Language ROS for Encodings for APL, BASIC, English

APL-BASIC-ENGLISH.ps1

<#
.SYNOPSIS
  Build an unabridged UTF-8 symbol index for APL and BASIC.
  Writes /mnt/data/symbol_index.json
#>

$ErrorActionPreference = 'Stop'

$dataset = @()

# ---------- BASIC ----------
$basic = @(
  'AND','OR','NOT','IF','THEN','ELSE','FOR','TO','STEP','NEXT',
  'GOTO','GOSUB','RETURN','PRINT','INPUT','DIM','DEF FN','END','REM',
  'DATA','READ','RESTORE','CHR$','ASC','LEN','MID$','LEFT$','RIGHT$',
  'RND','SIN','COS','TAN','LOG','EXP','PEEK','POKE','OPEN','CLOSE',
  'INPUT#','PRINT#','ON','ERROR','RESUME','+','-','*','/','^','=','<',
  '>','<=','>=','<>'
)
foreach ($sym in $basic) {
  $dataset += [ordered]@{
    lang='BASIC'
    symbol=$sym
    name="BASIC keyword or operator"
    ascii_equiv=$sym
    unicode=[string]::Join(' ',($sym.ToCharArray() | ForEach-Object {"U+{0:X4}" -f [int][char]$_}))
    invariant_id=":BASIC_$($sym.Replace(' ','_').Replace('$','S'))"
    variant_byte="0x{0:X2}" -f ([int]([byte][char]$sym[0]))
    state=":1"
  }
}

# ---------- APL ----------
# Use the Unicode block 2336–237A for primary glyphs.
$aplSymbols = @(
  '+','-','×','÷','⋆','⍟','⌹','○','<','≤','=','≥','>','≠','∨','∧',
  '⍲','⍱','∊','⊂','⊃','↑','↓','⍴','⍳','⍸','⌷','⍋','⍒','⍉','⍪','⍎',
  '⍕','⎕','⍞','⍠','⍣','⍤','⍥','⍨','⍫','⍬','⍺','⍵','∇','¯','⍹'
)

foreach ($sym in $aplSymbols) {
  $dataset += [ordered]@{
    lang='APL'
    symbol=$sym
    name='APL glyph'
    ascii_equiv='N/A'
    unicode="U+{0:X4}" -f [int][char]$sym
    invariant_id=":APL_{0:X4}" -f [int][char]$sym
    variant_byte="0x{0:X2}" -f ([int][byte][char]$sym[0])
    state=":1"
  }
}

# ---------- Write file ----------
$path = "/mnt/data/symbol_index.json"
$json = $dataset | ConvertTo-Json -Depth 4 -Compress
[IO.File]::WriteAllText($path,$json,[Text.Encoding]::UTF8)
Write-Host "Dataset written to $path (`$($dataset.Count) entries)."