(defpackage #:nekod.llm2asm
  (:use #:cl)
  (:export
   #:tokenize-intent
   #:intent-token
   #:token-id
   #:token-class
   #:intent-token-text
   #:intent-token-class
   #:intent-token-bytes
   #:lookup-kernel
   #:compile-intent
   #:safe-compile-intent
   #:infra-ast
   #:infra-ast-type
   #:infra-ast-target
   #:infra-ast-constraint
   #:infra-ast-priority))

(in-package #:nekod.llm2asm)

;; Byte-pair encoding for x86-64 / ARM64 opcodes
;; The kernel map from sovereign-shadow-compiler feeds this

;; Known x86-64 primitive opcodes (from sovereign-shadow-compiler KERNEL_MAP)
(defvar *x86-kernel-map*
  '(("ADD"     . #(#x48 #x89 #xf8 #x48 #x01 #xf0 #xc3))
    ("MUL"     . #(#x48 #x89 #xf8 #x48 #x0f #xaf #xc6 #xc3))
    ("XOR"     . #(#x48 #x89 #xf8 #x48 #x31 #xf0 #xc3))
    ("LOOP"    . #(#x48 #x31 #xc0 #x48 #xff #xc0 #x48 #x39 #xf8 #x7c #xf8 #xc3))
    ("MEMCPY"  . #(#x48 #x89 #xd1 #xf3 #xa4 #xc3))
    ("MEMSET"  . #(#x48 #x89 #xd1 #x40 #x88 #xf0 #xf3 #xaa #xc3))
    ("STRCMP"  . #(#x48 #x31 #xc0 #x8a #x0f #x8a #x16 #x38 #xd1 #x75 #x0c
                   #x84 #xc9 #x74 #x0a #x48 #xff #xc7 #x48 #xff #xc6 #xeb
                   #xee #x48 #x19 #xc0 #x48 #x83 #xc8 #x01 #xc3))))

(defstruct intent-token
  id
  text
  class    ; :op :operand :constraint :intent
  bytes    ; raw bytes if this is a known kernel op
  )

(defun tokenize-intent (intent-string)
  "Tokenize an intent string into intent-token list."
  (let* ((words (uiop:split-string (string-upcase intent-string)))
         (tokens (loop for word in words
                       for id from 0
                       collect (let ((kernel-bytes (cdr (assoc word *x86-kernel-map* :test #'string=))))
                                 (make-intent-token
                                  :id id
                                  :text word
                                  :class (if kernel-bytes :op :intent)
                                  :bytes kernel-bytes)))))
    tokens))

(defun lookup-kernel (op-name)
  "Look up raw bytes for a known kernel operation."
  (cdr (assoc (string-upcase op-name) *x86-kernel-map* :test #'string=)))