar_chars in their metadata will never be included in an autogenerated language set
  • Do NOT manually specify except on Noto families for the purposes of including an exhaustive language set
• Sample glyphs (sample_glyphs)
  • Autogenerated; can manually override by specifying in METADATA.pb
  • Filtered from unicode_sections using intersection with coverage
  • Avoid manually specifying except for families without meaningful codepoint support (e.g. Noto Emoji)

Catalog (fonts.google.com)

The primary consumer of lang metadata is Catalog. There are a couple areas where lang metadata influences how fonts are presented.

• Sample text
  • Used in both grid tiles and specimens
  • Possible sample texts are those of any support language of the font family
  • Assuming autogenerated language support metadata, may still display tofu if:
    • Coverage for font X includes base character set for language Y
    • Coverage for font X does not include all of the characters required to render the sample text of language Y
• Specimen – Glyphs
  • Shows the sample glyphs
  • We believe we have systematically eliminated the possibility of displaying tofu here

For information about how to update the lang metadata repository, see google/fonts repository explained.

This site is open source. Improve this page.