add shebang support to compiler driver

Author: robertmuth

FE/lexer.cc

@@ -64,6 +64,20 @@ typedef std::array<uint8_t, 129> TrieNode;
 
 LexerStats LexerRaw::stats;
 
+LexerRaw::LexerRaw(std::string_view input, Name file_id)
+    : input_(input), end_(input.size()) {
+  srcloc_.file = file_id;
+  if (input.starts_with("#!")) {
+    // shebang line, skip it
+    for (pos_ = 2; pos_ < input_.size(); ++pos_) {
+      if (input_[pos_] == '\n') {
+        ++line_no_;
+        break;
+      }
+    }
+  }
+}
+
 // Below is bunch of adhoc lexer helpers
 // These are quite horrible and the only excuse for them
 // is that they let us do without a depenency on a lexer library,
@@ -156,7 +170,7 @@ uint32_t LexerRaw::HandleSimpleStr() {
           return i + 1 - pos_;
           break;
         }
-        if(c == '\n') CompilerError(srcloc_) << "Newline in string literal";
+        if (c == '\n') CompilerError(srcloc_) << "Newline in string literal";
       }
     }
   }

FE/lexer.h

@@ -45,10 +45,7 @@ class LexerRaw {
   static LexerStats stats;
 
   // input is assumed to have a trailing zero byte
-  LexerRaw(std::string_view input, Name file_id)
-      : input_(input), end_(input.size()) {
-    srcloc_.file = file_id;
-  }
+  LexerRaw(std::string_view input, Name file_id);
 
   ~LexerRaw() {
     stats.num_files++;

FE/lexer_tab.py

@@ -406,6 +406,10 @@ def __init__(self: Any, filename: str, fp: io.TextIOWrapper):
         self._col_no = 0
         self._current_line = ""
         self._trie: TRIE = MakeTrie(False)
+        self._current_line = self._fill_line()
+        if self._current_line.startswith("#!"):
+            # ignore shebang mode
+            self._current_line = self._fill_line()
 
     def _GetSrcLoc(self) -> cwast.SrcLoc:
         return cwast.SrcLoc(self._fileamame, self._line_no)

cwerg.py

@@ -7,6 +7,7 @@
 import os
 import platform
 import sys
+import subprocess
 from typing import Any
 
 
@@ -130,6 +131,27 @@ def Diagnostics():
     print()
 
 
+def ShebangMode(script_path: str, extra_args: list[str]) -> int:
+    import uuid
+    import tempfile
+    arch = DEFAULT_ARCH
+    prefix = tempfile.gettempdir() + f"/cwerg_shebang_{uuid.uuid4().hex}"
+    ir_file = f"{prefix}.ir"
+    fe_cmd = GetFeCommand("c++", arch, script_path, ir_file)
+    if 0 != os.system(fe_cmd):
+        return 1
+    exe_file = f"{prefix}.exe"
+    be_cmd = GetBeCommand("c++", arch, ir_file, exe_file)
+    if 0 != os.system(be_cmd):
+        return 1
+    rm_cmd = f"chmod a+x {exe_file} ; rm -f {ir_file}"
+    if 0 != os.system(rm_cmd):
+        return 1
+    res = subprocess.run([exe_file] + extra_args).returncode
+    os.remove(exe_file)
+    return res
+
+
 def main():
     if sys.version_info[0] != 3 or sys.version_info[1] < 12:
         print("Cwerg requires Python 3.12 or higher")
@@ -138,6 +160,9 @@ def main():
         print("Cwerg currently requires Linux")
         return 1
 
+    if sys.argv[1] == "@script@":
+        exit(ShebangMode(sys.argv[2], sys.argv[3:]))
+
     parser = argparse.ArgumentParser(add_help=False)
     parser.add_argument('-diag', help='show diagnostics and exit',
                         action='store_true')