Module speakers 

Speaker recognition: the CAST block, speakers.json, and the built-in fallback list. Port of the speaker half of XILP001_script_parser.py.

Structs

CastEntry

One CAST-block entry.

SpeakerMatch

A speaker matched at the start of a line.

Speakers

Display names (match order) plus the display → key map.

Constants

BUILTIN 🔒

Built-in display names, already longest-first as the Python literal is.

Statics

NON_WORD 🔒

PAREN 🔒

ROLE_DASH 🔒

SPACES 🔒

Functions

builtin_key_order 🔒

builtin_speakers

The built-in table, for callers that need it without any script context.

display_to_key

Display name → normalized key: drop parentheticals and punctuation, lowercase, collapse whitespace to underscores.

extract_cast_from_script

Parse the CAST: bullet block from the script header.

key_to_display

Reverse map key → first display name, as generate_cast_config builds it from the module-level SPEAKER_KEYS.

load_speakers

Merge CAST-block entries with speakers.json, falling back to the built-ins only when neither exists.

load_speakers_registry

Full speakers.json entries keyed by speaker key, for cast skeletons.

read_speakers_json 🔒

resolve_speakers_file

Which speakers.json to load, without reading it.

try_match_speaker

Match a known speaker prefix, splitting off any (direction).