asm_compiler.py 8.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271
  1. #!/usr/bin/python3
  2. import sys
  3. import argparse
  4. from os import path
  5. def decode_byte(val: str):
  6. try:
  7. if val.endswith('h'):
  8. return int(val[:-1], 16)
  9. if val.startswith('0x'):
  10. return int(val[2:], 16)
  11. if val.startswith('b'):
  12. return int(val.replace('_', '')[1:], 2)
  13. except ValueError:
  14. raise ValueError(f"Invalid binary '{val}'")
  15. if val.isdigit():
  16. i = int(val)
  17. if i > 255 or i < 0:
  18. raise ValueError(f"Invalid binary '{val}', unsigned int out of bounds")
  19. return i
  20. if (val.startswith('+') or val.startswith('-')) and val[1:].isdigit():
  21. i = int(val)
  22. if i > 127 or i < -128:
  23. raise ValueError(f"Invalid binary '{val}', signed int out of bounds")
  24. if i < 0: # convert to unsigned
  25. i += 2**8
  26. return i
  27. if len(val) == 3 and ((val[0] == "'" and val[2] == "'") or (val[0] == '"' and val[2] == '"')):
  28. return ord(val[1])
  29. raise ValueError(f"Invalid binary '{val}'")
  30. def is_reg(r):
  31. if r.startswith('$'):
  32. r = r[1:]
  33. return len(r) == 2 and (r == 'ra' or r == 'rb' or r == 'rc' or r == 're')
  34. def decode_reg(r):
  35. if isinstance(r, int):
  36. if 0 <= r <= 3:
  37. return r
  38. raise ValueError(f"Invalid register value {r}")
  39. rl = r.lower()
  40. if rl.startswith('$'):
  41. rl = rl[1:]
  42. if rl == 'ra':
  43. return 0
  44. if rl == 'rb':
  45. return 1
  46. if rl == 'rc':
  47. return 2
  48. if rl == 're':
  49. return 3
  50. raise ValueError(f"Invalid register name '{r}'")
  51. def assemble(file):
  52. odata = []
  53. afile = open(file, 'r')
  54. failed = False
  55. refs = dict()
  56. for lnum, line in enumerate(afile.readlines()):
  57. lnum += 1 # Line numbers start from 1, not 0
  58. if '//' in line:
  59. line = line[:line.index('//')]
  60. if ':' in line:
  61. rsplit = line.split(':', 1)
  62. ref = rsplit[0]
  63. if not ref.isalnum():
  64. print(f"{file}:{lnum}: Invalid pointer reference '{ref}'")
  65. failed = True
  66. continue
  67. if ref in refs:
  68. print(f"{file}:{lnum}: Pointer reference '{ref}' is duplicated with {file}:{refs[ref][0]}")
  69. failed = True
  70. continue
  71. refs[ref] = [lnum, len(odata)]
  72. line = rsplit[1]
  73. line = line.replace('\n', '').replace('\r', '').replace('\t', '')
  74. line = line.strip(' ')
  75. if line == '':
  76. continue
  77. ops = line.split()
  78. instr = ops[0].upper()
  79. rops = 3
  80. if instr == 'CPY' or instr == 'COPY':
  81. iname = 'COPY'
  82. inibb = 0
  83. elif instr == 'ADD':
  84. iname = 'ADD'
  85. inibb = 1
  86. elif instr == 'SUB':
  87. iname = 'SUB'
  88. inibb = 2
  89. elif instr == 'AND':
  90. iname = 'AND'
  91. inibb = 3
  92. elif instr == 'OR':
  93. iname = 'OR'
  94. inibb = 4
  95. elif instr == 'XOR':
  96. iname = 'XOR'
  97. inibb = 5
  98. elif instr == 'GT' or instr == 'GRT':
  99. iname = 'GT'
  100. inibb = 6
  101. elif instr == 'EX' or instr == 'EXT':
  102. iname = 'EXT'
  103. inibb = 7
  104. elif instr == 'SHFL':
  105. iname = 'SHTL'
  106. inibb = 7
  107. ops.append(0)
  108. elif instr == 'SHFR':
  109. iname = 'SHTR'
  110. inibb = 7
  111. ops.append(1)
  112. elif instr == 'ROTR':
  113. iname = 'ROTR'
  114. inibb = 7
  115. ops.append(2)
  116. elif instr == 'LW':
  117. iname = 'LW'
  118. inibb = 8
  119. elif instr == 'SW':
  120. iname = 'SW'
  121. inibb = 9
  122. elif instr == 'JEQ':
  123. iname = 'JEQ'
  124. rops = 4
  125. inibb = 10
  126. elif instr == 'JMP' or instr == 'JUMP':
  127. iname = 'JUMP'
  128. rops = 2
  129. inibb = 11
  130. elif instr == 'PUSH':
  131. iname = 'PUSH'
  132. rops = 2
  133. inibb = 14
  134. elif instr == 'POP':
  135. iname = 'POP'
  136. rops = 2
  137. inibb = 15
  138. else:
  139. if len(ops) == 1:
