From 40bd3c1d5bc97675d32db7872e6f346aec288b82 Mon Sep 17 00:00:00 2001
From: Matthew Sotoudeh <matthew@masot.net>
Date: Sun, 1 Sep 2024 12:47:04 -0700
Subject: version of dietcc that compiles to a folder for global analyses

---
 scripts/foldercc | 129 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 129 insertions(+)
 create mode 100755 scripts/foldercc

(limited to 'scripts/foldercc')

diff --git a/scripts/foldercc b/scripts/foldercc
new file mode 100755
index 0000000..38b84d2
--- /dev/null
+++ b/scripts/foldercc
@@ -0,0 +1,129 @@
+#!/bin/python3
+import subprocess
+import re
+import tempfile
+import shlex
+import sys
+import os
+
+dietc_dir = os.path.dirname(os.path.dirname(os.path.realpath(__file__)))
+
+def check_result(spr, debug):
+  if not spr.returncode: return spr.stdout
+  print("Failed to run gcc wrapper:", ' '.join(sys.argv), file=sys.stderr)
+  if spr.stderr:
+    print(spr.stderr.decode("utf-8"), file=sys.stderr)
+  open("/tmp/dietc.error.c", "wb").write(debug)
+  assert False
+
+def main(argv):
+  argv = argv[1:]
+  def argument(flag): return argv[argv.index(flag) + 1]
+  CC = "gcc"
+  tmpdir = tempfile.TemporaryDirectory()
+
+  # get the arguments WITHOUT -c, -o, source files, etc.
+  stripped_args = []
+  i = 0
+  while i < len(argv):
+    if (argv[i] == "-c" or argv[i].endswith(".c") or argv[i].endswith(".a") or argv[i].endswith(".o")):
+      i += 1
+    elif argv[i] == "-o":
+      i += 2
+    else:
+      stripped_args.append(argv[i])
+      i += 1
+
+  # ====== HANDLE -c ======
+  if "-c" in argv:
+    c_file = next(a for a in argv if a.endswith(".c"))
+    o_file = c_file[:-2] + ".o"
+    if "-o" in argv: o_file = argument("-o")
+
+    # (1) RUN PREPROCESSOR
+    c = open(c_file, "r").read()
+    t_file = open(f"{tmpdir.name}/file.c", "wb")
+    t_file.write(f"#line 1 \"{c_file}\"\n".encode("utf-8"))
+    t_file.write(f"#include \"{dietc_dir}/scripts/stdincludes/dietc_defines.h\"\n".encode("utf-8"))
+    t_file.write(c.encode("utf-8"))
+    t_file.flush()
+    dirname = os.path.dirname(os.path.realpath(c_file))
+    argstr = list(map(shlex.quote, stripped_args))
+    argstr.insert(0, f"-I{dietc_dir}/scripts/stdincludes")
+    argstr = ' '.join(argstr)
+    preproc_c = check_result(
+      subprocess.run(f"{CC} -I{dirname} -E {argstr} {t_file.name}",
+                     shell=True, capture_output=True), c)
+
+    # (2) STRIP GCC-SPECIFIC STUFF AFTER PREPROCESSING
+    preproc_c = re.sub(rb"\[\[[^\]]*\]\]", b"", preproc_c)
+    preproc_c = re.sub(rb"__asm__", b"", preproc_c)
+
+    # (3) RUN DIETC
+    preproc_c = preproc_c.replace(b"# ", b"#line ")
+    t_file = open(f"{tmpdir.name}/file.c", "wb")
+    t_file.write(preproc_c)
+    t_file.flush()
+    dirname = os.path.dirname(os.path.realpath(c_file))
+    # TODO: args = ' '.join(meta["dietc_args"])
+    args = ""
+    diet_c = check_result(
+      subprocess.run(f"timeout 5s {dietc_dir}/dietc {t_file.name} {args}",
+                     shell=True, stdout=subprocess.PIPE, stderr=sys.stderr), preproc_c)
+
+    # (4) SAVE TO A .o
+    with open(o_file, "wb") as out:
+      argstr = list(map(shlex.quote, stripped_args))
+      argstr = ' '.join(argstr)
+      out.write(b"DIETC")
+      out.write(len(argstr).to_bytes(4, "little"))
+      out.write(argstr.encode("utf-8"))
+      out.write(diet_c)
+
+    return
+
+  # ======= If any .cs, first convert them to .os =======
+  for i in range(len(argv)):
+    if not argv[i].endswith(".c"): continue
+    o_file = f"{tmpdir.name}/arg_{i}.o"
+    main(["dietcc", *stripped_args, "-c", argv[i], "-o", o_file])
+    argv[i] = o_file
+
+  # ======= If any .as, first ar -x them into .os =======
+  all_os = []
+  for i in range(len(argv)):
+    if argv[i].endswith(".o"):
+      all_os.append(argv[i])
+    else:
+      subprocess.run(["ar", "-x", argv[i], "--output", f"{tmpdir.name}/{i}.unar"])
+      for path in os.listdir(f"{tmpdir.name}/{i}.unar"):
+        all_os.append(f"{tmpdir.name}/{i}.unar/{path}")
+
+  # ======= Put all the .os together into an output folder with Makefile =======
+  out_path = "a.out"
+  if "-o" in argv: out_path = argument("-o")
+  assert out_path not in ("", "/")
+
+  subprocess.run(["rm", "-rf", out_path])
+  os.mkdir(out_path)
+  makefile = open(f"{out_path}/Makefile", "w")
+  makefile.write("a.out: " + ' '.join([f"{i}.o" for i in range(len(all_os))]) + "\n")
+  argstr = list(map(shlex.quote, stripped_args))
+  argstr = ' '.join(argstr)
+  makefile.write(f"\t$(CC) $^ -o $@ {argstr}\n\n")
+  for i, o_file in enumerate(all_os):
+    f = open(o_file, "rb")
+    if f.read(5) == b"DIETC":
+      # dietc file!
+      argstr = f.read(int.from_bytes(f.read(4), "little")).decode("utf-8")
+      with open(f"{out_path}/{i}.c", "wb") as i_file:
+        i_file.write(f.read())
+      makefile.write(f"{i}.o: {i}.c\n")
+      makefile.write(f"\t$(CC) -c $^ -o $@ {argstr} -Wno-builtin-declaration-mismatch\n\n")
+      # TODO: include a postprocessing pass ...
+    else:
+      # regular .o
+      shutil.copy(o_file, f"{out_path}/{i}.o")
+  makefile.close()
+
+main(sys.argv.copy())
-- 
cgit v1.2.3