Merge branch 'lukeshu/better-cpp'

1 files changed, 186 insertions, 0 deletions

diff --git a/libmisc/wrap-cc b/libmisc/wrap-cc
new file mode 100755
index 0000000..e7a0b91
--- /dev/null
+++ b/libmisc/wrap-cc
@@ -0,0 +1,186 @@
+#!/usr/bin/env python3
+# libmisc/wrap-cc - Wrapper around GCC to enhance the preprocessor
+#
+# Copyright (C) 2025  Luke T. Shumaker <lukeshu@lukeshu.com>
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+import os
+import subprocess
+import sys
+import typing
+
+
+def scan_tuple(
+    text: str, beg: int, on_part: typing.Callable[[str], None] | None = None
+) -> int:
+    assert text[beg] == "("
+    pos = beg + 1
+    arg_start = pos
+    parens = 1
+    instring = False
+    while parens:
+        c = text[pos]
+        if instring:
+            match c:
+                case "\\":
+                    pos += 1
+                case '"':
+                    instring = False
+        else:
+            match c:
+                case "(":
+                    parens += 1
+                case ")":
+                    parens -= 1
+                    if on_part and parens == 0 and text[beg + 1 : pos].strip():
+                        on_part(text[arg_start:pos])
+                case ",":
+                    if on_part and parens == 1:
+                        on_part(text[arg_start:pos])
+                        arg_start = pos + 1
+                case '"':
+                    instring = True
+        pos += 1
+    assert text[pos - 1] == ")"
+    return pos - 1
+
+
+def unquote(cstr: str) -> str:
+    assert len(cstr) >= 2 and cstr[0] == '"' and cstr[-1] == '"'
+    cstr = cstr[1:-1]
+    out = ""
+    while cstr:
+        if cstr[0] == "\\":
+            match cstr[1]:
+                case "n":
+                    out += "\n"
+                    cstr = cstr[2:]
+                case "\\":
+                    out += "\\"
+                    cstr = cstr[2:]
+                case '"':
+                    out += '"'
+                    cstr = cstr[2:]
+        else:
+            out += cstr[0]
+            cstr = cstr[1:]
+    return out
+
+
+def preprocess(all_args: list[str]) -> typing.NoReturn:
+    # argparse #################################################################
+    _args = all_args
+
+    def shift(n: int) -> list[str]:
+        nonlocal _args
+        ret = _args[:n]
+        _args = _args[n:]
+        return ret
+
+    arg0 = shift(1)[0]
+    common_flags: list[str] = []
+    output_flags: list[str] = []
+    positional: list[str] = []
+    while _args:
+        if len(_args[0]) > 2 and _args[0][0] == "-" and _args[0][1] in "IDU":
+            _args = [_args[0][:2], _args[0][2:], *_args[1:]]
+        match _args[0]:
+            # Mode
+            case "-E" | "-quiet":
+                common_flags += shift(1)
+            case "-lang-asm":
+                os.execvp(all_args[0], all_args)
+            # Search path
+            case "-I" | "-imultilib" | "-isystem":
+                common_flags += shift(2)
+            # Define/Undefine
+            case "-D" | "-U":
+                common_flags += shift(2)
+            # Optimization
+            case "-O0" | "-O1" | "-O2" | "-O3" | "-Os" | "-Ofast" | "-Og" | "-Oz":
+                common_flags += shift(1)
+            case "-g":
+                common_flags += shift(1)
+            # Output files
+            case "-MD" | "-MF" | "-MT" | "-dumpbase" | "-dumpbase-ext":
+                output_flags += shift(2)
+            case "-o":
+                output_flags += shift(2)
+            # Other
+            case _:
+                if _args[0].startswith("-"):
+                    if _args[0].startswith("-std="):
+                        common_flags += shift(1)
+                    elif _args[0].startswith("-m"):
+                        common_flags += shift(1)
+                    elif _args[0].startswith("-f"):
+                        common_flags += shift(1)
+                    elif _args[0].startswith("-W"):
+                        common_flags += shift(1)
+                    else:
+                        raise ValueError(f"unknown flag: {_args!r}")
+                else:
+                    positional += shift(1)
+    if len(positional) != 1:
+        raise ValueError("expected 1 input file")
+    infile = positional[0]
+
+    # enhance ##################################################################
+
+    common_flags += ["-D", "__LIBMISC_ENHANCED_CPP__"]
+
+    text = subprocess.run(
+        [arg0, *common_flags, infile],
+        stdin=subprocess.DEVNULL,
+        stdout=subprocess.PIPE,
+        stderr=sys.stderr,
+        check=True,
+        text=True,
+    ).stdout
+
+    macros: dict[str, str] = {}
+
+    marker = "__xx__LM_DEFAPPEND__xx__"
+    pos = 0
+    while (marker_beg := text.find(marker, pos)) >= 0:
+        args: list[str] = []
+
+        def add_arg(arg: str) -> None:
+            nonlocal args
+            args.append(arg)
+
+        beg_paren = marker_beg + len(marker)
+        end_paren = scan_tuple(text, beg_paren, add_arg)
+
+        before = text[:marker_beg]
+        # old = text[marker_beg : end_paren + 1]
+        after = text[end_paren + 1 :]
+
+        assert len(args) == 2
+        k = unquote(args[0].strip())
+        v = unquote(args[1].strip())
+        if k not in macros:
+            macros[k] = v
+        else:
+            macros[k] += " " + v
+
+        text = before + after
+        pos = len(before)
+
+    common_flags += ["-D", marker + "=LM_EAT"]
+    for k, v in macros.items():
+        common_flags += ["-D", k + "=" + v]
+
+    # Run, for-real ############################################################
+    os.execvp(arg0, [arg0, *common_flags, *output_flags, infile])
+
+
+def main(all_args: list[str]) -> typing.NoReturn:
+    if len(all_args) >= 2 and all_args[0].endswith("cc1") and all_args[1] == "-E":
+        preprocess(all_args)
+    else:
+        os.execvp(all_args[0], all_args)
+
+
+if __name__ == "__main__":
+    main(sys.argv[1:])