Automate fixing BSS ordering (#2009)

* Automate fixing BSS ordering

* Typo

* Some cleanups

* Move pragma check after printing BSS info

* Some proofreading

* multiprocessing, require version, some colors

* Tweak output

* Black + mypy

* Move logging and sys.exit out of helper functions

* Use stdout instead of stderr in fix_bss.py

* Add suggestion to conflicting offsets error

Co-authored-by: Dragorn421 <Dragorn421@users.noreply.github.com>

* Remove var = list[T]()

* Improve error handling

Co-authored-by: Dragorn421 <Dragorn421@users.noreply.github.com>

* Add error if no pointers to BSS

* Add comment about process_file_worker

* Only print updates if stdout is a tty

* Use new binary-search-esque candidate generation algorithm

Co-authored-by: Dragorn421 <Dragorn421@users.noreply.github.com>

* Add Wikipedia link

* More comment tweaks

---------

Co-authored-by: Dragorn421 <Dragorn421@users.noreply.github.com>

tools/preprocess.py

@@ -3,13 +3,13 @@
 # SPDX-FileCopyrightText: © 2024 ZeldaRET
 # SPDX-License-Identifier: CC0-1.0
 
-# Usage: preprocess.py [compile command minus input file...] [single input file]
+# Usage: preprocess.py [flags] -- [compile command minus input file...] [single input file]
 # Preprocess a C file to:
 # * Re-encode from UTF-8 to EUC-JP (the repo uses UTF-8 for text encoding, but
 #   the strings in the ROM are encoded in EUC-JP)
-# * Replace `#pragma increment_block_number N` with `N` fake structs for
-#   controlling BSS ordering
+# * Replace `#pragma increment_block_number` with fake structs for controlling BSS ordering
 
+import argparse
 from pathlib import Path
 import os
 import tempfile
@@ -22,39 +22,59 @@ def fail(message):
     sys.exit(1)
 
 
-def process_file(filename, input, output):
+def process_file(version, filename, input, output):
     output.write(f'#line 1 "{filename}"\n')
     for i, line in enumerate(input, start=1):
-        if line.startswith("#pragma increment_block_number"):
-            parts = line.split()
-            if len(parts) != 3:
-                fail(
-                    f"{filename}:{i}: increment_block_number must be followed by an integer"
-                )
-            try:
-                amount = int(parts[2])
-            except ValueError:
-                fail(
-                    f"{filename}:{i}: increment_block_number must be followed by an integer"
-                )
+        if line.startswith("#pragma increment_block_number "):
+            # Grab pragma argument and remove quotes
+            arg = line.strip()[len("#pragma increment_block_number ") + 1 : -1]
+            amount = 0
+            for part in arg.split():
+                kv = part.split(":")
+                if len(kv) != 2:
+                    fail(
+                        f"{filename}:{i}: increment_block_number must be followed by a list of version:amount pairs"
+                    )
+                if kv[0] != version:
+                    continue
+                try:
+                    amount = int(kv[1])
+                except ValueError:
+                    fail(
+                        f"{filename}:{i}: increment_block_number amount must be an integer"
+                    )
+
+            # Always generate at least one struct so that fix_bss.py can know where the increment_block_number pragmas are
+            if amount == 0:
+                amount = 256
+
             # Write fake structs for BSS ordering
             for j in range(amount):
-                output.write(f"struct DummyStruct_{i:05}_{j:03};\n")
+                output.write(f"struct increment_block_number_{i:05}_{j:03};\n")
             output.write(f'#line {i + 1} "{filename}"\n')
         else:
             output.write(line)
 
 
 def main():
-    filename = Path(sys.argv[-1])
+    parser = argparse.ArgumentParser()
+    parser.add_argument("-v", "--oot-version", help="Which version should be processed")
+    parser.add_argument(
+        "args",
+        nargs="+",
+    )
+
+    args = parser.parse_args()
+
+    filename = Path(args.args[-1])
     with tempfile.TemporaryDirectory(prefix="oot_") as tmpdir:
         tmpfile = Path(tmpdir) / filename.name
 
         with open(filename, mode="r", encoding="utf-8") as input:
             with open(tmpfile, mode="w", encoding="euc-jp") as output:
-                process_file(filename, input, output)
+                process_file(args.oot_version, filename, input, output)
 
-        compile_command = sys.argv[1:-1] + ["-I", filename.parent, tmpfile]
+        compile_command = args.args[:-1] + ["-I", filename.parent, tmpfile]
         process = subprocess.run(compile_command)
         return process.returncode