Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 26 additions & 0 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,32 @@ Interscript.transliterate("odni-rus-Cyrl-Latn-2015", "привет")
# => "privet"
----

== Workflow extension (ISC-WF-1)

Maps can reuse rule blocks and condition rows on a language. A
`ruleset` is spliced at a `use` position (recursively, across
declared dependencies); `use ... override { ... }` replaces rows on
byte-exact from-equality; a `languages` declaration plus `when`
guards let the caller select a language per call — guards are inert
without a selection:

[source,ruby]
----
# alalc-tha and bgnpcgn-tha share their leading-vowel machinery
# through libs/thai-latn.isc — one ruleset, per-map overrides
Interscript.transliterate("bgnpcgn-tha-Thai-Latn-2002", "กรุงเทพ")

# language conditioning: select one of the map's declared languages;
# guards are inert when no (or an undeclared) language is passed
Interscript.transliterate("some-map-id", "...", {}, language: "kur")
----

The constructs are additive: maps that don't use them transliterate
identically. The corpus adoption so far is the Thai leading-vowel
shared ruleset (W1/W2); W3 conditioning is exercised by the
conformance fixtures. Normative text: interscript/maps
`docs/isc-wf-1.md`.

== Neural models (Interscript::ML)

The gem embeds the IMF v1 model runtime — resolve a model id through
Expand Down
Loading