From 44776befd8333e6cd0cd62bd13b2d7462ea4a24e Mon Sep 17 00:00:00 2001
From: Rui Ueyama <ruiu@cs.stanford.edu>
Date: Tue, 8 Sep 2020 22:15:00 +0900
Subject: [PATCH] Add include guard optimization

---
 preprocess.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 53 insertions(+)

diff --git a/preprocess.c b/preprocess.c
index a09ba6a..10a85e3 100644
--- a/preprocess.c
+++ b/preprocess.c
@@ -777,6 +777,45 @@ static char *read_include_path(Token **rest, Token *tok) {
   error_tok(tok, "expected a filename");
 }
 
+// Detect the following "include guard" pattern.
+//
+//   #ifndef FOO_H
+//   #define FOO_H
+//   ...
+//   #endif
+static char *detect_include_guard(Token *tok) {
+  // Detect the first two lines.
+  if (!is_hash(tok) || !equal(tok->next, "ifndef"))
+    return NULL;
+  tok = tok->next->next;
+
+  if (tok->kind != TK_IDENT)
+    return NULL;
+
+  char *macro = strndup(tok->loc, tok->len);
+  tok = tok->next;
+
+  if (!is_hash(tok) || !equal(tok->next, "define") || !equal(tok->next->next, macro))
+    return NULL;
+
+  // Read until the end of the file.
+  while (tok->kind != TK_EOF) {
+    if (!is_hash(tok)) {
+      tok = tok->next;
+      continue;
+    }
+
+    if (equal(tok->next, "endif") && tok->next->next->kind == TK_EOF)
+      return macro;
+
+    if (equal(tok, "if") || equal(tok, "ifdef") || equal(tok, "ifndef"))
+      tok = skip_cond_incl(tok->next);
+    else
+      tok = tok->next;
+  }
+  return NULL;
+}
+
 // Read #line arguments
 static void read_line_marker(Token **rest, Token *tok) {
   Token *start = tok;
@@ -820,9 +859,23 @@ static Token *preprocess2(Token *tok) {
 
     if (equal(tok, "include")) {
       char *path = read_include_path(&tok, tok->next);
+
+      // If we read the same file before, and if the file was guarded
+      // by the usual #ifndef ... #endif pattern, we may be able to
+      // skip the file without opening it.
+      static HashMap include_guards;
+      char *guard_name = hashmap_get(&include_guards, path);
+      if (guard_name && hashmap_get(&macros, guard_name))
+        continue;
+
       Token *tok2 = tokenize_file(path);
       if (!tok2)
         error_tok(start, "%s: cannot open file: %s", path, strerror(errno));
+
+      guard_name = detect_include_guard(tok2);
+      if (guard_name)
+        hashmap_put(&include_guards, path, guard_name);
+
       tok = append(tok2, tok);
       continue;
     }