fix: accept 4-hex-digit \U escapes

The corpus converter emits \U040E-style escapes (gki-bel, mvd-bel,
odni-bel); the Ruby grammar demanded 8 hex digits for \U and rejected
the files. Accept 4..8 in both the DSL grammar and the .parg.
Gate: 283/285 (remaining: un-mar has converter-corrupted escape output
needing a maps-repo fix; odni-kor has a 2-element comment-parallel
capture delta).

Author: ronaldtse

isc.artifact.json

@@ -62,7 +62,10 @@ module Primitives
               str('"').as(:dquote) |
               str("\\").as(:backslash) |
               (str("u") >> match(/[0-9a-fA-F]/).repeat(4, 4).as(:unicode)) |
-              (str("U") >> match(/[0-9a-fA-F]/).repeat(8, 8).as(:unicode))
+              # The corpus converter emits \U with 4 hex digits too
+              # (gki-bel "\U040E") — accept 4..8, don't be stricter than
+              # the corpus.
+              (str("U") >> match(/[0-9a-fA-F]/).repeat(4, 8).as(:unicode))
             )
           end
 

lib/interscript/isc/grammar/concerns/primitives.rb

@@ -38,7 +38,7 @@ grammar InterscriptIsc version "0.1.0" {
   esc_dq = "\"" as dquote
   esc_bs = "\\" as backslash
   esc_uni4 = "u" (4hexd) as unicode
-  esc_uni8 = "U" (8hexd) as unicode
+  esc_uni8 = "U" (4*8hexd) as unicode
   escape_sequence = "\\" (esc_nl / esc_cr / esc_tab / esc_dq / esc_bs / esc_uni4 / esc_uni8)
 
   raw_run = 1*(!("\\" / "\"") chr)