  140. try:
  141. odata.append(decode_byte(ops[0]))
  142. continue
  143. except ValueError:
  144. pass
  145. print(f"{file}:{lnum}: Instruction '{ops[0]}' not recognised")
  146. failed = True
  147. continue
  148. if len(ops) != rops:
  149. print(f"{file}:{lnum}: {iname} instruction requires {rops - 1} arguments")
  150. failed = True
  151. continue
  152. try:
  153. if iname == 'JUMP':
  154. odata.append(inibb << 4)
  155. try:
  156. odata.append(decode_byte(ops[1]))
  157. except ValueError:
  158. if not ops[1].isalnum():
  159. print(f"{file}:{lnum}: Invalid pointer reference '{ops[1]}'")
  160. failed = True
  161. continue
  162. if ops[1] in refs:
  163. odata.append(refs[ops[1]][1])
  164. else:
  165. refs[ops[1]] = [lnum, None]
  166. odata.append(ops[1])
  167. continue
  168. rd = decode_reg(ops[1])
  169. if iname == 'COPY' and not is_reg(ops[2]):
  170. imm = decode_byte(ops[2])
  171. odata.append((inibb << 4) | (rd << 2) | rd)
  172. odata.append(int(imm))
  173. continue
  174. if iname == 'PUSH' or iname == 'POP':
  175. odata.append((inibb << 4) | (rd << 2) | rd)
  176. continue
  177. rs = decode_reg(ops[2])
  178. if iname == 'COPY' and rd == rs:
  179. print(f"{file}:{lnum}: {iname} cannot copy register to itself")
  180. failed = True
  181. continue
  182. odata.append((inibb << 4) | (rd << 2) | rs)
  183. if iname == 'JEQ':
  184. try:
  185. odata.append(decode_byte(ops[3]))
  186. except ValueError:
  187. if not ops[3].isalnum():
  188. print(f"{file}:{lnum}: Invalid pointer reference '{ops[3]}'")
  189. failed = True
  190. continue
  191. if ops[3] in refs:
  192. odata.append(refs[ops[3]][1])
  193. else:
  194. refs[ops[3]] = [lnum, None]
  195. odata.append(ops[3])
  196. continue
  197. except ValueError as e:
  198. print(f"{file}:{lnum}: {e}")
  199. failed = True
  200. continue
  201. afile.close()
  202. # Convert jumps
  203. for i, l in enumerate(odata):
  204. if isinstance(l, str):
  205. if refs[l][1] is None:
  206. print(f"{file}:{refs[l][0]}: Pointer reference '{l}' does not exist!")
  207. failed = True
  208. continue
  209. odata[i] = refs[l][1]
  210. return not failed, odata
  211. if __name__ == '__main__':
  212. parser = argparse.ArgumentParser(description='Assembly compiler', add_help=True)
  213. parser.add_argument('file', help='Files to compile')
  214. parser.add_argument('-t', '--output_type', choices=['bin', 'mem', 'binary'], default='mem', help='Output type')
  215. parser.add_argument('-o', '--output', help='Output file')
  216. parser.add_argument('-f', '--force', action='store_true', help='Force override output file')
  217. args = parser.parse_args(sys.argv[1:])
  218. if not path.isfile(args.file):
  219. print(f'No file {args.file}!')
  220. sys.exit(1)
  221. output = args.output
  222. if not output:
  223. opath = path.dirname(args.file)
  224. bname = path.basename(args.file).rsplit('.', 1)[0]
  225. ext = '.out'
  226. if args.output_type == 'mem':
  227. ext = '.mem'
  228. elif args.output_type == 'bin':
  229. ext = '.bin'
  230. output = path.join(opath, bname + ext)
  231. if not args.force and path.isfile(output):
  232. print(f'Output file already exists {output}!')
  233. sys.exit(1)
  234. success, data = assemble(args.file)
  235. if success:
  236. print(f"Saving {args.output_type} data to {output}")
  237. with open(output, 'wb') as of:
  238. if args.output_type == 'binary':
  239. a = '\n'.join([format(i, '08b') for i in data])
  240. of.write(a.encode())
  241. elif args.output_type == 'mem':
  242. a = [format(i, '02x') for i in data]
  243. for i in range(int(len(a)/8)+1):
  244. of.write((' '.join(a[i*8:(i+1)*8]) + '\n').encode())
  245. elif args.output_type == 'bin':
  246. of.write(bytes(data))
  247. else:
  248. print(f'Failed to compile {args.file}!')
  249. sys.exit(1)
  250. sys.exit(0)