Catalogs¶

A catalog is a set of languages, families and scripts that know one another: a language's family, ancestors and scripts are looked up in the catalog it came from. The bundled catalog, English Wiktionary's own as it was when the version was made, is the one Language, LanguageFamily and Script look up.

import com.fenakhay.polyglossa.Catalog

val catalog = Catalog.BUNDLED
catalog.date                        // "2026-11-06"
catalog["en"]                       // English (en)
catalog.family("ine")               // Indo-European (ine)
catalog.script("Latn")              // Latin (Latn)
catalog.named("Old English")        // Old English (ang)
catalog.scriptNamed("Arabic")       // Arabic (Arab)
catalog.languages.size              // 8970

import Polyglossa

let catalog = Catalog.bundled
catalog.date                        // "2026-11-06"
catalog["en"]                       // English (en)
catalog.family("ine")               // Indo-European (ine)
catalog.script("Latn")              // Latin (Latn)
catalog.named("Old English")        // Old English (ang)
catalog.scriptNamed("Arabic")       // Arabic (Arab)
catalog.languages.count             // 8970

from polyglossa import Catalog

catalog = Catalog.bundled()
catalog.date                        # '2026-11-06'
catalog["en"].code                  # 'en'; catalog.get("xx") is None
catalog.family("ine").code          # 'ine'
catalog.script("Latn").code         # 'Latn'
catalog.named("Old English").code   # 'ang'
catalog.script_named("Arabic").code # 'Arab'
len(catalog.languages)              # 8970

In Kotlin the collections are in no particular order; in Swift and Python they are sorted by code.

The catalog as text¶

A catalog is written as catalog.tsv, tab-separated lines with a language, a family or a script to each, sorted by code, so two copies differ line by line where the data does. Every implementation reads it:

val text = Catalog.BUNDLED.toTsv()
val copy = Catalog.fromTsv(text)
copy["en"]?.family                 // Anglic (gmw-ang), from copy, not from BUNDLED

let text = try String(contentsOfFile: "catalog.tsv", encoding: .utf8)
let copy = try Catalog(tsv: text)
copy["en"]?.family                 // Anglic (gmw-ang), from copy

text = open("catalog.tsv", encoding="utf-8").read()
copy = Catalog(text)
copy["en"].family.code             # 'gmw-ang', from copy

A catalog read this way is a world of its own: its languages' links lead into it, not into the bundled one. That's how a program can use a newer catalog.tsv, from a later release's assets, without waiting for a new version of the library.

From Wiktionary's exports¶

The Kotlin library also reads what Wiktionary's Module:JSON data exports, which is how the bundled catalog is made. Catalog.EXPORTS lists the six invocations, and fromExports takes what the wiki expanded each to, in that order:

val texts = Catalog.EXPORTS.map { invocation -> expandOnTheWiki(invocation) }
val fresh = Catalog.fromExports(texts, date = "2026-11-12")

expandOnTheWiki stands for a call to the wiki's action=expandtemplates. The marks that tell Traditional from Simplified Han are not in Wiktionary's data: withScriptMarks adds them, by script code, as polyglossa's own tool does from Unihan:

val marked = fresh.withScriptMarks(mapOf("Hant" to traditional, "Hans" to simplified))

Each call makes a new catalog and leaves the one it came from as it was.